1 //===-- MachODump.cpp - Object file dumping utility for llvm --------------===// 2 // 3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. 4 // See https://llvm.org/LICENSE.txt for license information. 5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception 6 // 7 //===----------------------------------------------------------------------===// 8 // 9 // This file implements the MachO-specific dumper for llvm-objdump. 10 // 11 //===----------------------------------------------------------------------===// 12 13 #include "MachODump.h" 14 15 #include "llvm-objdump.h" 16 #include "llvm-c/Disassembler.h" 17 #include "llvm/ADT/STLExtras.h" 18 #include "llvm/ADT/StringExtras.h" 19 #include "llvm/ADT/Triple.h" 20 #include "llvm/BinaryFormat/MachO.h" 21 #include "llvm/Config/config.h" 22 #include "llvm/DebugInfo/DIContext.h" 23 #include "llvm/DebugInfo/DWARF/DWARFContext.h" 24 #include "llvm/Demangle/Demangle.h" 25 #include "llvm/MC/MCAsmInfo.h" 26 #include "llvm/MC/MCContext.h" 27 #include "llvm/MC/MCDisassembler/MCDisassembler.h" 28 #include "llvm/MC/MCInst.h" 29 #include "llvm/MC/MCInstPrinter.h" 30 #include "llvm/MC/MCInstrDesc.h" 31 #include "llvm/MC/MCInstrInfo.h" 32 #include "llvm/MC/MCRegisterInfo.h" 33 #include "llvm/MC/MCSubtargetInfo.h" 34 #include "llvm/MC/MCTargetOptions.h" 35 #include "llvm/Object/MachO.h" 36 #include "llvm/Object/MachOUniversal.h" 37 #include "llvm/Support/Casting.h" 38 #include "llvm/Support/CommandLine.h" 39 #include "llvm/Support/Debug.h" 40 #include "llvm/Support/Endian.h" 41 #include "llvm/Support/Format.h" 42 #include "llvm/Support/FormattedStream.h" 43 #include "llvm/Support/GraphWriter.h" 44 #include "llvm/Support/LEB128.h" 45 #include "llvm/Support/MemoryBuffer.h" 46 #include "llvm/Support/TargetRegistry.h" 47 #include "llvm/Support/TargetSelect.h" 48 #include "llvm/Support/ToolOutputFile.h" 49 #include "llvm/Support/WithColor.h" 50 #include "llvm/Support/raw_ostream.h" 51 #include <algorithm> 52 #include <cstring> 53 #include <system_error> 54 55 #ifdef HAVE_LIBXAR 56 extern "C" { 57 #include <xar/xar.h> 58 } 59 #endif 60 61 using namespace llvm; 62 using namespace llvm::object; 63 using namespace llvm::objdump; 64 65 cl::OptionCategory objdump::MachOCat("llvm-objdump MachO Specific Options"); 66 67 cl::opt<bool> objdump::FirstPrivateHeader( 68 "private-header", 69 cl::desc("Display only the first format specific file header"), 70 cl::cat(MachOCat)); 71 72 cl::opt<bool> objdump::ExportsTrie("exports-trie", 73 cl::desc("Display mach-o exported symbols"), 74 cl::cat(MachOCat)); 75 76 cl::opt<bool> objdump::Rebase("rebase", 77 cl::desc("Display mach-o rebasing info"), 78 cl::cat(MachOCat)); 79 80 cl::opt<bool> objdump::Bind("bind", cl::desc("Display mach-o binding info"), 81 cl::cat(MachOCat)); 82 83 cl::opt<bool> objdump::LazyBind("lazy-bind", 84 cl::desc("Display mach-o lazy binding info"), 85 cl::cat(MachOCat)); 86 87 cl::opt<bool> objdump::WeakBind("weak-bind", 88 cl::desc("Display mach-o weak binding info"), 89 cl::cat(MachOCat)); 90 91 static cl::opt<bool> 92 UseDbg("g", cl::Grouping, 93 cl::desc("Print line information from debug info if available"), 94 cl::cat(MachOCat)); 95 96 static cl::opt<std::string> DSYMFile("dsym", 97 cl::desc("Use .dSYM file for debug info"), 98 cl::cat(MachOCat)); 99 100 static cl::opt<bool> FullLeadingAddr("full-leading-addr", 101 cl::desc("Print full leading address"), 102 cl::cat(MachOCat)); 103 104 static cl::opt<bool> NoLeadingHeaders("no-leading-headers", 105 cl::desc("Print no leading headers"), 106 cl::cat(MachOCat)); 107 108 cl::opt<bool> objdump::UniversalHeaders( 109 "universal-headers", 110 cl::desc("Print Mach-O universal headers (requires -macho)"), 111 cl::cat(MachOCat)); 112 113 static cl::opt<bool> ArchiveMemberOffsets( 114 "archive-member-offsets", 115 cl::desc("Print the offset to each archive member for Mach-O archives " 116 "(requires -macho and -archive-headers)"), 117 cl::cat(MachOCat)); 118 119 cl::opt<bool> objdump::IndirectSymbols( 120 "indirect-symbols", 121 cl::desc( 122 "Print indirect symbol table for Mach-O objects (requires -macho)"), 123 cl::cat(MachOCat)); 124 125 cl::opt<bool> objdump::DataInCode( 126 "data-in-code", 127 cl::desc( 128 "Print the data in code table for Mach-O objects (requires -macho)"), 129 cl::cat(MachOCat)); 130 131 cl::opt<bool> 132 objdump::LinkOptHints("link-opt-hints", 133 cl::desc("Print the linker optimization hints for " 134 "Mach-O objects (requires -macho)"), 135 cl::cat(MachOCat)); 136 137 cl::opt<bool> 138 objdump::InfoPlist("info-plist", 139 cl::desc("Print the info plist section as strings for " 140 "Mach-O objects (requires -macho)"), 141 cl::cat(MachOCat)); 142 143 cl::opt<bool> 144 objdump::DylibsUsed("dylibs-used", 145 cl::desc("Print the shared libraries used for linked " 146 "Mach-O files (requires -macho)"), 147 cl::cat(MachOCat)); 148 149 cl::opt<bool> objdump::DylibId("dylib-id", 150 cl::desc("Print the shared library's id for the " 151 "dylib Mach-O file (requires -macho)"), 152 cl::cat(MachOCat)); 153 154 static cl::opt<bool> 155 NonVerbose("non-verbose", 156 cl::desc("Print the info for Mach-O objects in non-verbose or " 157 "numeric form (requires -macho)"), 158 cl::cat(MachOCat)); 159 160 cl::opt<bool> 161 objdump::ObjcMetaData("objc-meta-data", 162 cl::desc("Print the Objective-C runtime meta data " 163 "for Mach-O files (requires -macho)"), 164 cl::cat(MachOCat)); 165 166 static cl::opt<std::string> DisSymName( 167 "dis-symname", 168 cl::desc("disassemble just this symbol's instructions (requires -macho)"), 169 cl::cat(MachOCat)); 170 171 static cl::opt<bool> NoSymbolicOperands( 172 "no-symbolic-operands", 173 cl::desc("do not symbolic operands when disassembling (requires -macho)"), 174 cl::cat(MachOCat)); 175 176 static cl::list<std::string> 177 ArchFlags("arch", cl::desc("architecture(s) from a Mach-O file to dump"), 178 cl::ZeroOrMore, cl::cat(MachOCat)); 179 180 static bool ArchAll = false; 181 182 static std::string ThumbTripleName; 183 184 static const Target *GetTarget(const MachOObjectFile *MachOObj, 185 const char **McpuDefault, 186 const Target **ThumbTarget) { 187 // Figure out the target triple. 188 Triple TT(TripleName); 189 if (TripleName.empty()) { 190 TT = MachOObj->getArchTriple(McpuDefault); 191 TripleName = TT.str(); 192 } 193 194 if (TT.getArch() == Triple::arm) { 195 // We've inferred a 32-bit ARM target from the object file. All MachO CPUs 196 // that support ARM are also capable of Thumb mode. 197 Triple ThumbTriple = TT; 198 std::string ThumbName = (Twine("thumb") + TT.getArchName().substr(3)).str(); 199 ThumbTriple.setArchName(ThumbName); 200 ThumbTripleName = ThumbTriple.str(); 201 } 202 203 // Get the target specific parser. 204 std::string Error; 205 const Target *TheTarget = TargetRegistry::lookupTarget(TripleName, Error); 206 if (TheTarget && ThumbTripleName.empty()) 207 return TheTarget; 208 209 *ThumbTarget = TargetRegistry::lookupTarget(ThumbTripleName, Error); 210 if (*ThumbTarget) 211 return TheTarget; 212 213 WithColor::error(errs(), "llvm-objdump") << "unable to get target for '"; 214 if (!TheTarget) 215 errs() << TripleName; 216 else 217 errs() << ThumbTripleName; 218 errs() << "', see --version and --triple.\n"; 219 return nullptr; 220 } 221 222 namespace { 223 struct SymbolSorter { 224 bool operator()(const SymbolRef &A, const SymbolRef &B) { 225 Expected<SymbolRef::Type> ATypeOrErr = A.getType(); 226 if (!ATypeOrErr) 227 reportError(ATypeOrErr.takeError(), A.getObject()->getFileName()); 228 SymbolRef::Type AType = *ATypeOrErr; 229 Expected<SymbolRef::Type> BTypeOrErr = B.getType(); 230 if (!BTypeOrErr) 231 reportError(BTypeOrErr.takeError(), B.getObject()->getFileName()); 232 SymbolRef::Type BType = *BTypeOrErr; 233 uint64_t AAddr = (AType != SymbolRef::ST_Function) ? 0 : A.getValue(); 234 uint64_t BAddr = (BType != SymbolRef::ST_Function) ? 0 : B.getValue(); 235 return AAddr < BAddr; 236 } 237 }; 238 } // namespace 239 240 // Types for the storted data in code table that is built before disassembly 241 // and the predicate function to sort them. 242 typedef std::pair<uint64_t, DiceRef> DiceTableEntry; 243 typedef std::vector<DiceTableEntry> DiceTable; 244 typedef DiceTable::iterator dice_table_iterator; 245 246 #ifdef HAVE_LIBXAR 247 namespace { 248 struct ScopedXarFile { 249 xar_t xar; 250 ScopedXarFile(const char *filename, int32_t flags) 251 : xar(xar_open(filename, flags)) {} 252 ~ScopedXarFile() { 253 if (xar) 254 xar_close(xar); 255 } 256 ScopedXarFile(const ScopedXarFile &) = delete; 257 ScopedXarFile &operator=(const ScopedXarFile &) = delete; 258 operator xar_t() { return xar; } 259 }; 260 261 struct ScopedXarIter { 262 xar_iter_t iter; 263 ScopedXarIter() : iter(xar_iter_new()) {} 264 ~ScopedXarIter() { 265 if (iter) 266 xar_iter_free(iter); 267 } 268 ScopedXarIter(const ScopedXarIter &) = delete; 269 ScopedXarIter &operator=(const ScopedXarIter &) = delete; 270 operator xar_iter_t() { return iter; } 271 }; 272 } // namespace 273 #endif // defined(HAVE_LIBXAR) 274 275 // This is used to search for a data in code table entry for the PC being 276 // disassembled. The j parameter has the PC in j.first. A single data in code 277 // table entry can cover many bytes for each of its Kind's. So if the offset, 278 // aka the i.first value, of the data in code table entry plus its Length 279 // covers the PC being searched for this will return true. If not it will 280 // return false. 281 static bool compareDiceTableEntries(const DiceTableEntry &i, 282 const DiceTableEntry &j) { 283 uint16_t Length; 284 i.second.getLength(Length); 285 286 return j.first >= i.first && j.first < i.first + Length; 287 } 288 289 static uint64_t DumpDataInCode(const uint8_t *bytes, uint64_t Length, 290 unsigned short Kind) { 291 uint32_t Value, Size = 1; 292 293 switch (Kind) { 294 default: 295 case MachO::DICE_KIND_DATA: 296 if (Length >= 4) { 297 if (!NoShowRawInsn) 298 dumpBytes(makeArrayRef(bytes, 4), outs()); 299 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0]; 300 outs() << "\t.long " << Value; 301 Size = 4; 302 } else if (Length >= 2) { 303 if (!NoShowRawInsn) 304 dumpBytes(makeArrayRef(bytes, 2), outs()); 305 Value = bytes[1] << 8 | bytes[0]; 306 outs() << "\t.short " << Value; 307 Size = 2; 308 } else { 309 if (!NoShowRawInsn) 310 dumpBytes(makeArrayRef(bytes, 2), outs()); 311 Value = bytes[0]; 312 outs() << "\t.byte " << Value; 313 Size = 1; 314 } 315 if (Kind == MachO::DICE_KIND_DATA) 316 outs() << "\t@ KIND_DATA\n"; 317 else 318 outs() << "\t@ data in code kind = " << Kind << "\n"; 319 break; 320 case MachO::DICE_KIND_JUMP_TABLE8: 321 if (!NoShowRawInsn) 322 dumpBytes(makeArrayRef(bytes, 1), outs()); 323 Value = bytes[0]; 324 outs() << "\t.byte " << format("%3u", Value) << "\t@ KIND_JUMP_TABLE8\n"; 325 Size = 1; 326 break; 327 case MachO::DICE_KIND_JUMP_TABLE16: 328 if (!NoShowRawInsn) 329 dumpBytes(makeArrayRef(bytes, 2), outs()); 330 Value = bytes[1] << 8 | bytes[0]; 331 outs() << "\t.short " << format("%5u", Value & 0xffff) 332 << "\t@ KIND_JUMP_TABLE16\n"; 333 Size = 2; 334 break; 335 case MachO::DICE_KIND_JUMP_TABLE32: 336 case MachO::DICE_KIND_ABS_JUMP_TABLE32: 337 if (!NoShowRawInsn) 338 dumpBytes(makeArrayRef(bytes, 4), outs()); 339 Value = bytes[3] << 24 | bytes[2] << 16 | bytes[1] << 8 | bytes[0]; 340 outs() << "\t.long " << Value; 341 if (Kind == MachO::DICE_KIND_JUMP_TABLE32) 342 outs() << "\t@ KIND_JUMP_TABLE32\n"; 343 else 344 outs() << "\t@ KIND_ABS_JUMP_TABLE32\n"; 345 Size = 4; 346 break; 347 } 348 return Size; 349 } 350 351 static void getSectionsAndSymbols(MachOObjectFile *MachOObj, 352 std::vector<SectionRef> &Sections, 353 std::vector<SymbolRef> &Symbols, 354 SmallVectorImpl<uint64_t> &FoundFns, 355 uint64_t &BaseSegmentAddress) { 356 const StringRef FileName = MachOObj->getFileName(); 357 for (const SymbolRef &Symbol : MachOObj->symbols()) { 358 StringRef SymName = unwrapOrError(Symbol.getName(), FileName); 359 if (!SymName.startswith("ltmp")) 360 Symbols.push_back(Symbol); 361 } 362 363 for (const SectionRef &Section : MachOObj->sections()) 364 Sections.push_back(Section); 365 366 bool BaseSegmentAddressSet = false; 367 for (const auto &Command : MachOObj->load_commands()) { 368 if (Command.C.cmd == MachO::LC_FUNCTION_STARTS) { 369 // We found a function starts segment, parse the addresses for later 370 // consumption. 371 MachO::linkedit_data_command LLC = 372 MachOObj->getLinkeditDataLoadCommand(Command); 373 374 MachOObj->ReadULEB128s(LLC.dataoff, FoundFns); 375 } else if (Command.C.cmd == MachO::LC_SEGMENT) { 376 MachO::segment_command SLC = MachOObj->getSegmentLoadCommand(Command); 377 StringRef SegName = SLC.segname; 378 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") { 379 BaseSegmentAddressSet = true; 380 BaseSegmentAddress = SLC.vmaddr; 381 } 382 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) { 383 MachO::segment_command_64 SLC = MachOObj->getSegment64LoadCommand(Command); 384 StringRef SegName = SLC.segname; 385 if (!BaseSegmentAddressSet && SegName != "__PAGEZERO") { 386 BaseSegmentAddressSet = true; 387 BaseSegmentAddress = SLC.vmaddr; 388 } 389 } 390 } 391 } 392 393 static bool DumpAndSkipDataInCode(uint64_t PC, const uint8_t *bytes, 394 DiceTable &Dices, uint64_t &InstSize) { 395 // Check the data in code table here to see if this is data not an 396 // instruction to be disassembled. 397 DiceTable Dice; 398 Dice.push_back(std::make_pair(PC, DiceRef())); 399 dice_table_iterator DTI = 400 std::search(Dices.begin(), Dices.end(), Dice.begin(), Dice.end(), 401 compareDiceTableEntries); 402 if (DTI != Dices.end()) { 403 uint16_t Length; 404 DTI->second.getLength(Length); 405 uint16_t Kind; 406 DTI->second.getKind(Kind); 407 InstSize = DumpDataInCode(bytes, Length, Kind); 408 if ((Kind == MachO::DICE_KIND_JUMP_TABLE8) && 409 (PC == (DTI->first + Length - 1)) && (Length & 1)) 410 InstSize++; 411 return true; 412 } 413 return false; 414 } 415 416 static void printRelocationTargetName(const MachOObjectFile *O, 417 const MachO::any_relocation_info &RE, 418 raw_string_ostream &Fmt) { 419 // Target of a scattered relocation is an address. In the interest of 420 // generating pretty output, scan through the symbol table looking for a 421 // symbol that aligns with that address. If we find one, print it. 422 // Otherwise, we just print the hex address of the target. 423 const StringRef FileName = O->getFileName(); 424 if (O->isRelocationScattered(RE)) { 425 uint32_t Val = O->getPlainRelocationSymbolNum(RE); 426 427 for (const SymbolRef &Symbol : O->symbols()) { 428 uint64_t Addr = unwrapOrError(Symbol.getAddress(), FileName); 429 if (Addr != Val) 430 continue; 431 Fmt << unwrapOrError(Symbol.getName(), FileName); 432 return; 433 } 434 435 // If we couldn't find a symbol that this relocation refers to, try 436 // to find a section beginning instead. 437 for (const SectionRef &Section : ToolSectionFilter(*O)) { 438 uint64_t Addr = Section.getAddress(); 439 if (Addr != Val) 440 continue; 441 StringRef NameOrErr = unwrapOrError(Section.getName(), O->getFileName()); 442 Fmt << NameOrErr; 443 return; 444 } 445 446 Fmt << format("0x%x", Val); 447 return; 448 } 449 450 StringRef S; 451 bool isExtern = O->getPlainRelocationExternal(RE); 452 uint64_t Val = O->getPlainRelocationSymbolNum(RE); 453 454 if (O->getAnyRelocationType(RE) == MachO::ARM64_RELOC_ADDEND) { 455 Fmt << format("0x%0" PRIx64, Val); 456 return; 457 } 458 459 if (isExtern) { 460 symbol_iterator SI = O->symbol_begin(); 461 advance(SI, Val); 462 S = unwrapOrError(SI->getName(), FileName); 463 } else { 464 section_iterator SI = O->section_begin(); 465 // Adjust for the fact that sections are 1-indexed. 466 if (Val == 0) { 467 Fmt << "0 (?,?)"; 468 return; 469 } 470 uint32_t I = Val - 1; 471 while (I != 0 && SI != O->section_end()) { 472 --I; 473 advance(SI, 1); 474 } 475 if (SI == O->section_end()) { 476 Fmt << Val << " (?,?)"; 477 } else { 478 if (Expected<StringRef> NameOrErr = SI->getName()) 479 S = *NameOrErr; 480 else 481 consumeError(NameOrErr.takeError()); 482 } 483 } 484 485 Fmt << S; 486 } 487 488 Error objdump::getMachORelocationValueString(const MachOObjectFile *Obj, 489 const RelocationRef &RelRef, 490 SmallVectorImpl<char> &Result) { 491 DataRefImpl Rel = RelRef.getRawDataRefImpl(); 492 MachO::any_relocation_info RE = Obj->getRelocation(Rel); 493 494 unsigned Arch = Obj->getArch(); 495 496 std::string FmtBuf; 497 raw_string_ostream Fmt(FmtBuf); 498 unsigned Type = Obj->getAnyRelocationType(RE); 499 bool IsPCRel = Obj->getAnyRelocationPCRel(RE); 500 501 // Determine any addends that should be displayed with the relocation. 502 // These require decoding the relocation type, which is triple-specific. 503 504 // X86_64 has entirely custom relocation types. 505 if (Arch == Triple::x86_64) { 506 switch (Type) { 507 case MachO::X86_64_RELOC_GOT_LOAD: 508 case MachO::X86_64_RELOC_GOT: { 509 printRelocationTargetName(Obj, RE, Fmt); 510 Fmt << "@GOT"; 511 if (IsPCRel) 512 Fmt << "PCREL"; 513 break; 514 } 515 case MachO::X86_64_RELOC_SUBTRACTOR: { 516 DataRefImpl RelNext = Rel; 517 Obj->moveRelocationNext(RelNext); 518 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext); 519 520 // X86_64_RELOC_SUBTRACTOR must be followed by a relocation of type 521 // X86_64_RELOC_UNSIGNED. 522 // NOTE: Scattered relocations don't exist on x86_64. 523 unsigned RType = Obj->getAnyRelocationType(RENext); 524 if (RType != MachO::X86_64_RELOC_UNSIGNED) 525 reportError(Obj->getFileName(), "Expected X86_64_RELOC_UNSIGNED after " 526 "X86_64_RELOC_SUBTRACTOR."); 527 528 // The X86_64_RELOC_UNSIGNED contains the minuend symbol; 529 // X86_64_RELOC_SUBTRACTOR contains the subtrahend. 530 printRelocationTargetName(Obj, RENext, Fmt); 531 Fmt << "-"; 532 printRelocationTargetName(Obj, RE, Fmt); 533 break; 534 } 535 case MachO::X86_64_RELOC_TLV: 536 printRelocationTargetName(Obj, RE, Fmt); 537 Fmt << "@TLV"; 538 if (IsPCRel) 539 Fmt << "P"; 540 break; 541 case MachO::X86_64_RELOC_SIGNED_1: 542 printRelocationTargetName(Obj, RE, Fmt); 543 Fmt << "-1"; 544 break; 545 case MachO::X86_64_RELOC_SIGNED_2: 546 printRelocationTargetName(Obj, RE, Fmt); 547 Fmt << "-2"; 548 break; 549 case MachO::X86_64_RELOC_SIGNED_4: 550 printRelocationTargetName(Obj, RE, Fmt); 551 Fmt << "-4"; 552 break; 553 default: 554 printRelocationTargetName(Obj, RE, Fmt); 555 break; 556 } 557 // X86 and ARM share some relocation types in common. 558 } else if (Arch == Triple::x86 || Arch == Triple::arm || 559 Arch == Triple::ppc) { 560 // Generic relocation types... 561 switch (Type) { 562 case MachO::GENERIC_RELOC_PAIR: // prints no info 563 return Error::success(); 564 case MachO::GENERIC_RELOC_SECTDIFF: { 565 DataRefImpl RelNext = Rel; 566 Obj->moveRelocationNext(RelNext); 567 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext); 568 569 // X86 sect diff's must be followed by a relocation of type 570 // GENERIC_RELOC_PAIR. 571 unsigned RType = Obj->getAnyRelocationType(RENext); 572 573 if (RType != MachO::GENERIC_RELOC_PAIR) 574 reportError(Obj->getFileName(), "Expected GENERIC_RELOC_PAIR after " 575 "GENERIC_RELOC_SECTDIFF."); 576 577 printRelocationTargetName(Obj, RE, Fmt); 578 Fmt << "-"; 579 printRelocationTargetName(Obj, RENext, Fmt); 580 break; 581 } 582 } 583 584 if (Arch == Triple::x86 || Arch == Triple::ppc) { 585 switch (Type) { 586 case MachO::GENERIC_RELOC_LOCAL_SECTDIFF: { 587 DataRefImpl RelNext = Rel; 588 Obj->moveRelocationNext(RelNext); 589 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext); 590 591 // X86 sect diff's must be followed by a relocation of type 592 // GENERIC_RELOC_PAIR. 593 unsigned RType = Obj->getAnyRelocationType(RENext); 594 if (RType != MachO::GENERIC_RELOC_PAIR) 595 reportError(Obj->getFileName(), "Expected GENERIC_RELOC_PAIR after " 596 "GENERIC_RELOC_LOCAL_SECTDIFF."); 597 598 printRelocationTargetName(Obj, RE, Fmt); 599 Fmt << "-"; 600 printRelocationTargetName(Obj, RENext, Fmt); 601 break; 602 } 603 case MachO::GENERIC_RELOC_TLV: { 604 printRelocationTargetName(Obj, RE, Fmt); 605 Fmt << "@TLV"; 606 if (IsPCRel) 607 Fmt << "P"; 608 break; 609 } 610 default: 611 printRelocationTargetName(Obj, RE, Fmt); 612 } 613 } else { // ARM-specific relocations 614 switch (Type) { 615 case MachO::ARM_RELOC_HALF: 616 case MachO::ARM_RELOC_HALF_SECTDIFF: { 617 // Half relocations steal a bit from the length field to encode 618 // whether this is an upper16 or a lower16 relocation. 619 bool isUpper = (Obj->getAnyRelocationLength(RE) & 0x1) == 1; 620 621 if (isUpper) 622 Fmt << ":upper16:("; 623 else 624 Fmt << ":lower16:("; 625 printRelocationTargetName(Obj, RE, Fmt); 626 627 DataRefImpl RelNext = Rel; 628 Obj->moveRelocationNext(RelNext); 629 MachO::any_relocation_info RENext = Obj->getRelocation(RelNext); 630 631 // ARM half relocs must be followed by a relocation of type 632 // ARM_RELOC_PAIR. 633 unsigned RType = Obj->getAnyRelocationType(RENext); 634 if (RType != MachO::ARM_RELOC_PAIR) 635 reportError(Obj->getFileName(), "Expected ARM_RELOC_PAIR after " 636 "ARM_RELOC_HALF"); 637 638 // NOTE: The half of the target virtual address is stashed in the 639 // address field of the secondary relocation, but we can't reverse 640 // engineer the constant offset from it without decoding the movw/movt 641 // instruction to find the other half in its immediate field. 642 643 // ARM_RELOC_HALF_SECTDIFF encodes the second section in the 644 // symbol/section pointer of the follow-on relocation. 645 if (Type == MachO::ARM_RELOC_HALF_SECTDIFF) { 646 Fmt << "-"; 647 printRelocationTargetName(Obj, RENext, Fmt); 648 } 649 650 Fmt << ")"; 651 break; 652 } 653 default: { 654 printRelocationTargetName(Obj, RE, Fmt); 655 } 656 } 657 } 658 } else 659 printRelocationTargetName(Obj, RE, Fmt); 660 661 Fmt.flush(); 662 Result.append(FmtBuf.begin(), FmtBuf.end()); 663 return Error::success(); 664 } 665 666 static void PrintIndirectSymbolTable(MachOObjectFile *O, bool verbose, 667 uint32_t n, uint32_t count, 668 uint32_t stride, uint64_t addr) { 669 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand(); 670 uint32_t nindirectsyms = Dysymtab.nindirectsyms; 671 if (n > nindirectsyms) 672 outs() << " (entries start past the end of the indirect symbol " 673 "table) (reserved1 field greater than the table size)"; 674 else if (n + count > nindirectsyms) 675 outs() << " (entries extends past the end of the indirect symbol " 676 "table)"; 677 outs() << "\n"; 678 uint32_t cputype = O->getHeader().cputype; 679 if (cputype & MachO::CPU_ARCH_ABI64) 680 outs() << "address index"; 681 else 682 outs() << "address index"; 683 if (verbose) 684 outs() << " name\n"; 685 else 686 outs() << "\n"; 687 for (uint32_t j = 0; j < count && n + j < nindirectsyms; j++) { 688 if (cputype & MachO::CPU_ARCH_ABI64) 689 outs() << format("0x%016" PRIx64, addr + j * stride) << " "; 690 else 691 outs() << format("0x%08" PRIx32, (uint32_t)addr + j * stride) << " "; 692 MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand(); 693 uint32_t indirect_symbol = O->getIndirectSymbolTableEntry(Dysymtab, n + j); 694 if (indirect_symbol == MachO::INDIRECT_SYMBOL_LOCAL) { 695 outs() << "LOCAL\n"; 696 continue; 697 } 698 if (indirect_symbol == 699 (MachO::INDIRECT_SYMBOL_LOCAL | MachO::INDIRECT_SYMBOL_ABS)) { 700 outs() << "LOCAL ABSOLUTE\n"; 701 continue; 702 } 703 if (indirect_symbol == MachO::INDIRECT_SYMBOL_ABS) { 704 outs() << "ABSOLUTE\n"; 705 continue; 706 } 707 outs() << format("%5u ", indirect_symbol); 708 if (verbose) { 709 MachO::symtab_command Symtab = O->getSymtabLoadCommand(); 710 if (indirect_symbol < Symtab.nsyms) { 711 symbol_iterator Sym = O->getSymbolByIndex(indirect_symbol); 712 SymbolRef Symbol = *Sym; 713 outs() << unwrapOrError(Symbol.getName(), O->getFileName()); 714 } else { 715 outs() << "?"; 716 } 717 } 718 outs() << "\n"; 719 } 720 } 721 722 static void PrintIndirectSymbols(MachOObjectFile *O, bool verbose) { 723 for (const auto &Load : O->load_commands()) { 724 if (Load.C.cmd == MachO::LC_SEGMENT_64) { 725 MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load); 726 for (unsigned J = 0; J < Seg.nsects; ++J) { 727 MachO::section_64 Sec = O->getSection64(Load, J); 728 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE; 729 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS || 730 section_type == MachO::S_LAZY_SYMBOL_POINTERS || 731 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS || 732 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS || 733 section_type == MachO::S_SYMBOL_STUBS) { 734 uint32_t stride; 735 if (section_type == MachO::S_SYMBOL_STUBS) 736 stride = Sec.reserved2; 737 else 738 stride = 8; 739 if (stride == 0) { 740 outs() << "Can't print indirect symbols for (" << Sec.segname << "," 741 << Sec.sectname << ") " 742 << "(size of stubs in reserved2 field is zero)\n"; 743 continue; 744 } 745 uint32_t count = Sec.size / stride; 746 outs() << "Indirect symbols for (" << Sec.segname << "," 747 << Sec.sectname << ") " << count << " entries"; 748 uint32_t n = Sec.reserved1; 749 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr); 750 } 751 } 752 } else if (Load.C.cmd == MachO::LC_SEGMENT) { 753 MachO::segment_command Seg = O->getSegmentLoadCommand(Load); 754 for (unsigned J = 0; J < Seg.nsects; ++J) { 755 MachO::section Sec = O->getSection(Load, J); 756 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE; 757 if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS || 758 section_type == MachO::S_LAZY_SYMBOL_POINTERS || 759 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS || 760 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS || 761 section_type == MachO::S_SYMBOL_STUBS) { 762 uint32_t stride; 763 if (section_type == MachO::S_SYMBOL_STUBS) 764 stride = Sec.reserved2; 765 else 766 stride = 4; 767 if (stride == 0) { 768 outs() << "Can't print indirect symbols for (" << Sec.segname << "," 769 << Sec.sectname << ") " 770 << "(size of stubs in reserved2 field is zero)\n"; 771 continue; 772 } 773 uint32_t count = Sec.size / stride; 774 outs() << "Indirect symbols for (" << Sec.segname << "," 775 << Sec.sectname << ") " << count << " entries"; 776 uint32_t n = Sec.reserved1; 777 PrintIndirectSymbolTable(O, verbose, n, count, stride, Sec.addr); 778 } 779 } 780 } 781 } 782 } 783 784 static void PrintRType(const uint64_t cputype, const unsigned r_type) { 785 static char const *generic_r_types[] = { 786 "VANILLA ", "PAIR ", "SECTDIF ", "PBLAPTR ", "LOCSDIF ", "TLV ", 787 " 6 (?) ", " 7 (?) ", " 8 (?) ", " 9 (?) ", " 10 (?) ", " 11 (?) ", 788 " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) " 789 }; 790 static char const *x86_64_r_types[] = { 791 "UNSIGND ", "SIGNED ", "BRANCH ", "GOT_LD ", "GOT ", "SUB ", 792 "SIGNED1 ", "SIGNED2 ", "SIGNED4 ", "TLV ", " 10 (?) ", " 11 (?) ", 793 " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) " 794 }; 795 static char const *arm_r_types[] = { 796 "VANILLA ", "PAIR ", "SECTDIFF", "LOCSDIF ", "PBLAPTR ", 797 "BR24 ", "T_BR22 ", "T_BR32 ", "HALF ", "HALFDIF ", 798 " 10 (?) ", " 11 (?) ", " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) " 799 }; 800 static char const *arm64_r_types[] = { 801 "UNSIGND ", "SUB ", "BR26 ", "PAGE21 ", "PAGOF12 ", 802 "GOTLDP ", "GOTLDPOF", "PTRTGOT ", "TLVLDP ", "TLVLDPOF", 803 "ADDEND ", " 11 (?) ", " 12 (?) ", " 13 (?) ", " 14 (?) ", " 15 (?) " 804 }; 805 806 if (r_type > 0xf){ 807 outs() << format("%-7u", r_type) << " "; 808 return; 809 } 810 switch (cputype) { 811 case MachO::CPU_TYPE_I386: 812 outs() << generic_r_types[r_type]; 813 break; 814 case MachO::CPU_TYPE_X86_64: 815 outs() << x86_64_r_types[r_type]; 816 break; 817 case MachO::CPU_TYPE_ARM: 818 outs() << arm_r_types[r_type]; 819 break; 820 case MachO::CPU_TYPE_ARM64: 821 case MachO::CPU_TYPE_ARM64_32: 822 outs() << arm64_r_types[r_type]; 823 break; 824 default: 825 outs() << format("%-7u ", r_type); 826 } 827 } 828 829 static void PrintRLength(const uint64_t cputype, const unsigned r_type, 830 const unsigned r_length, const bool previous_arm_half){ 831 if (cputype == MachO::CPU_TYPE_ARM && 832 (r_type == MachO::ARM_RELOC_HALF || 833 r_type == MachO::ARM_RELOC_HALF_SECTDIFF || previous_arm_half == true)) { 834 if ((r_length & 0x1) == 0) 835 outs() << "lo/"; 836 else 837 outs() << "hi/"; 838 if ((r_length & 0x1) == 0) 839 outs() << "arm "; 840 else 841 outs() << "thm "; 842 } else { 843 switch (r_length) { 844 case 0: 845 outs() << "byte "; 846 break; 847 case 1: 848 outs() << "word "; 849 break; 850 case 2: 851 outs() << "long "; 852 break; 853 case 3: 854 if (cputype == MachO::CPU_TYPE_X86_64) 855 outs() << "quad "; 856 else 857 outs() << format("?(%2d) ", r_length); 858 break; 859 default: 860 outs() << format("?(%2d) ", r_length); 861 } 862 } 863 } 864 865 static void PrintRelocationEntries(const MachOObjectFile *O, 866 const relocation_iterator Begin, 867 const relocation_iterator End, 868 const uint64_t cputype, 869 const bool verbose) { 870 const MachO::symtab_command Symtab = O->getSymtabLoadCommand(); 871 bool previous_arm_half = false; 872 bool previous_sectdiff = false; 873 uint32_t sectdiff_r_type = 0; 874 875 for (relocation_iterator Reloc = Begin; Reloc != End; ++Reloc) { 876 const DataRefImpl Rel = Reloc->getRawDataRefImpl(); 877 const MachO::any_relocation_info RE = O->getRelocation(Rel); 878 const unsigned r_type = O->getAnyRelocationType(RE); 879 const bool r_scattered = O->isRelocationScattered(RE); 880 const unsigned r_pcrel = O->getAnyRelocationPCRel(RE); 881 const unsigned r_length = O->getAnyRelocationLength(RE); 882 const unsigned r_address = O->getAnyRelocationAddress(RE); 883 const bool r_extern = (r_scattered ? false : 884 O->getPlainRelocationExternal(RE)); 885 const uint32_t r_value = (r_scattered ? 886 O->getScatteredRelocationValue(RE) : 0); 887 const unsigned r_symbolnum = (r_scattered ? 0 : 888 O->getPlainRelocationSymbolNum(RE)); 889 890 if (r_scattered && cputype != MachO::CPU_TYPE_X86_64) { 891 if (verbose) { 892 // scattered: address 893 if ((cputype == MachO::CPU_TYPE_I386 && 894 r_type == MachO::GENERIC_RELOC_PAIR) || 895 (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR)) 896 outs() << " "; 897 else 898 outs() << format("%08x ", (unsigned int)r_address); 899 900 // scattered: pcrel 901 if (r_pcrel) 902 outs() << "True "; 903 else 904 outs() << "False "; 905 906 // scattered: length 907 PrintRLength(cputype, r_type, r_length, previous_arm_half); 908 909 // scattered: extern & type 910 outs() << "n/a "; 911 PrintRType(cputype, r_type); 912 913 // scattered: scattered & value 914 outs() << format("True 0x%08x", (unsigned int)r_value); 915 if (previous_sectdiff == false) { 916 if ((cputype == MachO::CPU_TYPE_ARM && 917 r_type == MachO::ARM_RELOC_PAIR)) 918 outs() << format(" half = 0x%04x ", (unsigned int)r_address); 919 } else if (cputype == MachO::CPU_TYPE_ARM && 920 sectdiff_r_type == MachO::ARM_RELOC_HALF_SECTDIFF) 921 outs() << format(" other_half = 0x%04x ", (unsigned int)r_address); 922 if ((cputype == MachO::CPU_TYPE_I386 && 923 (r_type == MachO::GENERIC_RELOC_SECTDIFF || 924 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) || 925 (cputype == MachO::CPU_TYPE_ARM && 926 (sectdiff_r_type == MachO::ARM_RELOC_SECTDIFF || 927 sectdiff_r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF || 928 sectdiff_r_type == MachO::ARM_RELOC_HALF_SECTDIFF))) { 929 previous_sectdiff = true; 930 sectdiff_r_type = r_type; 931 } else { 932 previous_sectdiff = false; 933 sectdiff_r_type = 0; 934 } 935 if (cputype == MachO::CPU_TYPE_ARM && 936 (r_type == MachO::ARM_RELOC_HALF || 937 r_type == MachO::ARM_RELOC_HALF_SECTDIFF)) 938 previous_arm_half = true; 939 else 940 previous_arm_half = false; 941 outs() << "\n"; 942 } 943 else { 944 // scattered: address pcrel length extern type scattered value 945 outs() << format("%08x %1d %-2d n/a %-7d 1 0x%08x\n", 946 (unsigned int)r_address, r_pcrel, r_length, r_type, 947 (unsigned int)r_value); 948 } 949 } 950 else { 951 if (verbose) { 952 // plain: address 953 if (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR) 954 outs() << " "; 955 else 956 outs() << format("%08x ", (unsigned int)r_address); 957 958 // plain: pcrel 959 if (r_pcrel) 960 outs() << "True "; 961 else 962 outs() << "False "; 963 964 // plain: length 965 PrintRLength(cputype, r_type, r_length, previous_arm_half); 966 967 if (r_extern) { 968 // plain: extern & type & scattered 969 outs() << "True "; 970 PrintRType(cputype, r_type); 971 outs() << "False "; 972 973 // plain: symbolnum/value 974 if (r_symbolnum > Symtab.nsyms) 975 outs() << format("?(%d)\n", r_symbolnum); 976 else { 977 SymbolRef Symbol = *O->getSymbolByIndex(r_symbolnum); 978 Expected<StringRef> SymNameNext = Symbol.getName(); 979 const char *name = NULL; 980 if (SymNameNext) 981 name = SymNameNext->data(); 982 if (name == NULL) 983 outs() << format("?(%d)\n", r_symbolnum); 984 else 985 outs() << name << "\n"; 986 } 987 } 988 else { 989 // plain: extern & type & scattered 990 outs() << "False "; 991 PrintRType(cputype, r_type); 992 outs() << "False "; 993 994 // plain: symbolnum/value 995 if (cputype == MachO::CPU_TYPE_ARM && r_type == MachO::ARM_RELOC_PAIR) 996 outs() << format("other_half = 0x%04x\n", (unsigned int)r_address); 997 else if ((cputype == MachO::CPU_TYPE_ARM64 || 998 cputype == MachO::CPU_TYPE_ARM64_32) && 999 r_type == MachO::ARM64_RELOC_ADDEND) 1000 outs() << format("addend = 0x%06x\n", (unsigned int)r_symbolnum); 1001 else { 1002 outs() << format("%d ", r_symbolnum); 1003 if (r_symbolnum == MachO::R_ABS) 1004 outs() << "R_ABS\n"; 1005 else { 1006 // in this case, r_symbolnum is actually a 1-based section number 1007 uint32_t nsects = O->section_end()->getRawDataRefImpl().d.a; 1008 if (r_symbolnum > 0 && r_symbolnum <= nsects) { 1009 object::DataRefImpl DRI; 1010 DRI.d.a = r_symbolnum-1; 1011 StringRef SegName = O->getSectionFinalSegmentName(DRI); 1012 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI)) 1013 outs() << "(" << SegName << "," << *NameOrErr << ")\n"; 1014 else 1015 outs() << "(?,?)\n"; 1016 } 1017 else { 1018 outs() << "(?,?)\n"; 1019 } 1020 } 1021 } 1022 } 1023 if (cputype == MachO::CPU_TYPE_ARM && 1024 (r_type == MachO::ARM_RELOC_HALF || 1025 r_type == MachO::ARM_RELOC_HALF_SECTDIFF)) 1026 previous_arm_half = true; 1027 else 1028 previous_arm_half = false; 1029 } 1030 else { 1031 // plain: address pcrel length extern type scattered symbolnum/section 1032 outs() << format("%08x %1d %-2d %1d %-7d 0 %d\n", 1033 (unsigned int)r_address, r_pcrel, r_length, r_extern, 1034 r_type, r_symbolnum); 1035 } 1036 } 1037 } 1038 } 1039 1040 static void PrintRelocations(const MachOObjectFile *O, const bool verbose) { 1041 const uint64_t cputype = O->getHeader().cputype; 1042 const MachO::dysymtab_command Dysymtab = O->getDysymtabLoadCommand(); 1043 if (Dysymtab.nextrel != 0) { 1044 outs() << "External relocation information " << Dysymtab.nextrel 1045 << " entries"; 1046 outs() << "\naddress pcrel length extern type scattered " 1047 "symbolnum/value\n"; 1048 PrintRelocationEntries(O, O->extrel_begin(), O->extrel_end(), cputype, 1049 verbose); 1050 } 1051 if (Dysymtab.nlocrel != 0) { 1052 outs() << format("Local relocation information %u entries", 1053 Dysymtab.nlocrel); 1054 outs() << "\naddress pcrel length extern type scattered " 1055 "symbolnum/value\n"; 1056 PrintRelocationEntries(O, O->locrel_begin(), O->locrel_end(), cputype, 1057 verbose); 1058 } 1059 for (const auto &Load : O->load_commands()) { 1060 if (Load.C.cmd == MachO::LC_SEGMENT_64) { 1061 const MachO::segment_command_64 Seg = O->getSegment64LoadCommand(Load); 1062 for (unsigned J = 0; J < Seg.nsects; ++J) { 1063 const MachO::section_64 Sec = O->getSection64(Load, J); 1064 if (Sec.nreloc != 0) { 1065 DataRefImpl DRI; 1066 DRI.d.a = J; 1067 const StringRef SegName = O->getSectionFinalSegmentName(DRI); 1068 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI)) 1069 outs() << "Relocation information (" << SegName << "," << *NameOrErr 1070 << format(") %u entries", Sec.nreloc); 1071 else 1072 outs() << "Relocation information (" << SegName << ",?) " 1073 << format("%u entries", Sec.nreloc); 1074 outs() << "\naddress pcrel length extern type scattered " 1075 "symbolnum/value\n"; 1076 PrintRelocationEntries(O, O->section_rel_begin(DRI), 1077 O->section_rel_end(DRI), cputype, verbose); 1078 } 1079 } 1080 } else if (Load.C.cmd == MachO::LC_SEGMENT) { 1081 const MachO::segment_command Seg = O->getSegmentLoadCommand(Load); 1082 for (unsigned J = 0; J < Seg.nsects; ++J) { 1083 const MachO::section Sec = O->getSection(Load, J); 1084 if (Sec.nreloc != 0) { 1085 DataRefImpl DRI; 1086 DRI.d.a = J; 1087 const StringRef SegName = O->getSectionFinalSegmentName(DRI); 1088 if (Expected<StringRef> NameOrErr = O->getSectionName(DRI)) 1089 outs() << "Relocation information (" << SegName << "," << *NameOrErr 1090 << format(") %u entries", Sec.nreloc); 1091 else 1092 outs() << "Relocation information (" << SegName << ",?) " 1093 << format("%u entries", Sec.nreloc); 1094 outs() << "\naddress pcrel length extern type scattered " 1095 "symbolnum/value\n"; 1096 PrintRelocationEntries(O, O->section_rel_begin(DRI), 1097 O->section_rel_end(DRI), cputype, verbose); 1098 } 1099 } 1100 } 1101 } 1102 } 1103 1104 static void PrintDataInCodeTable(MachOObjectFile *O, bool verbose) { 1105 MachO::linkedit_data_command DIC = O->getDataInCodeLoadCommand(); 1106 uint32_t nentries = DIC.datasize / sizeof(struct MachO::data_in_code_entry); 1107 outs() << "Data in code table (" << nentries << " entries)\n"; 1108 outs() << "offset length kind\n"; 1109 for (dice_iterator DI = O->begin_dices(), DE = O->end_dices(); DI != DE; 1110 ++DI) { 1111 uint32_t Offset; 1112 DI->getOffset(Offset); 1113 outs() << format("0x%08" PRIx32, Offset) << " "; 1114 uint16_t Length; 1115 DI->getLength(Length); 1116 outs() << format("%6u", Length) << " "; 1117 uint16_t Kind; 1118 DI->getKind(Kind); 1119 if (verbose) { 1120 switch (Kind) { 1121 case MachO::DICE_KIND_DATA: 1122 outs() << "DATA"; 1123 break; 1124 case MachO::DICE_KIND_JUMP_TABLE8: 1125 outs() << "JUMP_TABLE8"; 1126 break; 1127 case MachO::DICE_KIND_JUMP_TABLE16: 1128 outs() << "JUMP_TABLE16"; 1129 break; 1130 case MachO::DICE_KIND_JUMP_TABLE32: 1131 outs() << "JUMP_TABLE32"; 1132 break; 1133 case MachO::DICE_KIND_ABS_JUMP_TABLE32: 1134 outs() << "ABS_JUMP_TABLE32"; 1135 break; 1136 default: 1137 outs() << format("0x%04" PRIx32, Kind); 1138 break; 1139 } 1140 } else 1141 outs() << format("0x%04" PRIx32, Kind); 1142 outs() << "\n"; 1143 } 1144 } 1145 1146 static void PrintLinkOptHints(MachOObjectFile *O) { 1147 MachO::linkedit_data_command LohLC = O->getLinkOptHintsLoadCommand(); 1148 const char *loh = O->getData().substr(LohLC.dataoff, 1).data(); 1149 uint32_t nloh = LohLC.datasize; 1150 outs() << "Linker optimiztion hints (" << nloh << " total bytes)\n"; 1151 for (uint32_t i = 0; i < nloh;) { 1152 unsigned n; 1153 uint64_t identifier = decodeULEB128((const uint8_t *)(loh + i), &n); 1154 i += n; 1155 outs() << " identifier " << identifier << " "; 1156 if (i >= nloh) 1157 return; 1158 switch (identifier) { 1159 case 1: 1160 outs() << "AdrpAdrp\n"; 1161 break; 1162 case 2: 1163 outs() << "AdrpLdr\n"; 1164 break; 1165 case 3: 1166 outs() << "AdrpAddLdr\n"; 1167 break; 1168 case 4: 1169 outs() << "AdrpLdrGotLdr\n"; 1170 break; 1171 case 5: 1172 outs() << "AdrpAddStr\n"; 1173 break; 1174 case 6: 1175 outs() << "AdrpLdrGotStr\n"; 1176 break; 1177 case 7: 1178 outs() << "AdrpAdd\n"; 1179 break; 1180 case 8: 1181 outs() << "AdrpLdrGot\n"; 1182 break; 1183 default: 1184 outs() << "Unknown identifier value\n"; 1185 break; 1186 } 1187 uint64_t narguments = decodeULEB128((const uint8_t *)(loh + i), &n); 1188 i += n; 1189 outs() << " narguments " << narguments << "\n"; 1190 if (i >= nloh) 1191 return; 1192 1193 for (uint32_t j = 0; j < narguments; j++) { 1194 uint64_t value = decodeULEB128((const uint8_t *)(loh + i), &n); 1195 i += n; 1196 outs() << "\tvalue " << format("0x%" PRIx64, value) << "\n"; 1197 if (i >= nloh) 1198 return; 1199 } 1200 } 1201 } 1202 1203 static void PrintDylibs(MachOObjectFile *O, bool JustId) { 1204 unsigned Index = 0; 1205 for (const auto &Load : O->load_commands()) { 1206 if ((JustId && Load.C.cmd == MachO::LC_ID_DYLIB) || 1207 (!JustId && (Load.C.cmd == MachO::LC_ID_DYLIB || 1208 Load.C.cmd == MachO::LC_LOAD_DYLIB || 1209 Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB || 1210 Load.C.cmd == MachO::LC_REEXPORT_DYLIB || 1211 Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB || 1212 Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB))) { 1213 MachO::dylib_command dl = O->getDylibIDLoadCommand(Load); 1214 if (dl.dylib.name < dl.cmdsize) { 1215 const char *p = (const char *)(Load.Ptr) + dl.dylib.name; 1216 if (JustId) 1217 outs() << p << "\n"; 1218 else { 1219 outs() << "\t" << p; 1220 outs() << " (compatibility version " 1221 << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "." 1222 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "." 1223 << (dl.dylib.compatibility_version & 0xff) << ","; 1224 outs() << " current version " 1225 << ((dl.dylib.current_version >> 16) & 0xffff) << "." 1226 << ((dl.dylib.current_version >> 8) & 0xff) << "." 1227 << (dl.dylib.current_version & 0xff); 1228 if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB) 1229 outs() << ", weak"; 1230 if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB) 1231 outs() << ", reexport"; 1232 if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) 1233 outs() << ", upward"; 1234 if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB) 1235 outs() << ", lazy"; 1236 outs() << ")\n"; 1237 } 1238 } else { 1239 outs() << "\tBad offset (" << dl.dylib.name << ") for name of "; 1240 if (Load.C.cmd == MachO::LC_ID_DYLIB) 1241 outs() << "LC_ID_DYLIB "; 1242 else if (Load.C.cmd == MachO::LC_LOAD_DYLIB) 1243 outs() << "LC_LOAD_DYLIB "; 1244 else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB) 1245 outs() << "LC_LOAD_WEAK_DYLIB "; 1246 else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB) 1247 outs() << "LC_LAZY_LOAD_DYLIB "; 1248 else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB) 1249 outs() << "LC_REEXPORT_DYLIB "; 1250 else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) 1251 outs() << "LC_LOAD_UPWARD_DYLIB "; 1252 else 1253 outs() << "LC_??? "; 1254 outs() << "command " << Index++ << "\n"; 1255 } 1256 } 1257 } 1258 } 1259 1260 typedef DenseMap<uint64_t, StringRef> SymbolAddressMap; 1261 1262 static void CreateSymbolAddressMap(MachOObjectFile *O, 1263 SymbolAddressMap *AddrMap) { 1264 // Create a map of symbol addresses to symbol names. 1265 const StringRef FileName = O->getFileName(); 1266 for (const SymbolRef &Symbol : O->symbols()) { 1267 SymbolRef::Type ST = unwrapOrError(Symbol.getType(), FileName); 1268 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data || 1269 ST == SymbolRef::ST_Other) { 1270 uint64_t Address = Symbol.getValue(); 1271 StringRef SymName = unwrapOrError(Symbol.getName(), FileName); 1272 if (!SymName.startswith(".objc")) 1273 (*AddrMap)[Address] = SymName; 1274 } 1275 } 1276 } 1277 1278 // GuessSymbolName is passed the address of what might be a symbol and a 1279 // pointer to the SymbolAddressMap. It returns the name of a symbol 1280 // with that address or nullptr if no symbol is found with that address. 1281 static const char *GuessSymbolName(uint64_t value, SymbolAddressMap *AddrMap) { 1282 const char *SymbolName = nullptr; 1283 // A DenseMap can't lookup up some values. 1284 if (value != 0xffffffffffffffffULL && value != 0xfffffffffffffffeULL) { 1285 StringRef name = AddrMap->lookup(value); 1286 if (!name.empty()) 1287 SymbolName = name.data(); 1288 } 1289 return SymbolName; 1290 } 1291 1292 static void DumpCstringChar(const char c) { 1293 char p[2]; 1294 p[0] = c; 1295 p[1] = '\0'; 1296 outs().write_escaped(p); 1297 } 1298 1299 static void DumpCstringSection(MachOObjectFile *O, const char *sect, 1300 uint32_t sect_size, uint64_t sect_addr, 1301 bool print_addresses) { 1302 for (uint32_t i = 0; i < sect_size; i++) { 1303 if (print_addresses) { 1304 if (O->is64Bit()) 1305 outs() << format("%016" PRIx64, sect_addr + i) << " "; 1306 else 1307 outs() << format("%08" PRIx64, sect_addr + i) << " "; 1308 } 1309 for (; i < sect_size && sect[i] != '\0'; i++) 1310 DumpCstringChar(sect[i]); 1311 if (i < sect_size && sect[i] == '\0') 1312 outs() << "\n"; 1313 } 1314 } 1315 1316 static void DumpLiteral4(uint32_t l, float f) { 1317 outs() << format("0x%08" PRIx32, l); 1318 if ((l & 0x7f800000) != 0x7f800000) 1319 outs() << format(" (%.16e)\n", f); 1320 else { 1321 if (l == 0x7f800000) 1322 outs() << " (+Infinity)\n"; 1323 else if (l == 0xff800000) 1324 outs() << " (-Infinity)\n"; 1325 else if ((l & 0x00400000) == 0x00400000) 1326 outs() << " (non-signaling Not-a-Number)\n"; 1327 else 1328 outs() << " (signaling Not-a-Number)\n"; 1329 } 1330 } 1331 1332 static void DumpLiteral4Section(MachOObjectFile *O, const char *sect, 1333 uint32_t sect_size, uint64_t sect_addr, 1334 bool print_addresses) { 1335 for (uint32_t i = 0; i < sect_size; i += sizeof(float)) { 1336 if (print_addresses) { 1337 if (O->is64Bit()) 1338 outs() << format("%016" PRIx64, sect_addr + i) << " "; 1339 else 1340 outs() << format("%08" PRIx64, sect_addr + i) << " "; 1341 } 1342 float f; 1343 memcpy(&f, sect + i, sizeof(float)); 1344 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1345 sys::swapByteOrder(f); 1346 uint32_t l; 1347 memcpy(&l, sect + i, sizeof(uint32_t)); 1348 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1349 sys::swapByteOrder(l); 1350 DumpLiteral4(l, f); 1351 } 1352 } 1353 1354 static void DumpLiteral8(MachOObjectFile *O, uint32_t l0, uint32_t l1, 1355 double d) { 1356 outs() << format("0x%08" PRIx32, l0) << " " << format("0x%08" PRIx32, l1); 1357 uint32_t Hi, Lo; 1358 Hi = (O->isLittleEndian()) ? l1 : l0; 1359 Lo = (O->isLittleEndian()) ? l0 : l1; 1360 1361 // Hi is the high word, so this is equivalent to if(isfinite(d)) 1362 if ((Hi & 0x7ff00000) != 0x7ff00000) 1363 outs() << format(" (%.16e)\n", d); 1364 else { 1365 if (Hi == 0x7ff00000 && Lo == 0) 1366 outs() << " (+Infinity)\n"; 1367 else if (Hi == 0xfff00000 && Lo == 0) 1368 outs() << " (-Infinity)\n"; 1369 else if ((Hi & 0x00080000) == 0x00080000) 1370 outs() << " (non-signaling Not-a-Number)\n"; 1371 else 1372 outs() << " (signaling Not-a-Number)\n"; 1373 } 1374 } 1375 1376 static void DumpLiteral8Section(MachOObjectFile *O, const char *sect, 1377 uint32_t sect_size, uint64_t sect_addr, 1378 bool print_addresses) { 1379 for (uint32_t i = 0; i < sect_size; i += sizeof(double)) { 1380 if (print_addresses) { 1381 if (O->is64Bit()) 1382 outs() << format("%016" PRIx64, sect_addr + i) << " "; 1383 else 1384 outs() << format("%08" PRIx64, sect_addr + i) << " "; 1385 } 1386 double d; 1387 memcpy(&d, sect + i, sizeof(double)); 1388 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1389 sys::swapByteOrder(d); 1390 uint32_t l0, l1; 1391 memcpy(&l0, sect + i, sizeof(uint32_t)); 1392 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t)); 1393 if (O->isLittleEndian() != sys::IsLittleEndianHost) { 1394 sys::swapByteOrder(l0); 1395 sys::swapByteOrder(l1); 1396 } 1397 DumpLiteral8(O, l0, l1, d); 1398 } 1399 } 1400 1401 static void DumpLiteral16(uint32_t l0, uint32_t l1, uint32_t l2, uint32_t l3) { 1402 outs() << format("0x%08" PRIx32, l0) << " "; 1403 outs() << format("0x%08" PRIx32, l1) << " "; 1404 outs() << format("0x%08" PRIx32, l2) << " "; 1405 outs() << format("0x%08" PRIx32, l3) << "\n"; 1406 } 1407 1408 static void DumpLiteral16Section(MachOObjectFile *O, const char *sect, 1409 uint32_t sect_size, uint64_t sect_addr, 1410 bool print_addresses) { 1411 for (uint32_t i = 0; i < sect_size; i += 16) { 1412 if (print_addresses) { 1413 if (O->is64Bit()) 1414 outs() << format("%016" PRIx64, sect_addr + i) << " "; 1415 else 1416 outs() << format("%08" PRIx64, sect_addr + i) << " "; 1417 } 1418 uint32_t l0, l1, l2, l3; 1419 memcpy(&l0, sect + i, sizeof(uint32_t)); 1420 memcpy(&l1, sect + i + sizeof(uint32_t), sizeof(uint32_t)); 1421 memcpy(&l2, sect + i + 2 * sizeof(uint32_t), sizeof(uint32_t)); 1422 memcpy(&l3, sect + i + 3 * sizeof(uint32_t), sizeof(uint32_t)); 1423 if (O->isLittleEndian() != sys::IsLittleEndianHost) { 1424 sys::swapByteOrder(l0); 1425 sys::swapByteOrder(l1); 1426 sys::swapByteOrder(l2); 1427 sys::swapByteOrder(l3); 1428 } 1429 DumpLiteral16(l0, l1, l2, l3); 1430 } 1431 } 1432 1433 static void DumpLiteralPointerSection(MachOObjectFile *O, 1434 const SectionRef &Section, 1435 const char *sect, uint32_t sect_size, 1436 uint64_t sect_addr, 1437 bool print_addresses) { 1438 // Collect the literal sections in this Mach-O file. 1439 std::vector<SectionRef> LiteralSections; 1440 for (const SectionRef &Section : O->sections()) { 1441 DataRefImpl Ref = Section.getRawDataRefImpl(); 1442 uint32_t section_type; 1443 if (O->is64Bit()) { 1444 const MachO::section_64 Sec = O->getSection64(Ref); 1445 section_type = Sec.flags & MachO::SECTION_TYPE; 1446 } else { 1447 const MachO::section Sec = O->getSection(Ref); 1448 section_type = Sec.flags & MachO::SECTION_TYPE; 1449 } 1450 if (section_type == MachO::S_CSTRING_LITERALS || 1451 section_type == MachO::S_4BYTE_LITERALS || 1452 section_type == MachO::S_8BYTE_LITERALS || 1453 section_type == MachO::S_16BYTE_LITERALS) 1454 LiteralSections.push_back(Section); 1455 } 1456 1457 // Set the size of the literal pointer. 1458 uint32_t lp_size = O->is64Bit() ? 8 : 4; 1459 1460 // Collect the external relocation symbols for the literal pointers. 1461 std::vector<std::pair<uint64_t, SymbolRef>> Relocs; 1462 for (const RelocationRef &Reloc : Section.relocations()) { 1463 DataRefImpl Rel; 1464 MachO::any_relocation_info RE; 1465 bool isExtern = false; 1466 Rel = Reloc.getRawDataRefImpl(); 1467 RE = O->getRelocation(Rel); 1468 isExtern = O->getPlainRelocationExternal(RE); 1469 if (isExtern) { 1470 uint64_t RelocOffset = Reloc.getOffset(); 1471 symbol_iterator RelocSym = Reloc.getSymbol(); 1472 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym)); 1473 } 1474 } 1475 array_pod_sort(Relocs.begin(), Relocs.end()); 1476 1477 // Dump each literal pointer. 1478 for (uint32_t i = 0; i < sect_size; i += lp_size) { 1479 if (print_addresses) { 1480 if (O->is64Bit()) 1481 outs() << format("%016" PRIx64, sect_addr + i) << " "; 1482 else 1483 outs() << format("%08" PRIx64, sect_addr + i) << " "; 1484 } 1485 uint64_t lp; 1486 if (O->is64Bit()) { 1487 memcpy(&lp, sect + i, sizeof(uint64_t)); 1488 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1489 sys::swapByteOrder(lp); 1490 } else { 1491 uint32_t li; 1492 memcpy(&li, sect + i, sizeof(uint32_t)); 1493 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1494 sys::swapByteOrder(li); 1495 lp = li; 1496 } 1497 1498 // First look for an external relocation entry for this literal pointer. 1499 auto Reloc = find_if(Relocs, [&](const std::pair<uint64_t, SymbolRef> &P) { 1500 return P.first == i; 1501 }); 1502 if (Reloc != Relocs.end()) { 1503 symbol_iterator RelocSym = Reloc->second; 1504 StringRef SymName = unwrapOrError(RelocSym->getName(), O->getFileName()); 1505 outs() << "external relocation entry for symbol:" << SymName << "\n"; 1506 continue; 1507 } 1508 1509 // For local references see what the section the literal pointer points to. 1510 auto Sect = find_if(LiteralSections, [&](const SectionRef &R) { 1511 return lp >= R.getAddress() && lp < R.getAddress() + R.getSize(); 1512 }); 1513 if (Sect == LiteralSections.end()) { 1514 outs() << format("0x%" PRIx64, lp) << " (not in a literal section)\n"; 1515 continue; 1516 } 1517 1518 uint64_t SectAddress = Sect->getAddress(); 1519 uint64_t SectSize = Sect->getSize(); 1520 1521 StringRef SectName; 1522 Expected<StringRef> SectNameOrErr = Sect->getName(); 1523 if (SectNameOrErr) 1524 SectName = *SectNameOrErr; 1525 else 1526 consumeError(SectNameOrErr.takeError()); 1527 1528 DataRefImpl Ref = Sect->getRawDataRefImpl(); 1529 StringRef SegmentName = O->getSectionFinalSegmentName(Ref); 1530 outs() << SegmentName << ":" << SectName << ":"; 1531 1532 uint32_t section_type; 1533 if (O->is64Bit()) { 1534 const MachO::section_64 Sec = O->getSection64(Ref); 1535 section_type = Sec.flags & MachO::SECTION_TYPE; 1536 } else { 1537 const MachO::section Sec = O->getSection(Ref); 1538 section_type = Sec.flags & MachO::SECTION_TYPE; 1539 } 1540 1541 StringRef BytesStr = unwrapOrError(Sect->getContents(), O->getFileName()); 1542 1543 const char *Contents = reinterpret_cast<const char *>(BytesStr.data()); 1544 1545 switch (section_type) { 1546 case MachO::S_CSTRING_LITERALS: 1547 for (uint64_t i = lp - SectAddress; i < SectSize && Contents[i] != '\0'; 1548 i++) { 1549 DumpCstringChar(Contents[i]); 1550 } 1551 outs() << "\n"; 1552 break; 1553 case MachO::S_4BYTE_LITERALS: 1554 float f; 1555 memcpy(&f, Contents + (lp - SectAddress), sizeof(float)); 1556 uint32_t l; 1557 memcpy(&l, Contents + (lp - SectAddress), sizeof(uint32_t)); 1558 if (O->isLittleEndian() != sys::IsLittleEndianHost) { 1559 sys::swapByteOrder(f); 1560 sys::swapByteOrder(l); 1561 } 1562 DumpLiteral4(l, f); 1563 break; 1564 case MachO::S_8BYTE_LITERALS: { 1565 double d; 1566 memcpy(&d, Contents + (lp - SectAddress), sizeof(double)); 1567 uint32_t l0, l1; 1568 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t)); 1569 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t), 1570 sizeof(uint32_t)); 1571 if (O->isLittleEndian() != sys::IsLittleEndianHost) { 1572 sys::swapByteOrder(f); 1573 sys::swapByteOrder(l0); 1574 sys::swapByteOrder(l1); 1575 } 1576 DumpLiteral8(O, l0, l1, d); 1577 break; 1578 } 1579 case MachO::S_16BYTE_LITERALS: { 1580 uint32_t l0, l1, l2, l3; 1581 memcpy(&l0, Contents + (lp - SectAddress), sizeof(uint32_t)); 1582 memcpy(&l1, Contents + (lp - SectAddress) + sizeof(uint32_t), 1583 sizeof(uint32_t)); 1584 memcpy(&l2, Contents + (lp - SectAddress) + 2 * sizeof(uint32_t), 1585 sizeof(uint32_t)); 1586 memcpy(&l3, Contents + (lp - SectAddress) + 3 * sizeof(uint32_t), 1587 sizeof(uint32_t)); 1588 if (O->isLittleEndian() != sys::IsLittleEndianHost) { 1589 sys::swapByteOrder(l0); 1590 sys::swapByteOrder(l1); 1591 sys::swapByteOrder(l2); 1592 sys::swapByteOrder(l3); 1593 } 1594 DumpLiteral16(l0, l1, l2, l3); 1595 break; 1596 } 1597 } 1598 } 1599 } 1600 1601 static void DumpInitTermPointerSection(MachOObjectFile *O, 1602 const SectionRef &Section, 1603 const char *sect, 1604 uint32_t sect_size, uint64_t sect_addr, 1605 SymbolAddressMap *AddrMap, 1606 bool verbose) { 1607 uint32_t stride; 1608 stride = (O->is64Bit()) ? sizeof(uint64_t) : sizeof(uint32_t); 1609 1610 // Collect the external relocation symbols for the pointers. 1611 std::vector<std::pair<uint64_t, SymbolRef>> Relocs; 1612 for (const RelocationRef &Reloc : Section.relocations()) { 1613 DataRefImpl Rel; 1614 MachO::any_relocation_info RE; 1615 bool isExtern = false; 1616 Rel = Reloc.getRawDataRefImpl(); 1617 RE = O->getRelocation(Rel); 1618 isExtern = O->getPlainRelocationExternal(RE); 1619 if (isExtern) { 1620 uint64_t RelocOffset = Reloc.getOffset(); 1621 symbol_iterator RelocSym = Reloc.getSymbol(); 1622 Relocs.push_back(std::make_pair(RelocOffset, *RelocSym)); 1623 } 1624 } 1625 array_pod_sort(Relocs.begin(), Relocs.end()); 1626 1627 for (uint32_t i = 0; i < sect_size; i += stride) { 1628 const char *SymbolName = nullptr; 1629 uint64_t p; 1630 if (O->is64Bit()) { 1631 outs() << format("0x%016" PRIx64, sect_addr + i * stride) << " "; 1632 uint64_t pointer_value; 1633 memcpy(&pointer_value, sect + i, stride); 1634 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1635 sys::swapByteOrder(pointer_value); 1636 outs() << format("0x%016" PRIx64, pointer_value); 1637 p = pointer_value; 1638 } else { 1639 outs() << format("0x%08" PRIx64, sect_addr + i * stride) << " "; 1640 uint32_t pointer_value; 1641 memcpy(&pointer_value, sect + i, stride); 1642 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1643 sys::swapByteOrder(pointer_value); 1644 outs() << format("0x%08" PRIx32, pointer_value); 1645 p = pointer_value; 1646 } 1647 if (verbose) { 1648 // First look for an external relocation entry for this pointer. 1649 auto Reloc = find_if(Relocs, [&](const std::pair<uint64_t, SymbolRef> &P) { 1650 return P.first == i; 1651 }); 1652 if (Reloc != Relocs.end()) { 1653 symbol_iterator RelocSym = Reloc->second; 1654 outs() << " " << unwrapOrError(RelocSym->getName(), O->getFileName()); 1655 } else { 1656 SymbolName = GuessSymbolName(p, AddrMap); 1657 if (SymbolName) 1658 outs() << " " << SymbolName; 1659 } 1660 } 1661 outs() << "\n"; 1662 } 1663 } 1664 1665 static void DumpRawSectionContents(MachOObjectFile *O, const char *sect, 1666 uint32_t size, uint64_t addr) { 1667 uint32_t cputype = O->getHeader().cputype; 1668 if (cputype == MachO::CPU_TYPE_I386 || cputype == MachO::CPU_TYPE_X86_64) { 1669 uint32_t j; 1670 for (uint32_t i = 0; i < size; i += j, addr += j) { 1671 if (O->is64Bit()) 1672 outs() << format("%016" PRIx64, addr) << "\t"; 1673 else 1674 outs() << format("%08" PRIx64, addr) << "\t"; 1675 for (j = 0; j < 16 && i + j < size; j++) { 1676 uint8_t byte_word = *(sect + i + j); 1677 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " "; 1678 } 1679 outs() << "\n"; 1680 } 1681 } else { 1682 uint32_t j; 1683 for (uint32_t i = 0; i < size; i += j, addr += j) { 1684 if (O->is64Bit()) 1685 outs() << format("%016" PRIx64, addr) << "\t"; 1686 else 1687 outs() << format("%08" PRIx64, addr) << "\t"; 1688 for (j = 0; j < 4 * sizeof(int32_t) && i + j < size; 1689 j += sizeof(int32_t)) { 1690 if (i + j + sizeof(int32_t) <= size) { 1691 uint32_t long_word; 1692 memcpy(&long_word, sect + i + j, sizeof(int32_t)); 1693 if (O->isLittleEndian() != sys::IsLittleEndianHost) 1694 sys::swapByteOrder(long_word); 1695 outs() << format("%08" PRIx32, long_word) << " "; 1696 } else { 1697 for (uint32_t k = 0; i + j + k < size; k++) { 1698 uint8_t byte_word = *(sect + i + j + k); 1699 outs() << format("%02" PRIx32, (uint32_t)byte_word) << " "; 1700 } 1701 } 1702 } 1703 outs() << "\n"; 1704 } 1705 } 1706 } 1707 1708 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF, 1709 StringRef DisSegName, StringRef DisSectName); 1710 static void DumpProtocolSection(MachOObjectFile *O, const char *sect, 1711 uint32_t size, uint32_t addr); 1712 #ifdef HAVE_LIBXAR 1713 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect, 1714 uint32_t size, bool verbose, 1715 bool PrintXarHeader, bool PrintXarFileHeaders, 1716 std::string XarMemberName); 1717 #endif // defined(HAVE_LIBXAR) 1718 1719 static void DumpSectionContents(StringRef Filename, MachOObjectFile *O, 1720 bool verbose) { 1721 SymbolAddressMap AddrMap; 1722 if (verbose) 1723 CreateSymbolAddressMap(O, &AddrMap); 1724 1725 for (unsigned i = 0; i < FilterSections.size(); ++i) { 1726 StringRef DumpSection = FilterSections[i]; 1727 std::pair<StringRef, StringRef> DumpSegSectName; 1728 DumpSegSectName = DumpSection.split(','); 1729 StringRef DumpSegName, DumpSectName; 1730 if (!DumpSegSectName.second.empty()) { 1731 DumpSegName = DumpSegSectName.first; 1732 DumpSectName = DumpSegSectName.second; 1733 } else { 1734 DumpSegName = ""; 1735 DumpSectName = DumpSegSectName.first; 1736 } 1737 for (const SectionRef &Section : O->sections()) { 1738 StringRef SectName; 1739 Expected<StringRef> SecNameOrErr = Section.getName(); 1740 if (SecNameOrErr) 1741 SectName = *SecNameOrErr; 1742 else 1743 consumeError(SecNameOrErr.takeError()); 1744 1745 if (!DumpSection.empty()) 1746 FoundSectionSet.insert(DumpSection); 1747 1748 DataRefImpl Ref = Section.getRawDataRefImpl(); 1749 StringRef SegName = O->getSectionFinalSegmentName(Ref); 1750 if ((DumpSegName.empty() || SegName == DumpSegName) && 1751 (SectName == DumpSectName)) { 1752 1753 uint32_t section_flags; 1754 if (O->is64Bit()) { 1755 const MachO::section_64 Sec = O->getSection64(Ref); 1756 section_flags = Sec.flags; 1757 1758 } else { 1759 const MachO::section Sec = O->getSection(Ref); 1760 section_flags = Sec.flags; 1761 } 1762 uint32_t section_type = section_flags & MachO::SECTION_TYPE; 1763 1764 StringRef BytesStr = 1765 unwrapOrError(Section.getContents(), O->getFileName()); 1766 const char *sect = reinterpret_cast<const char *>(BytesStr.data()); 1767 uint32_t sect_size = BytesStr.size(); 1768 uint64_t sect_addr = Section.getAddress(); 1769 1770 if (!NoLeadingHeaders) 1771 outs() << "Contents of (" << SegName << "," << SectName 1772 << ") section\n"; 1773 1774 if (verbose) { 1775 if ((section_flags & MachO::S_ATTR_PURE_INSTRUCTIONS) || 1776 (section_flags & MachO::S_ATTR_SOME_INSTRUCTIONS)) { 1777 DisassembleMachO(Filename, O, SegName, SectName); 1778 continue; 1779 } 1780 if (SegName == "__TEXT" && SectName == "__info_plist") { 1781 outs() << sect; 1782 continue; 1783 } 1784 if (SegName == "__OBJC" && SectName == "__protocol") { 1785 DumpProtocolSection(O, sect, sect_size, sect_addr); 1786 continue; 1787 } 1788 #ifdef HAVE_LIBXAR 1789 if (SegName == "__LLVM" && SectName == "__bundle") { 1790 DumpBitcodeSection(O, sect, sect_size, verbose, !NoSymbolicOperands, 1791 ArchiveHeaders, ""); 1792 continue; 1793 } 1794 #endif // defined(HAVE_LIBXAR) 1795 switch (section_type) { 1796 case MachO::S_REGULAR: 1797 DumpRawSectionContents(O, sect, sect_size, sect_addr); 1798 break; 1799 case MachO::S_ZEROFILL: 1800 outs() << "zerofill section and has no contents in the file\n"; 1801 break; 1802 case MachO::S_CSTRING_LITERALS: 1803 DumpCstringSection(O, sect, sect_size, sect_addr, !NoLeadingAddr); 1804 break; 1805 case MachO::S_4BYTE_LITERALS: 1806 DumpLiteral4Section(O, sect, sect_size, sect_addr, !NoLeadingAddr); 1807 break; 1808 case MachO::S_8BYTE_LITERALS: 1809 DumpLiteral8Section(O, sect, sect_size, sect_addr, !NoLeadingAddr); 1810 break; 1811 case MachO::S_16BYTE_LITERALS: 1812 DumpLiteral16Section(O, sect, sect_size, sect_addr, !NoLeadingAddr); 1813 break; 1814 case MachO::S_LITERAL_POINTERS: 1815 DumpLiteralPointerSection(O, Section, sect, sect_size, sect_addr, 1816 !NoLeadingAddr); 1817 break; 1818 case MachO::S_MOD_INIT_FUNC_POINTERS: 1819 case MachO::S_MOD_TERM_FUNC_POINTERS: 1820 DumpInitTermPointerSection(O, Section, sect, sect_size, sect_addr, 1821 &AddrMap, verbose); 1822 break; 1823 default: 1824 outs() << "Unknown section type (" 1825 << format("0x%08" PRIx32, section_type) << ")\n"; 1826 DumpRawSectionContents(O, sect, sect_size, sect_addr); 1827 break; 1828 } 1829 } else { 1830 if (section_type == MachO::S_ZEROFILL) 1831 outs() << "zerofill section and has no contents in the file\n"; 1832 else 1833 DumpRawSectionContents(O, sect, sect_size, sect_addr); 1834 } 1835 } 1836 } 1837 } 1838 } 1839 1840 static void DumpInfoPlistSectionContents(StringRef Filename, 1841 MachOObjectFile *O) { 1842 for (const SectionRef &Section : O->sections()) { 1843 StringRef SectName; 1844 Expected<StringRef> SecNameOrErr = Section.getName(); 1845 if (SecNameOrErr) 1846 SectName = *SecNameOrErr; 1847 else 1848 consumeError(SecNameOrErr.takeError()); 1849 1850 DataRefImpl Ref = Section.getRawDataRefImpl(); 1851 StringRef SegName = O->getSectionFinalSegmentName(Ref); 1852 if (SegName == "__TEXT" && SectName == "__info_plist") { 1853 if (!NoLeadingHeaders) 1854 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 1855 StringRef BytesStr = 1856 unwrapOrError(Section.getContents(), O->getFileName()); 1857 const char *sect = reinterpret_cast<const char *>(BytesStr.data()); 1858 outs() << format("%.*s", BytesStr.size(), sect) << "\n"; 1859 return; 1860 } 1861 } 1862 } 1863 1864 // checkMachOAndArchFlags() checks to see if the ObjectFile is a Mach-O file 1865 // and if it is and there is a list of architecture flags is specified then 1866 // check to make sure this Mach-O file is one of those architectures or all 1867 // architectures were specified. If not then an error is generated and this 1868 // routine returns false. Else it returns true. 1869 static bool checkMachOAndArchFlags(ObjectFile *O, StringRef Filename) { 1870 auto *MachO = dyn_cast<MachOObjectFile>(O); 1871 1872 if (!MachO || ArchAll || ArchFlags.empty()) 1873 return true; 1874 1875 MachO::mach_header H; 1876 MachO::mach_header_64 H_64; 1877 Triple T; 1878 const char *McpuDefault, *ArchFlag; 1879 if (MachO->is64Bit()) { 1880 H_64 = MachO->MachOObjectFile::getHeader64(); 1881 T = MachOObjectFile::getArchTriple(H_64.cputype, H_64.cpusubtype, 1882 &McpuDefault, &ArchFlag); 1883 } else { 1884 H = MachO->MachOObjectFile::getHeader(); 1885 T = MachOObjectFile::getArchTriple(H.cputype, H.cpusubtype, 1886 &McpuDefault, &ArchFlag); 1887 } 1888 const std::string ArchFlagName(ArchFlag); 1889 if (none_of(ArchFlags, [&](const std::string &Name) { 1890 return Name == ArchFlagName; 1891 })) { 1892 WithColor::error(errs(), "llvm-objdump") 1893 << Filename << ": no architecture specified.\n"; 1894 return false; 1895 } 1896 return true; 1897 } 1898 1899 static void printObjcMetaData(MachOObjectFile *O, bool verbose); 1900 1901 // ProcessMachO() is passed a single opened Mach-O file, which may be an 1902 // archive member and or in a slice of a universal file. It prints the 1903 // the file name and header info and then processes it according to the 1904 // command line options. 1905 static void ProcessMachO(StringRef Name, MachOObjectFile *MachOOF, 1906 StringRef ArchiveMemberName = StringRef(), 1907 StringRef ArchitectureName = StringRef()) { 1908 // If we are doing some processing here on the Mach-O file print the header 1909 // info. And don't print it otherwise like in the case of printing the 1910 // UniversalHeaders or ArchiveHeaders. 1911 if (Disassemble || Relocations || PrivateHeaders || ExportsTrie || Rebase || 1912 Bind || SymbolTable || LazyBind || WeakBind || IndirectSymbols || 1913 DataInCode || LinkOptHints || DylibsUsed || DylibId || ObjcMetaData || 1914 (!FilterSections.empty())) { 1915 if (!NoLeadingHeaders) { 1916 outs() << Name; 1917 if (!ArchiveMemberName.empty()) 1918 outs() << '(' << ArchiveMemberName << ')'; 1919 if (!ArchitectureName.empty()) 1920 outs() << " (architecture " << ArchitectureName << ")"; 1921 outs() << ":\n"; 1922 } 1923 } 1924 // To use the report_error() form with an ArchiveName and FileName set 1925 // these up based on what is passed for Name and ArchiveMemberName. 1926 StringRef ArchiveName; 1927 StringRef FileName; 1928 if (!ArchiveMemberName.empty()) { 1929 ArchiveName = Name; 1930 FileName = ArchiveMemberName; 1931 } else { 1932 ArchiveName = StringRef(); 1933 FileName = Name; 1934 } 1935 1936 // If we need the symbol table to do the operation then check it here to 1937 // produce a good error message as to where the Mach-O file comes from in 1938 // the error message. 1939 if (Disassemble || IndirectSymbols || !FilterSections.empty() || UnwindInfo) 1940 if (Error Err = MachOOF->checkSymbolTable()) 1941 reportError(std::move(Err), FileName, ArchiveName, ArchitectureName); 1942 1943 if (DisassembleAll) { 1944 for (const SectionRef &Section : MachOOF->sections()) { 1945 StringRef SectName; 1946 if (Expected<StringRef> NameOrErr = Section.getName()) 1947 SectName = *NameOrErr; 1948 else 1949 consumeError(NameOrErr.takeError()); 1950 1951 if (SectName.equals("__text")) { 1952 DataRefImpl Ref = Section.getRawDataRefImpl(); 1953 StringRef SegName = MachOOF->getSectionFinalSegmentName(Ref); 1954 DisassembleMachO(FileName, MachOOF, SegName, SectName); 1955 } 1956 } 1957 } 1958 else if (Disassemble) { 1959 if (MachOOF->getHeader().filetype == MachO::MH_KEXT_BUNDLE && 1960 MachOOF->getHeader().cputype == MachO::CPU_TYPE_ARM64) 1961 DisassembleMachO(FileName, MachOOF, "__TEXT_EXEC", "__text"); 1962 else 1963 DisassembleMachO(FileName, MachOOF, "__TEXT", "__text"); 1964 } 1965 if (IndirectSymbols) 1966 PrintIndirectSymbols(MachOOF, !NonVerbose); 1967 if (DataInCode) 1968 PrintDataInCodeTable(MachOOF, !NonVerbose); 1969 if (LinkOptHints) 1970 PrintLinkOptHints(MachOOF); 1971 if (Relocations) 1972 PrintRelocations(MachOOF, !NonVerbose); 1973 if (SectionHeaders) 1974 printSectionHeaders(MachOOF); 1975 if (SectionContents) 1976 printSectionContents(MachOOF); 1977 if (!FilterSections.empty()) 1978 DumpSectionContents(FileName, MachOOF, !NonVerbose); 1979 if (InfoPlist) 1980 DumpInfoPlistSectionContents(FileName, MachOOF); 1981 if (DylibsUsed) 1982 PrintDylibs(MachOOF, false); 1983 if (DylibId) 1984 PrintDylibs(MachOOF, true); 1985 if (SymbolTable) 1986 printSymbolTable(MachOOF, ArchiveName, ArchitectureName); 1987 if (UnwindInfo) 1988 printMachOUnwindInfo(MachOOF); 1989 if (PrivateHeaders) { 1990 printMachOFileHeader(MachOOF); 1991 printMachOLoadCommands(MachOOF); 1992 } 1993 if (FirstPrivateHeader) 1994 printMachOFileHeader(MachOOF); 1995 if (ObjcMetaData) 1996 printObjcMetaData(MachOOF, !NonVerbose); 1997 if (ExportsTrie) 1998 printExportsTrie(MachOOF); 1999 if (Rebase) 2000 printRebaseTable(MachOOF); 2001 if (Bind) 2002 printBindTable(MachOOF); 2003 if (LazyBind) 2004 printLazyBindTable(MachOOF); 2005 if (WeakBind) 2006 printWeakBindTable(MachOOF); 2007 2008 if (DwarfDumpType != DIDT_Null) { 2009 std::unique_ptr<DIContext> DICtx = DWARFContext::create(*MachOOF); 2010 // Dump the complete DWARF structure. 2011 DIDumpOptions DumpOpts; 2012 DumpOpts.DumpType = DwarfDumpType; 2013 DICtx->dump(outs(), DumpOpts); 2014 } 2015 } 2016 2017 // printUnknownCPUType() helps print_fat_headers for unknown CPU's. 2018 static void printUnknownCPUType(uint32_t cputype, uint32_t cpusubtype) { 2019 outs() << " cputype (" << cputype << ")\n"; 2020 outs() << " cpusubtype (" << cpusubtype << ")\n"; 2021 } 2022 2023 // printCPUType() helps print_fat_headers by printing the cputype and 2024 // pusubtype (symbolically for the one's it knows about). 2025 static void printCPUType(uint32_t cputype, uint32_t cpusubtype) { 2026 switch (cputype) { 2027 case MachO::CPU_TYPE_I386: 2028 switch (cpusubtype) { 2029 case MachO::CPU_SUBTYPE_I386_ALL: 2030 outs() << " cputype CPU_TYPE_I386\n"; 2031 outs() << " cpusubtype CPU_SUBTYPE_I386_ALL\n"; 2032 break; 2033 default: 2034 printUnknownCPUType(cputype, cpusubtype); 2035 break; 2036 } 2037 break; 2038 case MachO::CPU_TYPE_X86_64: 2039 switch (cpusubtype) { 2040 case MachO::CPU_SUBTYPE_X86_64_ALL: 2041 outs() << " cputype CPU_TYPE_X86_64\n"; 2042 outs() << " cpusubtype CPU_SUBTYPE_X86_64_ALL\n"; 2043 break; 2044 case MachO::CPU_SUBTYPE_X86_64_H: 2045 outs() << " cputype CPU_TYPE_X86_64\n"; 2046 outs() << " cpusubtype CPU_SUBTYPE_X86_64_H\n"; 2047 break; 2048 default: 2049 printUnknownCPUType(cputype, cpusubtype); 2050 break; 2051 } 2052 break; 2053 case MachO::CPU_TYPE_ARM: 2054 switch (cpusubtype) { 2055 case MachO::CPU_SUBTYPE_ARM_ALL: 2056 outs() << " cputype CPU_TYPE_ARM\n"; 2057 outs() << " cpusubtype CPU_SUBTYPE_ARM_ALL\n"; 2058 break; 2059 case MachO::CPU_SUBTYPE_ARM_V4T: 2060 outs() << " cputype CPU_TYPE_ARM\n"; 2061 outs() << " cpusubtype CPU_SUBTYPE_ARM_V4T\n"; 2062 break; 2063 case MachO::CPU_SUBTYPE_ARM_V5TEJ: 2064 outs() << " cputype CPU_TYPE_ARM\n"; 2065 outs() << " cpusubtype CPU_SUBTYPE_ARM_V5TEJ\n"; 2066 break; 2067 case MachO::CPU_SUBTYPE_ARM_XSCALE: 2068 outs() << " cputype CPU_TYPE_ARM\n"; 2069 outs() << " cpusubtype CPU_SUBTYPE_ARM_XSCALE\n"; 2070 break; 2071 case MachO::CPU_SUBTYPE_ARM_V6: 2072 outs() << " cputype CPU_TYPE_ARM\n"; 2073 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6\n"; 2074 break; 2075 case MachO::CPU_SUBTYPE_ARM_V6M: 2076 outs() << " cputype CPU_TYPE_ARM\n"; 2077 outs() << " cpusubtype CPU_SUBTYPE_ARM_V6M\n"; 2078 break; 2079 case MachO::CPU_SUBTYPE_ARM_V7: 2080 outs() << " cputype CPU_TYPE_ARM\n"; 2081 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7\n"; 2082 break; 2083 case MachO::CPU_SUBTYPE_ARM_V7EM: 2084 outs() << " cputype CPU_TYPE_ARM\n"; 2085 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7EM\n"; 2086 break; 2087 case MachO::CPU_SUBTYPE_ARM_V7K: 2088 outs() << " cputype CPU_TYPE_ARM\n"; 2089 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7K\n"; 2090 break; 2091 case MachO::CPU_SUBTYPE_ARM_V7M: 2092 outs() << " cputype CPU_TYPE_ARM\n"; 2093 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7M\n"; 2094 break; 2095 case MachO::CPU_SUBTYPE_ARM_V7S: 2096 outs() << " cputype CPU_TYPE_ARM\n"; 2097 outs() << " cpusubtype CPU_SUBTYPE_ARM_V7S\n"; 2098 break; 2099 default: 2100 printUnknownCPUType(cputype, cpusubtype); 2101 break; 2102 } 2103 break; 2104 case MachO::CPU_TYPE_ARM64: 2105 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 2106 case MachO::CPU_SUBTYPE_ARM64_ALL: 2107 outs() << " cputype CPU_TYPE_ARM64\n"; 2108 outs() << " cpusubtype CPU_SUBTYPE_ARM64_ALL\n"; 2109 break; 2110 case MachO::CPU_SUBTYPE_ARM64E: 2111 outs() << " cputype CPU_TYPE_ARM64\n"; 2112 outs() << " cpusubtype CPU_SUBTYPE_ARM64E\n"; 2113 break; 2114 default: 2115 printUnknownCPUType(cputype, cpusubtype); 2116 break; 2117 } 2118 break; 2119 case MachO::CPU_TYPE_ARM64_32: 2120 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 2121 case MachO::CPU_SUBTYPE_ARM64_32_V8: 2122 outs() << " cputype CPU_TYPE_ARM64_32\n"; 2123 outs() << " cpusubtype CPU_SUBTYPE_ARM64_32_V8\n"; 2124 break; 2125 default: 2126 printUnknownCPUType(cputype, cpusubtype); 2127 break; 2128 } 2129 break; 2130 default: 2131 printUnknownCPUType(cputype, cpusubtype); 2132 break; 2133 } 2134 } 2135 2136 static void printMachOUniversalHeaders(const object::MachOUniversalBinary *UB, 2137 bool verbose) { 2138 outs() << "Fat headers\n"; 2139 if (verbose) { 2140 if (UB->getMagic() == MachO::FAT_MAGIC) 2141 outs() << "fat_magic FAT_MAGIC\n"; 2142 else // UB->getMagic() == MachO::FAT_MAGIC_64 2143 outs() << "fat_magic FAT_MAGIC_64\n"; 2144 } else 2145 outs() << "fat_magic " << format("0x%" PRIx32, MachO::FAT_MAGIC) << "\n"; 2146 2147 uint32_t nfat_arch = UB->getNumberOfObjects(); 2148 StringRef Buf = UB->getData(); 2149 uint64_t size = Buf.size(); 2150 uint64_t big_size = sizeof(struct MachO::fat_header) + 2151 nfat_arch * sizeof(struct MachO::fat_arch); 2152 outs() << "nfat_arch " << UB->getNumberOfObjects(); 2153 if (nfat_arch == 0) 2154 outs() << " (malformed, contains zero architecture types)\n"; 2155 else if (big_size > size) 2156 outs() << " (malformed, architectures past end of file)\n"; 2157 else 2158 outs() << "\n"; 2159 2160 for (uint32_t i = 0; i < nfat_arch; ++i) { 2161 MachOUniversalBinary::ObjectForArch OFA(UB, i); 2162 uint32_t cputype = OFA.getCPUType(); 2163 uint32_t cpusubtype = OFA.getCPUSubType(); 2164 outs() << "architecture "; 2165 for (uint32_t j = 0; i != 0 && j <= i - 1; j++) { 2166 MachOUniversalBinary::ObjectForArch other_OFA(UB, j); 2167 uint32_t other_cputype = other_OFA.getCPUType(); 2168 uint32_t other_cpusubtype = other_OFA.getCPUSubType(); 2169 if (cputype != 0 && cpusubtype != 0 && cputype == other_cputype && 2170 (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) == 2171 (other_cpusubtype & ~MachO::CPU_SUBTYPE_MASK)) { 2172 outs() << "(illegal duplicate architecture) "; 2173 break; 2174 } 2175 } 2176 if (verbose) { 2177 outs() << OFA.getArchFlagName() << "\n"; 2178 printCPUType(cputype, cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 2179 } else { 2180 outs() << i << "\n"; 2181 outs() << " cputype " << cputype << "\n"; 2182 outs() << " cpusubtype " << (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) 2183 << "\n"; 2184 } 2185 if (verbose && 2186 (cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) 2187 outs() << " capabilities CPU_SUBTYPE_LIB64\n"; 2188 else 2189 outs() << " capabilities " 2190 << format("0x%" PRIx32, 2191 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24) << "\n"; 2192 outs() << " offset " << OFA.getOffset(); 2193 if (OFA.getOffset() > size) 2194 outs() << " (past end of file)"; 2195 if (OFA.getOffset() % (1ull << OFA.getAlign()) != 0) 2196 outs() << " (not aligned on it's alignment (2^" << OFA.getAlign() << ")"; 2197 outs() << "\n"; 2198 outs() << " size " << OFA.getSize(); 2199 big_size = OFA.getOffset() + OFA.getSize(); 2200 if (big_size > size) 2201 outs() << " (past end of file)"; 2202 outs() << "\n"; 2203 outs() << " align 2^" << OFA.getAlign() << " (" << (1 << OFA.getAlign()) 2204 << ")\n"; 2205 } 2206 } 2207 2208 static void printArchiveChild(StringRef Filename, const Archive::Child &C, 2209 size_t ChildIndex, bool verbose, 2210 bool print_offset, 2211 StringRef ArchitectureName = StringRef()) { 2212 if (print_offset) 2213 outs() << C.getChildOffset() << "\t"; 2214 sys::fs::perms Mode = 2215 unwrapOrError(C.getAccessMode(), getFileNameForError(C, ChildIndex), 2216 Filename, ArchitectureName); 2217 if (verbose) { 2218 // FIXME: this first dash, "-", is for (Mode & S_IFMT) == S_IFREG. 2219 // But there is nothing in sys::fs::perms for S_IFMT or S_IFREG. 2220 outs() << "-"; 2221 outs() << ((Mode & sys::fs::owner_read) ? "r" : "-"); 2222 outs() << ((Mode & sys::fs::owner_write) ? "w" : "-"); 2223 outs() << ((Mode & sys::fs::owner_exe) ? "x" : "-"); 2224 outs() << ((Mode & sys::fs::group_read) ? "r" : "-"); 2225 outs() << ((Mode & sys::fs::group_write) ? "w" : "-"); 2226 outs() << ((Mode & sys::fs::group_exe) ? "x" : "-"); 2227 outs() << ((Mode & sys::fs::others_read) ? "r" : "-"); 2228 outs() << ((Mode & sys::fs::others_write) ? "w" : "-"); 2229 outs() << ((Mode & sys::fs::others_exe) ? "x" : "-"); 2230 } else { 2231 outs() << format("0%o ", Mode); 2232 } 2233 2234 outs() << format("%3d/%-3d %5" PRId64 " ", 2235 unwrapOrError(C.getUID(), getFileNameForError(C, ChildIndex), 2236 Filename, ArchitectureName), 2237 unwrapOrError(C.getGID(), getFileNameForError(C, ChildIndex), 2238 Filename, ArchitectureName), 2239 unwrapOrError(C.getRawSize(), 2240 getFileNameForError(C, ChildIndex), Filename, 2241 ArchitectureName)); 2242 2243 StringRef RawLastModified = C.getRawLastModified(); 2244 if (verbose) { 2245 unsigned Seconds; 2246 if (RawLastModified.getAsInteger(10, Seconds)) 2247 outs() << "(date: \"" << RawLastModified 2248 << "\" contains non-decimal chars) "; 2249 else { 2250 // Since cime(3) returns a 26 character string of the form: 2251 // "Sun Sep 16 01:03:52 1973\n\0" 2252 // just print 24 characters. 2253 time_t t = Seconds; 2254 outs() << format("%.24s ", ctime(&t)); 2255 } 2256 } else { 2257 outs() << RawLastModified << " "; 2258 } 2259 2260 if (verbose) { 2261 Expected<StringRef> NameOrErr = C.getName(); 2262 if (!NameOrErr) { 2263 consumeError(NameOrErr.takeError()); 2264 outs() << unwrapOrError(C.getRawName(), 2265 getFileNameForError(C, ChildIndex), Filename, 2266 ArchitectureName) 2267 << "\n"; 2268 } else { 2269 StringRef Name = NameOrErr.get(); 2270 outs() << Name << "\n"; 2271 } 2272 } else { 2273 outs() << unwrapOrError(C.getRawName(), getFileNameForError(C, ChildIndex), 2274 Filename, ArchitectureName) 2275 << "\n"; 2276 } 2277 } 2278 2279 static void printArchiveHeaders(StringRef Filename, Archive *A, bool verbose, 2280 bool print_offset, 2281 StringRef ArchitectureName = StringRef()) { 2282 Error Err = Error::success(); 2283 size_t I = 0; 2284 for (const auto &C : A->children(Err, false)) 2285 printArchiveChild(Filename, C, I++, verbose, print_offset, 2286 ArchitectureName); 2287 2288 if (Err) 2289 reportError(std::move(Err), Filename, "", ArchitectureName); 2290 } 2291 2292 static bool ValidateArchFlags() { 2293 // Check for -arch all and verifiy the -arch flags are valid. 2294 for (unsigned i = 0; i < ArchFlags.size(); ++i) { 2295 if (ArchFlags[i] == "all") { 2296 ArchAll = true; 2297 } else { 2298 if (!MachOObjectFile::isValidArch(ArchFlags[i])) { 2299 WithColor::error(errs(), "llvm-objdump") 2300 << "unknown architecture named '" + ArchFlags[i] + 2301 "'for the -arch option\n"; 2302 return false; 2303 } 2304 } 2305 } 2306 return true; 2307 } 2308 2309 // ParseInputMachO() parses the named Mach-O file in Filename and handles the 2310 // -arch flags selecting just those slices as specified by them and also parses 2311 // archive files. Then for each individual Mach-O file ProcessMachO() is 2312 // called to process the file based on the command line options. 2313 void objdump::parseInputMachO(StringRef Filename) { 2314 if (!ValidateArchFlags()) 2315 return; 2316 2317 // Attempt to open the binary. 2318 Expected<OwningBinary<Binary>> BinaryOrErr = createBinary(Filename); 2319 if (!BinaryOrErr) { 2320 if (Error E = isNotObjectErrorInvalidFileType(BinaryOrErr.takeError())) 2321 reportError(std::move(E), Filename); 2322 else 2323 outs() << Filename << ": is not an object file\n"; 2324 return; 2325 } 2326 Binary &Bin = *BinaryOrErr.get().getBinary(); 2327 2328 if (Archive *A = dyn_cast<Archive>(&Bin)) { 2329 outs() << "Archive : " << Filename << "\n"; 2330 if (ArchiveHeaders) 2331 printArchiveHeaders(Filename, A, !NonVerbose, ArchiveMemberOffsets); 2332 2333 Error Err = Error::success(); 2334 unsigned I = -1; 2335 for (auto &C : A->children(Err)) { 2336 ++I; 2337 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary(); 2338 if (!ChildOrErr) { 2339 if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError())) 2340 reportError(std::move(E), getFileNameForError(C, I), Filename); 2341 continue; 2342 } 2343 if (MachOObjectFile *O = dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) { 2344 if (!checkMachOAndArchFlags(O, Filename)) 2345 return; 2346 ProcessMachO(Filename, O, O->getFileName()); 2347 } 2348 } 2349 if (Err) 2350 reportError(std::move(Err), Filename); 2351 return; 2352 } 2353 if (MachOUniversalBinary *UB = dyn_cast<MachOUniversalBinary>(&Bin)) { 2354 parseInputMachO(UB); 2355 return; 2356 } 2357 if (ObjectFile *O = dyn_cast<ObjectFile>(&Bin)) { 2358 if (!checkMachOAndArchFlags(O, Filename)) 2359 return; 2360 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&*O)) 2361 ProcessMachO(Filename, MachOOF); 2362 else 2363 WithColor::error(errs(), "llvm-objdump") 2364 << Filename << "': " 2365 << "object is not a Mach-O file type.\n"; 2366 return; 2367 } 2368 llvm_unreachable("Input object can't be invalid at this point"); 2369 } 2370 2371 void objdump::parseInputMachO(MachOUniversalBinary *UB) { 2372 if (!ValidateArchFlags()) 2373 return; 2374 2375 auto Filename = UB->getFileName(); 2376 2377 if (UniversalHeaders) 2378 printMachOUniversalHeaders(UB, !NonVerbose); 2379 2380 // If we have a list of architecture flags specified dump only those. 2381 if (!ArchAll && !ArchFlags.empty()) { 2382 // Look for a slice in the universal binary that matches each ArchFlag. 2383 bool ArchFound; 2384 for (unsigned i = 0; i < ArchFlags.size(); ++i) { 2385 ArchFound = false; 2386 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(), 2387 E = UB->end_objects(); 2388 I != E; ++I) { 2389 if (ArchFlags[i] == I->getArchFlagName()) { 2390 ArchFound = true; 2391 Expected<std::unique_ptr<ObjectFile>> ObjOrErr = 2392 I->getAsObjectFile(); 2393 std::string ArchitectureName = ""; 2394 if (ArchFlags.size() > 1) 2395 ArchitectureName = I->getArchFlagName(); 2396 if (ObjOrErr) { 2397 ObjectFile &O = *ObjOrErr.get(); 2398 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O)) 2399 ProcessMachO(Filename, MachOOF, "", ArchitectureName); 2400 } else if (Error E = isNotObjectErrorInvalidFileType( 2401 ObjOrErr.takeError())) { 2402 reportError(std::move(E), "", Filename, ArchitectureName); 2403 continue; 2404 } else if (Expected<std::unique_ptr<Archive>> AOrErr = 2405 I->getAsArchive()) { 2406 std::unique_ptr<Archive> &A = *AOrErr; 2407 outs() << "Archive : " << Filename; 2408 if (!ArchitectureName.empty()) 2409 outs() << " (architecture " << ArchitectureName << ")"; 2410 outs() << "\n"; 2411 if (ArchiveHeaders) 2412 printArchiveHeaders(Filename, A.get(), !NonVerbose, 2413 ArchiveMemberOffsets, ArchitectureName); 2414 Error Err = Error::success(); 2415 unsigned I = -1; 2416 for (auto &C : A->children(Err)) { 2417 ++I; 2418 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary(); 2419 if (!ChildOrErr) { 2420 if (Error E = 2421 isNotObjectErrorInvalidFileType(ChildOrErr.takeError())) 2422 reportError(std::move(E), getFileNameForError(C, I), Filename, 2423 ArchitectureName); 2424 continue; 2425 } 2426 if (MachOObjectFile *O = 2427 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) 2428 ProcessMachO(Filename, O, O->getFileName(), ArchitectureName); 2429 } 2430 if (Err) 2431 reportError(std::move(Err), Filename); 2432 } else { 2433 consumeError(AOrErr.takeError()); 2434 reportError(Filename, 2435 "Mach-O universal file for architecture " + 2436 StringRef(I->getArchFlagName()) + 2437 " is not a Mach-O file or an archive file"); 2438 } 2439 } 2440 } 2441 if (!ArchFound) { 2442 WithColor::error(errs(), "llvm-objdump") 2443 << "file: " + Filename + " does not contain " 2444 << "architecture: " + ArchFlags[i] + "\n"; 2445 return; 2446 } 2447 } 2448 return; 2449 } 2450 // No architecture flags were specified so if this contains a slice that 2451 // matches the host architecture dump only that. 2452 if (!ArchAll) { 2453 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(), 2454 E = UB->end_objects(); 2455 I != E; ++I) { 2456 if (MachOObjectFile::getHostArch().getArchName() == 2457 I->getArchFlagName()) { 2458 Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile(); 2459 std::string ArchiveName; 2460 ArchiveName.clear(); 2461 if (ObjOrErr) { 2462 ObjectFile &O = *ObjOrErr.get(); 2463 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&O)) 2464 ProcessMachO(Filename, MachOOF); 2465 } else if (Error E = 2466 isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) { 2467 reportError(std::move(E), Filename); 2468 } else if (Expected<std::unique_ptr<Archive>> AOrErr = 2469 I->getAsArchive()) { 2470 std::unique_ptr<Archive> &A = *AOrErr; 2471 outs() << "Archive : " << Filename << "\n"; 2472 if (ArchiveHeaders) 2473 printArchiveHeaders(Filename, A.get(), !NonVerbose, 2474 ArchiveMemberOffsets); 2475 Error Err = Error::success(); 2476 unsigned I = -1; 2477 for (auto &C : A->children(Err)) { 2478 ++I; 2479 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary(); 2480 if (!ChildOrErr) { 2481 if (Error E = 2482 isNotObjectErrorInvalidFileType(ChildOrErr.takeError())) 2483 reportError(std::move(E), getFileNameForError(C, I), Filename); 2484 continue; 2485 } 2486 if (MachOObjectFile *O = 2487 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) 2488 ProcessMachO(Filename, O, O->getFileName()); 2489 } 2490 if (Err) 2491 reportError(std::move(Err), Filename); 2492 } else { 2493 consumeError(AOrErr.takeError()); 2494 reportError(Filename, "Mach-O universal file for architecture " + 2495 StringRef(I->getArchFlagName()) + 2496 " is not a Mach-O file or an archive file"); 2497 } 2498 return; 2499 } 2500 } 2501 } 2502 // Either all architectures have been specified or none have been specified 2503 // and this does not contain the host architecture so dump all the slices. 2504 bool moreThanOneArch = UB->getNumberOfObjects() > 1; 2505 for (MachOUniversalBinary::object_iterator I = UB->begin_objects(), 2506 E = UB->end_objects(); 2507 I != E; ++I) { 2508 Expected<std::unique_ptr<ObjectFile>> ObjOrErr = I->getAsObjectFile(); 2509 std::string ArchitectureName = ""; 2510 if (moreThanOneArch) 2511 ArchitectureName = I->getArchFlagName(); 2512 if (ObjOrErr) { 2513 ObjectFile &Obj = *ObjOrErr.get(); 2514 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(&Obj)) 2515 ProcessMachO(Filename, MachOOF, "", ArchitectureName); 2516 } else if (Error E = 2517 isNotObjectErrorInvalidFileType(ObjOrErr.takeError())) { 2518 reportError(std::move(E), Filename, "", ArchitectureName); 2519 } else if (Expected<std::unique_ptr<Archive>> AOrErr = I->getAsArchive()) { 2520 std::unique_ptr<Archive> &A = *AOrErr; 2521 outs() << "Archive : " << Filename; 2522 if (!ArchitectureName.empty()) 2523 outs() << " (architecture " << ArchitectureName << ")"; 2524 outs() << "\n"; 2525 if (ArchiveHeaders) 2526 printArchiveHeaders(Filename, A.get(), !NonVerbose, 2527 ArchiveMemberOffsets, ArchitectureName); 2528 Error Err = Error::success(); 2529 unsigned I = -1; 2530 for (auto &C : A->children(Err)) { 2531 ++I; 2532 Expected<std::unique_ptr<Binary>> ChildOrErr = C.getAsBinary(); 2533 if (!ChildOrErr) { 2534 if (Error E = isNotObjectErrorInvalidFileType(ChildOrErr.takeError())) 2535 reportError(std::move(E), getFileNameForError(C, I), Filename, 2536 ArchitectureName); 2537 continue; 2538 } 2539 if (MachOObjectFile *O = 2540 dyn_cast<MachOObjectFile>(&*ChildOrErr.get())) { 2541 if (MachOObjectFile *MachOOF = dyn_cast<MachOObjectFile>(O)) 2542 ProcessMachO(Filename, MachOOF, MachOOF->getFileName(), 2543 ArchitectureName); 2544 } 2545 } 2546 if (Err) 2547 reportError(std::move(Err), Filename); 2548 } else { 2549 consumeError(AOrErr.takeError()); 2550 reportError(Filename, "Mach-O universal file for architecture " + 2551 StringRef(I->getArchFlagName()) + 2552 " is not a Mach-O file or an archive file"); 2553 } 2554 } 2555 } 2556 2557 namespace { 2558 // The block of info used by the Symbolizer call backs. 2559 struct DisassembleInfo { 2560 DisassembleInfo(MachOObjectFile *O, SymbolAddressMap *AddrMap, 2561 std::vector<SectionRef> *Sections, bool verbose) 2562 : verbose(verbose), O(O), AddrMap(AddrMap), Sections(Sections) {} 2563 bool verbose; 2564 MachOObjectFile *O; 2565 SectionRef S; 2566 SymbolAddressMap *AddrMap; 2567 std::vector<SectionRef> *Sections; 2568 const char *class_name = nullptr; 2569 const char *selector_name = nullptr; 2570 std::unique_ptr<char[]> method = nullptr; 2571 char *demangled_name = nullptr; 2572 uint64_t adrp_addr = 0; 2573 uint32_t adrp_inst = 0; 2574 std::unique_ptr<SymbolAddressMap> bindtable; 2575 uint32_t depth = 0; 2576 }; 2577 } // namespace 2578 2579 // SymbolizerGetOpInfo() is the operand information call back function. 2580 // This is called to get the symbolic information for operand(s) of an 2581 // instruction when it is being done. This routine does this from 2582 // the relocation information, symbol table, etc. That block of information 2583 // is a pointer to the struct DisassembleInfo that was passed when the 2584 // disassembler context was created and passed to back to here when 2585 // called back by the disassembler for instruction operands that could have 2586 // relocation information. The address of the instruction containing operand is 2587 // at the Pc parameter. The immediate value the operand has is passed in 2588 // op_info->Value and is at Offset past the start of the instruction and has a 2589 // byte Size of 1, 2 or 4. The symbolc information is returned in TagBuf is the 2590 // LLVMOpInfo1 struct defined in the header "llvm-c/Disassembler.h" as symbol 2591 // names and addends of the symbolic expression to add for the operand. The 2592 // value of TagType is currently 1 (for the LLVMOpInfo1 struct). If symbolic 2593 // information is returned then this function returns 1 else it returns 0. 2594 static int SymbolizerGetOpInfo(void *DisInfo, uint64_t Pc, uint64_t Offset, 2595 uint64_t Size, int TagType, void *TagBuf) { 2596 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo; 2597 struct LLVMOpInfo1 *op_info = (struct LLVMOpInfo1 *)TagBuf; 2598 uint64_t value = op_info->Value; 2599 2600 // Make sure all fields returned are zero if we don't set them. 2601 memset((void *)op_info, '\0', sizeof(struct LLVMOpInfo1)); 2602 op_info->Value = value; 2603 2604 // If the TagType is not the value 1 which it code knows about or if no 2605 // verbose symbolic information is wanted then just return 0, indicating no 2606 // information is being returned. 2607 if (TagType != 1 || !info->verbose) 2608 return 0; 2609 2610 unsigned int Arch = info->O->getArch(); 2611 if (Arch == Triple::x86) { 2612 if (Size != 1 && Size != 2 && Size != 4 && Size != 0) 2613 return 0; 2614 if (info->O->getHeader().filetype != MachO::MH_OBJECT) { 2615 // TODO: 2616 // Search the external relocation entries of a fully linked image 2617 // (if any) for an entry that matches this segment offset. 2618 // uint32_t seg_offset = (Pc + Offset); 2619 return 0; 2620 } 2621 // In MH_OBJECT filetypes search the section's relocation entries (if any) 2622 // for an entry for this section offset. 2623 uint32_t sect_addr = info->S.getAddress(); 2624 uint32_t sect_offset = (Pc + Offset) - sect_addr; 2625 bool reloc_found = false; 2626 DataRefImpl Rel; 2627 MachO::any_relocation_info RE; 2628 bool isExtern = false; 2629 SymbolRef Symbol; 2630 bool r_scattered = false; 2631 uint32_t r_value, pair_r_value, r_type; 2632 for (const RelocationRef &Reloc : info->S.relocations()) { 2633 uint64_t RelocOffset = Reloc.getOffset(); 2634 if (RelocOffset == sect_offset) { 2635 Rel = Reloc.getRawDataRefImpl(); 2636 RE = info->O->getRelocation(Rel); 2637 r_type = info->O->getAnyRelocationType(RE); 2638 r_scattered = info->O->isRelocationScattered(RE); 2639 if (r_scattered) { 2640 r_value = info->O->getScatteredRelocationValue(RE); 2641 if (r_type == MachO::GENERIC_RELOC_SECTDIFF || 2642 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF) { 2643 DataRefImpl RelNext = Rel; 2644 info->O->moveRelocationNext(RelNext); 2645 MachO::any_relocation_info RENext; 2646 RENext = info->O->getRelocation(RelNext); 2647 if (info->O->isRelocationScattered(RENext)) 2648 pair_r_value = info->O->getScatteredRelocationValue(RENext); 2649 else 2650 return 0; 2651 } 2652 } else { 2653 isExtern = info->O->getPlainRelocationExternal(RE); 2654 if (isExtern) { 2655 symbol_iterator RelocSym = Reloc.getSymbol(); 2656 Symbol = *RelocSym; 2657 } 2658 } 2659 reloc_found = true; 2660 break; 2661 } 2662 } 2663 if (reloc_found && isExtern) { 2664 op_info->AddSymbol.Present = 1; 2665 op_info->AddSymbol.Name = 2666 unwrapOrError(Symbol.getName(), info->O->getFileName()).data(); 2667 // For i386 extern relocation entries the value in the instruction is 2668 // the offset from the symbol, and value is already set in op_info->Value. 2669 return 1; 2670 } 2671 if (reloc_found && (r_type == MachO::GENERIC_RELOC_SECTDIFF || 2672 r_type == MachO::GENERIC_RELOC_LOCAL_SECTDIFF)) { 2673 const char *add = GuessSymbolName(r_value, info->AddrMap); 2674 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap); 2675 uint32_t offset = value - (r_value - pair_r_value); 2676 op_info->AddSymbol.Present = 1; 2677 if (add != nullptr) 2678 op_info->AddSymbol.Name = add; 2679 else 2680 op_info->AddSymbol.Value = r_value; 2681 op_info->SubtractSymbol.Present = 1; 2682 if (sub != nullptr) 2683 op_info->SubtractSymbol.Name = sub; 2684 else 2685 op_info->SubtractSymbol.Value = pair_r_value; 2686 op_info->Value = offset; 2687 return 1; 2688 } 2689 return 0; 2690 } 2691 if (Arch == Triple::x86_64) { 2692 if (Size != 1 && Size != 2 && Size != 4 && Size != 0) 2693 return 0; 2694 // For non MH_OBJECT types, like MH_KEXT_BUNDLE, Search the external 2695 // relocation entries of a linked image (if any) for an entry that matches 2696 // this segment offset. 2697 if (info->O->getHeader().filetype != MachO::MH_OBJECT) { 2698 uint64_t seg_offset = Pc + Offset; 2699 bool reloc_found = false; 2700 DataRefImpl Rel; 2701 MachO::any_relocation_info RE; 2702 bool isExtern = false; 2703 SymbolRef Symbol; 2704 for (const RelocationRef &Reloc : info->O->external_relocations()) { 2705 uint64_t RelocOffset = Reloc.getOffset(); 2706 if (RelocOffset == seg_offset) { 2707 Rel = Reloc.getRawDataRefImpl(); 2708 RE = info->O->getRelocation(Rel); 2709 // external relocation entries should always be external. 2710 isExtern = info->O->getPlainRelocationExternal(RE); 2711 if (isExtern) { 2712 symbol_iterator RelocSym = Reloc.getSymbol(); 2713 Symbol = *RelocSym; 2714 } 2715 reloc_found = true; 2716 break; 2717 } 2718 } 2719 if (reloc_found && isExtern) { 2720 // The Value passed in will be adjusted by the Pc if the instruction 2721 // adds the Pc. But for x86_64 external relocation entries the Value 2722 // is the offset from the external symbol. 2723 if (info->O->getAnyRelocationPCRel(RE)) 2724 op_info->Value -= Pc + Offset + Size; 2725 const char *name = 2726 unwrapOrError(Symbol.getName(), info->O->getFileName()).data(); 2727 op_info->AddSymbol.Present = 1; 2728 op_info->AddSymbol.Name = name; 2729 return 1; 2730 } 2731 return 0; 2732 } 2733 // In MH_OBJECT filetypes search the section's relocation entries (if any) 2734 // for an entry for this section offset. 2735 uint64_t sect_addr = info->S.getAddress(); 2736 uint64_t sect_offset = (Pc + Offset) - sect_addr; 2737 bool reloc_found = false; 2738 DataRefImpl Rel; 2739 MachO::any_relocation_info RE; 2740 bool isExtern = false; 2741 SymbolRef Symbol; 2742 for (const RelocationRef &Reloc : info->S.relocations()) { 2743 uint64_t RelocOffset = Reloc.getOffset(); 2744 if (RelocOffset == sect_offset) { 2745 Rel = Reloc.getRawDataRefImpl(); 2746 RE = info->O->getRelocation(Rel); 2747 // NOTE: Scattered relocations don't exist on x86_64. 2748 isExtern = info->O->getPlainRelocationExternal(RE); 2749 if (isExtern) { 2750 symbol_iterator RelocSym = Reloc.getSymbol(); 2751 Symbol = *RelocSym; 2752 } 2753 reloc_found = true; 2754 break; 2755 } 2756 } 2757 if (reloc_found && isExtern) { 2758 // The Value passed in will be adjusted by the Pc if the instruction 2759 // adds the Pc. But for x86_64 external relocation entries the Value 2760 // is the offset from the external symbol. 2761 if (info->O->getAnyRelocationPCRel(RE)) 2762 op_info->Value -= Pc + Offset + Size; 2763 const char *name = 2764 unwrapOrError(Symbol.getName(), info->O->getFileName()).data(); 2765 unsigned Type = info->O->getAnyRelocationType(RE); 2766 if (Type == MachO::X86_64_RELOC_SUBTRACTOR) { 2767 DataRefImpl RelNext = Rel; 2768 info->O->moveRelocationNext(RelNext); 2769 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext); 2770 unsigned TypeNext = info->O->getAnyRelocationType(RENext); 2771 bool isExternNext = info->O->getPlainRelocationExternal(RENext); 2772 unsigned SymbolNum = info->O->getPlainRelocationSymbolNum(RENext); 2773 if (TypeNext == MachO::X86_64_RELOC_UNSIGNED && isExternNext) { 2774 op_info->SubtractSymbol.Present = 1; 2775 op_info->SubtractSymbol.Name = name; 2776 symbol_iterator RelocSymNext = info->O->getSymbolByIndex(SymbolNum); 2777 Symbol = *RelocSymNext; 2778 name = unwrapOrError(Symbol.getName(), info->O->getFileName()).data(); 2779 } 2780 } 2781 // TODO: add the VariantKinds to op_info->VariantKind for relocation types 2782 // like: X86_64_RELOC_TLV, X86_64_RELOC_GOT_LOAD and X86_64_RELOC_GOT. 2783 op_info->AddSymbol.Present = 1; 2784 op_info->AddSymbol.Name = name; 2785 return 1; 2786 } 2787 return 0; 2788 } 2789 if (Arch == Triple::arm) { 2790 if (Offset != 0 || (Size != 4 && Size != 2)) 2791 return 0; 2792 if (info->O->getHeader().filetype != MachO::MH_OBJECT) { 2793 // TODO: 2794 // Search the external relocation entries of a fully linked image 2795 // (if any) for an entry that matches this segment offset. 2796 // uint32_t seg_offset = (Pc + Offset); 2797 return 0; 2798 } 2799 // In MH_OBJECT filetypes search the section's relocation entries (if any) 2800 // for an entry for this section offset. 2801 uint32_t sect_addr = info->S.getAddress(); 2802 uint32_t sect_offset = (Pc + Offset) - sect_addr; 2803 DataRefImpl Rel; 2804 MachO::any_relocation_info RE; 2805 bool isExtern = false; 2806 SymbolRef Symbol; 2807 bool r_scattered = false; 2808 uint32_t r_value, pair_r_value, r_type, r_length, other_half; 2809 auto Reloc = 2810 find_if(info->S.relocations(), [&](const RelocationRef &Reloc) { 2811 uint64_t RelocOffset = Reloc.getOffset(); 2812 return RelocOffset == sect_offset; 2813 }); 2814 2815 if (Reloc == info->S.relocations().end()) 2816 return 0; 2817 2818 Rel = Reloc->getRawDataRefImpl(); 2819 RE = info->O->getRelocation(Rel); 2820 r_length = info->O->getAnyRelocationLength(RE); 2821 r_scattered = info->O->isRelocationScattered(RE); 2822 if (r_scattered) { 2823 r_value = info->O->getScatteredRelocationValue(RE); 2824 r_type = info->O->getScatteredRelocationType(RE); 2825 } else { 2826 r_type = info->O->getAnyRelocationType(RE); 2827 isExtern = info->O->getPlainRelocationExternal(RE); 2828 if (isExtern) { 2829 symbol_iterator RelocSym = Reloc->getSymbol(); 2830 Symbol = *RelocSym; 2831 } 2832 } 2833 if (r_type == MachO::ARM_RELOC_HALF || 2834 r_type == MachO::ARM_RELOC_SECTDIFF || 2835 r_type == MachO::ARM_RELOC_LOCAL_SECTDIFF || 2836 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { 2837 DataRefImpl RelNext = Rel; 2838 info->O->moveRelocationNext(RelNext); 2839 MachO::any_relocation_info RENext; 2840 RENext = info->O->getRelocation(RelNext); 2841 other_half = info->O->getAnyRelocationAddress(RENext) & 0xffff; 2842 if (info->O->isRelocationScattered(RENext)) 2843 pair_r_value = info->O->getScatteredRelocationValue(RENext); 2844 } 2845 2846 if (isExtern) { 2847 const char *name = 2848 unwrapOrError(Symbol.getName(), info->O->getFileName()).data(); 2849 op_info->AddSymbol.Present = 1; 2850 op_info->AddSymbol.Name = name; 2851 switch (r_type) { 2852 case MachO::ARM_RELOC_HALF: 2853 if ((r_length & 0x1) == 1) { 2854 op_info->Value = value << 16 | other_half; 2855 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; 2856 } else { 2857 op_info->Value = other_half << 16 | value; 2858 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; 2859 } 2860 break; 2861 default: 2862 break; 2863 } 2864 return 1; 2865 } 2866 // If we have a branch that is not an external relocation entry then 2867 // return 0 so the code in tryAddingSymbolicOperand() can use the 2868 // SymbolLookUp call back with the branch target address to look up the 2869 // symbol and possibility add an annotation for a symbol stub. 2870 if (isExtern == 0 && (r_type == MachO::ARM_RELOC_BR24 || 2871 r_type == MachO::ARM_THUMB_RELOC_BR22)) 2872 return 0; 2873 2874 uint32_t offset = 0; 2875 if (r_type == MachO::ARM_RELOC_HALF || 2876 r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { 2877 if ((r_length & 0x1) == 1) 2878 value = value << 16 | other_half; 2879 else 2880 value = other_half << 16 | value; 2881 } 2882 if (r_scattered && (r_type != MachO::ARM_RELOC_HALF && 2883 r_type != MachO::ARM_RELOC_HALF_SECTDIFF)) { 2884 offset = value - r_value; 2885 value = r_value; 2886 } 2887 2888 if (r_type == MachO::ARM_RELOC_HALF_SECTDIFF) { 2889 if ((r_length & 0x1) == 1) 2890 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; 2891 else 2892 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; 2893 const char *add = GuessSymbolName(r_value, info->AddrMap); 2894 const char *sub = GuessSymbolName(pair_r_value, info->AddrMap); 2895 int32_t offset = value - (r_value - pair_r_value); 2896 op_info->AddSymbol.Present = 1; 2897 if (add != nullptr) 2898 op_info->AddSymbol.Name = add; 2899 else 2900 op_info->AddSymbol.Value = r_value; 2901 op_info->SubtractSymbol.Present = 1; 2902 if (sub != nullptr) 2903 op_info->SubtractSymbol.Name = sub; 2904 else 2905 op_info->SubtractSymbol.Value = pair_r_value; 2906 op_info->Value = offset; 2907 return 1; 2908 } 2909 2910 op_info->AddSymbol.Present = 1; 2911 op_info->Value = offset; 2912 if (r_type == MachO::ARM_RELOC_HALF) { 2913 if ((r_length & 0x1) == 1) 2914 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_HI16; 2915 else 2916 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM_LO16; 2917 } 2918 const char *add = GuessSymbolName(value, info->AddrMap); 2919 if (add != nullptr) { 2920 op_info->AddSymbol.Name = add; 2921 return 1; 2922 } 2923 op_info->AddSymbol.Value = value; 2924 return 1; 2925 } 2926 if (Arch == Triple::aarch64) { 2927 if (Offset != 0 || Size != 4) 2928 return 0; 2929 if (info->O->getHeader().filetype != MachO::MH_OBJECT) { 2930 // TODO: 2931 // Search the external relocation entries of a fully linked image 2932 // (if any) for an entry that matches this segment offset. 2933 // uint64_t seg_offset = (Pc + Offset); 2934 return 0; 2935 } 2936 // In MH_OBJECT filetypes search the section's relocation entries (if any) 2937 // for an entry for this section offset. 2938 uint64_t sect_addr = info->S.getAddress(); 2939 uint64_t sect_offset = (Pc + Offset) - sect_addr; 2940 auto Reloc = 2941 find_if(info->S.relocations(), [&](const RelocationRef &Reloc) { 2942 uint64_t RelocOffset = Reloc.getOffset(); 2943 return RelocOffset == sect_offset; 2944 }); 2945 2946 if (Reloc == info->S.relocations().end()) 2947 return 0; 2948 2949 DataRefImpl Rel = Reloc->getRawDataRefImpl(); 2950 MachO::any_relocation_info RE = info->O->getRelocation(Rel); 2951 uint32_t r_type = info->O->getAnyRelocationType(RE); 2952 if (r_type == MachO::ARM64_RELOC_ADDEND) { 2953 DataRefImpl RelNext = Rel; 2954 info->O->moveRelocationNext(RelNext); 2955 MachO::any_relocation_info RENext = info->O->getRelocation(RelNext); 2956 if (value == 0) { 2957 value = info->O->getPlainRelocationSymbolNum(RENext); 2958 op_info->Value = value; 2959 } 2960 } 2961 // NOTE: Scattered relocations don't exist on arm64. 2962 if (!info->O->getPlainRelocationExternal(RE)) 2963 return 0; 2964 const char *name = 2965 unwrapOrError(Reloc->getSymbol()->getName(), info->O->getFileName()) 2966 .data(); 2967 op_info->AddSymbol.Present = 1; 2968 op_info->AddSymbol.Name = name; 2969 2970 switch (r_type) { 2971 case MachO::ARM64_RELOC_PAGE21: 2972 /* @page */ 2973 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGE; 2974 break; 2975 case MachO::ARM64_RELOC_PAGEOFF12: 2976 /* @pageoff */ 2977 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_PAGEOFF; 2978 break; 2979 case MachO::ARM64_RELOC_GOT_LOAD_PAGE21: 2980 /* @gotpage */ 2981 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGE; 2982 break; 2983 case MachO::ARM64_RELOC_GOT_LOAD_PAGEOFF12: 2984 /* @gotpageoff */ 2985 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_GOTPAGEOFF; 2986 break; 2987 case MachO::ARM64_RELOC_TLVP_LOAD_PAGE21: 2988 /* @tvlppage is not implemented in llvm-mc */ 2989 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVP; 2990 break; 2991 case MachO::ARM64_RELOC_TLVP_LOAD_PAGEOFF12: 2992 /* @tvlppageoff is not implemented in llvm-mc */ 2993 op_info->VariantKind = LLVMDisassembler_VariantKind_ARM64_TLVOFF; 2994 break; 2995 default: 2996 case MachO::ARM64_RELOC_BRANCH26: 2997 op_info->VariantKind = LLVMDisassembler_VariantKind_None; 2998 break; 2999 } 3000 return 1; 3001 } 3002 return 0; 3003 } 3004 3005 // GuessCstringPointer is passed the address of what might be a pointer to a 3006 // literal string in a cstring section. If that address is in a cstring section 3007 // it returns a pointer to that string. Else it returns nullptr. 3008 static const char *GuessCstringPointer(uint64_t ReferenceValue, 3009 struct DisassembleInfo *info) { 3010 for (const auto &Load : info->O->load_commands()) { 3011 if (Load.C.cmd == MachO::LC_SEGMENT_64) { 3012 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load); 3013 for (unsigned J = 0; J < Seg.nsects; ++J) { 3014 MachO::section_64 Sec = info->O->getSection64(Load, J); 3015 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE; 3016 if (section_type == MachO::S_CSTRING_LITERALS && 3017 ReferenceValue >= Sec.addr && 3018 ReferenceValue < Sec.addr + Sec.size) { 3019 uint64_t sect_offset = ReferenceValue - Sec.addr; 3020 uint64_t object_offset = Sec.offset + sect_offset; 3021 StringRef MachOContents = info->O->getData(); 3022 uint64_t object_size = MachOContents.size(); 3023 const char *object_addr = (const char *)MachOContents.data(); 3024 if (object_offset < object_size) { 3025 const char *name = object_addr + object_offset; 3026 return name; 3027 } else { 3028 return nullptr; 3029 } 3030 } 3031 } 3032 } else if (Load.C.cmd == MachO::LC_SEGMENT) { 3033 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load); 3034 for (unsigned J = 0; J < Seg.nsects; ++J) { 3035 MachO::section Sec = info->O->getSection(Load, J); 3036 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE; 3037 if (section_type == MachO::S_CSTRING_LITERALS && 3038 ReferenceValue >= Sec.addr && 3039 ReferenceValue < Sec.addr + Sec.size) { 3040 uint64_t sect_offset = ReferenceValue - Sec.addr; 3041 uint64_t object_offset = Sec.offset + sect_offset; 3042 StringRef MachOContents = info->O->getData(); 3043 uint64_t object_size = MachOContents.size(); 3044 const char *object_addr = (const char *)MachOContents.data(); 3045 if (object_offset < object_size) { 3046 const char *name = object_addr + object_offset; 3047 return name; 3048 } else { 3049 return nullptr; 3050 } 3051 } 3052 } 3053 } 3054 } 3055 return nullptr; 3056 } 3057 3058 // GuessIndirectSymbol returns the name of the indirect symbol for the 3059 // ReferenceValue passed in or nullptr. This is used when ReferenceValue maybe 3060 // an address of a symbol stub or a lazy or non-lazy pointer to associate the 3061 // symbol name being referenced by the stub or pointer. 3062 static const char *GuessIndirectSymbol(uint64_t ReferenceValue, 3063 struct DisassembleInfo *info) { 3064 MachO::dysymtab_command Dysymtab = info->O->getDysymtabLoadCommand(); 3065 MachO::symtab_command Symtab = info->O->getSymtabLoadCommand(); 3066 for (const auto &Load : info->O->load_commands()) { 3067 if (Load.C.cmd == MachO::LC_SEGMENT_64) { 3068 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load); 3069 for (unsigned J = 0; J < Seg.nsects; ++J) { 3070 MachO::section_64 Sec = info->O->getSection64(Load, J); 3071 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE; 3072 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS || 3073 section_type == MachO::S_LAZY_SYMBOL_POINTERS || 3074 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS || 3075 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS || 3076 section_type == MachO::S_SYMBOL_STUBS) && 3077 ReferenceValue >= Sec.addr && 3078 ReferenceValue < Sec.addr + Sec.size) { 3079 uint32_t stride; 3080 if (section_type == MachO::S_SYMBOL_STUBS) 3081 stride = Sec.reserved2; 3082 else 3083 stride = 8; 3084 if (stride == 0) 3085 return nullptr; 3086 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride; 3087 if (index < Dysymtab.nindirectsyms) { 3088 uint32_t indirect_symbol = 3089 info->O->getIndirectSymbolTableEntry(Dysymtab, index); 3090 if (indirect_symbol < Symtab.nsyms) { 3091 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol); 3092 return unwrapOrError(Sym->getName(), info->O->getFileName()) 3093 .data(); 3094 } 3095 } 3096 } 3097 } 3098 } else if (Load.C.cmd == MachO::LC_SEGMENT) { 3099 MachO::segment_command Seg = info->O->getSegmentLoadCommand(Load); 3100 for (unsigned J = 0; J < Seg.nsects; ++J) { 3101 MachO::section Sec = info->O->getSection(Load, J); 3102 uint32_t section_type = Sec.flags & MachO::SECTION_TYPE; 3103 if ((section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS || 3104 section_type == MachO::S_LAZY_SYMBOL_POINTERS || 3105 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS || 3106 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS || 3107 section_type == MachO::S_SYMBOL_STUBS) && 3108 ReferenceValue >= Sec.addr && 3109 ReferenceValue < Sec.addr + Sec.size) { 3110 uint32_t stride; 3111 if (section_type == MachO::S_SYMBOL_STUBS) 3112 stride = Sec.reserved2; 3113 else 3114 stride = 4; 3115 if (stride == 0) 3116 return nullptr; 3117 uint32_t index = Sec.reserved1 + (ReferenceValue - Sec.addr) / stride; 3118 if (index < Dysymtab.nindirectsyms) { 3119 uint32_t indirect_symbol = 3120 info->O->getIndirectSymbolTableEntry(Dysymtab, index); 3121 if (indirect_symbol < Symtab.nsyms) { 3122 symbol_iterator Sym = info->O->getSymbolByIndex(indirect_symbol); 3123 return unwrapOrError(Sym->getName(), info->O->getFileName()) 3124 .data(); 3125 } 3126 } 3127 } 3128 } 3129 } 3130 } 3131 return nullptr; 3132 } 3133 3134 // method_reference() is called passing it the ReferenceName that might be 3135 // a reference it to an Objective-C method call. If so then it allocates and 3136 // assembles a method call string with the values last seen and saved in 3137 // the DisassembleInfo's class_name and selector_name fields. This is saved 3138 // into the method field of the info and any previous string is free'ed. 3139 // Then the class_name field in the info is set to nullptr. The method call 3140 // string is set into ReferenceName and ReferenceType is set to 3141 // LLVMDisassembler_ReferenceType_Out_Objc_Message. If this not a method call 3142 // then both ReferenceType and ReferenceName are left unchanged. 3143 static void method_reference(struct DisassembleInfo *info, 3144 uint64_t *ReferenceType, 3145 const char **ReferenceName) { 3146 unsigned int Arch = info->O->getArch(); 3147 if (*ReferenceName != nullptr) { 3148 if (strcmp(*ReferenceName, "_objc_msgSend") == 0) { 3149 if (info->selector_name != nullptr) { 3150 if (info->class_name != nullptr) { 3151 info->method = std::make_unique<char[]>( 3152 5 + strlen(info->class_name) + strlen(info->selector_name)); 3153 char *method = info->method.get(); 3154 if (method != nullptr) { 3155 strcpy(method, "+["); 3156 strcat(method, info->class_name); 3157 strcat(method, " "); 3158 strcat(method, info->selector_name); 3159 strcat(method, "]"); 3160 *ReferenceName = method; 3161 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message; 3162 } 3163 } else { 3164 info->method = 3165 std::make_unique<char[]>(9 + strlen(info->selector_name)); 3166 char *method = info->method.get(); 3167 if (method != nullptr) { 3168 if (Arch == Triple::x86_64) 3169 strcpy(method, "-[%rdi "); 3170 else if (Arch == Triple::aarch64) 3171 strcpy(method, "-[x0 "); 3172 else 3173 strcpy(method, "-[r? "); 3174 strcat(method, info->selector_name); 3175 strcat(method, "]"); 3176 *ReferenceName = method; 3177 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message; 3178 } 3179 } 3180 info->class_name = nullptr; 3181 } 3182 } else if (strcmp(*ReferenceName, "_objc_msgSendSuper2") == 0) { 3183 if (info->selector_name != nullptr) { 3184 info->method = 3185 std::make_unique<char[]>(17 + strlen(info->selector_name)); 3186 char *method = info->method.get(); 3187 if (method != nullptr) { 3188 if (Arch == Triple::x86_64) 3189 strcpy(method, "-[[%rdi super] "); 3190 else if (Arch == Triple::aarch64) 3191 strcpy(method, "-[[x0 super] "); 3192 else 3193 strcpy(method, "-[[r? super] "); 3194 strcat(method, info->selector_name); 3195 strcat(method, "]"); 3196 *ReferenceName = method; 3197 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message; 3198 } 3199 info->class_name = nullptr; 3200 } 3201 } 3202 } 3203 } 3204 3205 // GuessPointerPointer() is passed the address of what might be a pointer to 3206 // a reference to an Objective-C class, selector, message ref or cfstring. 3207 // If so the value of the pointer is returned and one of the booleans are set 3208 // to true. If not zero is returned and all the booleans are set to false. 3209 static uint64_t GuessPointerPointer(uint64_t ReferenceValue, 3210 struct DisassembleInfo *info, 3211 bool &classref, bool &selref, bool &msgref, 3212 bool &cfstring) { 3213 classref = false; 3214 selref = false; 3215 msgref = false; 3216 cfstring = false; 3217 for (const auto &Load : info->O->load_commands()) { 3218 if (Load.C.cmd == MachO::LC_SEGMENT_64) { 3219 MachO::segment_command_64 Seg = info->O->getSegment64LoadCommand(Load); 3220 for (unsigned J = 0; J < Seg.nsects; ++J) { 3221 MachO::section_64 Sec = info->O->getSection64(Load, J); 3222 if ((strncmp(Sec.sectname, "__objc_selrefs", 16) == 0 || 3223 strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 || 3224 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0 || 3225 strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 || 3226 strncmp(Sec.sectname, "__cfstring", 16) == 0) && 3227 ReferenceValue >= Sec.addr && 3228 ReferenceValue < Sec.addr + Sec.size) { 3229 uint64_t sect_offset = ReferenceValue - Sec.addr; 3230 uint64_t object_offset = Sec.offset + sect_offset; 3231 StringRef MachOContents = info->O->getData(); 3232 uint64_t object_size = MachOContents.size(); 3233 const char *object_addr = (const char *)MachOContents.data(); 3234 if (object_offset < object_size) { 3235 uint64_t pointer_value; 3236 memcpy(&pointer_value, object_addr + object_offset, 3237 sizeof(uint64_t)); 3238 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 3239 sys::swapByteOrder(pointer_value); 3240 if (strncmp(Sec.sectname, "__objc_selrefs", 16) == 0) 3241 selref = true; 3242 else if (strncmp(Sec.sectname, "__objc_classrefs", 16) == 0 || 3243 strncmp(Sec.sectname, "__objc_superrefs", 16) == 0) 3244 classref = true; 3245 else if (strncmp(Sec.sectname, "__objc_msgrefs", 16) == 0 && 3246 ReferenceValue + 8 < Sec.addr + Sec.size) { 3247 msgref = true; 3248 memcpy(&pointer_value, object_addr + object_offset + 8, 3249 sizeof(uint64_t)); 3250 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 3251 sys::swapByteOrder(pointer_value); 3252 } else if (strncmp(Sec.sectname, "__cfstring", 16) == 0) 3253 cfstring = true; 3254 return pointer_value; 3255 } else { 3256 return 0; 3257 } 3258 } 3259 } 3260 } 3261 // TODO: Look for LC_SEGMENT for 32-bit Mach-O files. 3262 } 3263 return 0; 3264 } 3265 3266 // get_pointer_64 returns a pointer to the bytes in the object file at the 3267 // Address from a section in the Mach-O file. And indirectly returns the 3268 // offset into the section, number of bytes left in the section past the offset 3269 // and which section is was being referenced. If the Address is not in a 3270 // section nullptr is returned. 3271 static const char *get_pointer_64(uint64_t Address, uint32_t &offset, 3272 uint32_t &left, SectionRef &S, 3273 DisassembleInfo *info, 3274 bool objc_only = false) { 3275 offset = 0; 3276 left = 0; 3277 S = SectionRef(); 3278 for (unsigned SectIdx = 0; SectIdx != info->Sections->size(); SectIdx++) { 3279 uint64_t SectAddress = ((*(info->Sections))[SectIdx]).getAddress(); 3280 uint64_t SectSize = ((*(info->Sections))[SectIdx]).getSize(); 3281 if (SectSize == 0) 3282 continue; 3283 if (objc_only) { 3284 StringRef SectName; 3285 Expected<StringRef> SecNameOrErr = 3286 ((*(info->Sections))[SectIdx]).getName(); 3287 if (SecNameOrErr) 3288 SectName = *SecNameOrErr; 3289 else 3290 consumeError(SecNameOrErr.takeError()); 3291 3292 DataRefImpl Ref = ((*(info->Sections))[SectIdx]).getRawDataRefImpl(); 3293 StringRef SegName = info->O->getSectionFinalSegmentName(Ref); 3294 if (SegName != "__OBJC" && SectName != "__cstring") 3295 continue; 3296 } 3297 if (Address >= SectAddress && Address < SectAddress + SectSize) { 3298 S = (*(info->Sections))[SectIdx]; 3299 offset = Address - SectAddress; 3300 left = SectSize - offset; 3301 StringRef SectContents = unwrapOrError( 3302 ((*(info->Sections))[SectIdx]).getContents(), info->O->getFileName()); 3303 return SectContents.data() + offset; 3304 } 3305 } 3306 return nullptr; 3307 } 3308 3309 static const char *get_pointer_32(uint32_t Address, uint32_t &offset, 3310 uint32_t &left, SectionRef &S, 3311 DisassembleInfo *info, 3312 bool objc_only = false) { 3313 return get_pointer_64(Address, offset, left, S, info, objc_only); 3314 } 3315 3316 // get_symbol_64() returns the name of a symbol (or nullptr) and the address of 3317 // the symbol indirectly through n_value. Based on the relocation information 3318 // for the specified section offset in the specified section reference. 3319 // If no relocation information is found and a non-zero ReferenceValue for the 3320 // symbol is passed, look up that address in the info's AddrMap. 3321 static const char *get_symbol_64(uint32_t sect_offset, SectionRef S, 3322 DisassembleInfo *info, uint64_t &n_value, 3323 uint64_t ReferenceValue = 0) { 3324 n_value = 0; 3325 if (!info->verbose) 3326 return nullptr; 3327 3328 // See if there is an external relocation entry at the sect_offset. 3329 bool reloc_found = false; 3330 DataRefImpl Rel; 3331 MachO::any_relocation_info RE; 3332 bool isExtern = false; 3333 SymbolRef Symbol; 3334 for (const RelocationRef &Reloc : S.relocations()) { 3335 uint64_t RelocOffset = Reloc.getOffset(); 3336 if (RelocOffset == sect_offset) { 3337 Rel = Reloc.getRawDataRefImpl(); 3338 RE = info->O->getRelocation(Rel); 3339 if (info->O->isRelocationScattered(RE)) 3340 continue; 3341 isExtern = info->O->getPlainRelocationExternal(RE); 3342 if (isExtern) { 3343 symbol_iterator RelocSym = Reloc.getSymbol(); 3344 Symbol = *RelocSym; 3345 } 3346 reloc_found = true; 3347 break; 3348 } 3349 } 3350 // If there is an external relocation entry for a symbol in this section 3351 // at this section_offset then use that symbol's value for the n_value 3352 // and return its name. 3353 const char *SymbolName = nullptr; 3354 if (reloc_found && isExtern) { 3355 n_value = Symbol.getValue(); 3356 StringRef Name = unwrapOrError(Symbol.getName(), info->O->getFileName()); 3357 if (!Name.empty()) { 3358 SymbolName = Name.data(); 3359 return SymbolName; 3360 } 3361 } 3362 3363 // TODO: For fully linked images, look through the external relocation 3364 // entries off the dynamic symtab command. For these the r_offset is from the 3365 // start of the first writeable segment in the Mach-O file. So the offset 3366 // to this section from that segment is passed to this routine by the caller, 3367 // as the database_offset. Which is the difference of the section's starting 3368 // address and the first writable segment. 3369 // 3370 // NOTE: need add passing the database_offset to this routine. 3371 3372 // We did not find an external relocation entry so look up the ReferenceValue 3373 // as an address of a symbol and if found return that symbol's name. 3374 SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap); 3375 3376 return SymbolName; 3377 } 3378 3379 static const char *get_symbol_32(uint32_t sect_offset, SectionRef S, 3380 DisassembleInfo *info, 3381 uint32_t ReferenceValue) { 3382 uint64_t n_value64; 3383 return get_symbol_64(sect_offset, S, info, n_value64, ReferenceValue); 3384 } 3385 3386 namespace { 3387 3388 // These are structs in the Objective-C meta data and read to produce the 3389 // comments for disassembly. While these are part of the ABI they are no 3390 // public defintions. So the are here not in include/llvm/BinaryFormat/MachO.h 3391 // . 3392 3393 // The cfstring object in a 64-bit Mach-O file. 3394 struct cfstring64_t { 3395 uint64_t isa; // class64_t * (64-bit pointer) 3396 uint64_t flags; // flag bits 3397 uint64_t characters; // char * (64-bit pointer) 3398 uint64_t length; // number of non-NULL characters in above 3399 }; 3400 3401 // The class object in a 64-bit Mach-O file. 3402 struct class64_t { 3403 uint64_t isa; // class64_t * (64-bit pointer) 3404 uint64_t superclass; // class64_t * (64-bit pointer) 3405 uint64_t cache; // Cache (64-bit pointer) 3406 uint64_t vtable; // IMP * (64-bit pointer) 3407 uint64_t data; // class_ro64_t * (64-bit pointer) 3408 }; 3409 3410 struct class32_t { 3411 uint32_t isa; /* class32_t * (32-bit pointer) */ 3412 uint32_t superclass; /* class32_t * (32-bit pointer) */ 3413 uint32_t cache; /* Cache (32-bit pointer) */ 3414 uint32_t vtable; /* IMP * (32-bit pointer) */ 3415 uint32_t data; /* class_ro32_t * (32-bit pointer) */ 3416 }; 3417 3418 struct class_ro64_t { 3419 uint32_t flags; 3420 uint32_t instanceStart; 3421 uint32_t instanceSize; 3422 uint32_t reserved; 3423 uint64_t ivarLayout; // const uint8_t * (64-bit pointer) 3424 uint64_t name; // const char * (64-bit pointer) 3425 uint64_t baseMethods; // const method_list_t * (64-bit pointer) 3426 uint64_t baseProtocols; // const protocol_list_t * (64-bit pointer) 3427 uint64_t ivars; // const ivar_list_t * (64-bit pointer) 3428 uint64_t weakIvarLayout; // const uint8_t * (64-bit pointer) 3429 uint64_t baseProperties; // const struct objc_property_list (64-bit pointer) 3430 }; 3431 3432 struct class_ro32_t { 3433 uint32_t flags; 3434 uint32_t instanceStart; 3435 uint32_t instanceSize; 3436 uint32_t ivarLayout; /* const uint8_t * (32-bit pointer) */ 3437 uint32_t name; /* const char * (32-bit pointer) */ 3438 uint32_t baseMethods; /* const method_list_t * (32-bit pointer) */ 3439 uint32_t baseProtocols; /* const protocol_list_t * (32-bit pointer) */ 3440 uint32_t ivars; /* const ivar_list_t * (32-bit pointer) */ 3441 uint32_t weakIvarLayout; /* const uint8_t * (32-bit pointer) */ 3442 uint32_t baseProperties; /* const struct objc_property_list * 3443 (32-bit pointer) */ 3444 }; 3445 3446 /* Values for class_ro{64,32}_t->flags */ 3447 #define RO_META (1 << 0) 3448 #define RO_ROOT (1 << 1) 3449 #define RO_HAS_CXX_STRUCTORS (1 << 2) 3450 3451 struct method_list64_t { 3452 uint32_t entsize; 3453 uint32_t count; 3454 /* struct method64_t first; These structures follow inline */ 3455 }; 3456 3457 struct method_list32_t { 3458 uint32_t entsize; 3459 uint32_t count; 3460 /* struct method32_t first; These structures follow inline */ 3461 }; 3462 3463 struct method64_t { 3464 uint64_t name; /* SEL (64-bit pointer) */ 3465 uint64_t types; /* const char * (64-bit pointer) */ 3466 uint64_t imp; /* IMP (64-bit pointer) */ 3467 }; 3468 3469 struct method32_t { 3470 uint32_t name; /* SEL (32-bit pointer) */ 3471 uint32_t types; /* const char * (32-bit pointer) */ 3472 uint32_t imp; /* IMP (32-bit pointer) */ 3473 }; 3474 3475 struct protocol_list64_t { 3476 uint64_t count; /* uintptr_t (a 64-bit value) */ 3477 /* struct protocol64_t * list[0]; These pointers follow inline */ 3478 }; 3479 3480 struct protocol_list32_t { 3481 uint32_t count; /* uintptr_t (a 32-bit value) */ 3482 /* struct protocol32_t * list[0]; These pointers follow inline */ 3483 }; 3484 3485 struct protocol64_t { 3486 uint64_t isa; /* id * (64-bit pointer) */ 3487 uint64_t name; /* const char * (64-bit pointer) */ 3488 uint64_t protocols; /* struct protocol_list64_t * 3489 (64-bit pointer) */ 3490 uint64_t instanceMethods; /* method_list_t * (64-bit pointer) */ 3491 uint64_t classMethods; /* method_list_t * (64-bit pointer) */ 3492 uint64_t optionalInstanceMethods; /* method_list_t * (64-bit pointer) */ 3493 uint64_t optionalClassMethods; /* method_list_t * (64-bit pointer) */ 3494 uint64_t instanceProperties; /* struct objc_property_list * 3495 (64-bit pointer) */ 3496 }; 3497 3498 struct protocol32_t { 3499 uint32_t isa; /* id * (32-bit pointer) */ 3500 uint32_t name; /* const char * (32-bit pointer) */ 3501 uint32_t protocols; /* struct protocol_list_t * 3502 (32-bit pointer) */ 3503 uint32_t instanceMethods; /* method_list_t * (32-bit pointer) */ 3504 uint32_t classMethods; /* method_list_t * (32-bit pointer) */ 3505 uint32_t optionalInstanceMethods; /* method_list_t * (32-bit pointer) */ 3506 uint32_t optionalClassMethods; /* method_list_t * (32-bit pointer) */ 3507 uint32_t instanceProperties; /* struct objc_property_list * 3508 (32-bit pointer) */ 3509 }; 3510 3511 struct ivar_list64_t { 3512 uint32_t entsize; 3513 uint32_t count; 3514 /* struct ivar64_t first; These structures follow inline */ 3515 }; 3516 3517 struct ivar_list32_t { 3518 uint32_t entsize; 3519 uint32_t count; 3520 /* struct ivar32_t first; These structures follow inline */ 3521 }; 3522 3523 struct ivar64_t { 3524 uint64_t offset; /* uintptr_t * (64-bit pointer) */ 3525 uint64_t name; /* const char * (64-bit pointer) */ 3526 uint64_t type; /* const char * (64-bit pointer) */ 3527 uint32_t alignment; 3528 uint32_t size; 3529 }; 3530 3531 struct ivar32_t { 3532 uint32_t offset; /* uintptr_t * (32-bit pointer) */ 3533 uint32_t name; /* const char * (32-bit pointer) */ 3534 uint32_t type; /* const char * (32-bit pointer) */ 3535 uint32_t alignment; 3536 uint32_t size; 3537 }; 3538 3539 struct objc_property_list64 { 3540 uint32_t entsize; 3541 uint32_t count; 3542 /* struct objc_property64 first; These structures follow inline */ 3543 }; 3544 3545 struct objc_property_list32 { 3546 uint32_t entsize; 3547 uint32_t count; 3548 /* struct objc_property32 first; These structures follow inline */ 3549 }; 3550 3551 struct objc_property64 { 3552 uint64_t name; /* const char * (64-bit pointer) */ 3553 uint64_t attributes; /* const char * (64-bit pointer) */ 3554 }; 3555 3556 struct objc_property32 { 3557 uint32_t name; /* const char * (32-bit pointer) */ 3558 uint32_t attributes; /* const char * (32-bit pointer) */ 3559 }; 3560 3561 struct category64_t { 3562 uint64_t name; /* const char * (64-bit pointer) */ 3563 uint64_t cls; /* struct class_t * (64-bit pointer) */ 3564 uint64_t instanceMethods; /* struct method_list_t * (64-bit pointer) */ 3565 uint64_t classMethods; /* struct method_list_t * (64-bit pointer) */ 3566 uint64_t protocols; /* struct protocol_list_t * (64-bit pointer) */ 3567 uint64_t instanceProperties; /* struct objc_property_list * 3568 (64-bit pointer) */ 3569 }; 3570 3571 struct category32_t { 3572 uint32_t name; /* const char * (32-bit pointer) */ 3573 uint32_t cls; /* struct class_t * (32-bit pointer) */ 3574 uint32_t instanceMethods; /* struct method_list_t * (32-bit pointer) */ 3575 uint32_t classMethods; /* struct method_list_t * (32-bit pointer) */ 3576 uint32_t protocols; /* struct protocol_list_t * (32-bit pointer) */ 3577 uint32_t instanceProperties; /* struct objc_property_list * 3578 (32-bit pointer) */ 3579 }; 3580 3581 struct objc_image_info64 { 3582 uint32_t version; 3583 uint32_t flags; 3584 }; 3585 struct objc_image_info32 { 3586 uint32_t version; 3587 uint32_t flags; 3588 }; 3589 struct imageInfo_t { 3590 uint32_t version; 3591 uint32_t flags; 3592 }; 3593 /* masks for objc_image_info.flags */ 3594 #define OBJC_IMAGE_IS_REPLACEMENT (1 << 0) 3595 #define OBJC_IMAGE_SUPPORTS_GC (1 << 1) 3596 #define OBJC_IMAGE_IS_SIMULATED (1 << 5) 3597 #define OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES (1 << 6) 3598 3599 struct message_ref64 { 3600 uint64_t imp; /* IMP (64-bit pointer) */ 3601 uint64_t sel; /* SEL (64-bit pointer) */ 3602 }; 3603 3604 struct message_ref32 { 3605 uint32_t imp; /* IMP (32-bit pointer) */ 3606 uint32_t sel; /* SEL (32-bit pointer) */ 3607 }; 3608 3609 // Objective-C 1 (32-bit only) meta data structs. 3610 3611 struct objc_module_t { 3612 uint32_t version; 3613 uint32_t size; 3614 uint32_t name; /* char * (32-bit pointer) */ 3615 uint32_t symtab; /* struct objc_symtab * (32-bit pointer) */ 3616 }; 3617 3618 struct objc_symtab_t { 3619 uint32_t sel_ref_cnt; 3620 uint32_t refs; /* SEL * (32-bit pointer) */ 3621 uint16_t cls_def_cnt; 3622 uint16_t cat_def_cnt; 3623 // uint32_t defs[1]; /* void * (32-bit pointer) variable size */ 3624 }; 3625 3626 struct objc_class_t { 3627 uint32_t isa; /* struct objc_class * (32-bit pointer) */ 3628 uint32_t super_class; /* struct objc_class * (32-bit pointer) */ 3629 uint32_t name; /* const char * (32-bit pointer) */ 3630 int32_t version; 3631 int32_t info; 3632 int32_t instance_size; 3633 uint32_t ivars; /* struct objc_ivar_list * (32-bit pointer) */ 3634 uint32_t methodLists; /* struct objc_method_list ** (32-bit pointer) */ 3635 uint32_t cache; /* struct objc_cache * (32-bit pointer) */ 3636 uint32_t protocols; /* struct objc_protocol_list * (32-bit pointer) */ 3637 }; 3638 3639 #define CLS_GETINFO(cls, infomask) ((cls)->info & (infomask)) 3640 // class is not a metaclass 3641 #define CLS_CLASS 0x1 3642 // class is a metaclass 3643 #define CLS_META 0x2 3644 3645 struct objc_category_t { 3646 uint32_t category_name; /* char * (32-bit pointer) */ 3647 uint32_t class_name; /* char * (32-bit pointer) */ 3648 uint32_t instance_methods; /* struct objc_method_list * (32-bit pointer) */ 3649 uint32_t class_methods; /* struct objc_method_list * (32-bit pointer) */ 3650 uint32_t protocols; /* struct objc_protocol_list * (32-bit ptr) */ 3651 }; 3652 3653 struct objc_ivar_t { 3654 uint32_t ivar_name; /* char * (32-bit pointer) */ 3655 uint32_t ivar_type; /* char * (32-bit pointer) */ 3656 int32_t ivar_offset; 3657 }; 3658 3659 struct objc_ivar_list_t { 3660 int32_t ivar_count; 3661 // struct objc_ivar_t ivar_list[1]; /* variable length structure */ 3662 }; 3663 3664 struct objc_method_list_t { 3665 uint32_t obsolete; /* struct objc_method_list * (32-bit pointer) */ 3666 int32_t method_count; 3667 // struct objc_method_t method_list[1]; /* variable length structure */ 3668 }; 3669 3670 struct objc_method_t { 3671 uint32_t method_name; /* SEL, aka struct objc_selector * (32-bit pointer) */ 3672 uint32_t method_types; /* char * (32-bit pointer) */ 3673 uint32_t method_imp; /* IMP, aka function pointer, (*IMP)(id, SEL, ...) 3674 (32-bit pointer) */ 3675 }; 3676 3677 struct objc_protocol_list_t { 3678 uint32_t next; /* struct objc_protocol_list * (32-bit pointer) */ 3679 int32_t count; 3680 // uint32_t list[1]; /* Protocol *, aka struct objc_protocol_t * 3681 // (32-bit pointer) */ 3682 }; 3683 3684 struct objc_protocol_t { 3685 uint32_t isa; /* struct objc_class * (32-bit pointer) */ 3686 uint32_t protocol_name; /* char * (32-bit pointer) */ 3687 uint32_t protocol_list; /* struct objc_protocol_list * (32-bit pointer) */ 3688 uint32_t instance_methods; /* struct objc_method_description_list * 3689 (32-bit pointer) */ 3690 uint32_t class_methods; /* struct objc_method_description_list * 3691 (32-bit pointer) */ 3692 }; 3693 3694 struct objc_method_description_list_t { 3695 int32_t count; 3696 // struct objc_method_description_t list[1]; 3697 }; 3698 3699 struct objc_method_description_t { 3700 uint32_t name; /* SEL, aka struct objc_selector * (32-bit pointer) */ 3701 uint32_t types; /* char * (32-bit pointer) */ 3702 }; 3703 3704 inline void swapStruct(struct cfstring64_t &cfs) { 3705 sys::swapByteOrder(cfs.isa); 3706 sys::swapByteOrder(cfs.flags); 3707 sys::swapByteOrder(cfs.characters); 3708 sys::swapByteOrder(cfs.length); 3709 } 3710 3711 inline void swapStruct(struct class64_t &c) { 3712 sys::swapByteOrder(c.isa); 3713 sys::swapByteOrder(c.superclass); 3714 sys::swapByteOrder(c.cache); 3715 sys::swapByteOrder(c.vtable); 3716 sys::swapByteOrder(c.data); 3717 } 3718 3719 inline void swapStruct(struct class32_t &c) { 3720 sys::swapByteOrder(c.isa); 3721 sys::swapByteOrder(c.superclass); 3722 sys::swapByteOrder(c.cache); 3723 sys::swapByteOrder(c.vtable); 3724 sys::swapByteOrder(c.data); 3725 } 3726 3727 inline void swapStruct(struct class_ro64_t &cro) { 3728 sys::swapByteOrder(cro.flags); 3729 sys::swapByteOrder(cro.instanceStart); 3730 sys::swapByteOrder(cro.instanceSize); 3731 sys::swapByteOrder(cro.reserved); 3732 sys::swapByteOrder(cro.ivarLayout); 3733 sys::swapByteOrder(cro.name); 3734 sys::swapByteOrder(cro.baseMethods); 3735 sys::swapByteOrder(cro.baseProtocols); 3736 sys::swapByteOrder(cro.ivars); 3737 sys::swapByteOrder(cro.weakIvarLayout); 3738 sys::swapByteOrder(cro.baseProperties); 3739 } 3740 3741 inline void swapStruct(struct class_ro32_t &cro) { 3742 sys::swapByteOrder(cro.flags); 3743 sys::swapByteOrder(cro.instanceStart); 3744 sys::swapByteOrder(cro.instanceSize); 3745 sys::swapByteOrder(cro.ivarLayout); 3746 sys::swapByteOrder(cro.name); 3747 sys::swapByteOrder(cro.baseMethods); 3748 sys::swapByteOrder(cro.baseProtocols); 3749 sys::swapByteOrder(cro.ivars); 3750 sys::swapByteOrder(cro.weakIvarLayout); 3751 sys::swapByteOrder(cro.baseProperties); 3752 } 3753 3754 inline void swapStruct(struct method_list64_t &ml) { 3755 sys::swapByteOrder(ml.entsize); 3756 sys::swapByteOrder(ml.count); 3757 } 3758 3759 inline void swapStruct(struct method_list32_t &ml) { 3760 sys::swapByteOrder(ml.entsize); 3761 sys::swapByteOrder(ml.count); 3762 } 3763 3764 inline void swapStruct(struct method64_t &m) { 3765 sys::swapByteOrder(m.name); 3766 sys::swapByteOrder(m.types); 3767 sys::swapByteOrder(m.imp); 3768 } 3769 3770 inline void swapStruct(struct method32_t &m) { 3771 sys::swapByteOrder(m.name); 3772 sys::swapByteOrder(m.types); 3773 sys::swapByteOrder(m.imp); 3774 } 3775 3776 inline void swapStruct(struct protocol_list64_t &pl) { 3777 sys::swapByteOrder(pl.count); 3778 } 3779 3780 inline void swapStruct(struct protocol_list32_t &pl) { 3781 sys::swapByteOrder(pl.count); 3782 } 3783 3784 inline void swapStruct(struct protocol64_t &p) { 3785 sys::swapByteOrder(p.isa); 3786 sys::swapByteOrder(p.name); 3787 sys::swapByteOrder(p.protocols); 3788 sys::swapByteOrder(p.instanceMethods); 3789 sys::swapByteOrder(p.classMethods); 3790 sys::swapByteOrder(p.optionalInstanceMethods); 3791 sys::swapByteOrder(p.optionalClassMethods); 3792 sys::swapByteOrder(p.instanceProperties); 3793 } 3794 3795 inline void swapStruct(struct protocol32_t &p) { 3796 sys::swapByteOrder(p.isa); 3797 sys::swapByteOrder(p.name); 3798 sys::swapByteOrder(p.protocols); 3799 sys::swapByteOrder(p.instanceMethods); 3800 sys::swapByteOrder(p.classMethods); 3801 sys::swapByteOrder(p.optionalInstanceMethods); 3802 sys::swapByteOrder(p.optionalClassMethods); 3803 sys::swapByteOrder(p.instanceProperties); 3804 } 3805 3806 inline void swapStruct(struct ivar_list64_t &il) { 3807 sys::swapByteOrder(il.entsize); 3808 sys::swapByteOrder(il.count); 3809 } 3810 3811 inline void swapStruct(struct ivar_list32_t &il) { 3812 sys::swapByteOrder(il.entsize); 3813 sys::swapByteOrder(il.count); 3814 } 3815 3816 inline void swapStruct(struct ivar64_t &i) { 3817 sys::swapByteOrder(i.offset); 3818 sys::swapByteOrder(i.name); 3819 sys::swapByteOrder(i.type); 3820 sys::swapByteOrder(i.alignment); 3821 sys::swapByteOrder(i.size); 3822 } 3823 3824 inline void swapStruct(struct ivar32_t &i) { 3825 sys::swapByteOrder(i.offset); 3826 sys::swapByteOrder(i.name); 3827 sys::swapByteOrder(i.type); 3828 sys::swapByteOrder(i.alignment); 3829 sys::swapByteOrder(i.size); 3830 } 3831 3832 inline void swapStruct(struct objc_property_list64 &pl) { 3833 sys::swapByteOrder(pl.entsize); 3834 sys::swapByteOrder(pl.count); 3835 } 3836 3837 inline void swapStruct(struct objc_property_list32 &pl) { 3838 sys::swapByteOrder(pl.entsize); 3839 sys::swapByteOrder(pl.count); 3840 } 3841 3842 inline void swapStruct(struct objc_property64 &op) { 3843 sys::swapByteOrder(op.name); 3844 sys::swapByteOrder(op.attributes); 3845 } 3846 3847 inline void swapStruct(struct objc_property32 &op) { 3848 sys::swapByteOrder(op.name); 3849 sys::swapByteOrder(op.attributes); 3850 } 3851 3852 inline void swapStruct(struct category64_t &c) { 3853 sys::swapByteOrder(c.name); 3854 sys::swapByteOrder(c.cls); 3855 sys::swapByteOrder(c.instanceMethods); 3856 sys::swapByteOrder(c.classMethods); 3857 sys::swapByteOrder(c.protocols); 3858 sys::swapByteOrder(c.instanceProperties); 3859 } 3860 3861 inline void swapStruct(struct category32_t &c) { 3862 sys::swapByteOrder(c.name); 3863 sys::swapByteOrder(c.cls); 3864 sys::swapByteOrder(c.instanceMethods); 3865 sys::swapByteOrder(c.classMethods); 3866 sys::swapByteOrder(c.protocols); 3867 sys::swapByteOrder(c.instanceProperties); 3868 } 3869 3870 inline void swapStruct(struct objc_image_info64 &o) { 3871 sys::swapByteOrder(o.version); 3872 sys::swapByteOrder(o.flags); 3873 } 3874 3875 inline void swapStruct(struct objc_image_info32 &o) { 3876 sys::swapByteOrder(o.version); 3877 sys::swapByteOrder(o.flags); 3878 } 3879 3880 inline void swapStruct(struct imageInfo_t &o) { 3881 sys::swapByteOrder(o.version); 3882 sys::swapByteOrder(o.flags); 3883 } 3884 3885 inline void swapStruct(struct message_ref64 &mr) { 3886 sys::swapByteOrder(mr.imp); 3887 sys::swapByteOrder(mr.sel); 3888 } 3889 3890 inline void swapStruct(struct message_ref32 &mr) { 3891 sys::swapByteOrder(mr.imp); 3892 sys::swapByteOrder(mr.sel); 3893 } 3894 3895 inline void swapStruct(struct objc_module_t &module) { 3896 sys::swapByteOrder(module.version); 3897 sys::swapByteOrder(module.size); 3898 sys::swapByteOrder(module.name); 3899 sys::swapByteOrder(module.symtab); 3900 } 3901 3902 inline void swapStruct(struct objc_symtab_t &symtab) { 3903 sys::swapByteOrder(symtab.sel_ref_cnt); 3904 sys::swapByteOrder(symtab.refs); 3905 sys::swapByteOrder(symtab.cls_def_cnt); 3906 sys::swapByteOrder(symtab.cat_def_cnt); 3907 } 3908 3909 inline void swapStruct(struct objc_class_t &objc_class) { 3910 sys::swapByteOrder(objc_class.isa); 3911 sys::swapByteOrder(objc_class.super_class); 3912 sys::swapByteOrder(objc_class.name); 3913 sys::swapByteOrder(objc_class.version); 3914 sys::swapByteOrder(objc_class.info); 3915 sys::swapByteOrder(objc_class.instance_size); 3916 sys::swapByteOrder(objc_class.ivars); 3917 sys::swapByteOrder(objc_class.methodLists); 3918 sys::swapByteOrder(objc_class.cache); 3919 sys::swapByteOrder(objc_class.protocols); 3920 } 3921 3922 inline void swapStruct(struct objc_category_t &objc_category) { 3923 sys::swapByteOrder(objc_category.category_name); 3924 sys::swapByteOrder(objc_category.class_name); 3925 sys::swapByteOrder(objc_category.instance_methods); 3926 sys::swapByteOrder(objc_category.class_methods); 3927 sys::swapByteOrder(objc_category.protocols); 3928 } 3929 3930 inline void swapStruct(struct objc_ivar_list_t &objc_ivar_list) { 3931 sys::swapByteOrder(objc_ivar_list.ivar_count); 3932 } 3933 3934 inline void swapStruct(struct objc_ivar_t &objc_ivar) { 3935 sys::swapByteOrder(objc_ivar.ivar_name); 3936 sys::swapByteOrder(objc_ivar.ivar_type); 3937 sys::swapByteOrder(objc_ivar.ivar_offset); 3938 } 3939 3940 inline void swapStruct(struct objc_method_list_t &method_list) { 3941 sys::swapByteOrder(method_list.obsolete); 3942 sys::swapByteOrder(method_list.method_count); 3943 } 3944 3945 inline void swapStruct(struct objc_method_t &method) { 3946 sys::swapByteOrder(method.method_name); 3947 sys::swapByteOrder(method.method_types); 3948 sys::swapByteOrder(method.method_imp); 3949 } 3950 3951 inline void swapStruct(struct objc_protocol_list_t &protocol_list) { 3952 sys::swapByteOrder(protocol_list.next); 3953 sys::swapByteOrder(protocol_list.count); 3954 } 3955 3956 inline void swapStruct(struct objc_protocol_t &protocol) { 3957 sys::swapByteOrder(protocol.isa); 3958 sys::swapByteOrder(protocol.protocol_name); 3959 sys::swapByteOrder(protocol.protocol_list); 3960 sys::swapByteOrder(protocol.instance_methods); 3961 sys::swapByteOrder(protocol.class_methods); 3962 } 3963 3964 inline void swapStruct(struct objc_method_description_list_t &mdl) { 3965 sys::swapByteOrder(mdl.count); 3966 } 3967 3968 inline void swapStruct(struct objc_method_description_t &md) { 3969 sys::swapByteOrder(md.name); 3970 sys::swapByteOrder(md.types); 3971 } 3972 3973 } // namespace 3974 3975 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue, 3976 struct DisassembleInfo *info); 3977 3978 // get_objc2_64bit_class_name() is used for disassembly and is passed a pointer 3979 // to an Objective-C class and returns the class name. It is also passed the 3980 // address of the pointer, so when the pointer is zero as it can be in an .o 3981 // file, that is used to look for an external relocation entry with a symbol 3982 // name. 3983 static const char *get_objc2_64bit_class_name(uint64_t pointer_value, 3984 uint64_t ReferenceValue, 3985 struct DisassembleInfo *info) { 3986 const char *r; 3987 uint32_t offset, left; 3988 SectionRef S; 3989 3990 // The pointer_value can be 0 in an object file and have a relocation 3991 // entry for the class symbol at the ReferenceValue (the address of the 3992 // pointer). 3993 if (pointer_value == 0) { 3994 r = get_pointer_64(ReferenceValue, offset, left, S, info); 3995 if (r == nullptr || left < sizeof(uint64_t)) 3996 return nullptr; 3997 uint64_t n_value; 3998 const char *symbol_name = get_symbol_64(offset, S, info, n_value); 3999 if (symbol_name == nullptr) 4000 return nullptr; 4001 const char *class_name = strrchr(symbol_name, '$'); 4002 if (class_name != nullptr && class_name[1] == '_' && class_name[2] != '\0') 4003 return class_name + 2; 4004 else 4005 return nullptr; 4006 } 4007 4008 // The case were the pointer_value is non-zero and points to a class defined 4009 // in this Mach-O file. 4010 r = get_pointer_64(pointer_value, offset, left, S, info); 4011 if (r == nullptr || left < sizeof(struct class64_t)) 4012 return nullptr; 4013 struct class64_t c; 4014 memcpy(&c, r, sizeof(struct class64_t)); 4015 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4016 swapStruct(c); 4017 if (c.data == 0) 4018 return nullptr; 4019 r = get_pointer_64(c.data, offset, left, S, info); 4020 if (r == nullptr || left < sizeof(struct class_ro64_t)) 4021 return nullptr; 4022 struct class_ro64_t cro; 4023 memcpy(&cro, r, sizeof(struct class_ro64_t)); 4024 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4025 swapStruct(cro); 4026 if (cro.name == 0) 4027 return nullptr; 4028 const char *name = get_pointer_64(cro.name, offset, left, S, info); 4029 return name; 4030 } 4031 4032 // get_objc2_64bit_cfstring_name is used for disassembly and is passed a 4033 // pointer to a cfstring and returns its name or nullptr. 4034 static const char *get_objc2_64bit_cfstring_name(uint64_t ReferenceValue, 4035 struct DisassembleInfo *info) { 4036 const char *r, *name; 4037 uint32_t offset, left; 4038 SectionRef S; 4039 struct cfstring64_t cfs; 4040 uint64_t cfs_characters; 4041 4042 r = get_pointer_64(ReferenceValue, offset, left, S, info); 4043 if (r == nullptr || left < sizeof(struct cfstring64_t)) 4044 return nullptr; 4045 memcpy(&cfs, r, sizeof(struct cfstring64_t)); 4046 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4047 swapStruct(cfs); 4048 if (cfs.characters == 0) { 4049 uint64_t n_value; 4050 const char *symbol_name = get_symbol_64( 4051 offset + offsetof(struct cfstring64_t, characters), S, info, n_value); 4052 if (symbol_name == nullptr) 4053 return nullptr; 4054 cfs_characters = n_value; 4055 } else 4056 cfs_characters = cfs.characters; 4057 name = get_pointer_64(cfs_characters, offset, left, S, info); 4058 4059 return name; 4060 } 4061 4062 // get_objc2_64bit_selref() is used for disassembly and is passed a the address 4063 // of a pointer to an Objective-C selector reference when the pointer value is 4064 // zero as in a .o file and is likely to have a external relocation entry with 4065 // who's symbol's n_value is the real pointer to the selector name. If that is 4066 // the case the real pointer to the selector name is returned else 0 is 4067 // returned 4068 static uint64_t get_objc2_64bit_selref(uint64_t ReferenceValue, 4069 struct DisassembleInfo *info) { 4070 uint32_t offset, left; 4071 SectionRef S; 4072 4073 const char *r = get_pointer_64(ReferenceValue, offset, left, S, info); 4074 if (r == nullptr || left < sizeof(uint64_t)) 4075 return 0; 4076 uint64_t n_value; 4077 const char *symbol_name = get_symbol_64(offset, S, info, n_value); 4078 if (symbol_name == nullptr) 4079 return 0; 4080 return n_value; 4081 } 4082 4083 static const SectionRef get_section(MachOObjectFile *O, const char *segname, 4084 const char *sectname) { 4085 for (const SectionRef &Section : O->sections()) { 4086 StringRef SectName; 4087 Expected<StringRef> SecNameOrErr = Section.getName(); 4088 if (SecNameOrErr) 4089 SectName = *SecNameOrErr; 4090 else 4091 consumeError(SecNameOrErr.takeError()); 4092 4093 DataRefImpl Ref = Section.getRawDataRefImpl(); 4094 StringRef SegName = O->getSectionFinalSegmentName(Ref); 4095 if (SegName == segname && SectName == sectname) 4096 return Section; 4097 } 4098 return SectionRef(); 4099 } 4100 4101 static void 4102 walk_pointer_list_64(const char *listname, const SectionRef S, 4103 MachOObjectFile *O, struct DisassembleInfo *info, 4104 void (*func)(uint64_t, struct DisassembleInfo *info)) { 4105 if (S == SectionRef()) 4106 return; 4107 4108 StringRef SectName; 4109 Expected<StringRef> SecNameOrErr = S.getName(); 4110 if (SecNameOrErr) 4111 SectName = *SecNameOrErr; 4112 else 4113 consumeError(SecNameOrErr.takeError()); 4114 4115 DataRefImpl Ref = S.getRawDataRefImpl(); 4116 StringRef SegName = O->getSectionFinalSegmentName(Ref); 4117 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 4118 4119 StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName()); 4120 const char *Contents = reinterpret_cast<const char *>(BytesStr.data()); 4121 4122 for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint64_t)) { 4123 uint32_t left = S.getSize() - i; 4124 uint32_t size = left < sizeof(uint64_t) ? left : sizeof(uint64_t); 4125 uint64_t p = 0; 4126 memcpy(&p, Contents + i, size); 4127 if (i + sizeof(uint64_t) > S.getSize()) 4128 outs() << listname << " list pointer extends past end of (" << SegName 4129 << "," << SectName << ") section\n"; 4130 outs() << format("%016" PRIx64, S.getAddress() + i) << " "; 4131 4132 if (O->isLittleEndian() != sys::IsLittleEndianHost) 4133 sys::swapByteOrder(p); 4134 4135 uint64_t n_value = 0; 4136 const char *name = get_symbol_64(i, S, info, n_value, p); 4137 if (name == nullptr) 4138 name = get_dyld_bind_info_symbolname(S.getAddress() + i, info); 4139 4140 if (n_value != 0) { 4141 outs() << format("0x%" PRIx64, n_value); 4142 if (p != 0) 4143 outs() << " + " << format("0x%" PRIx64, p); 4144 } else 4145 outs() << format("0x%" PRIx64, p); 4146 if (name != nullptr) 4147 outs() << " " << name; 4148 outs() << "\n"; 4149 4150 p += n_value; 4151 if (func) 4152 func(p, info); 4153 } 4154 } 4155 4156 static void 4157 walk_pointer_list_32(const char *listname, const SectionRef S, 4158 MachOObjectFile *O, struct DisassembleInfo *info, 4159 void (*func)(uint32_t, struct DisassembleInfo *info)) { 4160 if (S == SectionRef()) 4161 return; 4162 4163 StringRef SectName = unwrapOrError(S.getName(), O->getFileName()); 4164 DataRefImpl Ref = S.getRawDataRefImpl(); 4165 StringRef SegName = O->getSectionFinalSegmentName(Ref); 4166 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 4167 4168 StringRef BytesStr = unwrapOrError(S.getContents(), O->getFileName()); 4169 const char *Contents = reinterpret_cast<const char *>(BytesStr.data()); 4170 4171 for (uint32_t i = 0; i < S.getSize(); i += sizeof(uint32_t)) { 4172 uint32_t left = S.getSize() - i; 4173 uint32_t size = left < sizeof(uint32_t) ? left : sizeof(uint32_t); 4174 uint32_t p = 0; 4175 memcpy(&p, Contents + i, size); 4176 if (i + sizeof(uint32_t) > S.getSize()) 4177 outs() << listname << " list pointer extends past end of (" << SegName 4178 << "," << SectName << ") section\n"; 4179 uint32_t Address = S.getAddress() + i; 4180 outs() << format("%08" PRIx32, Address) << " "; 4181 4182 if (O->isLittleEndian() != sys::IsLittleEndianHost) 4183 sys::swapByteOrder(p); 4184 outs() << format("0x%" PRIx32, p); 4185 4186 const char *name = get_symbol_32(i, S, info, p); 4187 if (name != nullptr) 4188 outs() << " " << name; 4189 outs() << "\n"; 4190 4191 if (func) 4192 func(p, info); 4193 } 4194 } 4195 4196 static void print_layout_map(const char *layout_map, uint32_t left) { 4197 if (layout_map == nullptr) 4198 return; 4199 outs() << " layout map: "; 4200 do { 4201 outs() << format("0x%02" PRIx32, (*layout_map) & 0xff) << " "; 4202 left--; 4203 layout_map++; 4204 } while (*layout_map != '\0' && left != 0); 4205 outs() << "\n"; 4206 } 4207 4208 static void print_layout_map64(uint64_t p, struct DisassembleInfo *info) { 4209 uint32_t offset, left; 4210 SectionRef S; 4211 const char *layout_map; 4212 4213 if (p == 0) 4214 return; 4215 layout_map = get_pointer_64(p, offset, left, S, info); 4216 print_layout_map(layout_map, left); 4217 } 4218 4219 static void print_layout_map32(uint32_t p, struct DisassembleInfo *info) { 4220 uint32_t offset, left; 4221 SectionRef S; 4222 const char *layout_map; 4223 4224 if (p == 0) 4225 return; 4226 layout_map = get_pointer_32(p, offset, left, S, info); 4227 print_layout_map(layout_map, left); 4228 } 4229 4230 static void print_method_list64_t(uint64_t p, struct DisassembleInfo *info, 4231 const char *indent) { 4232 struct method_list64_t ml; 4233 struct method64_t m; 4234 const char *r; 4235 uint32_t offset, xoffset, left, i; 4236 SectionRef S, xS; 4237 const char *name, *sym_name; 4238 uint64_t n_value; 4239 4240 r = get_pointer_64(p, offset, left, S, info); 4241 if (r == nullptr) 4242 return; 4243 memset(&ml, '\0', sizeof(struct method_list64_t)); 4244 if (left < sizeof(struct method_list64_t)) { 4245 memcpy(&ml, r, left); 4246 outs() << " (method_list_t entends past the end of the section)\n"; 4247 } else 4248 memcpy(&ml, r, sizeof(struct method_list64_t)); 4249 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4250 swapStruct(ml); 4251 outs() << indent << "\t\t entsize " << ml.entsize << "\n"; 4252 outs() << indent << "\t\t count " << ml.count << "\n"; 4253 4254 p += sizeof(struct method_list64_t); 4255 offset += sizeof(struct method_list64_t); 4256 for (i = 0; i < ml.count; i++) { 4257 r = get_pointer_64(p, offset, left, S, info); 4258 if (r == nullptr) 4259 return; 4260 memset(&m, '\0', sizeof(struct method64_t)); 4261 if (left < sizeof(struct method64_t)) { 4262 memcpy(&m, r, left); 4263 outs() << indent << " (method_t extends past the end of the section)\n"; 4264 } else 4265 memcpy(&m, r, sizeof(struct method64_t)); 4266 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4267 swapStruct(m); 4268 4269 outs() << indent << "\t\t name "; 4270 sym_name = get_symbol_64(offset + offsetof(struct method64_t, name), S, 4271 info, n_value, m.name); 4272 if (n_value != 0) { 4273 if (info->verbose && sym_name != nullptr) 4274 outs() << sym_name; 4275 else 4276 outs() << format("0x%" PRIx64, n_value); 4277 if (m.name != 0) 4278 outs() << " + " << format("0x%" PRIx64, m.name); 4279 } else 4280 outs() << format("0x%" PRIx64, m.name); 4281 name = get_pointer_64(m.name + n_value, xoffset, left, xS, info); 4282 if (name != nullptr) 4283 outs() << format(" %.*s", left, name); 4284 outs() << "\n"; 4285 4286 outs() << indent << "\t\t types "; 4287 sym_name = get_symbol_64(offset + offsetof(struct method64_t, types), S, 4288 info, n_value, m.types); 4289 if (n_value != 0) { 4290 if (info->verbose && sym_name != nullptr) 4291 outs() << sym_name; 4292 else 4293 outs() << format("0x%" PRIx64, n_value); 4294 if (m.types != 0) 4295 outs() << " + " << format("0x%" PRIx64, m.types); 4296 } else 4297 outs() << format("0x%" PRIx64, m.types); 4298 name = get_pointer_64(m.types + n_value, xoffset, left, xS, info); 4299 if (name != nullptr) 4300 outs() << format(" %.*s", left, name); 4301 outs() << "\n"; 4302 4303 outs() << indent << "\t\t imp "; 4304 name = get_symbol_64(offset + offsetof(struct method64_t, imp), S, info, 4305 n_value, m.imp); 4306 if (info->verbose && name == nullptr) { 4307 if (n_value != 0) { 4308 outs() << format("0x%" PRIx64, n_value) << " "; 4309 if (m.imp != 0) 4310 outs() << "+ " << format("0x%" PRIx64, m.imp) << " "; 4311 } else 4312 outs() << format("0x%" PRIx64, m.imp) << " "; 4313 } 4314 if (name != nullptr) 4315 outs() << name; 4316 outs() << "\n"; 4317 4318 p += sizeof(struct method64_t); 4319 offset += sizeof(struct method64_t); 4320 } 4321 } 4322 4323 static void print_method_list32_t(uint64_t p, struct DisassembleInfo *info, 4324 const char *indent) { 4325 struct method_list32_t ml; 4326 struct method32_t m; 4327 const char *r, *name; 4328 uint32_t offset, xoffset, left, i; 4329 SectionRef S, xS; 4330 4331 r = get_pointer_32(p, offset, left, S, info); 4332 if (r == nullptr) 4333 return; 4334 memset(&ml, '\0', sizeof(struct method_list32_t)); 4335 if (left < sizeof(struct method_list32_t)) { 4336 memcpy(&ml, r, left); 4337 outs() << " (method_list_t entends past the end of the section)\n"; 4338 } else 4339 memcpy(&ml, r, sizeof(struct method_list32_t)); 4340 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4341 swapStruct(ml); 4342 outs() << indent << "\t\t entsize " << ml.entsize << "\n"; 4343 outs() << indent << "\t\t count " << ml.count << "\n"; 4344 4345 p += sizeof(struct method_list32_t); 4346 offset += sizeof(struct method_list32_t); 4347 for (i = 0; i < ml.count; i++) { 4348 r = get_pointer_32(p, offset, left, S, info); 4349 if (r == nullptr) 4350 return; 4351 memset(&m, '\0', sizeof(struct method32_t)); 4352 if (left < sizeof(struct method32_t)) { 4353 memcpy(&ml, r, left); 4354 outs() << indent << " (method_t entends past the end of the section)\n"; 4355 } else 4356 memcpy(&m, r, sizeof(struct method32_t)); 4357 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4358 swapStruct(m); 4359 4360 outs() << indent << "\t\t name " << format("0x%" PRIx32, m.name); 4361 name = get_pointer_32(m.name, xoffset, left, xS, info); 4362 if (name != nullptr) 4363 outs() << format(" %.*s", left, name); 4364 outs() << "\n"; 4365 4366 outs() << indent << "\t\t types " << format("0x%" PRIx32, m.types); 4367 name = get_pointer_32(m.types, xoffset, left, xS, info); 4368 if (name != nullptr) 4369 outs() << format(" %.*s", left, name); 4370 outs() << "\n"; 4371 4372 outs() << indent << "\t\t imp " << format("0x%" PRIx32, m.imp); 4373 name = get_symbol_32(offset + offsetof(struct method32_t, imp), S, info, 4374 m.imp); 4375 if (name != nullptr) 4376 outs() << " " << name; 4377 outs() << "\n"; 4378 4379 p += sizeof(struct method32_t); 4380 offset += sizeof(struct method32_t); 4381 } 4382 } 4383 4384 static bool print_method_list(uint32_t p, struct DisassembleInfo *info) { 4385 uint32_t offset, left, xleft; 4386 SectionRef S; 4387 struct objc_method_list_t method_list; 4388 struct objc_method_t method; 4389 const char *r, *methods, *name, *SymbolName; 4390 int32_t i; 4391 4392 r = get_pointer_32(p, offset, left, S, info, true); 4393 if (r == nullptr) 4394 return true; 4395 4396 outs() << "\n"; 4397 if (left > sizeof(struct objc_method_list_t)) { 4398 memcpy(&method_list, r, sizeof(struct objc_method_list_t)); 4399 } else { 4400 outs() << "\t\t objc_method_list extends past end of the section\n"; 4401 memset(&method_list, '\0', sizeof(struct objc_method_list_t)); 4402 memcpy(&method_list, r, left); 4403 } 4404 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4405 swapStruct(method_list); 4406 4407 outs() << "\t\t obsolete " 4408 << format("0x%08" PRIx32, method_list.obsolete) << "\n"; 4409 outs() << "\t\t method_count " << method_list.method_count << "\n"; 4410 4411 methods = r + sizeof(struct objc_method_list_t); 4412 for (i = 0; i < method_list.method_count; i++) { 4413 if ((i + 1) * sizeof(struct objc_method_t) > left) { 4414 outs() << "\t\t remaining method's extend past the of the section\n"; 4415 break; 4416 } 4417 memcpy(&method, methods + i * sizeof(struct objc_method_t), 4418 sizeof(struct objc_method_t)); 4419 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4420 swapStruct(method); 4421 4422 outs() << "\t\t method_name " 4423 << format("0x%08" PRIx32, method.method_name); 4424 if (info->verbose) { 4425 name = get_pointer_32(method.method_name, offset, xleft, S, info, true); 4426 if (name != nullptr) 4427 outs() << format(" %.*s", xleft, name); 4428 else 4429 outs() << " (not in an __OBJC section)"; 4430 } 4431 outs() << "\n"; 4432 4433 outs() << "\t\t method_types " 4434 << format("0x%08" PRIx32, method.method_types); 4435 if (info->verbose) { 4436 name = get_pointer_32(method.method_types, offset, xleft, S, info, true); 4437 if (name != nullptr) 4438 outs() << format(" %.*s", xleft, name); 4439 else 4440 outs() << " (not in an __OBJC section)"; 4441 } 4442 outs() << "\n"; 4443 4444 outs() << "\t\t method_imp " 4445 << format("0x%08" PRIx32, method.method_imp) << " "; 4446 if (info->verbose) { 4447 SymbolName = GuessSymbolName(method.method_imp, info->AddrMap); 4448 if (SymbolName != nullptr) 4449 outs() << SymbolName; 4450 } 4451 outs() << "\n"; 4452 } 4453 return false; 4454 } 4455 4456 static void print_protocol_list64_t(uint64_t p, struct DisassembleInfo *info) { 4457 struct protocol_list64_t pl; 4458 uint64_t q, n_value; 4459 struct protocol64_t pc; 4460 const char *r; 4461 uint32_t offset, xoffset, left, i; 4462 SectionRef S, xS; 4463 const char *name, *sym_name; 4464 4465 r = get_pointer_64(p, offset, left, S, info); 4466 if (r == nullptr) 4467 return; 4468 memset(&pl, '\0', sizeof(struct protocol_list64_t)); 4469 if (left < sizeof(struct protocol_list64_t)) { 4470 memcpy(&pl, r, left); 4471 outs() << " (protocol_list_t entends past the end of the section)\n"; 4472 } else 4473 memcpy(&pl, r, sizeof(struct protocol_list64_t)); 4474 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4475 swapStruct(pl); 4476 outs() << " count " << pl.count << "\n"; 4477 4478 p += sizeof(struct protocol_list64_t); 4479 offset += sizeof(struct protocol_list64_t); 4480 for (i = 0; i < pl.count; i++) { 4481 r = get_pointer_64(p, offset, left, S, info); 4482 if (r == nullptr) 4483 return; 4484 q = 0; 4485 if (left < sizeof(uint64_t)) { 4486 memcpy(&q, r, left); 4487 outs() << " (protocol_t * entends past the end of the section)\n"; 4488 } else 4489 memcpy(&q, r, sizeof(uint64_t)); 4490 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4491 sys::swapByteOrder(q); 4492 4493 outs() << "\t\t list[" << i << "] "; 4494 sym_name = get_symbol_64(offset, S, info, n_value, q); 4495 if (n_value != 0) { 4496 if (info->verbose && sym_name != nullptr) 4497 outs() << sym_name; 4498 else 4499 outs() << format("0x%" PRIx64, n_value); 4500 if (q != 0) 4501 outs() << " + " << format("0x%" PRIx64, q); 4502 } else 4503 outs() << format("0x%" PRIx64, q); 4504 outs() << " (struct protocol_t *)\n"; 4505 4506 r = get_pointer_64(q + n_value, offset, left, S, info); 4507 if (r == nullptr) 4508 return; 4509 memset(&pc, '\0', sizeof(struct protocol64_t)); 4510 if (left < sizeof(struct protocol64_t)) { 4511 memcpy(&pc, r, left); 4512 outs() << " (protocol_t entends past the end of the section)\n"; 4513 } else 4514 memcpy(&pc, r, sizeof(struct protocol64_t)); 4515 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4516 swapStruct(pc); 4517 4518 outs() << "\t\t\t isa " << format("0x%" PRIx64, pc.isa) << "\n"; 4519 4520 outs() << "\t\t\t name "; 4521 sym_name = get_symbol_64(offset + offsetof(struct protocol64_t, name), S, 4522 info, n_value, pc.name); 4523 if (n_value != 0) { 4524 if (info->verbose && sym_name != nullptr) 4525 outs() << sym_name; 4526 else 4527 outs() << format("0x%" PRIx64, n_value); 4528 if (pc.name != 0) 4529 outs() << " + " << format("0x%" PRIx64, pc.name); 4530 } else 4531 outs() << format("0x%" PRIx64, pc.name); 4532 name = get_pointer_64(pc.name + n_value, xoffset, left, xS, info); 4533 if (name != nullptr) 4534 outs() << format(" %.*s", left, name); 4535 outs() << "\n"; 4536 4537 outs() << "\t\t\tprotocols " << format("0x%" PRIx64, pc.protocols) << "\n"; 4538 4539 outs() << "\t\t instanceMethods "; 4540 sym_name = 4541 get_symbol_64(offset + offsetof(struct protocol64_t, instanceMethods), 4542 S, info, n_value, pc.instanceMethods); 4543 if (n_value != 0) { 4544 if (info->verbose && sym_name != nullptr) 4545 outs() << sym_name; 4546 else 4547 outs() << format("0x%" PRIx64, n_value); 4548 if (pc.instanceMethods != 0) 4549 outs() << " + " << format("0x%" PRIx64, pc.instanceMethods); 4550 } else 4551 outs() << format("0x%" PRIx64, pc.instanceMethods); 4552 outs() << " (struct method_list_t *)\n"; 4553 if (pc.instanceMethods + n_value != 0) 4554 print_method_list64_t(pc.instanceMethods + n_value, info, "\t"); 4555 4556 outs() << "\t\t classMethods "; 4557 sym_name = 4558 get_symbol_64(offset + offsetof(struct protocol64_t, classMethods), S, 4559 info, n_value, pc.classMethods); 4560 if (n_value != 0) { 4561 if (info->verbose && sym_name != nullptr) 4562 outs() << sym_name; 4563 else 4564 outs() << format("0x%" PRIx64, n_value); 4565 if (pc.classMethods != 0) 4566 outs() << " + " << format("0x%" PRIx64, pc.classMethods); 4567 } else 4568 outs() << format("0x%" PRIx64, pc.classMethods); 4569 outs() << " (struct method_list_t *)\n"; 4570 if (pc.classMethods + n_value != 0) 4571 print_method_list64_t(pc.classMethods + n_value, info, "\t"); 4572 4573 outs() << "\t optionalInstanceMethods " 4574 << format("0x%" PRIx64, pc.optionalInstanceMethods) << "\n"; 4575 outs() << "\t optionalClassMethods " 4576 << format("0x%" PRIx64, pc.optionalClassMethods) << "\n"; 4577 outs() << "\t instanceProperties " 4578 << format("0x%" PRIx64, pc.instanceProperties) << "\n"; 4579 4580 p += sizeof(uint64_t); 4581 offset += sizeof(uint64_t); 4582 } 4583 } 4584 4585 static void print_protocol_list32_t(uint32_t p, struct DisassembleInfo *info) { 4586 struct protocol_list32_t pl; 4587 uint32_t q; 4588 struct protocol32_t pc; 4589 const char *r; 4590 uint32_t offset, xoffset, left, i; 4591 SectionRef S, xS; 4592 const char *name; 4593 4594 r = get_pointer_32(p, offset, left, S, info); 4595 if (r == nullptr) 4596 return; 4597 memset(&pl, '\0', sizeof(struct protocol_list32_t)); 4598 if (left < sizeof(struct protocol_list32_t)) { 4599 memcpy(&pl, r, left); 4600 outs() << " (protocol_list_t entends past the end of the section)\n"; 4601 } else 4602 memcpy(&pl, r, sizeof(struct protocol_list32_t)); 4603 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4604 swapStruct(pl); 4605 outs() << " count " << pl.count << "\n"; 4606 4607 p += sizeof(struct protocol_list32_t); 4608 offset += sizeof(struct protocol_list32_t); 4609 for (i = 0; i < pl.count; i++) { 4610 r = get_pointer_32(p, offset, left, S, info); 4611 if (r == nullptr) 4612 return; 4613 q = 0; 4614 if (left < sizeof(uint32_t)) { 4615 memcpy(&q, r, left); 4616 outs() << " (protocol_t * entends past the end of the section)\n"; 4617 } else 4618 memcpy(&q, r, sizeof(uint32_t)); 4619 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4620 sys::swapByteOrder(q); 4621 outs() << "\t\t list[" << i << "] " << format("0x%" PRIx32, q) 4622 << " (struct protocol_t *)\n"; 4623 r = get_pointer_32(q, offset, left, S, info); 4624 if (r == nullptr) 4625 return; 4626 memset(&pc, '\0', sizeof(struct protocol32_t)); 4627 if (left < sizeof(struct protocol32_t)) { 4628 memcpy(&pc, r, left); 4629 outs() << " (protocol_t entends past the end of the section)\n"; 4630 } else 4631 memcpy(&pc, r, sizeof(struct protocol32_t)); 4632 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4633 swapStruct(pc); 4634 outs() << "\t\t\t isa " << format("0x%" PRIx32, pc.isa) << "\n"; 4635 outs() << "\t\t\t name " << format("0x%" PRIx32, pc.name); 4636 name = get_pointer_32(pc.name, xoffset, left, xS, info); 4637 if (name != nullptr) 4638 outs() << format(" %.*s", left, name); 4639 outs() << "\n"; 4640 outs() << "\t\t\tprotocols " << format("0x%" PRIx32, pc.protocols) << "\n"; 4641 outs() << "\t\t instanceMethods " 4642 << format("0x%" PRIx32, pc.instanceMethods) 4643 << " (struct method_list_t *)\n"; 4644 if (pc.instanceMethods != 0) 4645 print_method_list32_t(pc.instanceMethods, info, "\t"); 4646 outs() << "\t\t classMethods " << format("0x%" PRIx32, pc.classMethods) 4647 << " (struct method_list_t *)\n"; 4648 if (pc.classMethods != 0) 4649 print_method_list32_t(pc.classMethods, info, "\t"); 4650 outs() << "\t optionalInstanceMethods " 4651 << format("0x%" PRIx32, pc.optionalInstanceMethods) << "\n"; 4652 outs() << "\t optionalClassMethods " 4653 << format("0x%" PRIx32, pc.optionalClassMethods) << "\n"; 4654 outs() << "\t instanceProperties " 4655 << format("0x%" PRIx32, pc.instanceProperties) << "\n"; 4656 p += sizeof(uint32_t); 4657 offset += sizeof(uint32_t); 4658 } 4659 } 4660 4661 static void print_indent(uint32_t indent) { 4662 for (uint32_t i = 0; i < indent;) { 4663 if (indent - i >= 8) { 4664 outs() << "\t"; 4665 i += 8; 4666 } else { 4667 for (uint32_t j = i; j < indent; j++) 4668 outs() << " "; 4669 return; 4670 } 4671 } 4672 } 4673 4674 static bool print_method_description_list(uint32_t p, uint32_t indent, 4675 struct DisassembleInfo *info) { 4676 uint32_t offset, left, xleft; 4677 SectionRef S; 4678 struct objc_method_description_list_t mdl; 4679 struct objc_method_description_t md; 4680 const char *r, *list, *name; 4681 int32_t i; 4682 4683 r = get_pointer_32(p, offset, left, S, info, true); 4684 if (r == nullptr) 4685 return true; 4686 4687 outs() << "\n"; 4688 if (left > sizeof(struct objc_method_description_list_t)) { 4689 memcpy(&mdl, r, sizeof(struct objc_method_description_list_t)); 4690 } else { 4691 print_indent(indent); 4692 outs() << " objc_method_description_list extends past end of the section\n"; 4693 memset(&mdl, '\0', sizeof(struct objc_method_description_list_t)); 4694 memcpy(&mdl, r, left); 4695 } 4696 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4697 swapStruct(mdl); 4698 4699 print_indent(indent); 4700 outs() << " count " << mdl.count << "\n"; 4701 4702 list = r + sizeof(struct objc_method_description_list_t); 4703 for (i = 0; i < mdl.count; i++) { 4704 if ((i + 1) * sizeof(struct objc_method_description_t) > left) { 4705 print_indent(indent); 4706 outs() << " remaining list entries extend past the of the section\n"; 4707 break; 4708 } 4709 print_indent(indent); 4710 outs() << " list[" << i << "]\n"; 4711 memcpy(&md, list + i * sizeof(struct objc_method_description_t), 4712 sizeof(struct objc_method_description_t)); 4713 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4714 swapStruct(md); 4715 4716 print_indent(indent); 4717 outs() << " name " << format("0x%08" PRIx32, md.name); 4718 if (info->verbose) { 4719 name = get_pointer_32(md.name, offset, xleft, S, info, true); 4720 if (name != nullptr) 4721 outs() << format(" %.*s", xleft, name); 4722 else 4723 outs() << " (not in an __OBJC section)"; 4724 } 4725 outs() << "\n"; 4726 4727 print_indent(indent); 4728 outs() << " types " << format("0x%08" PRIx32, md.types); 4729 if (info->verbose) { 4730 name = get_pointer_32(md.types, offset, xleft, S, info, true); 4731 if (name != nullptr) 4732 outs() << format(" %.*s", xleft, name); 4733 else 4734 outs() << " (not in an __OBJC section)"; 4735 } 4736 outs() << "\n"; 4737 } 4738 return false; 4739 } 4740 4741 static bool print_protocol_list(uint32_t p, uint32_t indent, 4742 struct DisassembleInfo *info); 4743 4744 static bool print_protocol(uint32_t p, uint32_t indent, 4745 struct DisassembleInfo *info) { 4746 uint32_t offset, left; 4747 SectionRef S; 4748 struct objc_protocol_t protocol; 4749 const char *r, *name; 4750 4751 r = get_pointer_32(p, offset, left, S, info, true); 4752 if (r == nullptr) 4753 return true; 4754 4755 outs() << "\n"; 4756 if (left >= sizeof(struct objc_protocol_t)) { 4757 memcpy(&protocol, r, sizeof(struct objc_protocol_t)); 4758 } else { 4759 print_indent(indent); 4760 outs() << " Protocol extends past end of the section\n"; 4761 memset(&protocol, '\0', sizeof(struct objc_protocol_t)); 4762 memcpy(&protocol, r, left); 4763 } 4764 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4765 swapStruct(protocol); 4766 4767 print_indent(indent); 4768 outs() << " isa " << format("0x%08" PRIx32, protocol.isa) 4769 << "\n"; 4770 4771 print_indent(indent); 4772 outs() << " protocol_name " 4773 << format("0x%08" PRIx32, protocol.protocol_name); 4774 if (info->verbose) { 4775 name = get_pointer_32(protocol.protocol_name, offset, left, S, info, true); 4776 if (name != nullptr) 4777 outs() << format(" %.*s", left, name); 4778 else 4779 outs() << " (not in an __OBJC section)"; 4780 } 4781 outs() << "\n"; 4782 4783 print_indent(indent); 4784 outs() << " protocol_list " 4785 << format("0x%08" PRIx32, protocol.protocol_list); 4786 if (print_protocol_list(protocol.protocol_list, indent + 4, info)) 4787 outs() << " (not in an __OBJC section)\n"; 4788 4789 print_indent(indent); 4790 outs() << " instance_methods " 4791 << format("0x%08" PRIx32, protocol.instance_methods); 4792 if (print_method_description_list(protocol.instance_methods, indent, info)) 4793 outs() << " (not in an __OBJC section)\n"; 4794 4795 print_indent(indent); 4796 outs() << " class_methods " 4797 << format("0x%08" PRIx32, protocol.class_methods); 4798 if (print_method_description_list(protocol.class_methods, indent, info)) 4799 outs() << " (not in an __OBJC section)\n"; 4800 4801 return false; 4802 } 4803 4804 static bool print_protocol_list(uint32_t p, uint32_t indent, 4805 struct DisassembleInfo *info) { 4806 uint32_t offset, left, l; 4807 SectionRef S; 4808 struct objc_protocol_list_t protocol_list; 4809 const char *r, *list; 4810 int32_t i; 4811 4812 r = get_pointer_32(p, offset, left, S, info, true); 4813 if (r == nullptr) 4814 return true; 4815 4816 outs() << "\n"; 4817 if (left > sizeof(struct objc_protocol_list_t)) { 4818 memcpy(&protocol_list, r, sizeof(struct objc_protocol_list_t)); 4819 } else { 4820 outs() << "\t\t objc_protocol_list_t extends past end of the section\n"; 4821 memset(&protocol_list, '\0', sizeof(struct objc_protocol_list_t)); 4822 memcpy(&protocol_list, r, left); 4823 } 4824 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4825 swapStruct(protocol_list); 4826 4827 print_indent(indent); 4828 outs() << " next " << format("0x%08" PRIx32, protocol_list.next) 4829 << "\n"; 4830 print_indent(indent); 4831 outs() << " count " << protocol_list.count << "\n"; 4832 4833 list = r + sizeof(struct objc_protocol_list_t); 4834 for (i = 0; i < protocol_list.count; i++) { 4835 if ((i + 1) * sizeof(uint32_t) > left) { 4836 outs() << "\t\t remaining list entries extend past the of the section\n"; 4837 break; 4838 } 4839 memcpy(&l, list + i * sizeof(uint32_t), sizeof(uint32_t)); 4840 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4841 sys::swapByteOrder(l); 4842 4843 print_indent(indent); 4844 outs() << " list[" << i << "] " << format("0x%08" PRIx32, l); 4845 if (print_protocol(l, indent, info)) 4846 outs() << "(not in an __OBJC section)\n"; 4847 } 4848 return false; 4849 } 4850 4851 static void print_ivar_list64_t(uint64_t p, struct DisassembleInfo *info) { 4852 struct ivar_list64_t il; 4853 struct ivar64_t i; 4854 const char *r; 4855 uint32_t offset, xoffset, left, j; 4856 SectionRef S, xS; 4857 const char *name, *sym_name, *ivar_offset_p; 4858 uint64_t ivar_offset, n_value; 4859 4860 r = get_pointer_64(p, offset, left, S, info); 4861 if (r == nullptr) 4862 return; 4863 memset(&il, '\0', sizeof(struct ivar_list64_t)); 4864 if (left < sizeof(struct ivar_list64_t)) { 4865 memcpy(&il, r, left); 4866 outs() << " (ivar_list_t entends past the end of the section)\n"; 4867 } else 4868 memcpy(&il, r, sizeof(struct ivar_list64_t)); 4869 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4870 swapStruct(il); 4871 outs() << " entsize " << il.entsize << "\n"; 4872 outs() << " count " << il.count << "\n"; 4873 4874 p += sizeof(struct ivar_list64_t); 4875 offset += sizeof(struct ivar_list64_t); 4876 for (j = 0; j < il.count; j++) { 4877 r = get_pointer_64(p, offset, left, S, info); 4878 if (r == nullptr) 4879 return; 4880 memset(&i, '\0', sizeof(struct ivar64_t)); 4881 if (left < sizeof(struct ivar64_t)) { 4882 memcpy(&i, r, left); 4883 outs() << " (ivar_t entends past the end of the section)\n"; 4884 } else 4885 memcpy(&i, r, sizeof(struct ivar64_t)); 4886 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4887 swapStruct(i); 4888 4889 outs() << "\t\t\t offset "; 4890 sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, offset), S, 4891 info, n_value, i.offset); 4892 if (n_value != 0) { 4893 if (info->verbose && sym_name != nullptr) 4894 outs() << sym_name; 4895 else 4896 outs() << format("0x%" PRIx64, n_value); 4897 if (i.offset != 0) 4898 outs() << " + " << format("0x%" PRIx64, i.offset); 4899 } else 4900 outs() << format("0x%" PRIx64, i.offset); 4901 ivar_offset_p = get_pointer_64(i.offset + n_value, xoffset, left, xS, info); 4902 if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) { 4903 memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset)); 4904 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4905 sys::swapByteOrder(ivar_offset); 4906 outs() << " " << ivar_offset << "\n"; 4907 } else 4908 outs() << "\n"; 4909 4910 outs() << "\t\t\t name "; 4911 sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, name), S, info, 4912 n_value, i.name); 4913 if (n_value != 0) { 4914 if (info->verbose && sym_name != nullptr) 4915 outs() << sym_name; 4916 else 4917 outs() << format("0x%" PRIx64, n_value); 4918 if (i.name != 0) 4919 outs() << " + " << format("0x%" PRIx64, i.name); 4920 } else 4921 outs() << format("0x%" PRIx64, i.name); 4922 name = get_pointer_64(i.name + n_value, xoffset, left, xS, info); 4923 if (name != nullptr) 4924 outs() << format(" %.*s", left, name); 4925 outs() << "\n"; 4926 4927 outs() << "\t\t\t type "; 4928 sym_name = get_symbol_64(offset + offsetof(struct ivar64_t, type), S, info, 4929 n_value, i.name); 4930 name = get_pointer_64(i.type + n_value, xoffset, left, xS, info); 4931 if (n_value != 0) { 4932 if (info->verbose && sym_name != nullptr) 4933 outs() << sym_name; 4934 else 4935 outs() << format("0x%" PRIx64, n_value); 4936 if (i.type != 0) 4937 outs() << " + " << format("0x%" PRIx64, i.type); 4938 } else 4939 outs() << format("0x%" PRIx64, i.type); 4940 if (name != nullptr) 4941 outs() << format(" %.*s", left, name); 4942 outs() << "\n"; 4943 4944 outs() << "\t\t\talignment " << i.alignment << "\n"; 4945 outs() << "\t\t\t size " << i.size << "\n"; 4946 4947 p += sizeof(struct ivar64_t); 4948 offset += sizeof(struct ivar64_t); 4949 } 4950 } 4951 4952 static void print_ivar_list32_t(uint32_t p, struct DisassembleInfo *info) { 4953 struct ivar_list32_t il; 4954 struct ivar32_t i; 4955 const char *r; 4956 uint32_t offset, xoffset, left, j; 4957 SectionRef S, xS; 4958 const char *name, *ivar_offset_p; 4959 uint32_t ivar_offset; 4960 4961 r = get_pointer_32(p, offset, left, S, info); 4962 if (r == nullptr) 4963 return; 4964 memset(&il, '\0', sizeof(struct ivar_list32_t)); 4965 if (left < sizeof(struct ivar_list32_t)) { 4966 memcpy(&il, r, left); 4967 outs() << " (ivar_list_t entends past the end of the section)\n"; 4968 } else 4969 memcpy(&il, r, sizeof(struct ivar_list32_t)); 4970 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4971 swapStruct(il); 4972 outs() << " entsize " << il.entsize << "\n"; 4973 outs() << " count " << il.count << "\n"; 4974 4975 p += sizeof(struct ivar_list32_t); 4976 offset += sizeof(struct ivar_list32_t); 4977 for (j = 0; j < il.count; j++) { 4978 r = get_pointer_32(p, offset, left, S, info); 4979 if (r == nullptr) 4980 return; 4981 memset(&i, '\0', sizeof(struct ivar32_t)); 4982 if (left < sizeof(struct ivar32_t)) { 4983 memcpy(&i, r, left); 4984 outs() << " (ivar_t entends past the end of the section)\n"; 4985 } else 4986 memcpy(&i, r, sizeof(struct ivar32_t)); 4987 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4988 swapStruct(i); 4989 4990 outs() << "\t\t\t offset " << format("0x%" PRIx32, i.offset); 4991 ivar_offset_p = get_pointer_32(i.offset, xoffset, left, xS, info); 4992 if (ivar_offset_p != nullptr && left >= sizeof(*ivar_offset_p)) { 4993 memcpy(&ivar_offset, ivar_offset_p, sizeof(ivar_offset)); 4994 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 4995 sys::swapByteOrder(ivar_offset); 4996 outs() << " " << ivar_offset << "\n"; 4997 } else 4998 outs() << "\n"; 4999 5000 outs() << "\t\t\t name " << format("0x%" PRIx32, i.name); 5001 name = get_pointer_32(i.name, xoffset, left, xS, info); 5002 if (name != nullptr) 5003 outs() << format(" %.*s", left, name); 5004 outs() << "\n"; 5005 5006 outs() << "\t\t\t type " << format("0x%" PRIx32, i.type); 5007 name = get_pointer_32(i.type, xoffset, left, xS, info); 5008 if (name != nullptr) 5009 outs() << format(" %.*s", left, name); 5010 outs() << "\n"; 5011 5012 outs() << "\t\t\talignment " << i.alignment << "\n"; 5013 outs() << "\t\t\t size " << i.size << "\n"; 5014 5015 p += sizeof(struct ivar32_t); 5016 offset += sizeof(struct ivar32_t); 5017 } 5018 } 5019 5020 static void print_objc_property_list64(uint64_t p, 5021 struct DisassembleInfo *info) { 5022 struct objc_property_list64 opl; 5023 struct objc_property64 op; 5024 const char *r; 5025 uint32_t offset, xoffset, left, j; 5026 SectionRef S, xS; 5027 const char *name, *sym_name; 5028 uint64_t n_value; 5029 5030 r = get_pointer_64(p, offset, left, S, info); 5031 if (r == nullptr) 5032 return; 5033 memset(&opl, '\0', sizeof(struct objc_property_list64)); 5034 if (left < sizeof(struct objc_property_list64)) { 5035 memcpy(&opl, r, left); 5036 outs() << " (objc_property_list entends past the end of the section)\n"; 5037 } else 5038 memcpy(&opl, r, sizeof(struct objc_property_list64)); 5039 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5040 swapStruct(opl); 5041 outs() << " entsize " << opl.entsize << "\n"; 5042 outs() << " count " << opl.count << "\n"; 5043 5044 p += sizeof(struct objc_property_list64); 5045 offset += sizeof(struct objc_property_list64); 5046 for (j = 0; j < opl.count; j++) { 5047 r = get_pointer_64(p, offset, left, S, info); 5048 if (r == nullptr) 5049 return; 5050 memset(&op, '\0', sizeof(struct objc_property64)); 5051 if (left < sizeof(struct objc_property64)) { 5052 memcpy(&op, r, left); 5053 outs() << " (objc_property entends past the end of the section)\n"; 5054 } else 5055 memcpy(&op, r, sizeof(struct objc_property64)); 5056 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5057 swapStruct(op); 5058 5059 outs() << "\t\t\t name "; 5060 sym_name = get_symbol_64(offset + offsetof(struct objc_property64, name), S, 5061 info, n_value, op.name); 5062 if (n_value != 0) { 5063 if (info->verbose && sym_name != nullptr) 5064 outs() << sym_name; 5065 else 5066 outs() << format("0x%" PRIx64, n_value); 5067 if (op.name != 0) 5068 outs() << " + " << format("0x%" PRIx64, op.name); 5069 } else 5070 outs() << format("0x%" PRIx64, op.name); 5071 name = get_pointer_64(op.name + n_value, xoffset, left, xS, info); 5072 if (name != nullptr) 5073 outs() << format(" %.*s", left, name); 5074 outs() << "\n"; 5075 5076 outs() << "\t\t\tattributes "; 5077 sym_name = 5078 get_symbol_64(offset + offsetof(struct objc_property64, attributes), S, 5079 info, n_value, op.attributes); 5080 if (n_value != 0) { 5081 if (info->verbose && sym_name != nullptr) 5082 outs() << sym_name; 5083 else 5084 outs() << format("0x%" PRIx64, n_value); 5085 if (op.attributes != 0) 5086 outs() << " + " << format("0x%" PRIx64, op.attributes); 5087 } else 5088 outs() << format("0x%" PRIx64, op.attributes); 5089 name = get_pointer_64(op.attributes + n_value, xoffset, left, xS, info); 5090 if (name != nullptr) 5091 outs() << format(" %.*s", left, name); 5092 outs() << "\n"; 5093 5094 p += sizeof(struct objc_property64); 5095 offset += sizeof(struct objc_property64); 5096 } 5097 } 5098 5099 static void print_objc_property_list32(uint32_t p, 5100 struct DisassembleInfo *info) { 5101 struct objc_property_list32 opl; 5102 struct objc_property32 op; 5103 const char *r; 5104 uint32_t offset, xoffset, left, j; 5105 SectionRef S, xS; 5106 const char *name; 5107 5108 r = get_pointer_32(p, offset, left, S, info); 5109 if (r == nullptr) 5110 return; 5111 memset(&opl, '\0', sizeof(struct objc_property_list32)); 5112 if (left < sizeof(struct objc_property_list32)) { 5113 memcpy(&opl, r, left); 5114 outs() << " (objc_property_list entends past the end of the section)\n"; 5115 } else 5116 memcpy(&opl, r, sizeof(struct objc_property_list32)); 5117 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5118 swapStruct(opl); 5119 outs() << " entsize " << opl.entsize << "\n"; 5120 outs() << " count " << opl.count << "\n"; 5121 5122 p += sizeof(struct objc_property_list32); 5123 offset += sizeof(struct objc_property_list32); 5124 for (j = 0; j < opl.count; j++) { 5125 r = get_pointer_32(p, offset, left, S, info); 5126 if (r == nullptr) 5127 return; 5128 memset(&op, '\0', sizeof(struct objc_property32)); 5129 if (left < sizeof(struct objc_property32)) { 5130 memcpy(&op, r, left); 5131 outs() << " (objc_property entends past the end of the section)\n"; 5132 } else 5133 memcpy(&op, r, sizeof(struct objc_property32)); 5134 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5135 swapStruct(op); 5136 5137 outs() << "\t\t\t name " << format("0x%" PRIx32, op.name); 5138 name = get_pointer_32(op.name, xoffset, left, xS, info); 5139 if (name != nullptr) 5140 outs() << format(" %.*s", left, name); 5141 outs() << "\n"; 5142 5143 outs() << "\t\t\tattributes " << format("0x%" PRIx32, op.attributes); 5144 name = get_pointer_32(op.attributes, xoffset, left, xS, info); 5145 if (name != nullptr) 5146 outs() << format(" %.*s", left, name); 5147 outs() << "\n"; 5148 5149 p += sizeof(struct objc_property32); 5150 offset += sizeof(struct objc_property32); 5151 } 5152 } 5153 5154 static bool print_class_ro64_t(uint64_t p, struct DisassembleInfo *info, 5155 bool &is_meta_class) { 5156 struct class_ro64_t cro; 5157 const char *r; 5158 uint32_t offset, xoffset, left; 5159 SectionRef S, xS; 5160 const char *name, *sym_name; 5161 uint64_t n_value; 5162 5163 r = get_pointer_64(p, offset, left, S, info); 5164 if (r == nullptr || left < sizeof(struct class_ro64_t)) 5165 return false; 5166 memcpy(&cro, r, sizeof(struct class_ro64_t)); 5167 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5168 swapStruct(cro); 5169 outs() << " flags " << format("0x%" PRIx32, cro.flags); 5170 if (cro.flags & RO_META) 5171 outs() << " RO_META"; 5172 if (cro.flags & RO_ROOT) 5173 outs() << " RO_ROOT"; 5174 if (cro.flags & RO_HAS_CXX_STRUCTORS) 5175 outs() << " RO_HAS_CXX_STRUCTORS"; 5176 outs() << "\n"; 5177 outs() << " instanceStart " << cro.instanceStart << "\n"; 5178 outs() << " instanceSize " << cro.instanceSize << "\n"; 5179 outs() << " reserved " << format("0x%" PRIx32, cro.reserved) 5180 << "\n"; 5181 outs() << " ivarLayout " << format("0x%" PRIx64, cro.ivarLayout) 5182 << "\n"; 5183 print_layout_map64(cro.ivarLayout, info); 5184 5185 outs() << " name "; 5186 sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, name), S, 5187 info, n_value, cro.name); 5188 if (n_value != 0) { 5189 if (info->verbose && sym_name != nullptr) 5190 outs() << sym_name; 5191 else 5192 outs() << format("0x%" PRIx64, n_value); 5193 if (cro.name != 0) 5194 outs() << " + " << format("0x%" PRIx64, cro.name); 5195 } else 5196 outs() << format("0x%" PRIx64, cro.name); 5197 name = get_pointer_64(cro.name + n_value, xoffset, left, xS, info); 5198 if (name != nullptr) 5199 outs() << format(" %.*s", left, name); 5200 outs() << "\n"; 5201 5202 outs() << " baseMethods "; 5203 sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, baseMethods), 5204 S, info, n_value, cro.baseMethods); 5205 if (n_value != 0) { 5206 if (info->verbose && sym_name != nullptr) 5207 outs() << sym_name; 5208 else 5209 outs() << format("0x%" PRIx64, n_value); 5210 if (cro.baseMethods != 0) 5211 outs() << " + " << format("0x%" PRIx64, cro.baseMethods); 5212 } else 5213 outs() << format("0x%" PRIx64, cro.baseMethods); 5214 outs() << " (struct method_list_t *)\n"; 5215 if (cro.baseMethods + n_value != 0) 5216 print_method_list64_t(cro.baseMethods + n_value, info, ""); 5217 5218 outs() << " baseProtocols "; 5219 sym_name = 5220 get_symbol_64(offset + offsetof(struct class_ro64_t, baseProtocols), S, 5221 info, n_value, cro.baseProtocols); 5222 if (n_value != 0) { 5223 if (info->verbose && sym_name != nullptr) 5224 outs() << sym_name; 5225 else 5226 outs() << format("0x%" PRIx64, n_value); 5227 if (cro.baseProtocols != 0) 5228 outs() << " + " << format("0x%" PRIx64, cro.baseProtocols); 5229 } else 5230 outs() << format("0x%" PRIx64, cro.baseProtocols); 5231 outs() << "\n"; 5232 if (cro.baseProtocols + n_value != 0) 5233 print_protocol_list64_t(cro.baseProtocols + n_value, info); 5234 5235 outs() << " ivars "; 5236 sym_name = get_symbol_64(offset + offsetof(struct class_ro64_t, ivars), S, 5237 info, n_value, cro.ivars); 5238 if (n_value != 0) { 5239 if (info->verbose && sym_name != nullptr) 5240 outs() << sym_name; 5241 else 5242 outs() << format("0x%" PRIx64, n_value); 5243 if (cro.ivars != 0) 5244 outs() << " + " << format("0x%" PRIx64, cro.ivars); 5245 } else 5246 outs() << format("0x%" PRIx64, cro.ivars); 5247 outs() << "\n"; 5248 if (cro.ivars + n_value != 0) 5249 print_ivar_list64_t(cro.ivars + n_value, info); 5250 5251 outs() << " weakIvarLayout "; 5252 sym_name = 5253 get_symbol_64(offset + offsetof(struct class_ro64_t, weakIvarLayout), S, 5254 info, n_value, cro.weakIvarLayout); 5255 if (n_value != 0) { 5256 if (info->verbose && sym_name != nullptr) 5257 outs() << sym_name; 5258 else 5259 outs() << format("0x%" PRIx64, n_value); 5260 if (cro.weakIvarLayout != 0) 5261 outs() << " + " << format("0x%" PRIx64, cro.weakIvarLayout); 5262 } else 5263 outs() << format("0x%" PRIx64, cro.weakIvarLayout); 5264 outs() << "\n"; 5265 print_layout_map64(cro.weakIvarLayout + n_value, info); 5266 5267 outs() << " baseProperties "; 5268 sym_name = 5269 get_symbol_64(offset + offsetof(struct class_ro64_t, baseProperties), S, 5270 info, n_value, cro.baseProperties); 5271 if (n_value != 0) { 5272 if (info->verbose && sym_name != nullptr) 5273 outs() << sym_name; 5274 else 5275 outs() << format("0x%" PRIx64, n_value); 5276 if (cro.baseProperties != 0) 5277 outs() << " + " << format("0x%" PRIx64, cro.baseProperties); 5278 } else 5279 outs() << format("0x%" PRIx64, cro.baseProperties); 5280 outs() << "\n"; 5281 if (cro.baseProperties + n_value != 0) 5282 print_objc_property_list64(cro.baseProperties + n_value, info); 5283 5284 is_meta_class = (cro.flags & RO_META) != 0; 5285 return true; 5286 } 5287 5288 static bool print_class_ro32_t(uint32_t p, struct DisassembleInfo *info, 5289 bool &is_meta_class) { 5290 struct class_ro32_t cro; 5291 const char *r; 5292 uint32_t offset, xoffset, left; 5293 SectionRef S, xS; 5294 const char *name; 5295 5296 r = get_pointer_32(p, offset, left, S, info); 5297 if (r == nullptr) 5298 return false; 5299 memset(&cro, '\0', sizeof(struct class_ro32_t)); 5300 if (left < sizeof(struct class_ro32_t)) { 5301 memcpy(&cro, r, left); 5302 outs() << " (class_ro_t entends past the end of the section)\n"; 5303 } else 5304 memcpy(&cro, r, sizeof(struct class_ro32_t)); 5305 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5306 swapStruct(cro); 5307 outs() << " flags " << format("0x%" PRIx32, cro.flags); 5308 if (cro.flags & RO_META) 5309 outs() << " RO_META"; 5310 if (cro.flags & RO_ROOT) 5311 outs() << " RO_ROOT"; 5312 if (cro.flags & RO_HAS_CXX_STRUCTORS) 5313 outs() << " RO_HAS_CXX_STRUCTORS"; 5314 outs() << "\n"; 5315 outs() << " instanceStart " << cro.instanceStart << "\n"; 5316 outs() << " instanceSize " << cro.instanceSize << "\n"; 5317 outs() << " ivarLayout " << format("0x%" PRIx32, cro.ivarLayout) 5318 << "\n"; 5319 print_layout_map32(cro.ivarLayout, info); 5320 5321 outs() << " name " << format("0x%" PRIx32, cro.name); 5322 name = get_pointer_32(cro.name, xoffset, left, xS, info); 5323 if (name != nullptr) 5324 outs() << format(" %.*s", left, name); 5325 outs() << "\n"; 5326 5327 outs() << " baseMethods " 5328 << format("0x%" PRIx32, cro.baseMethods) 5329 << " (struct method_list_t *)\n"; 5330 if (cro.baseMethods != 0) 5331 print_method_list32_t(cro.baseMethods, info, ""); 5332 5333 outs() << " baseProtocols " 5334 << format("0x%" PRIx32, cro.baseProtocols) << "\n"; 5335 if (cro.baseProtocols != 0) 5336 print_protocol_list32_t(cro.baseProtocols, info); 5337 outs() << " ivars " << format("0x%" PRIx32, cro.ivars) 5338 << "\n"; 5339 if (cro.ivars != 0) 5340 print_ivar_list32_t(cro.ivars, info); 5341 outs() << " weakIvarLayout " 5342 << format("0x%" PRIx32, cro.weakIvarLayout) << "\n"; 5343 print_layout_map32(cro.weakIvarLayout, info); 5344 outs() << " baseProperties " 5345 << format("0x%" PRIx32, cro.baseProperties) << "\n"; 5346 if (cro.baseProperties != 0) 5347 print_objc_property_list32(cro.baseProperties, info); 5348 is_meta_class = (cro.flags & RO_META) != 0; 5349 return true; 5350 } 5351 5352 static void print_class64_t(uint64_t p, struct DisassembleInfo *info) { 5353 struct class64_t c; 5354 const char *r; 5355 uint32_t offset, left; 5356 SectionRef S; 5357 const char *name; 5358 uint64_t isa_n_value, n_value; 5359 5360 r = get_pointer_64(p, offset, left, S, info); 5361 if (r == nullptr || left < sizeof(struct class64_t)) 5362 return; 5363 memcpy(&c, r, sizeof(struct class64_t)); 5364 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5365 swapStruct(c); 5366 5367 outs() << " isa " << format("0x%" PRIx64, c.isa); 5368 name = get_symbol_64(offset + offsetof(struct class64_t, isa), S, info, 5369 isa_n_value, c.isa); 5370 if (name != nullptr) 5371 outs() << " " << name; 5372 outs() << "\n"; 5373 5374 outs() << " superclass " << format("0x%" PRIx64, c.superclass); 5375 name = get_symbol_64(offset + offsetof(struct class64_t, superclass), S, info, 5376 n_value, c.superclass); 5377 if (name != nullptr) 5378 outs() << " " << name; 5379 else { 5380 name = get_dyld_bind_info_symbolname(S.getAddress() + 5381 offset + offsetof(struct class64_t, superclass), info); 5382 if (name != nullptr) 5383 outs() << " " << name; 5384 } 5385 outs() << "\n"; 5386 5387 outs() << " cache " << format("0x%" PRIx64, c.cache); 5388 name = get_symbol_64(offset + offsetof(struct class64_t, cache), S, info, 5389 n_value, c.cache); 5390 if (name != nullptr) 5391 outs() << " " << name; 5392 outs() << "\n"; 5393 5394 outs() << " vtable " << format("0x%" PRIx64, c.vtable); 5395 name = get_symbol_64(offset + offsetof(struct class64_t, vtable), S, info, 5396 n_value, c.vtable); 5397 if (name != nullptr) 5398 outs() << " " << name; 5399 outs() << "\n"; 5400 5401 name = get_symbol_64(offset + offsetof(struct class64_t, data), S, info, 5402 n_value, c.data); 5403 outs() << " data "; 5404 if (n_value != 0) { 5405 if (info->verbose && name != nullptr) 5406 outs() << name; 5407 else 5408 outs() << format("0x%" PRIx64, n_value); 5409 if (c.data != 0) 5410 outs() << " + " << format("0x%" PRIx64, c.data); 5411 } else 5412 outs() << format("0x%" PRIx64, c.data); 5413 outs() << " (struct class_ro_t *)"; 5414 5415 // This is a Swift class if some of the low bits of the pointer are set. 5416 if ((c.data + n_value) & 0x7) 5417 outs() << " Swift class"; 5418 outs() << "\n"; 5419 bool is_meta_class; 5420 if (!print_class_ro64_t((c.data + n_value) & ~0x7, info, is_meta_class)) 5421 return; 5422 5423 if (!is_meta_class && 5424 c.isa + isa_n_value != p && 5425 c.isa + isa_n_value != 0 && 5426 info->depth < 100) { 5427 info->depth++; 5428 outs() << "Meta Class\n"; 5429 print_class64_t(c.isa + isa_n_value, info); 5430 } 5431 } 5432 5433 static void print_class32_t(uint32_t p, struct DisassembleInfo *info) { 5434 struct class32_t c; 5435 const char *r; 5436 uint32_t offset, left; 5437 SectionRef S; 5438 const char *name; 5439 5440 r = get_pointer_32(p, offset, left, S, info); 5441 if (r == nullptr) 5442 return; 5443 memset(&c, '\0', sizeof(struct class32_t)); 5444 if (left < sizeof(struct class32_t)) { 5445 memcpy(&c, r, left); 5446 outs() << " (class_t entends past the end of the section)\n"; 5447 } else 5448 memcpy(&c, r, sizeof(struct class32_t)); 5449 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5450 swapStruct(c); 5451 5452 outs() << " isa " << format("0x%" PRIx32, c.isa); 5453 name = 5454 get_symbol_32(offset + offsetof(struct class32_t, isa), S, info, c.isa); 5455 if (name != nullptr) 5456 outs() << " " << name; 5457 outs() << "\n"; 5458 5459 outs() << " superclass " << format("0x%" PRIx32, c.superclass); 5460 name = get_symbol_32(offset + offsetof(struct class32_t, superclass), S, info, 5461 c.superclass); 5462 if (name != nullptr) 5463 outs() << " " << name; 5464 outs() << "\n"; 5465 5466 outs() << " cache " << format("0x%" PRIx32, c.cache); 5467 name = get_symbol_32(offset + offsetof(struct class32_t, cache), S, info, 5468 c.cache); 5469 if (name != nullptr) 5470 outs() << " " << name; 5471 outs() << "\n"; 5472 5473 outs() << " vtable " << format("0x%" PRIx32, c.vtable); 5474 name = get_symbol_32(offset + offsetof(struct class32_t, vtable), S, info, 5475 c.vtable); 5476 if (name != nullptr) 5477 outs() << " " << name; 5478 outs() << "\n"; 5479 5480 name = 5481 get_symbol_32(offset + offsetof(struct class32_t, data), S, info, c.data); 5482 outs() << " data " << format("0x%" PRIx32, c.data) 5483 << " (struct class_ro_t *)"; 5484 5485 // This is a Swift class if some of the low bits of the pointer are set. 5486 if (c.data & 0x3) 5487 outs() << " Swift class"; 5488 outs() << "\n"; 5489 bool is_meta_class; 5490 if (!print_class_ro32_t(c.data & ~0x3, info, is_meta_class)) 5491 return; 5492 5493 if (!is_meta_class) { 5494 outs() << "Meta Class\n"; 5495 print_class32_t(c.isa, info); 5496 } 5497 } 5498 5499 static void print_objc_class_t(struct objc_class_t *objc_class, 5500 struct DisassembleInfo *info) { 5501 uint32_t offset, left, xleft; 5502 const char *name, *p, *ivar_list; 5503 SectionRef S; 5504 int32_t i; 5505 struct objc_ivar_list_t objc_ivar_list; 5506 struct objc_ivar_t ivar; 5507 5508 outs() << "\t\t isa " << format("0x%08" PRIx32, objc_class->isa); 5509 if (info->verbose && CLS_GETINFO(objc_class, CLS_META)) { 5510 name = get_pointer_32(objc_class->isa, offset, left, S, info, true); 5511 if (name != nullptr) 5512 outs() << format(" %.*s", left, name); 5513 else 5514 outs() << " (not in an __OBJC section)"; 5515 } 5516 outs() << "\n"; 5517 5518 outs() << "\t super_class " 5519 << format("0x%08" PRIx32, objc_class->super_class); 5520 if (info->verbose) { 5521 name = get_pointer_32(objc_class->super_class, offset, left, S, info, true); 5522 if (name != nullptr) 5523 outs() << format(" %.*s", left, name); 5524 else 5525 outs() << " (not in an __OBJC section)"; 5526 } 5527 outs() << "\n"; 5528 5529 outs() << "\t\t name " << format("0x%08" PRIx32, objc_class->name); 5530 if (info->verbose) { 5531 name = get_pointer_32(objc_class->name, offset, left, S, info, true); 5532 if (name != nullptr) 5533 outs() << format(" %.*s", left, name); 5534 else 5535 outs() << " (not in an __OBJC section)"; 5536 } 5537 outs() << "\n"; 5538 5539 outs() << "\t\t version " << format("0x%08" PRIx32, objc_class->version) 5540 << "\n"; 5541 5542 outs() << "\t\t info " << format("0x%08" PRIx32, objc_class->info); 5543 if (info->verbose) { 5544 if (CLS_GETINFO(objc_class, CLS_CLASS)) 5545 outs() << " CLS_CLASS"; 5546 else if (CLS_GETINFO(objc_class, CLS_META)) 5547 outs() << " CLS_META"; 5548 } 5549 outs() << "\n"; 5550 5551 outs() << "\t instance_size " 5552 << format("0x%08" PRIx32, objc_class->instance_size) << "\n"; 5553 5554 p = get_pointer_32(objc_class->ivars, offset, left, S, info, true); 5555 outs() << "\t\t ivars " << format("0x%08" PRIx32, objc_class->ivars); 5556 if (p != nullptr) { 5557 if (left > sizeof(struct objc_ivar_list_t)) { 5558 outs() << "\n"; 5559 memcpy(&objc_ivar_list, p, sizeof(struct objc_ivar_list_t)); 5560 } else { 5561 outs() << " (entends past the end of the section)\n"; 5562 memset(&objc_ivar_list, '\0', sizeof(struct objc_ivar_list_t)); 5563 memcpy(&objc_ivar_list, p, left); 5564 } 5565 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5566 swapStruct(objc_ivar_list); 5567 outs() << "\t\t ivar_count " << objc_ivar_list.ivar_count << "\n"; 5568 ivar_list = p + sizeof(struct objc_ivar_list_t); 5569 for (i = 0; i < objc_ivar_list.ivar_count; i++) { 5570 if ((i + 1) * sizeof(struct objc_ivar_t) > left) { 5571 outs() << "\t\t remaining ivar's extend past the of the section\n"; 5572 break; 5573 } 5574 memcpy(&ivar, ivar_list + i * sizeof(struct objc_ivar_t), 5575 sizeof(struct objc_ivar_t)); 5576 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5577 swapStruct(ivar); 5578 5579 outs() << "\t\t\tivar_name " << format("0x%08" PRIx32, ivar.ivar_name); 5580 if (info->verbose) { 5581 name = get_pointer_32(ivar.ivar_name, offset, xleft, S, info, true); 5582 if (name != nullptr) 5583 outs() << format(" %.*s", xleft, name); 5584 else 5585 outs() << " (not in an __OBJC section)"; 5586 } 5587 outs() << "\n"; 5588 5589 outs() << "\t\t\tivar_type " << format("0x%08" PRIx32, ivar.ivar_type); 5590 if (info->verbose) { 5591 name = get_pointer_32(ivar.ivar_type, offset, xleft, S, info, true); 5592 if (name != nullptr) 5593 outs() << format(" %.*s", xleft, name); 5594 else 5595 outs() << " (not in an __OBJC section)"; 5596 } 5597 outs() << "\n"; 5598 5599 outs() << "\t\t ivar_offset " 5600 << format("0x%08" PRIx32, ivar.ivar_offset) << "\n"; 5601 } 5602 } else { 5603 outs() << " (not in an __OBJC section)\n"; 5604 } 5605 5606 outs() << "\t\t methods " << format("0x%08" PRIx32, objc_class->methodLists); 5607 if (print_method_list(objc_class->methodLists, info)) 5608 outs() << " (not in an __OBJC section)\n"; 5609 5610 outs() << "\t\t cache " << format("0x%08" PRIx32, objc_class->cache) 5611 << "\n"; 5612 5613 outs() << "\t\tprotocols " << format("0x%08" PRIx32, objc_class->protocols); 5614 if (print_protocol_list(objc_class->protocols, 16, info)) 5615 outs() << " (not in an __OBJC section)\n"; 5616 } 5617 5618 static void print_objc_objc_category_t(struct objc_category_t *objc_category, 5619 struct DisassembleInfo *info) { 5620 uint32_t offset, left; 5621 const char *name; 5622 SectionRef S; 5623 5624 outs() << "\t category name " 5625 << format("0x%08" PRIx32, objc_category->category_name); 5626 if (info->verbose) { 5627 name = get_pointer_32(objc_category->category_name, offset, left, S, info, 5628 true); 5629 if (name != nullptr) 5630 outs() << format(" %.*s", left, name); 5631 else 5632 outs() << " (not in an __OBJC section)"; 5633 } 5634 outs() << "\n"; 5635 5636 outs() << "\t\t class name " 5637 << format("0x%08" PRIx32, objc_category->class_name); 5638 if (info->verbose) { 5639 name = 5640 get_pointer_32(objc_category->class_name, offset, left, S, info, true); 5641 if (name != nullptr) 5642 outs() << format(" %.*s", left, name); 5643 else 5644 outs() << " (not in an __OBJC section)"; 5645 } 5646 outs() << "\n"; 5647 5648 outs() << "\t instance methods " 5649 << format("0x%08" PRIx32, objc_category->instance_methods); 5650 if (print_method_list(objc_category->instance_methods, info)) 5651 outs() << " (not in an __OBJC section)\n"; 5652 5653 outs() << "\t class methods " 5654 << format("0x%08" PRIx32, objc_category->class_methods); 5655 if (print_method_list(objc_category->class_methods, info)) 5656 outs() << " (not in an __OBJC section)\n"; 5657 } 5658 5659 static void print_category64_t(uint64_t p, struct DisassembleInfo *info) { 5660 struct category64_t c; 5661 const char *r; 5662 uint32_t offset, xoffset, left; 5663 SectionRef S, xS; 5664 const char *name, *sym_name; 5665 uint64_t n_value; 5666 5667 r = get_pointer_64(p, offset, left, S, info); 5668 if (r == nullptr) 5669 return; 5670 memset(&c, '\0', sizeof(struct category64_t)); 5671 if (left < sizeof(struct category64_t)) { 5672 memcpy(&c, r, left); 5673 outs() << " (category_t entends past the end of the section)\n"; 5674 } else 5675 memcpy(&c, r, sizeof(struct category64_t)); 5676 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5677 swapStruct(c); 5678 5679 outs() << " name "; 5680 sym_name = get_symbol_64(offset + offsetof(struct category64_t, name), S, 5681 info, n_value, c.name); 5682 if (n_value != 0) { 5683 if (info->verbose && sym_name != nullptr) 5684 outs() << sym_name; 5685 else 5686 outs() << format("0x%" PRIx64, n_value); 5687 if (c.name != 0) 5688 outs() << " + " << format("0x%" PRIx64, c.name); 5689 } else 5690 outs() << format("0x%" PRIx64, c.name); 5691 name = get_pointer_64(c.name + n_value, xoffset, left, xS, info); 5692 if (name != nullptr) 5693 outs() << format(" %.*s", left, name); 5694 outs() << "\n"; 5695 5696 outs() << " cls "; 5697 sym_name = get_symbol_64(offset + offsetof(struct category64_t, cls), S, info, 5698 n_value, c.cls); 5699 if (n_value != 0) { 5700 if (info->verbose && sym_name != nullptr) 5701 outs() << sym_name; 5702 else 5703 outs() << format("0x%" PRIx64, n_value); 5704 if (c.cls != 0) 5705 outs() << " + " << format("0x%" PRIx64, c.cls); 5706 } else 5707 outs() << format("0x%" PRIx64, c.cls); 5708 outs() << "\n"; 5709 if (c.cls + n_value != 0) 5710 print_class64_t(c.cls + n_value, info); 5711 5712 outs() << " instanceMethods "; 5713 sym_name = 5714 get_symbol_64(offset + offsetof(struct category64_t, instanceMethods), S, 5715 info, n_value, c.instanceMethods); 5716 if (n_value != 0) { 5717 if (info->verbose && sym_name != nullptr) 5718 outs() << sym_name; 5719 else 5720 outs() << format("0x%" PRIx64, n_value); 5721 if (c.instanceMethods != 0) 5722 outs() << " + " << format("0x%" PRIx64, c.instanceMethods); 5723 } else 5724 outs() << format("0x%" PRIx64, c.instanceMethods); 5725 outs() << "\n"; 5726 if (c.instanceMethods + n_value != 0) 5727 print_method_list64_t(c.instanceMethods + n_value, info, ""); 5728 5729 outs() << " classMethods "; 5730 sym_name = get_symbol_64(offset + offsetof(struct category64_t, classMethods), 5731 S, info, n_value, c.classMethods); 5732 if (n_value != 0) { 5733 if (info->verbose && sym_name != nullptr) 5734 outs() << sym_name; 5735 else 5736 outs() << format("0x%" PRIx64, n_value); 5737 if (c.classMethods != 0) 5738 outs() << " + " << format("0x%" PRIx64, c.classMethods); 5739 } else 5740 outs() << format("0x%" PRIx64, c.classMethods); 5741 outs() << "\n"; 5742 if (c.classMethods + n_value != 0) 5743 print_method_list64_t(c.classMethods + n_value, info, ""); 5744 5745 outs() << " protocols "; 5746 sym_name = get_symbol_64(offset + offsetof(struct category64_t, protocols), S, 5747 info, n_value, c.protocols); 5748 if (n_value != 0) { 5749 if (info->verbose && sym_name != nullptr) 5750 outs() << sym_name; 5751 else 5752 outs() << format("0x%" PRIx64, n_value); 5753 if (c.protocols != 0) 5754 outs() << " + " << format("0x%" PRIx64, c.protocols); 5755 } else 5756 outs() << format("0x%" PRIx64, c.protocols); 5757 outs() << "\n"; 5758 if (c.protocols + n_value != 0) 5759 print_protocol_list64_t(c.protocols + n_value, info); 5760 5761 outs() << "instanceProperties "; 5762 sym_name = 5763 get_symbol_64(offset + offsetof(struct category64_t, instanceProperties), 5764 S, info, n_value, c.instanceProperties); 5765 if (n_value != 0) { 5766 if (info->verbose && sym_name != nullptr) 5767 outs() << sym_name; 5768 else 5769 outs() << format("0x%" PRIx64, n_value); 5770 if (c.instanceProperties != 0) 5771 outs() << " + " << format("0x%" PRIx64, c.instanceProperties); 5772 } else 5773 outs() << format("0x%" PRIx64, c.instanceProperties); 5774 outs() << "\n"; 5775 if (c.instanceProperties + n_value != 0) 5776 print_objc_property_list64(c.instanceProperties + n_value, info); 5777 } 5778 5779 static void print_category32_t(uint32_t p, struct DisassembleInfo *info) { 5780 struct category32_t c; 5781 const char *r; 5782 uint32_t offset, left; 5783 SectionRef S, xS; 5784 const char *name; 5785 5786 r = get_pointer_32(p, offset, left, S, info); 5787 if (r == nullptr) 5788 return; 5789 memset(&c, '\0', sizeof(struct category32_t)); 5790 if (left < sizeof(struct category32_t)) { 5791 memcpy(&c, r, left); 5792 outs() << " (category_t entends past the end of the section)\n"; 5793 } else 5794 memcpy(&c, r, sizeof(struct category32_t)); 5795 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5796 swapStruct(c); 5797 5798 outs() << " name " << format("0x%" PRIx32, c.name); 5799 name = get_symbol_32(offset + offsetof(struct category32_t, name), S, info, 5800 c.name); 5801 if (name) 5802 outs() << " " << name; 5803 outs() << "\n"; 5804 5805 outs() << " cls " << format("0x%" PRIx32, c.cls) << "\n"; 5806 if (c.cls != 0) 5807 print_class32_t(c.cls, info); 5808 outs() << " instanceMethods " << format("0x%" PRIx32, c.instanceMethods) 5809 << "\n"; 5810 if (c.instanceMethods != 0) 5811 print_method_list32_t(c.instanceMethods, info, ""); 5812 outs() << " classMethods " << format("0x%" PRIx32, c.classMethods) 5813 << "\n"; 5814 if (c.classMethods != 0) 5815 print_method_list32_t(c.classMethods, info, ""); 5816 outs() << " protocols " << format("0x%" PRIx32, c.protocols) << "\n"; 5817 if (c.protocols != 0) 5818 print_protocol_list32_t(c.protocols, info); 5819 outs() << "instanceProperties " << format("0x%" PRIx32, c.instanceProperties) 5820 << "\n"; 5821 if (c.instanceProperties != 0) 5822 print_objc_property_list32(c.instanceProperties, info); 5823 } 5824 5825 static void print_message_refs64(SectionRef S, struct DisassembleInfo *info) { 5826 uint32_t i, left, offset, xoffset; 5827 uint64_t p, n_value; 5828 struct message_ref64 mr; 5829 const char *name, *sym_name; 5830 const char *r; 5831 SectionRef xS; 5832 5833 if (S == SectionRef()) 5834 return; 5835 5836 StringRef SectName; 5837 Expected<StringRef> SecNameOrErr = S.getName(); 5838 if (SecNameOrErr) 5839 SectName = *SecNameOrErr; 5840 else 5841 consumeError(SecNameOrErr.takeError()); 5842 5843 DataRefImpl Ref = S.getRawDataRefImpl(); 5844 StringRef SegName = info->O->getSectionFinalSegmentName(Ref); 5845 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 5846 offset = 0; 5847 for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) { 5848 p = S.getAddress() + i; 5849 r = get_pointer_64(p, offset, left, S, info); 5850 if (r == nullptr) 5851 return; 5852 memset(&mr, '\0', sizeof(struct message_ref64)); 5853 if (left < sizeof(struct message_ref64)) { 5854 memcpy(&mr, r, left); 5855 outs() << " (message_ref entends past the end of the section)\n"; 5856 } else 5857 memcpy(&mr, r, sizeof(struct message_ref64)); 5858 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5859 swapStruct(mr); 5860 5861 outs() << " imp "; 5862 name = get_symbol_64(offset + offsetof(struct message_ref64, imp), S, info, 5863 n_value, mr.imp); 5864 if (n_value != 0) { 5865 outs() << format("0x%" PRIx64, n_value) << " "; 5866 if (mr.imp != 0) 5867 outs() << "+ " << format("0x%" PRIx64, mr.imp) << " "; 5868 } else 5869 outs() << format("0x%" PRIx64, mr.imp) << " "; 5870 if (name != nullptr) 5871 outs() << " " << name; 5872 outs() << "\n"; 5873 5874 outs() << " sel "; 5875 sym_name = get_symbol_64(offset + offsetof(struct message_ref64, sel), S, 5876 info, n_value, mr.sel); 5877 if (n_value != 0) { 5878 if (info->verbose && sym_name != nullptr) 5879 outs() << sym_name; 5880 else 5881 outs() << format("0x%" PRIx64, n_value); 5882 if (mr.sel != 0) 5883 outs() << " + " << format("0x%" PRIx64, mr.sel); 5884 } else 5885 outs() << format("0x%" PRIx64, mr.sel); 5886 name = get_pointer_64(mr.sel + n_value, xoffset, left, xS, info); 5887 if (name != nullptr) 5888 outs() << format(" %.*s", left, name); 5889 outs() << "\n"; 5890 5891 offset += sizeof(struct message_ref64); 5892 } 5893 } 5894 5895 static void print_message_refs32(SectionRef S, struct DisassembleInfo *info) { 5896 uint32_t i, left, offset, xoffset, p; 5897 struct message_ref32 mr; 5898 const char *name, *r; 5899 SectionRef xS; 5900 5901 if (S == SectionRef()) 5902 return; 5903 5904 StringRef SectName; 5905 Expected<StringRef> SecNameOrErr = S.getName(); 5906 if (SecNameOrErr) 5907 SectName = *SecNameOrErr; 5908 else 5909 consumeError(SecNameOrErr.takeError()); 5910 5911 DataRefImpl Ref = S.getRawDataRefImpl(); 5912 StringRef SegName = info->O->getSectionFinalSegmentName(Ref); 5913 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 5914 offset = 0; 5915 for (i = 0; i < S.getSize(); i += sizeof(struct message_ref64)) { 5916 p = S.getAddress() + i; 5917 r = get_pointer_32(p, offset, left, S, info); 5918 if (r == nullptr) 5919 return; 5920 memset(&mr, '\0', sizeof(struct message_ref32)); 5921 if (left < sizeof(struct message_ref32)) { 5922 memcpy(&mr, r, left); 5923 outs() << " (message_ref entends past the end of the section)\n"; 5924 } else 5925 memcpy(&mr, r, sizeof(struct message_ref32)); 5926 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5927 swapStruct(mr); 5928 5929 outs() << " imp " << format("0x%" PRIx32, mr.imp); 5930 name = get_symbol_32(offset + offsetof(struct message_ref32, imp), S, info, 5931 mr.imp); 5932 if (name != nullptr) 5933 outs() << " " << name; 5934 outs() << "\n"; 5935 5936 outs() << " sel " << format("0x%" PRIx32, mr.sel); 5937 name = get_pointer_32(mr.sel, xoffset, left, xS, info); 5938 if (name != nullptr) 5939 outs() << " " << name; 5940 outs() << "\n"; 5941 5942 offset += sizeof(struct message_ref32); 5943 } 5944 } 5945 5946 static void print_image_info64(SectionRef S, struct DisassembleInfo *info) { 5947 uint32_t left, offset, swift_version; 5948 uint64_t p; 5949 struct objc_image_info64 o; 5950 const char *r; 5951 5952 if (S == SectionRef()) 5953 return; 5954 5955 StringRef SectName; 5956 Expected<StringRef> SecNameOrErr = S.getName(); 5957 if (SecNameOrErr) 5958 SectName = *SecNameOrErr; 5959 else 5960 consumeError(SecNameOrErr.takeError()); 5961 5962 DataRefImpl Ref = S.getRawDataRefImpl(); 5963 StringRef SegName = info->O->getSectionFinalSegmentName(Ref); 5964 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 5965 p = S.getAddress(); 5966 r = get_pointer_64(p, offset, left, S, info); 5967 if (r == nullptr) 5968 return; 5969 memset(&o, '\0', sizeof(struct objc_image_info64)); 5970 if (left < sizeof(struct objc_image_info64)) { 5971 memcpy(&o, r, left); 5972 outs() << " (objc_image_info entends past the end of the section)\n"; 5973 } else 5974 memcpy(&o, r, sizeof(struct objc_image_info64)); 5975 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 5976 swapStruct(o); 5977 outs() << " version " << o.version << "\n"; 5978 outs() << " flags " << format("0x%" PRIx32, o.flags); 5979 if (o.flags & OBJC_IMAGE_IS_REPLACEMENT) 5980 outs() << " OBJC_IMAGE_IS_REPLACEMENT"; 5981 if (o.flags & OBJC_IMAGE_SUPPORTS_GC) 5982 outs() << " OBJC_IMAGE_SUPPORTS_GC"; 5983 if (o.flags & OBJC_IMAGE_IS_SIMULATED) 5984 outs() << " OBJC_IMAGE_IS_SIMULATED"; 5985 if (o.flags & OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES) 5986 outs() << " OBJC_IMAGE_HAS_CATEGORY_CLASS_PROPERTIES"; 5987 swift_version = (o.flags >> 8) & 0xff; 5988 if (swift_version != 0) { 5989 if (swift_version == 1) 5990 outs() << " Swift 1.0"; 5991 else if (swift_version == 2) 5992 outs() << " Swift 1.1"; 5993 else if(swift_version == 3) 5994 outs() << " Swift 2.0"; 5995 else if(swift_version == 4) 5996 outs() << " Swift 3.0"; 5997 else if(swift_version == 5) 5998 outs() << " Swift 4.0"; 5999 else if(swift_version == 6) 6000 outs() << " Swift 4.1/Swift 4.2"; 6001 else if(swift_version == 7) 6002 outs() << " Swift 5 or later"; 6003 else 6004 outs() << " unknown future Swift version (" << swift_version << ")"; 6005 } 6006 outs() << "\n"; 6007 } 6008 6009 static void print_image_info32(SectionRef S, struct DisassembleInfo *info) { 6010 uint32_t left, offset, swift_version, p; 6011 struct objc_image_info32 o; 6012 const char *r; 6013 6014 if (S == SectionRef()) 6015 return; 6016 6017 StringRef SectName; 6018 Expected<StringRef> SecNameOrErr = S.getName(); 6019 if (SecNameOrErr) 6020 SectName = *SecNameOrErr; 6021 else 6022 consumeError(SecNameOrErr.takeError()); 6023 6024 DataRefImpl Ref = S.getRawDataRefImpl(); 6025 StringRef SegName = info->O->getSectionFinalSegmentName(Ref); 6026 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 6027 p = S.getAddress(); 6028 r = get_pointer_32(p, offset, left, S, info); 6029 if (r == nullptr) 6030 return; 6031 memset(&o, '\0', sizeof(struct objc_image_info32)); 6032 if (left < sizeof(struct objc_image_info32)) { 6033 memcpy(&o, r, left); 6034 outs() << " (objc_image_info entends past the end of the section)\n"; 6035 } else 6036 memcpy(&o, r, sizeof(struct objc_image_info32)); 6037 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 6038 swapStruct(o); 6039 outs() << " version " << o.version << "\n"; 6040 outs() << " flags " << format("0x%" PRIx32, o.flags); 6041 if (o.flags & OBJC_IMAGE_IS_REPLACEMENT) 6042 outs() << " OBJC_IMAGE_IS_REPLACEMENT"; 6043 if (o.flags & OBJC_IMAGE_SUPPORTS_GC) 6044 outs() << " OBJC_IMAGE_SUPPORTS_GC"; 6045 swift_version = (o.flags >> 8) & 0xff; 6046 if (swift_version != 0) { 6047 if (swift_version == 1) 6048 outs() << " Swift 1.0"; 6049 else if (swift_version == 2) 6050 outs() << " Swift 1.1"; 6051 else if(swift_version == 3) 6052 outs() << " Swift 2.0"; 6053 else if(swift_version == 4) 6054 outs() << " Swift 3.0"; 6055 else if(swift_version == 5) 6056 outs() << " Swift 4.0"; 6057 else if(swift_version == 6) 6058 outs() << " Swift 4.1/Swift 4.2"; 6059 else if(swift_version == 7) 6060 outs() << " Swift 5 or later"; 6061 else 6062 outs() << " unknown future Swift version (" << swift_version << ")"; 6063 } 6064 outs() << "\n"; 6065 } 6066 6067 static void print_image_info(SectionRef S, struct DisassembleInfo *info) { 6068 uint32_t left, offset, p; 6069 struct imageInfo_t o; 6070 const char *r; 6071 6072 StringRef SectName; 6073 Expected<StringRef> SecNameOrErr = S.getName(); 6074 if (SecNameOrErr) 6075 SectName = *SecNameOrErr; 6076 else 6077 consumeError(SecNameOrErr.takeError()); 6078 6079 DataRefImpl Ref = S.getRawDataRefImpl(); 6080 StringRef SegName = info->O->getSectionFinalSegmentName(Ref); 6081 outs() << "Contents of (" << SegName << "," << SectName << ") section\n"; 6082 p = S.getAddress(); 6083 r = get_pointer_32(p, offset, left, S, info); 6084 if (r == nullptr) 6085 return; 6086 memset(&o, '\0', sizeof(struct imageInfo_t)); 6087 if (left < sizeof(struct imageInfo_t)) { 6088 memcpy(&o, r, left); 6089 outs() << " (imageInfo entends past the end of the section)\n"; 6090 } else 6091 memcpy(&o, r, sizeof(struct imageInfo_t)); 6092 if (info->O->isLittleEndian() != sys::IsLittleEndianHost) 6093 swapStruct(o); 6094 outs() << " version " << o.version << "\n"; 6095 outs() << " flags " << format("0x%" PRIx32, o.flags); 6096 if (o.flags & 0x1) 6097 outs() << " F&C"; 6098 if (o.flags & 0x2) 6099 outs() << " GC"; 6100 if (o.flags & 0x4) 6101 outs() << " GC-only"; 6102 else 6103 outs() << " RR"; 6104 outs() << "\n"; 6105 } 6106 6107 static void printObjc2_64bit_MetaData(MachOObjectFile *O, bool verbose) { 6108 SymbolAddressMap AddrMap; 6109 if (verbose) 6110 CreateSymbolAddressMap(O, &AddrMap); 6111 6112 std::vector<SectionRef> Sections; 6113 for (const SectionRef &Section : O->sections()) 6114 Sections.push_back(Section); 6115 6116 struct DisassembleInfo info(O, &AddrMap, &Sections, verbose); 6117 6118 SectionRef CL = get_section(O, "__OBJC2", "__class_list"); 6119 if (CL == SectionRef()) 6120 CL = get_section(O, "__DATA", "__objc_classlist"); 6121 if (CL == SectionRef()) 6122 CL = get_section(O, "__DATA_CONST", "__objc_classlist"); 6123 if (CL == SectionRef()) 6124 CL = get_section(O, "__DATA_DIRTY", "__objc_classlist"); 6125 info.S = CL; 6126 walk_pointer_list_64("class", CL, O, &info, print_class64_t); 6127 6128 SectionRef CR = get_section(O, "__OBJC2", "__class_refs"); 6129 if (CR == SectionRef()) 6130 CR = get_section(O, "__DATA", "__objc_classrefs"); 6131 if (CR == SectionRef()) 6132 CR = get_section(O, "__DATA_CONST", "__objc_classrefs"); 6133 if (CR == SectionRef()) 6134 CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs"); 6135 info.S = CR; 6136 walk_pointer_list_64("class refs", CR, O, &info, nullptr); 6137 6138 SectionRef SR = get_section(O, "__OBJC2", "__super_refs"); 6139 if (SR == SectionRef()) 6140 SR = get_section(O, "__DATA", "__objc_superrefs"); 6141 if (SR == SectionRef()) 6142 SR = get_section(O, "__DATA_CONST", "__objc_superrefs"); 6143 if (SR == SectionRef()) 6144 SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs"); 6145 info.S = SR; 6146 walk_pointer_list_64("super refs", SR, O, &info, nullptr); 6147 6148 SectionRef CA = get_section(O, "__OBJC2", "__category_list"); 6149 if (CA == SectionRef()) 6150 CA = get_section(O, "__DATA", "__objc_catlist"); 6151 if (CA == SectionRef()) 6152 CA = get_section(O, "__DATA_CONST", "__objc_catlist"); 6153 if (CA == SectionRef()) 6154 CA = get_section(O, "__DATA_DIRTY", "__objc_catlist"); 6155 info.S = CA; 6156 walk_pointer_list_64("category", CA, O, &info, print_category64_t); 6157 6158 SectionRef PL = get_section(O, "__OBJC2", "__protocol_list"); 6159 if (PL == SectionRef()) 6160 PL = get_section(O, "__DATA", "__objc_protolist"); 6161 if (PL == SectionRef()) 6162 PL = get_section(O, "__DATA_CONST", "__objc_protolist"); 6163 if (PL == SectionRef()) 6164 PL = get_section(O, "__DATA_DIRTY", "__objc_protolist"); 6165 info.S = PL; 6166 walk_pointer_list_64("protocol", PL, O, &info, nullptr); 6167 6168 SectionRef MR = get_section(O, "__OBJC2", "__message_refs"); 6169 if (MR == SectionRef()) 6170 MR = get_section(O, "__DATA", "__objc_msgrefs"); 6171 if (MR == SectionRef()) 6172 MR = get_section(O, "__DATA_CONST", "__objc_msgrefs"); 6173 if (MR == SectionRef()) 6174 MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs"); 6175 info.S = MR; 6176 print_message_refs64(MR, &info); 6177 6178 SectionRef II = get_section(O, "__OBJC2", "__image_info"); 6179 if (II == SectionRef()) 6180 II = get_section(O, "__DATA", "__objc_imageinfo"); 6181 if (II == SectionRef()) 6182 II = get_section(O, "__DATA_CONST", "__objc_imageinfo"); 6183 if (II == SectionRef()) 6184 II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo"); 6185 info.S = II; 6186 print_image_info64(II, &info); 6187 } 6188 6189 static void printObjc2_32bit_MetaData(MachOObjectFile *O, bool verbose) { 6190 SymbolAddressMap AddrMap; 6191 if (verbose) 6192 CreateSymbolAddressMap(O, &AddrMap); 6193 6194 std::vector<SectionRef> Sections; 6195 for (const SectionRef &Section : O->sections()) 6196 Sections.push_back(Section); 6197 6198 struct DisassembleInfo info(O, &AddrMap, &Sections, verbose); 6199 6200 SectionRef CL = get_section(O, "__OBJC2", "__class_list"); 6201 if (CL == SectionRef()) 6202 CL = get_section(O, "__DATA", "__objc_classlist"); 6203 if (CL == SectionRef()) 6204 CL = get_section(O, "__DATA_CONST", "__objc_classlist"); 6205 if (CL == SectionRef()) 6206 CL = get_section(O, "__DATA_DIRTY", "__objc_classlist"); 6207 info.S = CL; 6208 walk_pointer_list_32("class", CL, O, &info, print_class32_t); 6209 6210 SectionRef CR = get_section(O, "__OBJC2", "__class_refs"); 6211 if (CR == SectionRef()) 6212 CR = get_section(O, "__DATA", "__objc_classrefs"); 6213 if (CR == SectionRef()) 6214 CR = get_section(O, "__DATA_CONST", "__objc_classrefs"); 6215 if (CR == SectionRef()) 6216 CR = get_section(O, "__DATA_DIRTY", "__objc_classrefs"); 6217 info.S = CR; 6218 walk_pointer_list_32("class refs", CR, O, &info, nullptr); 6219 6220 SectionRef SR = get_section(O, "__OBJC2", "__super_refs"); 6221 if (SR == SectionRef()) 6222 SR = get_section(O, "__DATA", "__objc_superrefs"); 6223 if (SR == SectionRef()) 6224 SR = get_section(O, "__DATA_CONST", "__objc_superrefs"); 6225 if (SR == SectionRef()) 6226 SR = get_section(O, "__DATA_DIRTY", "__objc_superrefs"); 6227 info.S = SR; 6228 walk_pointer_list_32("super refs", SR, O, &info, nullptr); 6229 6230 SectionRef CA = get_section(O, "__OBJC2", "__category_list"); 6231 if (CA == SectionRef()) 6232 CA = get_section(O, "__DATA", "__objc_catlist"); 6233 if (CA == SectionRef()) 6234 CA = get_section(O, "__DATA_CONST", "__objc_catlist"); 6235 if (CA == SectionRef()) 6236 CA = get_section(O, "__DATA_DIRTY", "__objc_catlist"); 6237 info.S = CA; 6238 walk_pointer_list_32("category", CA, O, &info, print_category32_t); 6239 6240 SectionRef PL = get_section(O, "__OBJC2", "__protocol_list"); 6241 if (PL == SectionRef()) 6242 PL = get_section(O, "__DATA", "__objc_protolist"); 6243 if (PL == SectionRef()) 6244 PL = get_section(O, "__DATA_CONST", "__objc_protolist"); 6245 if (PL == SectionRef()) 6246 PL = get_section(O, "__DATA_DIRTY", "__objc_protolist"); 6247 info.S = PL; 6248 walk_pointer_list_32("protocol", PL, O, &info, nullptr); 6249 6250 SectionRef MR = get_section(O, "__OBJC2", "__message_refs"); 6251 if (MR == SectionRef()) 6252 MR = get_section(O, "__DATA", "__objc_msgrefs"); 6253 if (MR == SectionRef()) 6254 MR = get_section(O, "__DATA_CONST", "__objc_msgrefs"); 6255 if (MR == SectionRef()) 6256 MR = get_section(O, "__DATA_DIRTY", "__objc_msgrefs"); 6257 info.S = MR; 6258 print_message_refs32(MR, &info); 6259 6260 SectionRef II = get_section(O, "__OBJC2", "__image_info"); 6261 if (II == SectionRef()) 6262 II = get_section(O, "__DATA", "__objc_imageinfo"); 6263 if (II == SectionRef()) 6264 II = get_section(O, "__DATA_CONST", "__objc_imageinfo"); 6265 if (II == SectionRef()) 6266 II = get_section(O, "__DATA_DIRTY", "__objc_imageinfo"); 6267 info.S = II; 6268 print_image_info32(II, &info); 6269 } 6270 6271 static bool printObjc1_32bit_MetaData(MachOObjectFile *O, bool verbose) { 6272 uint32_t i, j, p, offset, xoffset, left, defs_left, def; 6273 const char *r, *name, *defs; 6274 struct objc_module_t module; 6275 SectionRef S, xS; 6276 struct objc_symtab_t symtab; 6277 struct objc_class_t objc_class; 6278 struct objc_category_t objc_category; 6279 6280 outs() << "Objective-C segment\n"; 6281 S = get_section(O, "__OBJC", "__module_info"); 6282 if (S == SectionRef()) 6283 return false; 6284 6285 SymbolAddressMap AddrMap; 6286 if (verbose) 6287 CreateSymbolAddressMap(O, &AddrMap); 6288 6289 std::vector<SectionRef> Sections; 6290 for (const SectionRef &Section : O->sections()) 6291 Sections.push_back(Section); 6292 6293 struct DisassembleInfo info(O, &AddrMap, &Sections, verbose); 6294 6295 for (i = 0; i < S.getSize(); i += sizeof(struct objc_module_t)) { 6296 p = S.getAddress() + i; 6297 r = get_pointer_32(p, offset, left, S, &info, true); 6298 if (r == nullptr) 6299 return true; 6300 memset(&module, '\0', sizeof(struct objc_module_t)); 6301 if (left < sizeof(struct objc_module_t)) { 6302 memcpy(&module, r, left); 6303 outs() << " (module extends past end of __module_info section)\n"; 6304 } else 6305 memcpy(&module, r, sizeof(struct objc_module_t)); 6306 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6307 swapStruct(module); 6308 6309 outs() << "Module " << format("0x%" PRIx32, p) << "\n"; 6310 outs() << " version " << module.version << "\n"; 6311 outs() << " size " << module.size << "\n"; 6312 outs() << " name "; 6313 name = get_pointer_32(module.name, xoffset, left, xS, &info, true); 6314 if (name != nullptr) 6315 outs() << format("%.*s", left, name); 6316 else 6317 outs() << format("0x%08" PRIx32, module.name) 6318 << "(not in an __OBJC section)"; 6319 outs() << "\n"; 6320 6321 r = get_pointer_32(module.symtab, xoffset, left, xS, &info, true); 6322 if (module.symtab == 0 || r == nullptr) { 6323 outs() << " symtab " << format("0x%08" PRIx32, module.symtab) 6324 << " (not in an __OBJC section)\n"; 6325 continue; 6326 } 6327 outs() << " symtab " << format("0x%08" PRIx32, module.symtab) << "\n"; 6328 memset(&symtab, '\0', sizeof(struct objc_symtab_t)); 6329 defs_left = 0; 6330 defs = nullptr; 6331 if (left < sizeof(struct objc_symtab_t)) { 6332 memcpy(&symtab, r, left); 6333 outs() << "\tsymtab extends past end of an __OBJC section)\n"; 6334 } else { 6335 memcpy(&symtab, r, sizeof(struct objc_symtab_t)); 6336 if (left > sizeof(struct objc_symtab_t)) { 6337 defs_left = left - sizeof(struct objc_symtab_t); 6338 defs = r + sizeof(struct objc_symtab_t); 6339 } 6340 } 6341 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6342 swapStruct(symtab); 6343 6344 outs() << "\tsel_ref_cnt " << symtab.sel_ref_cnt << "\n"; 6345 r = get_pointer_32(symtab.refs, xoffset, left, xS, &info, true); 6346 outs() << "\trefs " << format("0x%08" PRIx32, symtab.refs); 6347 if (r == nullptr) 6348 outs() << " (not in an __OBJC section)"; 6349 outs() << "\n"; 6350 outs() << "\tcls_def_cnt " << symtab.cls_def_cnt << "\n"; 6351 outs() << "\tcat_def_cnt " << symtab.cat_def_cnt << "\n"; 6352 if (symtab.cls_def_cnt > 0) 6353 outs() << "\tClass Definitions\n"; 6354 for (j = 0; j < symtab.cls_def_cnt; j++) { 6355 if ((j + 1) * sizeof(uint32_t) > defs_left) { 6356 outs() << "\t(remaining class defs entries entends past the end of the " 6357 << "section)\n"; 6358 break; 6359 } 6360 memcpy(&def, defs + j * sizeof(uint32_t), sizeof(uint32_t)); 6361 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6362 sys::swapByteOrder(def); 6363 6364 r = get_pointer_32(def, xoffset, left, xS, &info, true); 6365 outs() << "\tdefs[" << j << "] " << format("0x%08" PRIx32, def); 6366 if (r != nullptr) { 6367 if (left > sizeof(struct objc_class_t)) { 6368 outs() << "\n"; 6369 memcpy(&objc_class, r, sizeof(struct objc_class_t)); 6370 } else { 6371 outs() << " (entends past the end of the section)\n"; 6372 memset(&objc_class, '\0', sizeof(struct objc_class_t)); 6373 memcpy(&objc_class, r, left); 6374 } 6375 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6376 swapStruct(objc_class); 6377 print_objc_class_t(&objc_class, &info); 6378 } else { 6379 outs() << "(not in an __OBJC section)\n"; 6380 } 6381 6382 if (CLS_GETINFO(&objc_class, CLS_CLASS)) { 6383 outs() << "\tMeta Class"; 6384 r = get_pointer_32(objc_class.isa, xoffset, left, xS, &info, true); 6385 if (r != nullptr) { 6386 if (left > sizeof(struct objc_class_t)) { 6387 outs() << "\n"; 6388 memcpy(&objc_class, r, sizeof(struct objc_class_t)); 6389 } else { 6390 outs() << " (entends past the end of the section)\n"; 6391 memset(&objc_class, '\0', sizeof(struct objc_class_t)); 6392 memcpy(&objc_class, r, left); 6393 } 6394 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6395 swapStruct(objc_class); 6396 print_objc_class_t(&objc_class, &info); 6397 } else { 6398 outs() << "(not in an __OBJC section)\n"; 6399 } 6400 } 6401 } 6402 if (symtab.cat_def_cnt > 0) 6403 outs() << "\tCategory Definitions\n"; 6404 for (j = 0; j < symtab.cat_def_cnt; j++) { 6405 if ((j + symtab.cls_def_cnt + 1) * sizeof(uint32_t) > defs_left) { 6406 outs() << "\t(remaining category defs entries entends past the end of " 6407 << "the section)\n"; 6408 break; 6409 } 6410 memcpy(&def, defs + (j + symtab.cls_def_cnt) * sizeof(uint32_t), 6411 sizeof(uint32_t)); 6412 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6413 sys::swapByteOrder(def); 6414 6415 r = get_pointer_32(def, xoffset, left, xS, &info, true); 6416 outs() << "\tdefs[" << j + symtab.cls_def_cnt << "] " 6417 << format("0x%08" PRIx32, def); 6418 if (r != nullptr) { 6419 if (left > sizeof(struct objc_category_t)) { 6420 outs() << "\n"; 6421 memcpy(&objc_category, r, sizeof(struct objc_category_t)); 6422 } else { 6423 outs() << " (entends past the end of the section)\n"; 6424 memset(&objc_category, '\0', sizeof(struct objc_category_t)); 6425 memcpy(&objc_category, r, left); 6426 } 6427 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6428 swapStruct(objc_category); 6429 print_objc_objc_category_t(&objc_category, &info); 6430 } else { 6431 outs() << "(not in an __OBJC section)\n"; 6432 } 6433 } 6434 } 6435 const SectionRef II = get_section(O, "__OBJC", "__image_info"); 6436 if (II != SectionRef()) 6437 print_image_info(II, &info); 6438 6439 return true; 6440 } 6441 6442 static void DumpProtocolSection(MachOObjectFile *O, const char *sect, 6443 uint32_t size, uint32_t addr) { 6444 SymbolAddressMap AddrMap; 6445 CreateSymbolAddressMap(O, &AddrMap); 6446 6447 std::vector<SectionRef> Sections; 6448 for (const SectionRef &Section : O->sections()) 6449 Sections.push_back(Section); 6450 6451 struct DisassembleInfo info(O, &AddrMap, &Sections, true); 6452 6453 const char *p; 6454 struct objc_protocol_t protocol; 6455 uint32_t left, paddr; 6456 for (p = sect; p < sect + size; p += sizeof(struct objc_protocol_t)) { 6457 memset(&protocol, '\0', sizeof(struct objc_protocol_t)); 6458 left = size - (p - sect); 6459 if (left < sizeof(struct objc_protocol_t)) { 6460 outs() << "Protocol extends past end of __protocol section\n"; 6461 memcpy(&protocol, p, left); 6462 } else 6463 memcpy(&protocol, p, sizeof(struct objc_protocol_t)); 6464 if (O->isLittleEndian() != sys::IsLittleEndianHost) 6465 swapStruct(protocol); 6466 paddr = addr + (p - sect); 6467 outs() << "Protocol " << format("0x%" PRIx32, paddr); 6468 if (print_protocol(paddr, 0, &info)) 6469 outs() << "(not in an __OBJC section)\n"; 6470 } 6471 } 6472 6473 #ifdef HAVE_LIBXAR 6474 static inline void swapStruct(struct xar_header &xar) { 6475 sys::swapByteOrder(xar.magic); 6476 sys::swapByteOrder(xar.size); 6477 sys::swapByteOrder(xar.version); 6478 sys::swapByteOrder(xar.toc_length_compressed); 6479 sys::swapByteOrder(xar.toc_length_uncompressed); 6480 sys::swapByteOrder(xar.cksum_alg); 6481 } 6482 6483 static void PrintModeVerbose(uint32_t mode) { 6484 switch(mode & S_IFMT){ 6485 case S_IFDIR: 6486 outs() << "d"; 6487 break; 6488 case S_IFCHR: 6489 outs() << "c"; 6490 break; 6491 case S_IFBLK: 6492 outs() << "b"; 6493 break; 6494 case S_IFREG: 6495 outs() << "-"; 6496 break; 6497 case S_IFLNK: 6498 outs() << "l"; 6499 break; 6500 case S_IFSOCK: 6501 outs() << "s"; 6502 break; 6503 default: 6504 outs() << "?"; 6505 break; 6506 } 6507 6508 /* owner permissions */ 6509 if(mode & S_IREAD) 6510 outs() << "r"; 6511 else 6512 outs() << "-"; 6513 if(mode & S_IWRITE) 6514 outs() << "w"; 6515 else 6516 outs() << "-"; 6517 if(mode & S_ISUID) 6518 outs() << "s"; 6519 else if(mode & S_IEXEC) 6520 outs() << "x"; 6521 else 6522 outs() << "-"; 6523 6524 /* group permissions */ 6525 if(mode & (S_IREAD >> 3)) 6526 outs() << "r"; 6527 else 6528 outs() << "-"; 6529 if(mode & (S_IWRITE >> 3)) 6530 outs() << "w"; 6531 else 6532 outs() << "-"; 6533 if(mode & S_ISGID) 6534 outs() << "s"; 6535 else if(mode & (S_IEXEC >> 3)) 6536 outs() << "x"; 6537 else 6538 outs() << "-"; 6539 6540 /* other permissions */ 6541 if(mode & (S_IREAD >> 6)) 6542 outs() << "r"; 6543 else 6544 outs() << "-"; 6545 if(mode & (S_IWRITE >> 6)) 6546 outs() << "w"; 6547 else 6548 outs() << "-"; 6549 if(mode & S_ISVTX) 6550 outs() << "t"; 6551 else if(mode & (S_IEXEC >> 6)) 6552 outs() << "x"; 6553 else 6554 outs() << "-"; 6555 } 6556 6557 static void PrintXarFilesSummary(const char *XarFilename, xar_t xar) { 6558 xar_file_t xf; 6559 const char *key, *type, *mode, *user, *group, *size, *mtime, *name, *m; 6560 char *endp; 6561 uint32_t mode_value; 6562 6563 ScopedXarIter xi; 6564 if (!xi) { 6565 WithColor::error(errs(), "llvm-objdump") 6566 << "can't obtain an xar iterator for xar archive " << XarFilename 6567 << "\n"; 6568 return; 6569 } 6570 6571 // Go through the xar's files. 6572 for (xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)) { 6573 ScopedXarIter xp; 6574 if(!xp){ 6575 WithColor::error(errs(), "llvm-objdump") 6576 << "can't obtain an xar iterator for xar archive " << XarFilename 6577 << "\n"; 6578 return; 6579 } 6580 type = nullptr; 6581 mode = nullptr; 6582 user = nullptr; 6583 group = nullptr; 6584 size = nullptr; 6585 mtime = nullptr; 6586 name = nullptr; 6587 for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){ 6588 const char *val = nullptr; 6589 xar_prop_get(xf, key, &val); 6590 #if 0 // Useful for debugging. 6591 outs() << "key: " << key << " value: " << val << "\n"; 6592 #endif 6593 if(strcmp(key, "type") == 0) 6594 type = val; 6595 if(strcmp(key, "mode") == 0) 6596 mode = val; 6597 if(strcmp(key, "user") == 0) 6598 user = val; 6599 if(strcmp(key, "group") == 0) 6600 group = val; 6601 if(strcmp(key, "data/size") == 0) 6602 size = val; 6603 if(strcmp(key, "mtime") == 0) 6604 mtime = val; 6605 if(strcmp(key, "name") == 0) 6606 name = val; 6607 } 6608 if(mode != nullptr){ 6609 mode_value = strtoul(mode, &endp, 8); 6610 if(*endp != '\0') 6611 outs() << "(mode: \"" << mode << "\" contains non-octal chars) "; 6612 if(strcmp(type, "file") == 0) 6613 mode_value |= S_IFREG; 6614 PrintModeVerbose(mode_value); 6615 outs() << " "; 6616 } 6617 if(user != nullptr) 6618 outs() << format("%10s/", user); 6619 if(group != nullptr) 6620 outs() << format("%-10s ", group); 6621 if(size != nullptr) 6622 outs() << format("%7s ", size); 6623 if(mtime != nullptr){ 6624 for(m = mtime; *m != 'T' && *m != '\0'; m++) 6625 outs() << *m; 6626 if(*m == 'T') 6627 m++; 6628 outs() << " "; 6629 for( ; *m != 'Z' && *m != '\0'; m++) 6630 outs() << *m; 6631 outs() << " "; 6632 } 6633 if(name != nullptr) 6634 outs() << name; 6635 outs() << "\n"; 6636 } 6637 } 6638 6639 static void DumpBitcodeSection(MachOObjectFile *O, const char *sect, 6640 uint32_t size, bool verbose, 6641 bool PrintXarHeader, bool PrintXarFileHeaders, 6642 std::string XarMemberName) { 6643 if(size < sizeof(struct xar_header)) { 6644 outs() << "size of (__LLVM,__bundle) section too small (smaller than size " 6645 "of struct xar_header)\n"; 6646 return; 6647 } 6648 struct xar_header XarHeader; 6649 memcpy(&XarHeader, sect, sizeof(struct xar_header)); 6650 if (sys::IsLittleEndianHost) 6651 swapStruct(XarHeader); 6652 if (PrintXarHeader) { 6653 if (!XarMemberName.empty()) 6654 outs() << "In xar member " << XarMemberName << ": "; 6655 else 6656 outs() << "For (__LLVM,__bundle) section: "; 6657 outs() << "xar header\n"; 6658 if (XarHeader.magic == XAR_HEADER_MAGIC) 6659 outs() << " magic XAR_HEADER_MAGIC\n"; 6660 else 6661 outs() << " magic " 6662 << format_hex(XarHeader.magic, 10, true) 6663 << " (not XAR_HEADER_MAGIC)\n"; 6664 outs() << " size " << XarHeader.size << "\n"; 6665 outs() << " version " << XarHeader.version << "\n"; 6666 outs() << " toc_length_compressed " << XarHeader.toc_length_compressed 6667 << "\n"; 6668 outs() << "toc_length_uncompressed " << XarHeader.toc_length_uncompressed 6669 << "\n"; 6670 outs() << " cksum_alg "; 6671 switch (XarHeader.cksum_alg) { 6672 case XAR_CKSUM_NONE: 6673 outs() << "XAR_CKSUM_NONE\n"; 6674 break; 6675 case XAR_CKSUM_SHA1: 6676 outs() << "XAR_CKSUM_SHA1\n"; 6677 break; 6678 case XAR_CKSUM_MD5: 6679 outs() << "XAR_CKSUM_MD5\n"; 6680 break; 6681 #ifdef XAR_CKSUM_SHA256 6682 case XAR_CKSUM_SHA256: 6683 outs() << "XAR_CKSUM_SHA256\n"; 6684 break; 6685 #endif 6686 #ifdef XAR_CKSUM_SHA512 6687 case XAR_CKSUM_SHA512: 6688 outs() << "XAR_CKSUM_SHA512\n"; 6689 break; 6690 #endif 6691 default: 6692 outs() << XarHeader.cksum_alg << "\n"; 6693 } 6694 } 6695 6696 SmallString<128> XarFilename; 6697 int FD; 6698 std::error_code XarEC = 6699 sys::fs::createTemporaryFile("llvm-objdump", "xar", FD, XarFilename); 6700 if (XarEC) { 6701 WithColor::error(errs(), "llvm-objdump") << XarEC.message() << "\n"; 6702 return; 6703 } 6704 ToolOutputFile XarFile(XarFilename, FD); 6705 raw_fd_ostream &XarOut = XarFile.os(); 6706 StringRef XarContents(sect, size); 6707 XarOut << XarContents; 6708 XarOut.close(); 6709 if (XarOut.has_error()) 6710 return; 6711 6712 ScopedXarFile xar(XarFilename.c_str(), READ); 6713 if (!xar) { 6714 WithColor::error(errs(), "llvm-objdump") 6715 << "can't create temporary xar archive " << XarFilename << "\n"; 6716 return; 6717 } 6718 6719 SmallString<128> TocFilename; 6720 std::error_code TocEC = 6721 sys::fs::createTemporaryFile("llvm-objdump", "toc", TocFilename); 6722 if (TocEC) { 6723 WithColor::error(errs(), "llvm-objdump") << TocEC.message() << "\n"; 6724 return; 6725 } 6726 xar_serialize(xar, TocFilename.c_str()); 6727 6728 if (PrintXarFileHeaders) { 6729 if (!XarMemberName.empty()) 6730 outs() << "In xar member " << XarMemberName << ": "; 6731 else 6732 outs() << "For (__LLVM,__bundle) section: "; 6733 outs() << "xar archive files:\n"; 6734 PrintXarFilesSummary(XarFilename.c_str(), xar); 6735 } 6736 6737 ErrorOr<std::unique_ptr<MemoryBuffer>> FileOrErr = 6738 MemoryBuffer::getFileOrSTDIN(TocFilename.c_str()); 6739 if (std::error_code EC = FileOrErr.getError()) { 6740 WithColor::error(errs(), "llvm-objdump") << EC.message() << "\n"; 6741 return; 6742 } 6743 std::unique_ptr<MemoryBuffer> &Buffer = FileOrErr.get(); 6744 6745 if (!XarMemberName.empty()) 6746 outs() << "In xar member " << XarMemberName << ": "; 6747 else 6748 outs() << "For (__LLVM,__bundle) section: "; 6749 outs() << "xar table of contents:\n"; 6750 outs() << Buffer->getBuffer() << "\n"; 6751 6752 // TODO: Go through the xar's files. 6753 ScopedXarIter xi; 6754 if(!xi){ 6755 WithColor::error(errs(), "llvm-objdump") 6756 << "can't obtain an xar iterator for xar archive " 6757 << XarFilename.c_str() << "\n"; 6758 return; 6759 } 6760 for(xar_file_t xf = xar_file_first(xar, xi); xf; xf = xar_file_next(xi)){ 6761 const char *key; 6762 const char *member_name, *member_type, *member_size_string; 6763 size_t member_size; 6764 6765 ScopedXarIter xp; 6766 if(!xp){ 6767 WithColor::error(errs(), "llvm-objdump") 6768 << "can't obtain an xar iterator for xar archive " 6769 << XarFilename.c_str() << "\n"; 6770 return; 6771 } 6772 member_name = NULL; 6773 member_type = NULL; 6774 member_size_string = NULL; 6775 for(key = xar_prop_first(xf, xp); key; key = xar_prop_next(xp)){ 6776 const char *val = nullptr; 6777 xar_prop_get(xf, key, &val); 6778 #if 0 // Useful for debugging. 6779 outs() << "key: " << key << " value: " << val << "\n"; 6780 #endif 6781 if (strcmp(key, "name") == 0) 6782 member_name = val; 6783 if (strcmp(key, "type") == 0) 6784 member_type = val; 6785 if (strcmp(key, "data/size") == 0) 6786 member_size_string = val; 6787 } 6788 /* 6789 * If we find a file with a name, date/size and type properties 6790 * and with the type being "file" see if that is a xar file. 6791 */ 6792 if (member_name != NULL && member_type != NULL && 6793 strcmp(member_type, "file") == 0 && 6794 member_size_string != NULL){ 6795 // Extract the file into a buffer. 6796 char *endptr; 6797 member_size = strtoul(member_size_string, &endptr, 10); 6798 if (*endptr == '\0' && member_size != 0) { 6799 char *buffer; 6800 if (xar_extract_tobuffersz(xar, xf, &buffer, &member_size) == 0) { 6801 #if 0 // Useful for debugging. 6802 outs() << "xar member: " << member_name << " extracted\n"; 6803 #endif 6804 // Set the XarMemberName we want to see printed in the header. 6805 std::string OldXarMemberName; 6806 // If XarMemberName is already set this is nested. So 6807 // save the old name and create the nested name. 6808 if (!XarMemberName.empty()) { 6809 OldXarMemberName = XarMemberName; 6810 XarMemberName = 6811 (Twine("[") + XarMemberName + "]" + member_name).str(); 6812 } else { 6813 OldXarMemberName = ""; 6814 XarMemberName = member_name; 6815 } 6816 // See if this is could be a xar file (nested). 6817 if (member_size >= sizeof(struct xar_header)) { 6818 #if 0 // Useful for debugging. 6819 outs() << "could be a xar file: " << member_name << "\n"; 6820 #endif 6821 memcpy((char *)&XarHeader, buffer, sizeof(struct xar_header)); 6822 if (sys::IsLittleEndianHost) 6823 swapStruct(XarHeader); 6824 if (XarHeader.magic == XAR_HEADER_MAGIC) 6825 DumpBitcodeSection(O, buffer, member_size, verbose, 6826 PrintXarHeader, PrintXarFileHeaders, 6827 XarMemberName); 6828 } 6829 XarMemberName = OldXarMemberName; 6830 delete buffer; 6831 } 6832 } 6833 } 6834 } 6835 } 6836 #endif // defined(HAVE_LIBXAR) 6837 6838 static void printObjcMetaData(MachOObjectFile *O, bool verbose) { 6839 if (O->is64Bit()) 6840 printObjc2_64bit_MetaData(O, verbose); 6841 else { 6842 MachO::mach_header H; 6843 H = O->getHeader(); 6844 if (H.cputype == MachO::CPU_TYPE_ARM) 6845 printObjc2_32bit_MetaData(O, verbose); 6846 else { 6847 // This is the 32-bit non-arm cputype case. Which is normally 6848 // the first Objective-C ABI. But it may be the case of a 6849 // binary for the iOS simulator which is the second Objective-C 6850 // ABI. In that case printObjc1_32bit_MetaData() will determine that 6851 // and return false. 6852 if (!printObjc1_32bit_MetaData(O, verbose)) 6853 printObjc2_32bit_MetaData(O, verbose); 6854 } 6855 } 6856 } 6857 6858 // GuessLiteralPointer returns a string which for the item in the Mach-O file 6859 // for the address passed in as ReferenceValue for printing as a comment with 6860 // the instruction and also returns the corresponding type of that item 6861 // indirectly through ReferenceType. 6862 // 6863 // If ReferenceValue is an address of literal cstring then a pointer to the 6864 // cstring is returned and ReferenceType is set to 6865 // LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr . 6866 // 6867 // If ReferenceValue is an address of an Objective-C CFString, Selector ref or 6868 // Class ref that name is returned and the ReferenceType is set accordingly. 6869 // 6870 // Lastly, literals which are Symbol address in a literal pool are looked for 6871 // and if found the symbol name is returned and ReferenceType is set to 6872 // LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr . 6873 // 6874 // If there is no item in the Mach-O file for the address passed in as 6875 // ReferenceValue nullptr is returned and ReferenceType is unchanged. 6876 static const char *GuessLiteralPointer(uint64_t ReferenceValue, 6877 uint64_t ReferencePC, 6878 uint64_t *ReferenceType, 6879 struct DisassembleInfo *info) { 6880 // First see if there is an external relocation entry at the ReferencePC. 6881 if (info->O->getHeader().filetype == MachO::MH_OBJECT) { 6882 uint64_t sect_addr = info->S.getAddress(); 6883 uint64_t sect_offset = ReferencePC - sect_addr; 6884 bool reloc_found = false; 6885 DataRefImpl Rel; 6886 MachO::any_relocation_info RE; 6887 bool isExtern = false; 6888 SymbolRef Symbol; 6889 for (const RelocationRef &Reloc : info->S.relocations()) { 6890 uint64_t RelocOffset = Reloc.getOffset(); 6891 if (RelocOffset == sect_offset) { 6892 Rel = Reloc.getRawDataRefImpl(); 6893 RE = info->O->getRelocation(Rel); 6894 if (info->O->isRelocationScattered(RE)) 6895 continue; 6896 isExtern = info->O->getPlainRelocationExternal(RE); 6897 if (isExtern) { 6898 symbol_iterator RelocSym = Reloc.getSymbol(); 6899 Symbol = *RelocSym; 6900 } 6901 reloc_found = true; 6902 break; 6903 } 6904 } 6905 // If there is an external relocation entry for a symbol in a section 6906 // then used that symbol's value for the value of the reference. 6907 if (reloc_found && isExtern) { 6908 if (info->O->getAnyRelocationPCRel(RE)) { 6909 unsigned Type = info->O->getAnyRelocationType(RE); 6910 if (Type == MachO::X86_64_RELOC_SIGNED) { 6911 ReferenceValue = Symbol.getValue(); 6912 } 6913 } 6914 } 6915 } 6916 6917 // Look for literals such as Objective-C CFStrings refs, Selector refs, 6918 // Message refs and Class refs. 6919 bool classref, selref, msgref, cfstring; 6920 uint64_t pointer_value = GuessPointerPointer(ReferenceValue, info, classref, 6921 selref, msgref, cfstring); 6922 if (classref && pointer_value == 0) { 6923 // Note the ReferenceValue is a pointer into the __objc_classrefs section. 6924 // And the pointer_value in that section is typically zero as it will be 6925 // set by dyld as part of the "bind information". 6926 const char *name = get_dyld_bind_info_symbolname(ReferenceValue, info); 6927 if (name != nullptr) { 6928 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref; 6929 const char *class_name = strrchr(name, '$'); 6930 if (class_name != nullptr && class_name[1] == '_' && 6931 class_name[2] != '\0') { 6932 info->class_name = class_name + 2; 6933 return name; 6934 } 6935 } 6936 } 6937 6938 if (classref) { 6939 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Class_Ref; 6940 const char *name = 6941 get_objc2_64bit_class_name(pointer_value, ReferenceValue, info); 6942 if (name != nullptr) 6943 info->class_name = name; 6944 else 6945 name = "bad class ref"; 6946 return name; 6947 } 6948 6949 if (cfstring) { 6950 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_CFString_Ref; 6951 const char *name = get_objc2_64bit_cfstring_name(ReferenceValue, info); 6952 return name; 6953 } 6954 6955 if (selref && pointer_value == 0) 6956 pointer_value = get_objc2_64bit_selref(ReferenceValue, info); 6957 6958 if (pointer_value != 0) 6959 ReferenceValue = pointer_value; 6960 6961 const char *name = GuessCstringPointer(ReferenceValue, info); 6962 if (name) { 6963 if (pointer_value != 0 && selref) { 6964 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Selector_Ref; 6965 info->selector_name = name; 6966 } else if (pointer_value != 0 && msgref) { 6967 info->class_name = nullptr; 6968 *ReferenceType = LLVMDisassembler_ReferenceType_Out_Objc_Message_Ref; 6969 info->selector_name = name; 6970 } else 6971 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_CstrAddr; 6972 return name; 6973 } 6974 6975 // Lastly look for an indirect symbol with this ReferenceValue which is in 6976 // a literal pool. If found return that symbol name. 6977 name = GuessIndirectSymbol(ReferenceValue, info); 6978 if (name) { 6979 *ReferenceType = LLVMDisassembler_ReferenceType_Out_LitPool_SymAddr; 6980 return name; 6981 } 6982 6983 return nullptr; 6984 } 6985 6986 // SymbolizerSymbolLookUp is the symbol lookup function passed when creating 6987 // the Symbolizer. It looks up the ReferenceValue using the info passed via the 6988 // pointer to the struct DisassembleInfo that was passed when MCSymbolizer 6989 // is created and returns the symbol name that matches the ReferenceValue or 6990 // nullptr if none. The ReferenceType is passed in for the IN type of 6991 // reference the instruction is making from the values in defined in the header 6992 // "llvm-c/Disassembler.h". On return the ReferenceType can set to a specific 6993 // Out type and the ReferenceName will also be set which is added as a comment 6994 // to the disassembled instruction. 6995 // 6996 // If the symbol name is a C++ mangled name then the demangled name is 6997 // returned through ReferenceName and ReferenceType is set to 6998 // LLVMDisassembler_ReferenceType_DeMangled_Name . 6999 // 7000 // When this is called to get a symbol name for a branch target then the 7001 // ReferenceType will be LLVMDisassembler_ReferenceType_In_Branch and then 7002 // SymbolValue will be looked for in the indirect symbol table to determine if 7003 // it is an address for a symbol stub. If so then the symbol name for that 7004 // stub is returned indirectly through ReferenceName and then ReferenceType is 7005 // set to LLVMDisassembler_ReferenceType_Out_SymbolStub. 7006 // 7007 // When this is called with an value loaded via a PC relative load then 7008 // ReferenceType will be LLVMDisassembler_ReferenceType_In_PCrel_Load then the 7009 // SymbolValue is checked to be an address of literal pointer, symbol pointer, 7010 // or an Objective-C meta data reference. If so the output ReferenceType is 7011 // set to correspond to that as well as setting the ReferenceName. 7012 static const char *SymbolizerSymbolLookUp(void *DisInfo, 7013 uint64_t ReferenceValue, 7014 uint64_t *ReferenceType, 7015 uint64_t ReferencePC, 7016 const char **ReferenceName) { 7017 struct DisassembleInfo *info = (struct DisassembleInfo *)DisInfo; 7018 // If no verbose symbolic information is wanted then just return nullptr. 7019 if (!info->verbose) { 7020 *ReferenceName = nullptr; 7021 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7022 return nullptr; 7023 } 7024 7025 const char *SymbolName = GuessSymbolName(ReferenceValue, info->AddrMap); 7026 7027 if (*ReferenceType == LLVMDisassembler_ReferenceType_In_Branch) { 7028 *ReferenceName = GuessIndirectSymbol(ReferenceValue, info); 7029 if (*ReferenceName != nullptr) { 7030 method_reference(info, ReferenceType, ReferenceName); 7031 if (*ReferenceType != LLVMDisassembler_ReferenceType_Out_Objc_Message) 7032 *ReferenceType = LLVMDisassembler_ReferenceType_Out_SymbolStub; 7033 } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) { 7034 if (info->demangled_name != nullptr) 7035 free(info->demangled_name); 7036 int status; 7037 info->demangled_name = 7038 itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status); 7039 if (info->demangled_name != nullptr) { 7040 *ReferenceName = info->demangled_name; 7041 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name; 7042 } else 7043 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7044 } else 7045 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7046 } else if (*ReferenceType == LLVMDisassembler_ReferenceType_In_PCrel_Load) { 7047 *ReferenceName = 7048 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info); 7049 if (*ReferenceName) 7050 method_reference(info, ReferenceType, ReferenceName); 7051 else 7052 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7053 // If this is arm64 and the reference is an adrp instruction save the 7054 // instruction, passed in ReferenceValue and the address of the instruction 7055 // for use later if we see and add immediate instruction. 7056 } else if (info->O->getArch() == Triple::aarch64 && 7057 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) { 7058 info->adrp_inst = ReferenceValue; 7059 info->adrp_addr = ReferencePC; 7060 SymbolName = nullptr; 7061 *ReferenceName = nullptr; 7062 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7063 // If this is arm64 and reference is an add immediate instruction and we 7064 // have 7065 // seen an adrp instruction just before it and the adrp's Xd register 7066 // matches 7067 // this add's Xn register reconstruct the value being referenced and look to 7068 // see if it is a literal pointer. Note the add immediate instruction is 7069 // passed in ReferenceValue. 7070 } else if (info->O->getArch() == Triple::aarch64 && 7071 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADDXri && 7072 ReferencePC - 4 == info->adrp_addr && 7073 (info->adrp_inst & 0x9f000000) == 0x90000000 && 7074 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) { 7075 uint32_t addxri_inst; 7076 uint64_t adrp_imm, addxri_imm; 7077 7078 adrp_imm = 7079 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3); 7080 if (info->adrp_inst & 0x0200000) 7081 adrp_imm |= 0xfffffffffc000000LL; 7082 7083 addxri_inst = ReferenceValue; 7084 addxri_imm = (addxri_inst >> 10) & 0xfff; 7085 if (((addxri_inst >> 22) & 0x3) == 1) 7086 addxri_imm <<= 12; 7087 7088 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) + 7089 (adrp_imm << 12) + addxri_imm; 7090 7091 *ReferenceName = 7092 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info); 7093 if (*ReferenceName == nullptr) 7094 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7095 // If this is arm64 and the reference is a load register instruction and we 7096 // have seen an adrp instruction just before it and the adrp's Xd register 7097 // matches this add's Xn register reconstruct the value being referenced and 7098 // look to see if it is a literal pointer. Note the load register 7099 // instruction is passed in ReferenceValue. 7100 } else if (info->O->getArch() == Triple::aarch64 && 7101 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXui && 7102 ReferencePC - 4 == info->adrp_addr && 7103 (info->adrp_inst & 0x9f000000) == 0x90000000 && 7104 (info->adrp_inst & 0x1f) == ((ReferenceValue >> 5) & 0x1f)) { 7105 uint32_t ldrxui_inst; 7106 uint64_t adrp_imm, ldrxui_imm; 7107 7108 adrp_imm = 7109 ((info->adrp_inst & 0x00ffffe0) >> 3) | ((info->adrp_inst >> 29) & 0x3); 7110 if (info->adrp_inst & 0x0200000) 7111 adrp_imm |= 0xfffffffffc000000LL; 7112 7113 ldrxui_inst = ReferenceValue; 7114 ldrxui_imm = (ldrxui_inst >> 10) & 0xfff; 7115 7116 ReferenceValue = (info->adrp_addr & 0xfffffffffffff000LL) + 7117 (adrp_imm << 12) + (ldrxui_imm << 3); 7118 7119 *ReferenceName = 7120 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info); 7121 if (*ReferenceName == nullptr) 7122 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7123 } 7124 // If this arm64 and is an load register (PC-relative) instruction the 7125 // ReferenceValue is the PC plus the immediate value. 7126 else if (info->O->getArch() == Triple::aarch64 && 7127 (*ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_LDRXl || 7128 *ReferenceType == LLVMDisassembler_ReferenceType_In_ARM64_ADR)) { 7129 *ReferenceName = 7130 GuessLiteralPointer(ReferenceValue, ReferencePC, ReferenceType, info); 7131 if (*ReferenceName == nullptr) 7132 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7133 } else if (SymbolName != nullptr && strncmp(SymbolName, "__Z", 3) == 0) { 7134 if (info->demangled_name != nullptr) 7135 free(info->demangled_name); 7136 int status; 7137 info->demangled_name = 7138 itaniumDemangle(SymbolName + 1, nullptr, nullptr, &status); 7139 if (info->demangled_name != nullptr) { 7140 *ReferenceName = info->demangled_name; 7141 *ReferenceType = LLVMDisassembler_ReferenceType_DeMangled_Name; 7142 } 7143 } 7144 else { 7145 *ReferenceName = nullptr; 7146 *ReferenceType = LLVMDisassembler_ReferenceType_InOut_None; 7147 } 7148 7149 return SymbolName; 7150 } 7151 7152 /// Emits the comments that are stored in the CommentStream. 7153 /// Each comment in the CommentStream must end with a newline. 7154 static void emitComments(raw_svector_ostream &CommentStream, 7155 SmallString<128> &CommentsToEmit, 7156 formatted_raw_ostream &FormattedOS, 7157 const MCAsmInfo &MAI) { 7158 // Flush the stream before taking its content. 7159 StringRef Comments = CommentsToEmit.str(); 7160 // Get the default information for printing a comment. 7161 StringRef CommentBegin = MAI.getCommentString(); 7162 unsigned CommentColumn = MAI.getCommentColumn(); 7163 bool IsFirst = true; 7164 while (!Comments.empty()) { 7165 if (!IsFirst) 7166 FormattedOS << '\n'; 7167 // Emit a line of comments. 7168 FormattedOS.PadToColumn(CommentColumn); 7169 size_t Position = Comments.find('\n'); 7170 FormattedOS << CommentBegin << ' ' << Comments.substr(0, Position); 7171 // Move after the newline character. 7172 Comments = Comments.substr(Position + 1); 7173 IsFirst = false; 7174 } 7175 FormattedOS.flush(); 7176 7177 // Tell the comment stream that the vector changed underneath it. 7178 CommentsToEmit.clear(); 7179 } 7180 7181 static void DisassembleMachO(StringRef Filename, MachOObjectFile *MachOOF, 7182 StringRef DisSegName, StringRef DisSectName) { 7183 const char *McpuDefault = nullptr; 7184 const Target *ThumbTarget = nullptr; 7185 const Target *TheTarget = GetTarget(MachOOF, &McpuDefault, &ThumbTarget); 7186 if (!TheTarget) { 7187 // GetTarget prints out stuff. 7188 return; 7189 } 7190 std::string MachOMCPU; 7191 if (MCPU.empty() && McpuDefault) 7192 MachOMCPU = McpuDefault; 7193 else 7194 MachOMCPU = MCPU; 7195 7196 std::unique_ptr<const MCInstrInfo> InstrInfo(TheTarget->createMCInstrInfo()); 7197 std::unique_ptr<const MCInstrInfo> ThumbInstrInfo; 7198 if (ThumbTarget) 7199 ThumbInstrInfo.reset(ThumbTarget->createMCInstrInfo()); 7200 7201 // Package up features to be passed to target/subtarget 7202 std::string FeaturesStr; 7203 if (!MAttrs.empty()) { 7204 SubtargetFeatures Features; 7205 for (unsigned i = 0; i != MAttrs.size(); ++i) 7206 Features.AddFeature(MAttrs[i]); 7207 FeaturesStr = Features.getString(); 7208 } 7209 7210 MCTargetOptions MCOptions; 7211 // Set up disassembler. 7212 std::unique_ptr<const MCRegisterInfo> MRI( 7213 TheTarget->createMCRegInfo(TripleName)); 7214 std::unique_ptr<const MCAsmInfo> AsmInfo( 7215 TheTarget->createMCAsmInfo(*MRI, TripleName, MCOptions)); 7216 std::unique_ptr<const MCSubtargetInfo> STI( 7217 TheTarget->createMCSubtargetInfo(TripleName, MachOMCPU, FeaturesStr)); 7218 MCContext Ctx(AsmInfo.get(), MRI.get(), nullptr); 7219 std::unique_ptr<MCDisassembler> DisAsm( 7220 TheTarget->createMCDisassembler(*STI, Ctx)); 7221 std::unique_ptr<MCSymbolizer> Symbolizer; 7222 struct DisassembleInfo SymbolizerInfo(nullptr, nullptr, nullptr, false); 7223 std::unique_ptr<MCRelocationInfo> RelInfo( 7224 TheTarget->createMCRelocationInfo(TripleName, Ctx)); 7225 if (RelInfo) { 7226 Symbolizer.reset(TheTarget->createMCSymbolizer( 7227 TripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp, 7228 &SymbolizerInfo, &Ctx, std::move(RelInfo))); 7229 DisAsm->setSymbolizer(std::move(Symbolizer)); 7230 } 7231 int AsmPrinterVariant = AsmInfo->getAssemblerDialect(); 7232 std::unique_ptr<MCInstPrinter> IP(TheTarget->createMCInstPrinter( 7233 Triple(TripleName), AsmPrinterVariant, *AsmInfo, *InstrInfo, *MRI)); 7234 // Set the display preference for hex vs. decimal immediates. 7235 IP->setPrintImmHex(PrintImmHex); 7236 // Comment stream and backing vector. 7237 SmallString<128> CommentsToEmit; 7238 raw_svector_ostream CommentStream(CommentsToEmit); 7239 // FIXME: Setting the CommentStream in the InstPrinter is problematic in that 7240 // if it is done then arm64 comments for string literals don't get printed 7241 // and some constant get printed instead and not setting it causes intel 7242 // (32-bit and 64-bit) comments printed with different spacing before the 7243 // comment causing different diffs with the 'C' disassembler library API. 7244 // IP->setCommentStream(CommentStream); 7245 7246 if (!AsmInfo || !STI || !DisAsm || !IP) { 7247 WithColor::error(errs(), "llvm-objdump") 7248 << "couldn't initialize disassembler for target " << TripleName << '\n'; 7249 return; 7250 } 7251 7252 // Set up separate thumb disassembler if needed. 7253 std::unique_ptr<const MCRegisterInfo> ThumbMRI; 7254 std::unique_ptr<const MCAsmInfo> ThumbAsmInfo; 7255 std::unique_ptr<const MCSubtargetInfo> ThumbSTI; 7256 std::unique_ptr<MCDisassembler> ThumbDisAsm; 7257 std::unique_ptr<MCInstPrinter> ThumbIP; 7258 std::unique_ptr<MCContext> ThumbCtx; 7259 std::unique_ptr<MCSymbolizer> ThumbSymbolizer; 7260 struct DisassembleInfo ThumbSymbolizerInfo(nullptr, nullptr, nullptr, false); 7261 std::unique_ptr<MCRelocationInfo> ThumbRelInfo; 7262 if (ThumbTarget) { 7263 ThumbMRI.reset(ThumbTarget->createMCRegInfo(ThumbTripleName)); 7264 ThumbAsmInfo.reset( 7265 ThumbTarget->createMCAsmInfo(*ThumbMRI, ThumbTripleName, MCOptions)); 7266 ThumbSTI.reset( 7267 ThumbTarget->createMCSubtargetInfo(ThumbTripleName, MachOMCPU, 7268 FeaturesStr)); 7269 ThumbCtx.reset(new MCContext(ThumbAsmInfo.get(), ThumbMRI.get(), nullptr)); 7270 ThumbDisAsm.reset(ThumbTarget->createMCDisassembler(*ThumbSTI, *ThumbCtx)); 7271 MCContext *PtrThumbCtx = ThumbCtx.get(); 7272 ThumbRelInfo.reset( 7273 ThumbTarget->createMCRelocationInfo(ThumbTripleName, *PtrThumbCtx)); 7274 if (ThumbRelInfo) { 7275 ThumbSymbolizer.reset(ThumbTarget->createMCSymbolizer( 7276 ThumbTripleName, SymbolizerGetOpInfo, SymbolizerSymbolLookUp, 7277 &ThumbSymbolizerInfo, PtrThumbCtx, std::move(ThumbRelInfo))); 7278 ThumbDisAsm->setSymbolizer(std::move(ThumbSymbolizer)); 7279 } 7280 int ThumbAsmPrinterVariant = ThumbAsmInfo->getAssemblerDialect(); 7281 ThumbIP.reset(ThumbTarget->createMCInstPrinter( 7282 Triple(ThumbTripleName), ThumbAsmPrinterVariant, *ThumbAsmInfo, 7283 *ThumbInstrInfo, *ThumbMRI)); 7284 // Set the display preference for hex vs. decimal immediates. 7285 ThumbIP->setPrintImmHex(PrintImmHex); 7286 } 7287 7288 if (ThumbTarget && (!ThumbAsmInfo || !ThumbSTI || !ThumbDisAsm || !ThumbIP)) { 7289 WithColor::error(errs(), "llvm-objdump") 7290 << "couldn't initialize disassembler for target " << ThumbTripleName 7291 << '\n'; 7292 return; 7293 } 7294 7295 MachO::mach_header Header = MachOOF->getHeader(); 7296 7297 // FIXME: Using the -cfg command line option, this code used to be able to 7298 // annotate relocations with the referenced symbol's name, and if this was 7299 // inside a __[cf]string section, the data it points to. This is now replaced 7300 // by the upcoming MCSymbolizer, which needs the appropriate setup done above. 7301 std::vector<SectionRef> Sections; 7302 std::vector<SymbolRef> Symbols; 7303 SmallVector<uint64_t, 8> FoundFns; 7304 uint64_t BaseSegmentAddress = 0; 7305 7306 getSectionsAndSymbols(MachOOF, Sections, Symbols, FoundFns, 7307 BaseSegmentAddress); 7308 7309 // Sort the symbols by address, just in case they didn't come in that way. 7310 llvm::sort(Symbols, SymbolSorter()); 7311 7312 // Build a data in code table that is sorted on by the address of each entry. 7313 uint64_t BaseAddress = 0; 7314 if (Header.filetype == MachO::MH_OBJECT) 7315 BaseAddress = Sections[0].getAddress(); 7316 else 7317 BaseAddress = BaseSegmentAddress; 7318 DiceTable Dices; 7319 for (dice_iterator DI = MachOOF->begin_dices(), DE = MachOOF->end_dices(); 7320 DI != DE; ++DI) { 7321 uint32_t Offset; 7322 DI->getOffset(Offset); 7323 Dices.push_back(std::make_pair(BaseAddress + Offset, *DI)); 7324 } 7325 array_pod_sort(Dices.begin(), Dices.end()); 7326 7327 // Try to find debug info and set up the DIContext for it. 7328 std::unique_ptr<DIContext> diContext; 7329 std::unique_ptr<Binary> DSYMBinary; 7330 std::unique_ptr<MemoryBuffer> DSYMBuf; 7331 if (UseDbg) { 7332 ObjectFile *DbgObj = MachOOF; 7333 7334 // A separate DSym file path was specified, parse it as a macho file, 7335 // get the sections and supply it to the section name parsing machinery. 7336 if (!DSYMFile.empty()) { 7337 std::string DSYMPath(DSYMFile); 7338 7339 // If DSYMPath is a .dSYM directory, append the Mach-O file. 7340 if (llvm::sys::fs::is_directory(DSYMPath) && 7341 llvm::sys::path::extension(DSYMPath) == ".dSYM") { 7342 SmallString<128> ShortName(llvm::sys::path::filename(DSYMPath)); 7343 llvm::sys::path::replace_extension(ShortName, ""); 7344 SmallString<1024> FullPath(DSYMPath); 7345 llvm::sys::path::append(FullPath, "Contents", "Resources", "DWARF", 7346 ShortName); 7347 DSYMPath = std::string(FullPath.str()); 7348 } 7349 7350 // Load the file. 7351 ErrorOr<std::unique_ptr<MemoryBuffer>> BufOrErr = 7352 MemoryBuffer::getFileOrSTDIN(DSYMPath); 7353 if (std::error_code EC = BufOrErr.getError()) { 7354 reportError(errorCodeToError(EC), DSYMPath); 7355 return; 7356 } 7357 7358 // We need to keep the file alive, because we're replacing DbgObj with it. 7359 DSYMBuf = std::move(BufOrErr.get()); 7360 7361 Expected<std::unique_ptr<Binary>> BinaryOrErr = 7362 createBinary(DSYMBuf.get()->getMemBufferRef()); 7363 if (!BinaryOrErr) { 7364 reportError(BinaryOrErr.takeError(), DSYMPath); 7365 return; 7366 } 7367 7368 // We need to keep the Binary alive with the buffer 7369 DSYMBinary = std::move(BinaryOrErr.get()); 7370 if (ObjectFile *O = dyn_cast<ObjectFile>(DSYMBinary.get())) { 7371 // this is a Mach-O object file, use it 7372 if (MachOObjectFile *MachDSYM = dyn_cast<MachOObjectFile>(&*O)) { 7373 DbgObj = MachDSYM; 7374 } 7375 else { 7376 WithColor::error(errs(), "llvm-objdump") 7377 << DSYMPath << " is not a Mach-O file type.\n"; 7378 return; 7379 } 7380 } 7381 else if (auto UB = dyn_cast<MachOUniversalBinary>(DSYMBinary.get())){ 7382 // this is a Universal Binary, find a Mach-O for this architecture 7383 uint32_t CPUType, CPUSubType; 7384 const char *ArchFlag; 7385 if (MachOOF->is64Bit()) { 7386 const MachO::mach_header_64 H_64 = MachOOF->getHeader64(); 7387 CPUType = H_64.cputype; 7388 CPUSubType = H_64.cpusubtype; 7389 } else { 7390 const MachO::mach_header H = MachOOF->getHeader(); 7391 CPUType = H.cputype; 7392 CPUSubType = H.cpusubtype; 7393 } 7394 Triple T = MachOObjectFile::getArchTriple(CPUType, CPUSubType, nullptr, 7395 &ArchFlag); 7396 Expected<std::unique_ptr<MachOObjectFile>> MachDSYM = 7397 UB->getMachOObjectForArch(ArchFlag); 7398 if (!MachDSYM) { 7399 reportError(MachDSYM.takeError(), DSYMPath); 7400 return; 7401 } 7402 7403 // We need to keep the Binary alive with the buffer 7404 DbgObj = &*MachDSYM.get(); 7405 DSYMBinary = std::move(*MachDSYM); 7406 } 7407 else { 7408 WithColor::error(errs(), "llvm-objdump") 7409 << DSYMPath << " is not a Mach-O or Universal file type.\n"; 7410 return; 7411 } 7412 } 7413 7414 // Setup the DIContext 7415 diContext = DWARFContext::create(*DbgObj); 7416 } 7417 7418 if (FilterSections.empty()) 7419 outs() << "(" << DisSegName << "," << DisSectName << ") section\n"; 7420 7421 for (unsigned SectIdx = 0; SectIdx != Sections.size(); SectIdx++) { 7422 Expected<StringRef> SecNameOrErr = Sections[SectIdx].getName(); 7423 if (!SecNameOrErr) { 7424 consumeError(SecNameOrErr.takeError()); 7425 continue; 7426 } 7427 if (*SecNameOrErr != DisSectName) 7428 continue; 7429 7430 DataRefImpl DR = Sections[SectIdx].getRawDataRefImpl(); 7431 7432 StringRef SegmentName = MachOOF->getSectionFinalSegmentName(DR); 7433 if (SegmentName != DisSegName) 7434 continue; 7435 7436 StringRef BytesStr = 7437 unwrapOrError(Sections[SectIdx].getContents(), Filename); 7438 ArrayRef<uint8_t> Bytes = arrayRefFromStringRef(BytesStr); 7439 uint64_t SectAddress = Sections[SectIdx].getAddress(); 7440 7441 bool symbolTableWorked = false; 7442 7443 // Create a map of symbol addresses to symbol names for use by 7444 // the SymbolizerSymbolLookUp() routine. 7445 SymbolAddressMap AddrMap; 7446 bool DisSymNameFound = false; 7447 for (const SymbolRef &Symbol : MachOOF->symbols()) { 7448 SymbolRef::Type ST = 7449 unwrapOrError(Symbol.getType(), MachOOF->getFileName()); 7450 if (ST == SymbolRef::ST_Function || ST == SymbolRef::ST_Data || 7451 ST == SymbolRef::ST_Other) { 7452 uint64_t Address = Symbol.getValue(); 7453 StringRef SymName = 7454 unwrapOrError(Symbol.getName(), MachOOF->getFileName()); 7455 AddrMap[Address] = SymName; 7456 if (!DisSymName.empty() && DisSymName == SymName) 7457 DisSymNameFound = true; 7458 } 7459 } 7460 if (!DisSymName.empty() && !DisSymNameFound) { 7461 outs() << "Can't find -dis-symname: " << DisSymName << "\n"; 7462 return; 7463 } 7464 // Set up the block of info used by the Symbolizer call backs. 7465 SymbolizerInfo.verbose = !NoSymbolicOperands; 7466 SymbolizerInfo.O = MachOOF; 7467 SymbolizerInfo.S = Sections[SectIdx]; 7468 SymbolizerInfo.AddrMap = &AddrMap; 7469 SymbolizerInfo.Sections = &Sections; 7470 // Same for the ThumbSymbolizer 7471 ThumbSymbolizerInfo.verbose = !NoSymbolicOperands; 7472 ThumbSymbolizerInfo.O = MachOOF; 7473 ThumbSymbolizerInfo.S = Sections[SectIdx]; 7474 ThumbSymbolizerInfo.AddrMap = &AddrMap; 7475 ThumbSymbolizerInfo.Sections = &Sections; 7476 7477 unsigned int Arch = MachOOF->getArch(); 7478 7479 // Skip all symbols if this is a stubs file. 7480 if (Bytes.empty()) 7481 return; 7482 7483 // If the section has symbols but no symbol at the start of the section 7484 // these are used to make sure the bytes before the first symbol are 7485 // disassembled. 7486 bool FirstSymbol = true; 7487 bool FirstSymbolAtSectionStart = true; 7488 7489 // Disassemble symbol by symbol. 7490 for (unsigned SymIdx = 0; SymIdx != Symbols.size(); SymIdx++) { 7491 StringRef SymName = 7492 unwrapOrError(Symbols[SymIdx].getName(), MachOOF->getFileName()); 7493 SymbolRef::Type ST = 7494 unwrapOrError(Symbols[SymIdx].getType(), MachOOF->getFileName()); 7495 if (ST != SymbolRef::ST_Function && ST != SymbolRef::ST_Data) 7496 continue; 7497 7498 // Make sure the symbol is defined in this section. 7499 bool containsSym = Sections[SectIdx].containsSymbol(Symbols[SymIdx]); 7500 if (!containsSym) { 7501 if (!DisSymName.empty() && DisSymName == SymName) { 7502 outs() << "-dis-symname: " << DisSymName << " not in the section\n"; 7503 return; 7504 } 7505 continue; 7506 } 7507 // The __mh_execute_header is special and we need to deal with that fact 7508 // this symbol is before the start of the (__TEXT,__text) section and at the 7509 // address of the start of the __TEXT segment. This is because this symbol 7510 // is an N_SECT symbol in the (__TEXT,__text) but its address is before the 7511 // start of the section in a standard MH_EXECUTE filetype. 7512 if (!DisSymName.empty() && DisSymName == "__mh_execute_header") { 7513 outs() << "-dis-symname: __mh_execute_header not in any section\n"; 7514 return; 7515 } 7516 // When this code is trying to disassemble a symbol at a time and in the 7517 // case there is only the __mh_execute_header symbol left as in a stripped 7518 // executable, we need to deal with this by ignoring this symbol so the 7519 // whole section is disassembled and this symbol is then not displayed. 7520 if (SymName == "__mh_execute_header" || SymName == "__mh_dylib_header" || 7521 SymName == "__mh_bundle_header" || SymName == "__mh_object_header" || 7522 SymName == "__mh_preload_header" || SymName == "__mh_dylinker_header") 7523 continue; 7524 7525 // If we are only disassembling one symbol see if this is that symbol. 7526 if (!DisSymName.empty() && DisSymName != SymName) 7527 continue; 7528 7529 // Start at the address of the symbol relative to the section's address. 7530 uint64_t SectSize = Sections[SectIdx].getSize(); 7531 uint64_t Start = Symbols[SymIdx].getValue(); 7532 uint64_t SectionAddress = Sections[SectIdx].getAddress(); 7533 Start -= SectionAddress; 7534 7535 if (Start > SectSize) { 7536 outs() << "section data ends, " << SymName 7537 << " lies outside valid range\n"; 7538 return; 7539 } 7540 7541 // Stop disassembling either at the beginning of the next symbol or at 7542 // the end of the section. 7543 bool containsNextSym = false; 7544 uint64_t NextSym = 0; 7545 uint64_t NextSymIdx = SymIdx + 1; 7546 while (Symbols.size() > NextSymIdx) { 7547 SymbolRef::Type NextSymType = unwrapOrError( 7548 Symbols[NextSymIdx].getType(), MachOOF->getFileName()); 7549 if (NextSymType == SymbolRef::ST_Function) { 7550 containsNextSym = 7551 Sections[SectIdx].containsSymbol(Symbols[NextSymIdx]); 7552 NextSym = Symbols[NextSymIdx].getValue(); 7553 NextSym -= SectionAddress; 7554 break; 7555 } 7556 ++NextSymIdx; 7557 } 7558 7559 uint64_t End = containsNextSym ? std::min(NextSym, SectSize) : SectSize; 7560 uint64_t Size; 7561 7562 symbolTableWorked = true; 7563 7564 DataRefImpl Symb = Symbols[SymIdx].getRawDataRefImpl(); 7565 bool IsThumb = MachOOF->getSymbolFlags(Symb) & SymbolRef::SF_Thumb; 7566 7567 // We only need the dedicated Thumb target if there's a real choice 7568 // (i.e. we're not targeting M-class) and the function is Thumb. 7569 bool UseThumbTarget = IsThumb && ThumbTarget; 7570 7571 // If we are not specifying a symbol to start disassembly with and this 7572 // is the first symbol in the section but not at the start of the section 7573 // then move the disassembly index to the start of the section and 7574 // don't print the symbol name just yet. This is so the bytes before the 7575 // first symbol are disassembled. 7576 uint64_t SymbolStart = Start; 7577 if (DisSymName.empty() && FirstSymbol && Start != 0) { 7578 FirstSymbolAtSectionStart = false; 7579 Start = 0; 7580 } 7581 else 7582 outs() << SymName << ":\n"; 7583 7584 DILineInfo lastLine; 7585 for (uint64_t Index = Start; Index < End; Index += Size) { 7586 MCInst Inst; 7587 7588 // If this is the first symbol in the section and it was not at the 7589 // start of the section, see if we are at its Index now and if so print 7590 // the symbol name. 7591 if (FirstSymbol && !FirstSymbolAtSectionStart && Index == SymbolStart) 7592 outs() << SymName << ":\n"; 7593 7594 uint64_t PC = SectAddress + Index; 7595 if (!NoLeadingAddr) { 7596 if (FullLeadingAddr) { 7597 if (MachOOF->is64Bit()) 7598 outs() << format("%016" PRIx64, PC); 7599 else 7600 outs() << format("%08" PRIx64, PC); 7601 } else { 7602 outs() << format("%8" PRIx64 ":", PC); 7603 } 7604 } 7605 if (!NoShowRawInsn || Arch == Triple::arm) 7606 outs() << "\t"; 7607 7608 if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, Size)) 7609 continue; 7610 7611 SmallVector<char, 64> AnnotationsBytes; 7612 raw_svector_ostream Annotations(AnnotationsBytes); 7613 7614 bool gotInst; 7615 if (UseThumbTarget) 7616 gotInst = ThumbDisAsm->getInstruction(Inst, Size, Bytes.slice(Index), 7617 PC, Annotations); 7618 else 7619 gotInst = DisAsm->getInstruction(Inst, Size, Bytes.slice(Index), PC, 7620 Annotations); 7621 if (gotInst) { 7622 if (!NoShowRawInsn || Arch == Triple::arm) { 7623 dumpBytes(makeArrayRef(Bytes.data() + Index, Size), outs()); 7624 } 7625 formatted_raw_ostream FormattedOS(outs()); 7626 StringRef AnnotationsStr = Annotations.str(); 7627 if (UseThumbTarget) 7628 ThumbIP->printInst(&Inst, PC, AnnotationsStr, *ThumbSTI, 7629 FormattedOS); 7630 else 7631 IP->printInst(&Inst, PC, AnnotationsStr, *STI, FormattedOS); 7632 emitComments(CommentStream, CommentsToEmit, FormattedOS, *AsmInfo); 7633 7634 // Print debug info. 7635 if (diContext) { 7636 DILineInfo dli = diContext->getLineInfoForAddress({PC, SectIdx}); 7637 // Print valid line info if it changed. 7638 if (dli != lastLine && dli.Line != 0) 7639 outs() << "\t## " << dli.FileName << ':' << dli.Line << ':' 7640 << dli.Column; 7641 lastLine = dli; 7642 } 7643 outs() << "\n"; 7644 } else { 7645 if (MachOOF->getArchTriple().isX86()) { 7646 outs() << format("\t.byte 0x%02x #bad opcode\n", 7647 *(Bytes.data() + Index) & 0xff); 7648 Size = 1; // skip exactly one illegible byte and move on. 7649 } else if (Arch == Triple::aarch64 || 7650 (Arch == Triple::arm && !IsThumb)) { 7651 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) | 7652 (*(Bytes.data() + Index + 1) & 0xff) << 8 | 7653 (*(Bytes.data() + Index + 2) & 0xff) << 16 | 7654 (*(Bytes.data() + Index + 3) & 0xff) << 24; 7655 outs() << format("\t.long\t0x%08x\n", opcode); 7656 Size = 4; 7657 } else if (Arch == Triple::arm) { 7658 assert(IsThumb && "ARM mode should have been dealt with above"); 7659 uint32_t opcode = (*(Bytes.data() + Index) & 0xff) | 7660 (*(Bytes.data() + Index + 1) & 0xff) << 8; 7661 outs() << format("\t.short\t0x%04x\n", opcode); 7662 Size = 2; 7663 } else{ 7664 WithColor::warning(errs(), "llvm-objdump") 7665 << "invalid instruction encoding\n"; 7666 if (Size == 0) 7667 Size = 1; // skip illegible bytes 7668 } 7669 } 7670 } 7671 // Now that we are done disassembled the first symbol set the bool that 7672 // were doing this to false. 7673 FirstSymbol = false; 7674 } 7675 if (!symbolTableWorked) { 7676 // Reading the symbol table didn't work, disassemble the whole section. 7677 uint64_t SectAddress = Sections[SectIdx].getAddress(); 7678 uint64_t SectSize = Sections[SectIdx].getSize(); 7679 uint64_t InstSize; 7680 for (uint64_t Index = 0; Index < SectSize; Index += InstSize) { 7681 MCInst Inst; 7682 7683 uint64_t PC = SectAddress + Index; 7684 7685 if (DumpAndSkipDataInCode(PC, Bytes.data() + Index, Dices, InstSize)) 7686 continue; 7687 7688 SmallVector<char, 64> AnnotationsBytes; 7689 raw_svector_ostream Annotations(AnnotationsBytes); 7690 if (DisAsm->getInstruction(Inst, InstSize, Bytes.slice(Index), PC, 7691 Annotations)) { 7692 if (!NoLeadingAddr) { 7693 if (FullLeadingAddr) { 7694 if (MachOOF->is64Bit()) 7695 outs() << format("%016" PRIx64, PC); 7696 else 7697 outs() << format("%08" PRIx64, PC); 7698 } else { 7699 outs() << format("%8" PRIx64 ":", PC); 7700 } 7701 } 7702 if (!NoShowRawInsn || Arch == Triple::arm) { 7703 outs() << "\t"; 7704 dumpBytes(makeArrayRef(Bytes.data() + Index, InstSize), outs()); 7705 } 7706 StringRef AnnotationsStr = Annotations.str(); 7707 IP->printInst(&Inst, PC, AnnotationsStr, *STI, outs()); 7708 outs() << "\n"; 7709 } else { 7710 if (MachOOF->getArchTriple().isX86()) { 7711 outs() << format("\t.byte 0x%02x #bad opcode\n", 7712 *(Bytes.data() + Index) & 0xff); 7713 InstSize = 1; // skip exactly one illegible byte and move on. 7714 } else { 7715 WithColor::warning(errs(), "llvm-objdump") 7716 << "invalid instruction encoding\n"; 7717 if (InstSize == 0) 7718 InstSize = 1; // skip illegible bytes 7719 } 7720 } 7721 } 7722 } 7723 // The TripleName's need to be reset if we are called again for a different 7724 // architecture. 7725 TripleName = ""; 7726 ThumbTripleName = ""; 7727 7728 if (SymbolizerInfo.demangled_name != nullptr) 7729 free(SymbolizerInfo.demangled_name); 7730 if (ThumbSymbolizerInfo.demangled_name != nullptr) 7731 free(ThumbSymbolizerInfo.demangled_name); 7732 } 7733 } 7734 7735 //===----------------------------------------------------------------------===// 7736 // __compact_unwind section dumping 7737 //===----------------------------------------------------------------------===// 7738 7739 namespace { 7740 7741 template <typename T> 7742 static uint64_t read(StringRef Contents, ptrdiff_t Offset) { 7743 using llvm::support::little; 7744 using llvm::support::unaligned; 7745 7746 if (Offset + sizeof(T) > Contents.size()) { 7747 outs() << "warning: attempt to read past end of buffer\n"; 7748 return T(); 7749 } 7750 7751 uint64_t Val = 7752 support::endian::read<T, little, unaligned>(Contents.data() + Offset); 7753 return Val; 7754 } 7755 7756 template <typename T> 7757 static uint64_t readNext(StringRef Contents, ptrdiff_t &Offset) { 7758 T Val = read<T>(Contents, Offset); 7759 Offset += sizeof(T); 7760 return Val; 7761 } 7762 7763 struct CompactUnwindEntry { 7764 uint32_t OffsetInSection; 7765 7766 uint64_t FunctionAddr; 7767 uint32_t Length; 7768 uint32_t CompactEncoding; 7769 uint64_t PersonalityAddr; 7770 uint64_t LSDAAddr; 7771 7772 RelocationRef FunctionReloc; 7773 RelocationRef PersonalityReloc; 7774 RelocationRef LSDAReloc; 7775 7776 CompactUnwindEntry(StringRef Contents, unsigned Offset, bool Is64) 7777 : OffsetInSection(Offset) { 7778 if (Is64) 7779 read<uint64_t>(Contents, Offset); 7780 else 7781 read<uint32_t>(Contents, Offset); 7782 } 7783 7784 private: 7785 template <typename UIntPtr> void read(StringRef Contents, ptrdiff_t Offset) { 7786 FunctionAddr = readNext<UIntPtr>(Contents, Offset); 7787 Length = readNext<uint32_t>(Contents, Offset); 7788 CompactEncoding = readNext<uint32_t>(Contents, Offset); 7789 PersonalityAddr = readNext<UIntPtr>(Contents, Offset); 7790 LSDAAddr = readNext<UIntPtr>(Contents, Offset); 7791 } 7792 }; 7793 } 7794 7795 /// Given a relocation from __compact_unwind, consisting of the RelocationRef 7796 /// and data being relocated, determine the best base Name and Addend to use for 7797 /// display purposes. 7798 /// 7799 /// 1. An Extern relocation will directly reference a symbol (and the data is 7800 /// then already an addend), so use that. 7801 /// 2. Otherwise the data is an offset in the object file's layout; try to find 7802 // a symbol before it in the same section, and use the offset from there. 7803 /// 3. Finally, if all that fails, fall back to an offset from the start of the 7804 /// referenced section. 7805 static void findUnwindRelocNameAddend(const MachOObjectFile *Obj, 7806 std::map<uint64_t, SymbolRef> &Symbols, 7807 const RelocationRef &Reloc, uint64_t Addr, 7808 StringRef &Name, uint64_t &Addend) { 7809 if (Reloc.getSymbol() != Obj->symbol_end()) { 7810 Name = unwrapOrError(Reloc.getSymbol()->getName(), Obj->getFileName()); 7811 Addend = Addr; 7812 return; 7813 } 7814 7815 auto RE = Obj->getRelocation(Reloc.getRawDataRefImpl()); 7816 SectionRef RelocSection = Obj->getAnyRelocationSection(RE); 7817 7818 uint64_t SectionAddr = RelocSection.getAddress(); 7819 7820 auto Sym = Symbols.upper_bound(Addr); 7821 if (Sym == Symbols.begin()) { 7822 // The first symbol in the object is after this reference, the best we can 7823 // do is section-relative notation. 7824 if (Expected<StringRef> NameOrErr = RelocSection.getName()) 7825 Name = *NameOrErr; 7826 else 7827 consumeError(NameOrErr.takeError()); 7828 7829 Addend = Addr - SectionAddr; 7830 return; 7831 } 7832 7833 // Go back one so that SymbolAddress <= Addr. 7834 --Sym; 7835 7836 section_iterator SymSection = 7837 unwrapOrError(Sym->second.getSection(), Obj->getFileName()); 7838 if (RelocSection == *SymSection) { 7839 // There's a valid symbol in the same section before this reference. 7840 Name = unwrapOrError(Sym->second.getName(), Obj->getFileName()); 7841 Addend = Addr - Sym->first; 7842 return; 7843 } 7844 7845 // There is a symbol before this reference, but it's in a different 7846 // section. Probably not helpful to mention it, so use the section name. 7847 if (Expected<StringRef> NameOrErr = RelocSection.getName()) 7848 Name = *NameOrErr; 7849 else 7850 consumeError(NameOrErr.takeError()); 7851 7852 Addend = Addr - SectionAddr; 7853 } 7854 7855 static void printUnwindRelocDest(const MachOObjectFile *Obj, 7856 std::map<uint64_t, SymbolRef> &Symbols, 7857 const RelocationRef &Reloc, uint64_t Addr) { 7858 StringRef Name; 7859 uint64_t Addend; 7860 7861 if (!Reloc.getObject()) 7862 return; 7863 7864 findUnwindRelocNameAddend(Obj, Symbols, Reloc, Addr, Name, Addend); 7865 7866 outs() << Name; 7867 if (Addend) 7868 outs() << " + " << format("0x%" PRIx64, Addend); 7869 } 7870 7871 static void 7872 printMachOCompactUnwindSection(const MachOObjectFile *Obj, 7873 std::map<uint64_t, SymbolRef> &Symbols, 7874 const SectionRef &CompactUnwind) { 7875 7876 if (!Obj->isLittleEndian()) { 7877 outs() << "Skipping big-endian __compact_unwind section\n"; 7878 return; 7879 } 7880 7881 bool Is64 = Obj->is64Bit(); 7882 uint32_t PointerSize = Is64 ? sizeof(uint64_t) : sizeof(uint32_t); 7883 uint32_t EntrySize = 3 * PointerSize + 2 * sizeof(uint32_t); 7884 7885 StringRef Contents = 7886 unwrapOrError(CompactUnwind.getContents(), Obj->getFileName()); 7887 SmallVector<CompactUnwindEntry, 4> CompactUnwinds; 7888 7889 // First populate the initial raw offsets, encodings and so on from the entry. 7890 for (unsigned Offset = 0; Offset < Contents.size(); Offset += EntrySize) { 7891 CompactUnwindEntry Entry(Contents, Offset, Is64); 7892 CompactUnwinds.push_back(Entry); 7893 } 7894 7895 // Next we need to look at the relocations to find out what objects are 7896 // actually being referred to. 7897 for (const RelocationRef &Reloc : CompactUnwind.relocations()) { 7898 uint64_t RelocAddress = Reloc.getOffset(); 7899 7900 uint32_t EntryIdx = RelocAddress / EntrySize; 7901 uint32_t OffsetInEntry = RelocAddress - EntryIdx * EntrySize; 7902 CompactUnwindEntry &Entry = CompactUnwinds[EntryIdx]; 7903 7904 if (OffsetInEntry == 0) 7905 Entry.FunctionReloc = Reloc; 7906 else if (OffsetInEntry == PointerSize + 2 * sizeof(uint32_t)) 7907 Entry.PersonalityReloc = Reloc; 7908 else if (OffsetInEntry == 2 * PointerSize + 2 * sizeof(uint32_t)) 7909 Entry.LSDAReloc = Reloc; 7910 else { 7911 outs() << "Invalid relocation in __compact_unwind section\n"; 7912 return; 7913 } 7914 } 7915 7916 // Finally, we're ready to print the data we've gathered. 7917 outs() << "Contents of __compact_unwind section:\n"; 7918 for (auto &Entry : CompactUnwinds) { 7919 outs() << " Entry at offset " 7920 << format("0x%" PRIx32, Entry.OffsetInSection) << ":\n"; 7921 7922 // 1. Start of the region this entry applies to. 7923 outs() << " start: " << format("0x%" PRIx64, 7924 Entry.FunctionAddr) << ' '; 7925 printUnwindRelocDest(Obj, Symbols, Entry.FunctionReloc, Entry.FunctionAddr); 7926 outs() << '\n'; 7927 7928 // 2. Length of the region this entry applies to. 7929 outs() << " length: " << format("0x%" PRIx32, Entry.Length) 7930 << '\n'; 7931 // 3. The 32-bit compact encoding. 7932 outs() << " compact encoding: " 7933 << format("0x%08" PRIx32, Entry.CompactEncoding) << '\n'; 7934 7935 // 4. The personality function, if present. 7936 if (Entry.PersonalityReloc.getObject()) { 7937 outs() << " personality function: " 7938 << format("0x%" PRIx64, Entry.PersonalityAddr) << ' '; 7939 printUnwindRelocDest(Obj, Symbols, Entry.PersonalityReloc, 7940 Entry.PersonalityAddr); 7941 outs() << '\n'; 7942 } 7943 7944 // 5. This entry's language-specific data area. 7945 if (Entry.LSDAReloc.getObject()) { 7946 outs() << " LSDA: " << format("0x%" PRIx64, 7947 Entry.LSDAAddr) << ' '; 7948 printUnwindRelocDest(Obj, Symbols, Entry.LSDAReloc, Entry.LSDAAddr); 7949 outs() << '\n'; 7950 } 7951 } 7952 } 7953 7954 //===----------------------------------------------------------------------===// 7955 // __unwind_info section dumping 7956 //===----------------------------------------------------------------------===// 7957 7958 static void printRegularSecondLevelUnwindPage(StringRef PageData) { 7959 ptrdiff_t Pos = 0; 7960 uint32_t Kind = readNext<uint32_t>(PageData, Pos); 7961 (void)Kind; 7962 assert(Kind == 2 && "kind for a regular 2nd level index should be 2"); 7963 7964 uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos); 7965 uint16_t NumEntries = readNext<uint16_t>(PageData, Pos); 7966 7967 Pos = EntriesStart; 7968 for (unsigned i = 0; i < NumEntries; ++i) { 7969 uint32_t FunctionOffset = readNext<uint32_t>(PageData, Pos); 7970 uint32_t Encoding = readNext<uint32_t>(PageData, Pos); 7971 7972 outs() << " [" << i << "]: " 7973 << "function offset=" << format("0x%08" PRIx32, FunctionOffset) 7974 << ", " 7975 << "encoding=" << format("0x%08" PRIx32, Encoding) << '\n'; 7976 } 7977 } 7978 7979 static void printCompressedSecondLevelUnwindPage( 7980 StringRef PageData, uint32_t FunctionBase, 7981 const SmallVectorImpl<uint32_t> &CommonEncodings) { 7982 ptrdiff_t Pos = 0; 7983 uint32_t Kind = readNext<uint32_t>(PageData, Pos); 7984 (void)Kind; 7985 assert(Kind == 3 && "kind for a compressed 2nd level index should be 3"); 7986 7987 uint16_t EntriesStart = readNext<uint16_t>(PageData, Pos); 7988 uint16_t NumEntries = readNext<uint16_t>(PageData, Pos); 7989 7990 uint16_t EncodingsStart = readNext<uint16_t>(PageData, Pos); 7991 readNext<uint16_t>(PageData, Pos); 7992 StringRef PageEncodings = PageData.substr(EncodingsStart, StringRef::npos); 7993 7994 Pos = EntriesStart; 7995 for (unsigned i = 0; i < NumEntries; ++i) { 7996 uint32_t Entry = readNext<uint32_t>(PageData, Pos); 7997 uint32_t FunctionOffset = FunctionBase + (Entry & 0xffffff); 7998 uint32_t EncodingIdx = Entry >> 24; 7999 8000 uint32_t Encoding; 8001 if (EncodingIdx < CommonEncodings.size()) 8002 Encoding = CommonEncodings[EncodingIdx]; 8003 else 8004 Encoding = read<uint32_t>(PageEncodings, 8005 sizeof(uint32_t) * 8006 (EncodingIdx - CommonEncodings.size())); 8007 8008 outs() << " [" << i << "]: " 8009 << "function offset=" << format("0x%08" PRIx32, FunctionOffset) 8010 << ", " 8011 << "encoding[" << EncodingIdx 8012 << "]=" << format("0x%08" PRIx32, Encoding) << '\n'; 8013 } 8014 } 8015 8016 static void printMachOUnwindInfoSection(const MachOObjectFile *Obj, 8017 std::map<uint64_t, SymbolRef> &Symbols, 8018 const SectionRef &UnwindInfo) { 8019 8020 if (!Obj->isLittleEndian()) { 8021 outs() << "Skipping big-endian __unwind_info section\n"; 8022 return; 8023 } 8024 8025 outs() << "Contents of __unwind_info section:\n"; 8026 8027 StringRef Contents = 8028 unwrapOrError(UnwindInfo.getContents(), Obj->getFileName()); 8029 ptrdiff_t Pos = 0; 8030 8031 //===---------------------------------- 8032 // Section header 8033 //===---------------------------------- 8034 8035 uint32_t Version = readNext<uint32_t>(Contents, Pos); 8036 outs() << " Version: " 8037 << format("0x%" PRIx32, Version) << '\n'; 8038 if (Version != 1) { 8039 outs() << " Skipping section with unknown version\n"; 8040 return; 8041 } 8042 8043 uint32_t CommonEncodingsStart = readNext<uint32_t>(Contents, Pos); 8044 outs() << " Common encodings array section offset: " 8045 << format("0x%" PRIx32, CommonEncodingsStart) << '\n'; 8046 uint32_t NumCommonEncodings = readNext<uint32_t>(Contents, Pos); 8047 outs() << " Number of common encodings in array: " 8048 << format("0x%" PRIx32, NumCommonEncodings) << '\n'; 8049 8050 uint32_t PersonalitiesStart = readNext<uint32_t>(Contents, Pos); 8051 outs() << " Personality function array section offset: " 8052 << format("0x%" PRIx32, PersonalitiesStart) << '\n'; 8053 uint32_t NumPersonalities = readNext<uint32_t>(Contents, Pos); 8054 outs() << " Number of personality functions in array: " 8055 << format("0x%" PRIx32, NumPersonalities) << '\n'; 8056 8057 uint32_t IndicesStart = readNext<uint32_t>(Contents, Pos); 8058 outs() << " Index array section offset: " 8059 << format("0x%" PRIx32, IndicesStart) << '\n'; 8060 uint32_t NumIndices = readNext<uint32_t>(Contents, Pos); 8061 outs() << " Number of indices in array: " 8062 << format("0x%" PRIx32, NumIndices) << '\n'; 8063 8064 //===---------------------------------- 8065 // A shared list of common encodings 8066 //===---------------------------------- 8067 8068 // These occupy indices in the range [0, N] whenever an encoding is referenced 8069 // from a compressed 2nd level index table. In practice the linker only 8070 // creates ~128 of these, so that indices are available to embed encodings in 8071 // the 2nd level index. 8072 8073 SmallVector<uint32_t, 64> CommonEncodings; 8074 outs() << " Common encodings: (count = " << NumCommonEncodings << ")\n"; 8075 Pos = CommonEncodingsStart; 8076 for (unsigned i = 0; i < NumCommonEncodings; ++i) { 8077 uint32_t Encoding = readNext<uint32_t>(Contents, Pos); 8078 CommonEncodings.push_back(Encoding); 8079 8080 outs() << " encoding[" << i << "]: " << format("0x%08" PRIx32, Encoding) 8081 << '\n'; 8082 } 8083 8084 //===---------------------------------- 8085 // Personality functions used in this executable 8086 //===---------------------------------- 8087 8088 // There should be only a handful of these (one per source language, 8089 // roughly). Particularly since they only get 2 bits in the compact encoding. 8090 8091 outs() << " Personality functions: (count = " << NumPersonalities << ")\n"; 8092 Pos = PersonalitiesStart; 8093 for (unsigned i = 0; i < NumPersonalities; ++i) { 8094 uint32_t PersonalityFn = readNext<uint32_t>(Contents, Pos); 8095 outs() << " personality[" << i + 1 8096 << "]: " << format("0x%08" PRIx32, PersonalityFn) << '\n'; 8097 } 8098 8099 //===---------------------------------- 8100 // The level 1 index entries 8101 //===---------------------------------- 8102 8103 // These specify an approximate place to start searching for the more detailed 8104 // information, sorted by PC. 8105 8106 struct IndexEntry { 8107 uint32_t FunctionOffset; 8108 uint32_t SecondLevelPageStart; 8109 uint32_t LSDAStart; 8110 }; 8111 8112 SmallVector<IndexEntry, 4> IndexEntries; 8113 8114 outs() << " Top level indices: (count = " << NumIndices << ")\n"; 8115 Pos = IndicesStart; 8116 for (unsigned i = 0; i < NumIndices; ++i) { 8117 IndexEntry Entry; 8118 8119 Entry.FunctionOffset = readNext<uint32_t>(Contents, Pos); 8120 Entry.SecondLevelPageStart = readNext<uint32_t>(Contents, Pos); 8121 Entry.LSDAStart = readNext<uint32_t>(Contents, Pos); 8122 IndexEntries.push_back(Entry); 8123 8124 outs() << " [" << i << "]: " 8125 << "function offset=" << format("0x%08" PRIx32, Entry.FunctionOffset) 8126 << ", " 8127 << "2nd level page offset=" 8128 << format("0x%08" PRIx32, Entry.SecondLevelPageStart) << ", " 8129 << "LSDA offset=" << format("0x%08" PRIx32, Entry.LSDAStart) << '\n'; 8130 } 8131 8132 //===---------------------------------- 8133 // Next come the LSDA tables 8134 //===---------------------------------- 8135 8136 // The LSDA layout is rather implicit: it's a contiguous array of entries from 8137 // the first top-level index's LSDAOffset to the last (sentinel). 8138 8139 outs() << " LSDA descriptors:\n"; 8140 Pos = IndexEntries[0].LSDAStart; 8141 const uint32_t LSDASize = 2 * sizeof(uint32_t); 8142 int NumLSDAs = 8143 (IndexEntries.back().LSDAStart - IndexEntries[0].LSDAStart) / LSDASize; 8144 8145 for (int i = 0; i < NumLSDAs; ++i) { 8146 uint32_t FunctionOffset = readNext<uint32_t>(Contents, Pos); 8147 uint32_t LSDAOffset = readNext<uint32_t>(Contents, Pos); 8148 outs() << " [" << i << "]: " 8149 << "function offset=" << format("0x%08" PRIx32, FunctionOffset) 8150 << ", " 8151 << "LSDA offset=" << format("0x%08" PRIx32, LSDAOffset) << '\n'; 8152 } 8153 8154 //===---------------------------------- 8155 // Finally, the 2nd level indices 8156 //===---------------------------------- 8157 8158 // Generally these are 4K in size, and have 2 possible forms: 8159 // + Regular stores up to 511 entries with disparate encodings 8160 // + Compressed stores up to 1021 entries if few enough compact encoding 8161 // values are used. 8162 outs() << " Second level indices:\n"; 8163 for (unsigned i = 0; i < IndexEntries.size() - 1; ++i) { 8164 // The final sentinel top-level index has no associated 2nd level page 8165 if (IndexEntries[i].SecondLevelPageStart == 0) 8166 break; 8167 8168 outs() << " Second level index[" << i << "]: " 8169 << "offset in section=" 8170 << format("0x%08" PRIx32, IndexEntries[i].SecondLevelPageStart) 8171 << ", " 8172 << "base function offset=" 8173 << format("0x%08" PRIx32, IndexEntries[i].FunctionOffset) << '\n'; 8174 8175 Pos = IndexEntries[i].SecondLevelPageStart; 8176 if (Pos + sizeof(uint32_t) > Contents.size()) { 8177 outs() << "warning: invalid offset for second level page: " << Pos << '\n'; 8178 continue; 8179 } 8180 8181 uint32_t Kind = 8182 *reinterpret_cast<const support::ulittle32_t *>(Contents.data() + Pos); 8183 if (Kind == 2) 8184 printRegularSecondLevelUnwindPage(Contents.substr(Pos, 4096)); 8185 else if (Kind == 3) 8186 printCompressedSecondLevelUnwindPage(Contents.substr(Pos, 4096), 8187 IndexEntries[i].FunctionOffset, 8188 CommonEncodings); 8189 else 8190 outs() << " Skipping 2nd level page with unknown kind " << Kind 8191 << '\n'; 8192 } 8193 } 8194 8195 void objdump::printMachOUnwindInfo(const MachOObjectFile *Obj) { 8196 std::map<uint64_t, SymbolRef> Symbols; 8197 for (const SymbolRef &SymRef : Obj->symbols()) { 8198 // Discard any undefined or absolute symbols. They're not going to take part 8199 // in the convenience lookup for unwind info and just take up resources. 8200 auto SectOrErr = SymRef.getSection(); 8201 if (!SectOrErr) { 8202 // TODO: Actually report errors helpfully. 8203 consumeError(SectOrErr.takeError()); 8204 continue; 8205 } 8206 section_iterator Section = *SectOrErr; 8207 if (Section == Obj->section_end()) 8208 continue; 8209 8210 uint64_t Addr = SymRef.getValue(); 8211 Symbols.insert(std::make_pair(Addr, SymRef)); 8212 } 8213 8214 for (const SectionRef &Section : Obj->sections()) { 8215 StringRef SectName; 8216 if (Expected<StringRef> NameOrErr = Section.getName()) 8217 SectName = *NameOrErr; 8218 else 8219 consumeError(NameOrErr.takeError()); 8220 8221 if (SectName == "__compact_unwind") 8222 printMachOCompactUnwindSection(Obj, Symbols, Section); 8223 else if (SectName == "__unwind_info") 8224 printMachOUnwindInfoSection(Obj, Symbols, Section); 8225 } 8226 } 8227 8228 static void PrintMachHeader(uint32_t magic, uint32_t cputype, 8229 uint32_t cpusubtype, uint32_t filetype, 8230 uint32_t ncmds, uint32_t sizeofcmds, uint32_t flags, 8231 bool verbose) { 8232 outs() << "Mach header\n"; 8233 outs() << " magic cputype cpusubtype caps filetype ncmds " 8234 "sizeofcmds flags\n"; 8235 if (verbose) { 8236 if (magic == MachO::MH_MAGIC) 8237 outs() << " MH_MAGIC"; 8238 else if (magic == MachO::MH_MAGIC_64) 8239 outs() << "MH_MAGIC_64"; 8240 else 8241 outs() << format(" 0x%08" PRIx32, magic); 8242 switch (cputype) { 8243 case MachO::CPU_TYPE_I386: 8244 outs() << " I386"; 8245 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8246 case MachO::CPU_SUBTYPE_I386_ALL: 8247 outs() << " ALL"; 8248 break; 8249 default: 8250 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8251 break; 8252 } 8253 break; 8254 case MachO::CPU_TYPE_X86_64: 8255 outs() << " X86_64"; 8256 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8257 case MachO::CPU_SUBTYPE_X86_64_ALL: 8258 outs() << " ALL"; 8259 break; 8260 case MachO::CPU_SUBTYPE_X86_64_H: 8261 outs() << " Haswell"; 8262 break; 8263 default: 8264 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8265 break; 8266 } 8267 break; 8268 case MachO::CPU_TYPE_ARM: 8269 outs() << " ARM"; 8270 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8271 case MachO::CPU_SUBTYPE_ARM_ALL: 8272 outs() << " ALL"; 8273 break; 8274 case MachO::CPU_SUBTYPE_ARM_V4T: 8275 outs() << " V4T"; 8276 break; 8277 case MachO::CPU_SUBTYPE_ARM_V5TEJ: 8278 outs() << " V5TEJ"; 8279 break; 8280 case MachO::CPU_SUBTYPE_ARM_XSCALE: 8281 outs() << " XSCALE"; 8282 break; 8283 case MachO::CPU_SUBTYPE_ARM_V6: 8284 outs() << " V6"; 8285 break; 8286 case MachO::CPU_SUBTYPE_ARM_V6M: 8287 outs() << " V6M"; 8288 break; 8289 case MachO::CPU_SUBTYPE_ARM_V7: 8290 outs() << " V7"; 8291 break; 8292 case MachO::CPU_SUBTYPE_ARM_V7EM: 8293 outs() << " V7EM"; 8294 break; 8295 case MachO::CPU_SUBTYPE_ARM_V7K: 8296 outs() << " V7K"; 8297 break; 8298 case MachO::CPU_SUBTYPE_ARM_V7M: 8299 outs() << " V7M"; 8300 break; 8301 case MachO::CPU_SUBTYPE_ARM_V7S: 8302 outs() << " V7S"; 8303 break; 8304 default: 8305 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8306 break; 8307 } 8308 break; 8309 case MachO::CPU_TYPE_ARM64: 8310 outs() << " ARM64"; 8311 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8312 case MachO::CPU_SUBTYPE_ARM64_ALL: 8313 outs() << " ALL"; 8314 break; 8315 case MachO::CPU_SUBTYPE_ARM64E: 8316 outs() << " E"; 8317 break; 8318 default: 8319 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8320 break; 8321 } 8322 break; 8323 case MachO::CPU_TYPE_ARM64_32: 8324 outs() << " ARM64_32"; 8325 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8326 case MachO::CPU_SUBTYPE_ARM64_32_V8: 8327 outs() << " V8"; 8328 break; 8329 default: 8330 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8331 break; 8332 } 8333 break; 8334 case MachO::CPU_TYPE_POWERPC: 8335 outs() << " PPC"; 8336 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8337 case MachO::CPU_SUBTYPE_POWERPC_ALL: 8338 outs() << " ALL"; 8339 break; 8340 default: 8341 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8342 break; 8343 } 8344 break; 8345 case MachO::CPU_TYPE_POWERPC64: 8346 outs() << " PPC64"; 8347 switch (cpusubtype & ~MachO::CPU_SUBTYPE_MASK) { 8348 case MachO::CPU_SUBTYPE_POWERPC_ALL: 8349 outs() << " ALL"; 8350 break; 8351 default: 8352 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8353 break; 8354 } 8355 break; 8356 default: 8357 outs() << format(" %7d", cputype); 8358 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8359 break; 8360 } 8361 if ((cpusubtype & MachO::CPU_SUBTYPE_MASK) == MachO::CPU_SUBTYPE_LIB64) { 8362 outs() << " LIB64"; 8363 } else { 8364 outs() << format(" 0x%02" PRIx32, 8365 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24); 8366 } 8367 switch (filetype) { 8368 case MachO::MH_OBJECT: 8369 outs() << " OBJECT"; 8370 break; 8371 case MachO::MH_EXECUTE: 8372 outs() << " EXECUTE"; 8373 break; 8374 case MachO::MH_FVMLIB: 8375 outs() << " FVMLIB"; 8376 break; 8377 case MachO::MH_CORE: 8378 outs() << " CORE"; 8379 break; 8380 case MachO::MH_PRELOAD: 8381 outs() << " PRELOAD"; 8382 break; 8383 case MachO::MH_DYLIB: 8384 outs() << " DYLIB"; 8385 break; 8386 case MachO::MH_DYLIB_STUB: 8387 outs() << " DYLIB_STUB"; 8388 break; 8389 case MachO::MH_DYLINKER: 8390 outs() << " DYLINKER"; 8391 break; 8392 case MachO::MH_BUNDLE: 8393 outs() << " BUNDLE"; 8394 break; 8395 case MachO::MH_DSYM: 8396 outs() << " DSYM"; 8397 break; 8398 case MachO::MH_KEXT_BUNDLE: 8399 outs() << " KEXTBUNDLE"; 8400 break; 8401 default: 8402 outs() << format(" %10u", filetype); 8403 break; 8404 } 8405 outs() << format(" %5u", ncmds); 8406 outs() << format(" %10u", sizeofcmds); 8407 uint32_t f = flags; 8408 if (f & MachO::MH_NOUNDEFS) { 8409 outs() << " NOUNDEFS"; 8410 f &= ~MachO::MH_NOUNDEFS; 8411 } 8412 if (f & MachO::MH_INCRLINK) { 8413 outs() << " INCRLINK"; 8414 f &= ~MachO::MH_INCRLINK; 8415 } 8416 if (f & MachO::MH_DYLDLINK) { 8417 outs() << " DYLDLINK"; 8418 f &= ~MachO::MH_DYLDLINK; 8419 } 8420 if (f & MachO::MH_BINDATLOAD) { 8421 outs() << " BINDATLOAD"; 8422 f &= ~MachO::MH_BINDATLOAD; 8423 } 8424 if (f & MachO::MH_PREBOUND) { 8425 outs() << " PREBOUND"; 8426 f &= ~MachO::MH_PREBOUND; 8427 } 8428 if (f & MachO::MH_SPLIT_SEGS) { 8429 outs() << " SPLIT_SEGS"; 8430 f &= ~MachO::MH_SPLIT_SEGS; 8431 } 8432 if (f & MachO::MH_LAZY_INIT) { 8433 outs() << " LAZY_INIT"; 8434 f &= ~MachO::MH_LAZY_INIT; 8435 } 8436 if (f & MachO::MH_TWOLEVEL) { 8437 outs() << " TWOLEVEL"; 8438 f &= ~MachO::MH_TWOLEVEL; 8439 } 8440 if (f & MachO::MH_FORCE_FLAT) { 8441 outs() << " FORCE_FLAT"; 8442 f &= ~MachO::MH_FORCE_FLAT; 8443 } 8444 if (f & MachO::MH_NOMULTIDEFS) { 8445 outs() << " NOMULTIDEFS"; 8446 f &= ~MachO::MH_NOMULTIDEFS; 8447 } 8448 if (f & MachO::MH_NOFIXPREBINDING) { 8449 outs() << " NOFIXPREBINDING"; 8450 f &= ~MachO::MH_NOFIXPREBINDING; 8451 } 8452 if (f & MachO::MH_PREBINDABLE) { 8453 outs() << " PREBINDABLE"; 8454 f &= ~MachO::MH_PREBINDABLE; 8455 } 8456 if (f & MachO::MH_ALLMODSBOUND) { 8457 outs() << " ALLMODSBOUND"; 8458 f &= ~MachO::MH_ALLMODSBOUND; 8459 } 8460 if (f & MachO::MH_SUBSECTIONS_VIA_SYMBOLS) { 8461 outs() << " SUBSECTIONS_VIA_SYMBOLS"; 8462 f &= ~MachO::MH_SUBSECTIONS_VIA_SYMBOLS; 8463 } 8464 if (f & MachO::MH_CANONICAL) { 8465 outs() << " CANONICAL"; 8466 f &= ~MachO::MH_CANONICAL; 8467 } 8468 if (f & MachO::MH_WEAK_DEFINES) { 8469 outs() << " WEAK_DEFINES"; 8470 f &= ~MachO::MH_WEAK_DEFINES; 8471 } 8472 if (f & MachO::MH_BINDS_TO_WEAK) { 8473 outs() << " BINDS_TO_WEAK"; 8474 f &= ~MachO::MH_BINDS_TO_WEAK; 8475 } 8476 if (f & MachO::MH_ALLOW_STACK_EXECUTION) { 8477 outs() << " ALLOW_STACK_EXECUTION"; 8478 f &= ~MachO::MH_ALLOW_STACK_EXECUTION; 8479 } 8480 if (f & MachO::MH_DEAD_STRIPPABLE_DYLIB) { 8481 outs() << " DEAD_STRIPPABLE_DYLIB"; 8482 f &= ~MachO::MH_DEAD_STRIPPABLE_DYLIB; 8483 } 8484 if (f & MachO::MH_PIE) { 8485 outs() << " PIE"; 8486 f &= ~MachO::MH_PIE; 8487 } 8488 if (f & MachO::MH_NO_REEXPORTED_DYLIBS) { 8489 outs() << " NO_REEXPORTED_DYLIBS"; 8490 f &= ~MachO::MH_NO_REEXPORTED_DYLIBS; 8491 } 8492 if (f & MachO::MH_HAS_TLV_DESCRIPTORS) { 8493 outs() << " MH_HAS_TLV_DESCRIPTORS"; 8494 f &= ~MachO::MH_HAS_TLV_DESCRIPTORS; 8495 } 8496 if (f & MachO::MH_NO_HEAP_EXECUTION) { 8497 outs() << " MH_NO_HEAP_EXECUTION"; 8498 f &= ~MachO::MH_NO_HEAP_EXECUTION; 8499 } 8500 if (f & MachO::MH_APP_EXTENSION_SAFE) { 8501 outs() << " APP_EXTENSION_SAFE"; 8502 f &= ~MachO::MH_APP_EXTENSION_SAFE; 8503 } 8504 if (f & MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO) { 8505 outs() << " NLIST_OUTOFSYNC_WITH_DYLDINFO"; 8506 f &= ~MachO::MH_NLIST_OUTOFSYNC_WITH_DYLDINFO; 8507 } 8508 if (f != 0 || flags == 0) 8509 outs() << format(" 0x%08" PRIx32, f); 8510 } else { 8511 outs() << format(" 0x%08" PRIx32, magic); 8512 outs() << format(" %7d", cputype); 8513 outs() << format(" %10d", cpusubtype & ~MachO::CPU_SUBTYPE_MASK); 8514 outs() << format(" 0x%02" PRIx32, 8515 (cpusubtype & MachO::CPU_SUBTYPE_MASK) >> 24); 8516 outs() << format(" %10u", filetype); 8517 outs() << format(" %5u", ncmds); 8518 outs() << format(" %10u", sizeofcmds); 8519 outs() << format(" 0x%08" PRIx32, flags); 8520 } 8521 outs() << "\n"; 8522 } 8523 8524 static void PrintSegmentCommand(uint32_t cmd, uint32_t cmdsize, 8525 StringRef SegName, uint64_t vmaddr, 8526 uint64_t vmsize, uint64_t fileoff, 8527 uint64_t filesize, uint32_t maxprot, 8528 uint32_t initprot, uint32_t nsects, 8529 uint32_t flags, uint32_t object_size, 8530 bool verbose) { 8531 uint64_t expected_cmdsize; 8532 if (cmd == MachO::LC_SEGMENT) { 8533 outs() << " cmd LC_SEGMENT\n"; 8534 expected_cmdsize = nsects; 8535 expected_cmdsize *= sizeof(struct MachO::section); 8536 expected_cmdsize += sizeof(struct MachO::segment_command); 8537 } else { 8538 outs() << " cmd LC_SEGMENT_64\n"; 8539 expected_cmdsize = nsects; 8540 expected_cmdsize *= sizeof(struct MachO::section_64); 8541 expected_cmdsize += sizeof(struct MachO::segment_command_64); 8542 } 8543 outs() << " cmdsize " << cmdsize; 8544 if (cmdsize != expected_cmdsize) 8545 outs() << " Inconsistent size\n"; 8546 else 8547 outs() << "\n"; 8548 outs() << " segname " << SegName << "\n"; 8549 if (cmd == MachO::LC_SEGMENT_64) { 8550 outs() << " vmaddr " << format("0x%016" PRIx64, vmaddr) << "\n"; 8551 outs() << " vmsize " << format("0x%016" PRIx64, vmsize) << "\n"; 8552 } else { 8553 outs() << " vmaddr " << format("0x%08" PRIx64, vmaddr) << "\n"; 8554 outs() << " vmsize " << format("0x%08" PRIx64, vmsize) << "\n"; 8555 } 8556 outs() << " fileoff " << fileoff; 8557 if (fileoff > object_size) 8558 outs() << " (past end of file)\n"; 8559 else 8560 outs() << "\n"; 8561 outs() << " filesize " << filesize; 8562 if (fileoff + filesize > object_size) 8563 outs() << " (past end of file)\n"; 8564 else 8565 outs() << "\n"; 8566 if (verbose) { 8567 if ((maxprot & 8568 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE | 8569 MachO::VM_PROT_EXECUTE)) != 0) 8570 outs() << " maxprot ?" << format("0x%08" PRIx32, maxprot) << "\n"; 8571 else { 8572 outs() << " maxprot "; 8573 outs() << ((maxprot & MachO::VM_PROT_READ) ? "r" : "-"); 8574 outs() << ((maxprot & MachO::VM_PROT_WRITE) ? "w" : "-"); 8575 outs() << ((maxprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n"); 8576 } 8577 if ((initprot & 8578 ~(MachO::VM_PROT_READ | MachO::VM_PROT_WRITE | 8579 MachO::VM_PROT_EXECUTE)) != 0) 8580 outs() << " initprot ?" << format("0x%08" PRIx32, initprot) << "\n"; 8581 else { 8582 outs() << " initprot "; 8583 outs() << ((initprot & MachO::VM_PROT_READ) ? "r" : "-"); 8584 outs() << ((initprot & MachO::VM_PROT_WRITE) ? "w" : "-"); 8585 outs() << ((initprot & MachO::VM_PROT_EXECUTE) ? "x\n" : "-\n"); 8586 } 8587 } else { 8588 outs() << " maxprot " << format("0x%08" PRIx32, maxprot) << "\n"; 8589 outs() << " initprot " << format("0x%08" PRIx32, initprot) << "\n"; 8590 } 8591 outs() << " nsects " << nsects << "\n"; 8592 if (verbose) { 8593 outs() << " flags"; 8594 if (flags == 0) 8595 outs() << " (none)\n"; 8596 else { 8597 if (flags & MachO::SG_HIGHVM) { 8598 outs() << " HIGHVM"; 8599 flags &= ~MachO::SG_HIGHVM; 8600 } 8601 if (flags & MachO::SG_FVMLIB) { 8602 outs() << " FVMLIB"; 8603 flags &= ~MachO::SG_FVMLIB; 8604 } 8605 if (flags & MachO::SG_NORELOC) { 8606 outs() << " NORELOC"; 8607 flags &= ~MachO::SG_NORELOC; 8608 } 8609 if (flags & MachO::SG_PROTECTED_VERSION_1) { 8610 outs() << " PROTECTED_VERSION_1"; 8611 flags &= ~MachO::SG_PROTECTED_VERSION_1; 8612 } 8613 if (flags) 8614 outs() << format(" 0x%08" PRIx32, flags) << " (unknown flags)\n"; 8615 else 8616 outs() << "\n"; 8617 } 8618 } else { 8619 outs() << " flags " << format("0x%" PRIx32, flags) << "\n"; 8620 } 8621 } 8622 8623 static void PrintSection(const char *sectname, const char *segname, 8624 uint64_t addr, uint64_t size, uint32_t offset, 8625 uint32_t align, uint32_t reloff, uint32_t nreloc, 8626 uint32_t flags, uint32_t reserved1, uint32_t reserved2, 8627 uint32_t cmd, const char *sg_segname, 8628 uint32_t filetype, uint32_t object_size, 8629 bool verbose) { 8630 outs() << "Section\n"; 8631 outs() << " sectname " << format("%.16s\n", sectname); 8632 outs() << " segname " << format("%.16s", segname); 8633 if (filetype != MachO::MH_OBJECT && strncmp(sg_segname, segname, 16) != 0) 8634 outs() << " (does not match segment)\n"; 8635 else 8636 outs() << "\n"; 8637 if (cmd == MachO::LC_SEGMENT_64) { 8638 outs() << " addr " << format("0x%016" PRIx64, addr) << "\n"; 8639 outs() << " size " << format("0x%016" PRIx64, size); 8640 } else { 8641 outs() << " addr " << format("0x%08" PRIx64, addr) << "\n"; 8642 outs() << " size " << format("0x%08" PRIx64, size); 8643 } 8644 if ((flags & MachO::S_ZEROFILL) != 0 && offset + size > object_size) 8645 outs() << " (past end of file)\n"; 8646 else 8647 outs() << "\n"; 8648 outs() << " offset " << offset; 8649 if (offset > object_size) 8650 outs() << " (past end of file)\n"; 8651 else 8652 outs() << "\n"; 8653 uint32_t align_shifted = 1 << align; 8654 outs() << " align 2^" << align << " (" << align_shifted << ")\n"; 8655 outs() << " reloff " << reloff; 8656 if (reloff > object_size) 8657 outs() << " (past end of file)\n"; 8658 else 8659 outs() << "\n"; 8660 outs() << " nreloc " << nreloc; 8661 if (reloff + nreloc * sizeof(struct MachO::relocation_info) > object_size) 8662 outs() << " (past end of file)\n"; 8663 else 8664 outs() << "\n"; 8665 uint32_t section_type = flags & MachO::SECTION_TYPE; 8666 if (verbose) { 8667 outs() << " type"; 8668 if (section_type == MachO::S_REGULAR) 8669 outs() << " S_REGULAR\n"; 8670 else if (section_type == MachO::S_ZEROFILL) 8671 outs() << " S_ZEROFILL\n"; 8672 else if (section_type == MachO::S_CSTRING_LITERALS) 8673 outs() << " S_CSTRING_LITERALS\n"; 8674 else if (section_type == MachO::S_4BYTE_LITERALS) 8675 outs() << " S_4BYTE_LITERALS\n"; 8676 else if (section_type == MachO::S_8BYTE_LITERALS) 8677 outs() << " S_8BYTE_LITERALS\n"; 8678 else if (section_type == MachO::S_16BYTE_LITERALS) 8679 outs() << " S_16BYTE_LITERALS\n"; 8680 else if (section_type == MachO::S_LITERAL_POINTERS) 8681 outs() << " S_LITERAL_POINTERS\n"; 8682 else if (section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS) 8683 outs() << " S_NON_LAZY_SYMBOL_POINTERS\n"; 8684 else if (section_type == MachO::S_LAZY_SYMBOL_POINTERS) 8685 outs() << " S_LAZY_SYMBOL_POINTERS\n"; 8686 else if (section_type == MachO::S_SYMBOL_STUBS) 8687 outs() << " S_SYMBOL_STUBS\n"; 8688 else if (section_type == MachO::S_MOD_INIT_FUNC_POINTERS) 8689 outs() << " S_MOD_INIT_FUNC_POINTERS\n"; 8690 else if (section_type == MachO::S_MOD_TERM_FUNC_POINTERS) 8691 outs() << " S_MOD_TERM_FUNC_POINTERS\n"; 8692 else if (section_type == MachO::S_COALESCED) 8693 outs() << " S_COALESCED\n"; 8694 else if (section_type == MachO::S_INTERPOSING) 8695 outs() << " S_INTERPOSING\n"; 8696 else if (section_type == MachO::S_DTRACE_DOF) 8697 outs() << " S_DTRACE_DOF\n"; 8698 else if (section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS) 8699 outs() << " S_LAZY_DYLIB_SYMBOL_POINTERS\n"; 8700 else if (section_type == MachO::S_THREAD_LOCAL_REGULAR) 8701 outs() << " S_THREAD_LOCAL_REGULAR\n"; 8702 else if (section_type == MachO::S_THREAD_LOCAL_ZEROFILL) 8703 outs() << " S_THREAD_LOCAL_ZEROFILL\n"; 8704 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLES) 8705 outs() << " S_THREAD_LOCAL_VARIABLES\n"; 8706 else if (section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS) 8707 outs() << " S_THREAD_LOCAL_VARIABLE_POINTERS\n"; 8708 else if (section_type == MachO::S_THREAD_LOCAL_INIT_FUNCTION_POINTERS) 8709 outs() << " S_THREAD_LOCAL_INIT_FUNCTION_POINTERS\n"; 8710 else 8711 outs() << format("0x%08" PRIx32, section_type) << "\n"; 8712 outs() << "attributes"; 8713 uint32_t section_attributes = flags & MachO::SECTION_ATTRIBUTES; 8714 if (section_attributes & MachO::S_ATTR_PURE_INSTRUCTIONS) 8715 outs() << " PURE_INSTRUCTIONS"; 8716 if (section_attributes & MachO::S_ATTR_NO_TOC) 8717 outs() << " NO_TOC"; 8718 if (section_attributes & MachO::S_ATTR_STRIP_STATIC_SYMS) 8719 outs() << " STRIP_STATIC_SYMS"; 8720 if (section_attributes & MachO::S_ATTR_NO_DEAD_STRIP) 8721 outs() << " NO_DEAD_STRIP"; 8722 if (section_attributes & MachO::S_ATTR_LIVE_SUPPORT) 8723 outs() << " LIVE_SUPPORT"; 8724 if (section_attributes & MachO::S_ATTR_SELF_MODIFYING_CODE) 8725 outs() << " SELF_MODIFYING_CODE"; 8726 if (section_attributes & MachO::S_ATTR_DEBUG) 8727 outs() << " DEBUG"; 8728 if (section_attributes & MachO::S_ATTR_SOME_INSTRUCTIONS) 8729 outs() << " SOME_INSTRUCTIONS"; 8730 if (section_attributes & MachO::S_ATTR_EXT_RELOC) 8731 outs() << " EXT_RELOC"; 8732 if (section_attributes & MachO::S_ATTR_LOC_RELOC) 8733 outs() << " LOC_RELOC"; 8734 if (section_attributes == 0) 8735 outs() << " (none)"; 8736 outs() << "\n"; 8737 } else 8738 outs() << " flags " << format("0x%08" PRIx32, flags) << "\n"; 8739 outs() << " reserved1 " << reserved1; 8740 if (section_type == MachO::S_SYMBOL_STUBS || 8741 section_type == MachO::S_LAZY_SYMBOL_POINTERS || 8742 section_type == MachO::S_LAZY_DYLIB_SYMBOL_POINTERS || 8743 section_type == MachO::S_NON_LAZY_SYMBOL_POINTERS || 8744 section_type == MachO::S_THREAD_LOCAL_VARIABLE_POINTERS) 8745 outs() << " (index into indirect symbol table)\n"; 8746 else 8747 outs() << "\n"; 8748 outs() << " reserved2 " << reserved2; 8749 if (section_type == MachO::S_SYMBOL_STUBS) 8750 outs() << " (size of stubs)\n"; 8751 else 8752 outs() << "\n"; 8753 } 8754 8755 static void PrintSymtabLoadCommand(MachO::symtab_command st, bool Is64Bit, 8756 uint32_t object_size) { 8757 outs() << " cmd LC_SYMTAB\n"; 8758 outs() << " cmdsize " << st.cmdsize; 8759 if (st.cmdsize != sizeof(struct MachO::symtab_command)) 8760 outs() << " Incorrect size\n"; 8761 else 8762 outs() << "\n"; 8763 outs() << " symoff " << st.symoff; 8764 if (st.symoff > object_size) 8765 outs() << " (past end of file)\n"; 8766 else 8767 outs() << "\n"; 8768 outs() << " nsyms " << st.nsyms; 8769 uint64_t big_size; 8770 if (Is64Bit) { 8771 big_size = st.nsyms; 8772 big_size *= sizeof(struct MachO::nlist_64); 8773 big_size += st.symoff; 8774 if (big_size > object_size) 8775 outs() << " (past end of file)\n"; 8776 else 8777 outs() << "\n"; 8778 } else { 8779 big_size = st.nsyms; 8780 big_size *= sizeof(struct MachO::nlist); 8781 big_size += st.symoff; 8782 if (big_size > object_size) 8783 outs() << " (past end of file)\n"; 8784 else 8785 outs() << "\n"; 8786 } 8787 outs() << " stroff " << st.stroff; 8788 if (st.stroff > object_size) 8789 outs() << " (past end of file)\n"; 8790 else 8791 outs() << "\n"; 8792 outs() << " strsize " << st.strsize; 8793 big_size = st.stroff; 8794 big_size += st.strsize; 8795 if (big_size > object_size) 8796 outs() << " (past end of file)\n"; 8797 else 8798 outs() << "\n"; 8799 } 8800 8801 static void PrintDysymtabLoadCommand(MachO::dysymtab_command dyst, 8802 uint32_t nsyms, uint32_t object_size, 8803 bool Is64Bit) { 8804 outs() << " cmd LC_DYSYMTAB\n"; 8805 outs() << " cmdsize " << dyst.cmdsize; 8806 if (dyst.cmdsize != sizeof(struct MachO::dysymtab_command)) 8807 outs() << " Incorrect size\n"; 8808 else 8809 outs() << "\n"; 8810 outs() << " ilocalsym " << dyst.ilocalsym; 8811 if (dyst.ilocalsym > nsyms) 8812 outs() << " (greater than the number of symbols)\n"; 8813 else 8814 outs() << "\n"; 8815 outs() << " nlocalsym " << dyst.nlocalsym; 8816 uint64_t big_size; 8817 big_size = dyst.ilocalsym; 8818 big_size += dyst.nlocalsym; 8819 if (big_size > nsyms) 8820 outs() << " (past the end of the symbol table)\n"; 8821 else 8822 outs() << "\n"; 8823 outs() << " iextdefsym " << dyst.iextdefsym; 8824 if (dyst.iextdefsym > nsyms) 8825 outs() << " (greater than the number of symbols)\n"; 8826 else 8827 outs() << "\n"; 8828 outs() << " nextdefsym " << dyst.nextdefsym; 8829 big_size = dyst.iextdefsym; 8830 big_size += dyst.nextdefsym; 8831 if (big_size > nsyms) 8832 outs() << " (past the end of the symbol table)\n"; 8833 else 8834 outs() << "\n"; 8835 outs() << " iundefsym " << dyst.iundefsym; 8836 if (dyst.iundefsym > nsyms) 8837 outs() << " (greater than the number of symbols)\n"; 8838 else 8839 outs() << "\n"; 8840 outs() << " nundefsym " << dyst.nundefsym; 8841 big_size = dyst.iundefsym; 8842 big_size += dyst.nundefsym; 8843 if (big_size > nsyms) 8844 outs() << " (past the end of the symbol table)\n"; 8845 else 8846 outs() << "\n"; 8847 outs() << " tocoff " << dyst.tocoff; 8848 if (dyst.tocoff > object_size) 8849 outs() << " (past end of file)\n"; 8850 else 8851 outs() << "\n"; 8852 outs() << " ntoc " << dyst.ntoc; 8853 big_size = dyst.ntoc; 8854 big_size *= sizeof(struct MachO::dylib_table_of_contents); 8855 big_size += dyst.tocoff; 8856 if (big_size > object_size) 8857 outs() << " (past end of file)\n"; 8858 else 8859 outs() << "\n"; 8860 outs() << " modtaboff " << dyst.modtaboff; 8861 if (dyst.modtaboff > object_size) 8862 outs() << " (past end of file)\n"; 8863 else 8864 outs() << "\n"; 8865 outs() << " nmodtab " << dyst.nmodtab; 8866 uint64_t modtabend; 8867 if (Is64Bit) { 8868 modtabend = dyst.nmodtab; 8869 modtabend *= sizeof(struct MachO::dylib_module_64); 8870 modtabend += dyst.modtaboff; 8871 } else { 8872 modtabend = dyst.nmodtab; 8873 modtabend *= sizeof(struct MachO::dylib_module); 8874 modtabend += dyst.modtaboff; 8875 } 8876 if (modtabend > object_size) 8877 outs() << " (past end of file)\n"; 8878 else 8879 outs() << "\n"; 8880 outs() << " extrefsymoff " << dyst.extrefsymoff; 8881 if (dyst.extrefsymoff > object_size) 8882 outs() << " (past end of file)\n"; 8883 else 8884 outs() << "\n"; 8885 outs() << " nextrefsyms " << dyst.nextrefsyms; 8886 big_size = dyst.nextrefsyms; 8887 big_size *= sizeof(struct MachO::dylib_reference); 8888 big_size += dyst.extrefsymoff; 8889 if (big_size > object_size) 8890 outs() << " (past end of file)\n"; 8891 else 8892 outs() << "\n"; 8893 outs() << " indirectsymoff " << dyst.indirectsymoff; 8894 if (dyst.indirectsymoff > object_size) 8895 outs() << " (past end of file)\n"; 8896 else 8897 outs() << "\n"; 8898 outs() << " nindirectsyms " << dyst.nindirectsyms; 8899 big_size = dyst.nindirectsyms; 8900 big_size *= sizeof(uint32_t); 8901 big_size += dyst.indirectsymoff; 8902 if (big_size > object_size) 8903 outs() << " (past end of file)\n"; 8904 else 8905 outs() << "\n"; 8906 outs() << " extreloff " << dyst.extreloff; 8907 if (dyst.extreloff > object_size) 8908 outs() << " (past end of file)\n"; 8909 else 8910 outs() << "\n"; 8911 outs() << " nextrel " << dyst.nextrel; 8912 big_size = dyst.nextrel; 8913 big_size *= sizeof(struct MachO::relocation_info); 8914 big_size += dyst.extreloff; 8915 if (big_size > object_size) 8916 outs() << " (past end of file)\n"; 8917 else 8918 outs() << "\n"; 8919 outs() << " locreloff " << dyst.locreloff; 8920 if (dyst.locreloff > object_size) 8921 outs() << " (past end of file)\n"; 8922 else 8923 outs() << "\n"; 8924 outs() << " nlocrel " << dyst.nlocrel; 8925 big_size = dyst.nlocrel; 8926 big_size *= sizeof(struct MachO::relocation_info); 8927 big_size += dyst.locreloff; 8928 if (big_size > object_size) 8929 outs() << " (past end of file)\n"; 8930 else 8931 outs() << "\n"; 8932 } 8933 8934 static void PrintDyldInfoLoadCommand(MachO::dyld_info_command dc, 8935 uint32_t object_size) { 8936 if (dc.cmd == MachO::LC_DYLD_INFO) 8937 outs() << " cmd LC_DYLD_INFO\n"; 8938 else 8939 outs() << " cmd LC_DYLD_INFO_ONLY\n"; 8940 outs() << " cmdsize " << dc.cmdsize; 8941 if (dc.cmdsize != sizeof(struct MachO::dyld_info_command)) 8942 outs() << " Incorrect size\n"; 8943 else 8944 outs() << "\n"; 8945 outs() << " rebase_off " << dc.rebase_off; 8946 if (dc.rebase_off > object_size) 8947 outs() << " (past end of file)\n"; 8948 else 8949 outs() << "\n"; 8950 outs() << " rebase_size " << dc.rebase_size; 8951 uint64_t big_size; 8952 big_size = dc.rebase_off; 8953 big_size += dc.rebase_size; 8954 if (big_size > object_size) 8955 outs() << " (past end of file)\n"; 8956 else 8957 outs() << "\n"; 8958 outs() << " bind_off " << dc.bind_off; 8959 if (dc.bind_off > object_size) 8960 outs() << " (past end of file)\n"; 8961 else 8962 outs() << "\n"; 8963 outs() << " bind_size " << dc.bind_size; 8964 big_size = dc.bind_off; 8965 big_size += dc.bind_size; 8966 if (big_size > object_size) 8967 outs() << " (past end of file)\n"; 8968 else 8969 outs() << "\n"; 8970 outs() << " weak_bind_off " << dc.weak_bind_off; 8971 if (dc.weak_bind_off > object_size) 8972 outs() << " (past end of file)\n"; 8973 else 8974 outs() << "\n"; 8975 outs() << " weak_bind_size " << dc.weak_bind_size; 8976 big_size = dc.weak_bind_off; 8977 big_size += dc.weak_bind_size; 8978 if (big_size > object_size) 8979 outs() << " (past end of file)\n"; 8980 else 8981 outs() << "\n"; 8982 outs() << " lazy_bind_off " << dc.lazy_bind_off; 8983 if (dc.lazy_bind_off > object_size) 8984 outs() << " (past end of file)\n"; 8985 else 8986 outs() << "\n"; 8987 outs() << " lazy_bind_size " << dc.lazy_bind_size; 8988 big_size = dc.lazy_bind_off; 8989 big_size += dc.lazy_bind_size; 8990 if (big_size > object_size) 8991 outs() << " (past end of file)\n"; 8992 else 8993 outs() << "\n"; 8994 outs() << " export_off " << dc.export_off; 8995 if (dc.export_off > object_size) 8996 outs() << " (past end of file)\n"; 8997 else 8998 outs() << "\n"; 8999 outs() << " export_size " << dc.export_size; 9000 big_size = dc.export_off; 9001 big_size += dc.export_size; 9002 if (big_size > object_size) 9003 outs() << " (past end of file)\n"; 9004 else 9005 outs() << "\n"; 9006 } 9007 9008 static void PrintDyldLoadCommand(MachO::dylinker_command dyld, 9009 const char *Ptr) { 9010 if (dyld.cmd == MachO::LC_ID_DYLINKER) 9011 outs() << " cmd LC_ID_DYLINKER\n"; 9012 else if (dyld.cmd == MachO::LC_LOAD_DYLINKER) 9013 outs() << " cmd LC_LOAD_DYLINKER\n"; 9014 else if (dyld.cmd == MachO::LC_DYLD_ENVIRONMENT) 9015 outs() << " cmd LC_DYLD_ENVIRONMENT\n"; 9016 else 9017 outs() << " cmd ?(" << dyld.cmd << ")\n"; 9018 outs() << " cmdsize " << dyld.cmdsize; 9019 if (dyld.cmdsize < sizeof(struct MachO::dylinker_command)) 9020 outs() << " Incorrect size\n"; 9021 else 9022 outs() << "\n"; 9023 if (dyld.name >= dyld.cmdsize) 9024 outs() << " name ?(bad offset " << dyld.name << ")\n"; 9025 else { 9026 const char *P = (const char *)(Ptr) + dyld.name; 9027 outs() << " name " << P << " (offset " << dyld.name << ")\n"; 9028 } 9029 } 9030 9031 static void PrintUuidLoadCommand(MachO::uuid_command uuid) { 9032 outs() << " cmd LC_UUID\n"; 9033 outs() << " cmdsize " << uuid.cmdsize; 9034 if (uuid.cmdsize != sizeof(struct MachO::uuid_command)) 9035 outs() << " Incorrect size\n"; 9036 else 9037 outs() << "\n"; 9038 outs() << " uuid "; 9039 for (int i = 0; i < 16; ++i) { 9040 outs() << format("%02" PRIX32, uuid.uuid[i]); 9041 if (i == 3 || i == 5 || i == 7 || i == 9) 9042 outs() << "-"; 9043 } 9044 outs() << "\n"; 9045 } 9046 9047 static void PrintRpathLoadCommand(MachO::rpath_command rpath, const char *Ptr) { 9048 outs() << " cmd LC_RPATH\n"; 9049 outs() << " cmdsize " << rpath.cmdsize; 9050 if (rpath.cmdsize < sizeof(struct MachO::rpath_command)) 9051 outs() << " Incorrect size\n"; 9052 else 9053 outs() << "\n"; 9054 if (rpath.path >= rpath.cmdsize) 9055 outs() << " path ?(bad offset " << rpath.path << ")\n"; 9056 else { 9057 const char *P = (const char *)(Ptr) + rpath.path; 9058 outs() << " path " << P << " (offset " << rpath.path << ")\n"; 9059 } 9060 } 9061 9062 static void PrintVersionMinLoadCommand(MachO::version_min_command vd) { 9063 StringRef LoadCmdName; 9064 switch (vd.cmd) { 9065 case MachO::LC_VERSION_MIN_MACOSX: 9066 LoadCmdName = "LC_VERSION_MIN_MACOSX"; 9067 break; 9068 case MachO::LC_VERSION_MIN_IPHONEOS: 9069 LoadCmdName = "LC_VERSION_MIN_IPHONEOS"; 9070 break; 9071 case MachO::LC_VERSION_MIN_TVOS: 9072 LoadCmdName = "LC_VERSION_MIN_TVOS"; 9073 break; 9074 case MachO::LC_VERSION_MIN_WATCHOS: 9075 LoadCmdName = "LC_VERSION_MIN_WATCHOS"; 9076 break; 9077 default: 9078 llvm_unreachable("Unknown version min load command"); 9079 } 9080 9081 outs() << " cmd " << LoadCmdName << '\n'; 9082 outs() << " cmdsize " << vd.cmdsize; 9083 if (vd.cmdsize != sizeof(struct MachO::version_min_command)) 9084 outs() << " Incorrect size\n"; 9085 else 9086 outs() << "\n"; 9087 outs() << " version " 9088 << MachOObjectFile::getVersionMinMajor(vd, false) << "." 9089 << MachOObjectFile::getVersionMinMinor(vd, false); 9090 uint32_t Update = MachOObjectFile::getVersionMinUpdate(vd, false); 9091 if (Update != 0) 9092 outs() << "." << Update; 9093 outs() << "\n"; 9094 if (vd.sdk == 0) 9095 outs() << " sdk n/a"; 9096 else { 9097 outs() << " sdk " 9098 << MachOObjectFile::getVersionMinMajor(vd, true) << "." 9099 << MachOObjectFile::getVersionMinMinor(vd, true); 9100 } 9101 Update = MachOObjectFile::getVersionMinUpdate(vd, true); 9102 if (Update != 0) 9103 outs() << "." << Update; 9104 outs() << "\n"; 9105 } 9106 9107 static void PrintNoteLoadCommand(MachO::note_command Nt) { 9108 outs() << " cmd LC_NOTE\n"; 9109 outs() << " cmdsize " << Nt.cmdsize; 9110 if (Nt.cmdsize != sizeof(struct MachO::note_command)) 9111 outs() << " Incorrect size\n"; 9112 else 9113 outs() << "\n"; 9114 const char *d = Nt.data_owner; 9115 outs() << "data_owner " << format("%.16s\n", d); 9116 outs() << " offset " << Nt.offset << "\n"; 9117 outs() << " size " << Nt.size << "\n"; 9118 } 9119 9120 static void PrintBuildToolVersion(MachO::build_tool_version bv) { 9121 outs() << " tool " << MachOObjectFile::getBuildTool(bv.tool) << "\n"; 9122 outs() << " version " << MachOObjectFile::getVersionString(bv.version) 9123 << "\n"; 9124 } 9125 9126 static void PrintBuildVersionLoadCommand(const MachOObjectFile *obj, 9127 MachO::build_version_command bd) { 9128 outs() << " cmd LC_BUILD_VERSION\n"; 9129 outs() << " cmdsize " << bd.cmdsize; 9130 if (bd.cmdsize != 9131 sizeof(struct MachO::build_version_command) + 9132 bd.ntools * sizeof(struct MachO::build_tool_version)) 9133 outs() << " Incorrect size\n"; 9134 else 9135 outs() << "\n"; 9136 outs() << " platform " << MachOObjectFile::getBuildPlatform(bd.platform) 9137 << "\n"; 9138 if (bd.sdk) 9139 outs() << " sdk " << MachOObjectFile::getVersionString(bd.sdk) 9140 << "\n"; 9141 else 9142 outs() << " sdk n/a\n"; 9143 outs() << " minos " << MachOObjectFile::getVersionString(bd.minos) 9144 << "\n"; 9145 outs() << " ntools " << bd.ntools << "\n"; 9146 for (unsigned i = 0; i < bd.ntools; ++i) { 9147 MachO::build_tool_version bv = obj->getBuildToolVersion(i); 9148 PrintBuildToolVersion(bv); 9149 } 9150 } 9151 9152 static void PrintSourceVersionCommand(MachO::source_version_command sd) { 9153 outs() << " cmd LC_SOURCE_VERSION\n"; 9154 outs() << " cmdsize " << sd.cmdsize; 9155 if (sd.cmdsize != sizeof(struct MachO::source_version_command)) 9156 outs() << " Incorrect size\n"; 9157 else 9158 outs() << "\n"; 9159 uint64_t a = (sd.version >> 40) & 0xffffff; 9160 uint64_t b = (sd.version >> 30) & 0x3ff; 9161 uint64_t c = (sd.version >> 20) & 0x3ff; 9162 uint64_t d = (sd.version >> 10) & 0x3ff; 9163 uint64_t e = sd.version & 0x3ff; 9164 outs() << " version " << a << "." << b; 9165 if (e != 0) 9166 outs() << "." << c << "." << d << "." << e; 9167 else if (d != 0) 9168 outs() << "." << c << "." << d; 9169 else if (c != 0) 9170 outs() << "." << c; 9171 outs() << "\n"; 9172 } 9173 9174 static void PrintEntryPointCommand(MachO::entry_point_command ep) { 9175 outs() << " cmd LC_MAIN\n"; 9176 outs() << " cmdsize " << ep.cmdsize; 9177 if (ep.cmdsize != sizeof(struct MachO::entry_point_command)) 9178 outs() << " Incorrect size\n"; 9179 else 9180 outs() << "\n"; 9181 outs() << " entryoff " << ep.entryoff << "\n"; 9182 outs() << " stacksize " << ep.stacksize << "\n"; 9183 } 9184 9185 static void PrintEncryptionInfoCommand(MachO::encryption_info_command ec, 9186 uint32_t object_size) { 9187 outs() << " cmd LC_ENCRYPTION_INFO\n"; 9188 outs() << " cmdsize " << ec.cmdsize; 9189 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command)) 9190 outs() << " Incorrect size\n"; 9191 else 9192 outs() << "\n"; 9193 outs() << " cryptoff " << ec.cryptoff; 9194 if (ec.cryptoff > object_size) 9195 outs() << " (past end of file)\n"; 9196 else 9197 outs() << "\n"; 9198 outs() << " cryptsize " << ec.cryptsize; 9199 if (ec.cryptsize > object_size) 9200 outs() << " (past end of file)\n"; 9201 else 9202 outs() << "\n"; 9203 outs() << " cryptid " << ec.cryptid << "\n"; 9204 } 9205 9206 static void PrintEncryptionInfoCommand64(MachO::encryption_info_command_64 ec, 9207 uint32_t object_size) { 9208 outs() << " cmd LC_ENCRYPTION_INFO_64\n"; 9209 outs() << " cmdsize " << ec.cmdsize; 9210 if (ec.cmdsize != sizeof(struct MachO::encryption_info_command_64)) 9211 outs() << " Incorrect size\n"; 9212 else 9213 outs() << "\n"; 9214 outs() << " cryptoff " << ec.cryptoff; 9215 if (ec.cryptoff > object_size) 9216 outs() << " (past end of file)\n"; 9217 else 9218 outs() << "\n"; 9219 outs() << " cryptsize " << ec.cryptsize; 9220 if (ec.cryptsize > object_size) 9221 outs() << " (past end of file)\n"; 9222 else 9223 outs() << "\n"; 9224 outs() << " cryptid " << ec.cryptid << "\n"; 9225 outs() << " pad " << ec.pad << "\n"; 9226 } 9227 9228 static void PrintLinkerOptionCommand(MachO::linker_option_command lo, 9229 const char *Ptr) { 9230 outs() << " cmd LC_LINKER_OPTION\n"; 9231 outs() << " cmdsize " << lo.cmdsize; 9232 if (lo.cmdsize < sizeof(struct MachO::linker_option_command)) 9233 outs() << " Incorrect size\n"; 9234 else 9235 outs() << "\n"; 9236 outs() << " count " << lo.count << "\n"; 9237 const char *string = Ptr + sizeof(struct MachO::linker_option_command); 9238 uint32_t left = lo.cmdsize - sizeof(struct MachO::linker_option_command); 9239 uint32_t i = 0; 9240 while (left > 0) { 9241 while (*string == '\0' && left > 0) { 9242 string++; 9243 left--; 9244 } 9245 if (left > 0) { 9246 i++; 9247 outs() << " string #" << i << " " << format("%.*s\n", left, string); 9248 uint32_t NullPos = StringRef(string, left).find('\0'); 9249 uint32_t len = std::min(NullPos, left) + 1; 9250 string += len; 9251 left -= len; 9252 } 9253 } 9254 if (lo.count != i) 9255 outs() << " count " << lo.count << " does not match number of strings " 9256 << i << "\n"; 9257 } 9258 9259 static void PrintSubFrameworkCommand(MachO::sub_framework_command sub, 9260 const char *Ptr) { 9261 outs() << " cmd LC_SUB_FRAMEWORK\n"; 9262 outs() << " cmdsize " << sub.cmdsize; 9263 if (sub.cmdsize < sizeof(struct MachO::sub_framework_command)) 9264 outs() << " Incorrect size\n"; 9265 else 9266 outs() << "\n"; 9267 if (sub.umbrella < sub.cmdsize) { 9268 const char *P = Ptr + sub.umbrella; 9269 outs() << " umbrella " << P << " (offset " << sub.umbrella << ")\n"; 9270 } else { 9271 outs() << " umbrella ?(bad offset " << sub.umbrella << ")\n"; 9272 } 9273 } 9274 9275 static void PrintSubUmbrellaCommand(MachO::sub_umbrella_command sub, 9276 const char *Ptr) { 9277 outs() << " cmd LC_SUB_UMBRELLA\n"; 9278 outs() << " cmdsize " << sub.cmdsize; 9279 if (sub.cmdsize < sizeof(struct MachO::sub_umbrella_command)) 9280 outs() << " Incorrect size\n"; 9281 else 9282 outs() << "\n"; 9283 if (sub.sub_umbrella < sub.cmdsize) { 9284 const char *P = Ptr + sub.sub_umbrella; 9285 outs() << " sub_umbrella " << P << " (offset " << sub.sub_umbrella << ")\n"; 9286 } else { 9287 outs() << " sub_umbrella ?(bad offset " << sub.sub_umbrella << ")\n"; 9288 } 9289 } 9290 9291 static void PrintSubLibraryCommand(MachO::sub_library_command sub, 9292 const char *Ptr) { 9293 outs() << " cmd LC_SUB_LIBRARY\n"; 9294 outs() << " cmdsize " << sub.cmdsize; 9295 if (sub.cmdsize < sizeof(struct MachO::sub_library_command)) 9296 outs() << " Incorrect size\n"; 9297 else 9298 outs() << "\n"; 9299 if (sub.sub_library < sub.cmdsize) { 9300 const char *P = Ptr + sub.sub_library; 9301 outs() << " sub_library " << P << " (offset " << sub.sub_library << ")\n"; 9302 } else { 9303 outs() << " sub_library ?(bad offset " << sub.sub_library << ")\n"; 9304 } 9305 } 9306 9307 static void PrintSubClientCommand(MachO::sub_client_command sub, 9308 const char *Ptr) { 9309 outs() << " cmd LC_SUB_CLIENT\n"; 9310 outs() << " cmdsize " << sub.cmdsize; 9311 if (sub.cmdsize < sizeof(struct MachO::sub_client_command)) 9312 outs() << " Incorrect size\n"; 9313 else 9314 outs() << "\n"; 9315 if (sub.client < sub.cmdsize) { 9316 const char *P = Ptr + sub.client; 9317 outs() << " client " << P << " (offset " << sub.client << ")\n"; 9318 } else { 9319 outs() << " client ?(bad offset " << sub.client << ")\n"; 9320 } 9321 } 9322 9323 static void PrintRoutinesCommand(MachO::routines_command r) { 9324 outs() << " cmd LC_ROUTINES\n"; 9325 outs() << " cmdsize " << r.cmdsize; 9326 if (r.cmdsize != sizeof(struct MachO::routines_command)) 9327 outs() << " Incorrect size\n"; 9328 else 9329 outs() << "\n"; 9330 outs() << " init_address " << format("0x%08" PRIx32, r.init_address) << "\n"; 9331 outs() << " init_module " << r.init_module << "\n"; 9332 outs() << " reserved1 " << r.reserved1 << "\n"; 9333 outs() << " reserved2 " << r.reserved2 << "\n"; 9334 outs() << " reserved3 " << r.reserved3 << "\n"; 9335 outs() << " reserved4 " << r.reserved4 << "\n"; 9336 outs() << " reserved5 " << r.reserved5 << "\n"; 9337 outs() << " reserved6 " << r.reserved6 << "\n"; 9338 } 9339 9340 static void PrintRoutinesCommand64(MachO::routines_command_64 r) { 9341 outs() << " cmd LC_ROUTINES_64\n"; 9342 outs() << " cmdsize " << r.cmdsize; 9343 if (r.cmdsize != sizeof(struct MachO::routines_command_64)) 9344 outs() << " Incorrect size\n"; 9345 else 9346 outs() << "\n"; 9347 outs() << " init_address " << format("0x%016" PRIx64, r.init_address) << "\n"; 9348 outs() << " init_module " << r.init_module << "\n"; 9349 outs() << " reserved1 " << r.reserved1 << "\n"; 9350 outs() << " reserved2 " << r.reserved2 << "\n"; 9351 outs() << " reserved3 " << r.reserved3 << "\n"; 9352 outs() << " reserved4 " << r.reserved4 << "\n"; 9353 outs() << " reserved5 " << r.reserved5 << "\n"; 9354 outs() << " reserved6 " << r.reserved6 << "\n"; 9355 } 9356 9357 static void Print_x86_thread_state32_t(MachO::x86_thread_state32_t &cpu32) { 9358 outs() << "\t eax " << format("0x%08" PRIx32, cpu32.eax); 9359 outs() << " ebx " << format("0x%08" PRIx32, cpu32.ebx); 9360 outs() << " ecx " << format("0x%08" PRIx32, cpu32.ecx); 9361 outs() << " edx " << format("0x%08" PRIx32, cpu32.edx) << "\n"; 9362 outs() << "\t edi " << format("0x%08" PRIx32, cpu32.edi); 9363 outs() << " esi " << format("0x%08" PRIx32, cpu32.esi); 9364 outs() << " ebp " << format("0x%08" PRIx32, cpu32.ebp); 9365 outs() << " esp " << format("0x%08" PRIx32, cpu32.esp) << "\n"; 9366 outs() << "\t ss " << format("0x%08" PRIx32, cpu32.ss); 9367 outs() << " eflags " << format("0x%08" PRIx32, cpu32.eflags); 9368 outs() << " eip " << format("0x%08" PRIx32, cpu32.eip); 9369 outs() << " cs " << format("0x%08" PRIx32, cpu32.cs) << "\n"; 9370 outs() << "\t ds " << format("0x%08" PRIx32, cpu32.ds); 9371 outs() << " es " << format("0x%08" PRIx32, cpu32.es); 9372 outs() << " fs " << format("0x%08" PRIx32, cpu32.fs); 9373 outs() << " gs " << format("0x%08" PRIx32, cpu32.gs) << "\n"; 9374 } 9375 9376 static void Print_x86_thread_state64_t(MachO::x86_thread_state64_t &cpu64) { 9377 outs() << " rax " << format("0x%016" PRIx64, cpu64.rax); 9378 outs() << " rbx " << format("0x%016" PRIx64, cpu64.rbx); 9379 outs() << " rcx " << format("0x%016" PRIx64, cpu64.rcx) << "\n"; 9380 outs() << " rdx " << format("0x%016" PRIx64, cpu64.rdx); 9381 outs() << " rdi " << format("0x%016" PRIx64, cpu64.rdi); 9382 outs() << " rsi " << format("0x%016" PRIx64, cpu64.rsi) << "\n"; 9383 outs() << " rbp " << format("0x%016" PRIx64, cpu64.rbp); 9384 outs() << " rsp " << format("0x%016" PRIx64, cpu64.rsp); 9385 outs() << " r8 " << format("0x%016" PRIx64, cpu64.r8) << "\n"; 9386 outs() << " r9 " << format("0x%016" PRIx64, cpu64.r9); 9387 outs() << " r10 " << format("0x%016" PRIx64, cpu64.r10); 9388 outs() << " r11 " << format("0x%016" PRIx64, cpu64.r11) << "\n"; 9389 outs() << " r12 " << format("0x%016" PRIx64, cpu64.r12); 9390 outs() << " r13 " << format("0x%016" PRIx64, cpu64.r13); 9391 outs() << " r14 " << format("0x%016" PRIx64, cpu64.r14) << "\n"; 9392 outs() << " r15 " << format("0x%016" PRIx64, cpu64.r15); 9393 outs() << " rip " << format("0x%016" PRIx64, cpu64.rip) << "\n"; 9394 outs() << "rflags " << format("0x%016" PRIx64, cpu64.rflags); 9395 outs() << " cs " << format("0x%016" PRIx64, cpu64.cs); 9396 outs() << " fs " << format("0x%016" PRIx64, cpu64.fs) << "\n"; 9397 outs() << " gs " << format("0x%016" PRIx64, cpu64.gs) << "\n"; 9398 } 9399 9400 static void Print_mmst_reg(MachO::mmst_reg_t &r) { 9401 uint32_t f; 9402 outs() << "\t mmst_reg "; 9403 for (f = 0; f < 10; f++) 9404 outs() << format("%02" PRIx32, (r.mmst_reg[f] & 0xff)) << " "; 9405 outs() << "\n"; 9406 outs() << "\t mmst_rsrv "; 9407 for (f = 0; f < 6; f++) 9408 outs() << format("%02" PRIx32, (r.mmst_rsrv[f] & 0xff)) << " "; 9409 outs() << "\n"; 9410 } 9411 9412 static void Print_xmm_reg(MachO::xmm_reg_t &r) { 9413 uint32_t f; 9414 outs() << "\t xmm_reg "; 9415 for (f = 0; f < 16; f++) 9416 outs() << format("%02" PRIx32, (r.xmm_reg[f] & 0xff)) << " "; 9417 outs() << "\n"; 9418 } 9419 9420 static void Print_x86_float_state_t(MachO::x86_float_state64_t &fpu) { 9421 outs() << "\t fpu_reserved[0] " << fpu.fpu_reserved[0]; 9422 outs() << " fpu_reserved[1] " << fpu.fpu_reserved[1] << "\n"; 9423 outs() << "\t control: invalid " << fpu.fpu_fcw.invalid; 9424 outs() << " denorm " << fpu.fpu_fcw.denorm; 9425 outs() << " zdiv " << fpu.fpu_fcw.zdiv; 9426 outs() << " ovrfl " << fpu.fpu_fcw.ovrfl; 9427 outs() << " undfl " << fpu.fpu_fcw.undfl; 9428 outs() << " precis " << fpu.fpu_fcw.precis << "\n"; 9429 outs() << "\t\t pc "; 9430 if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_24B) 9431 outs() << "FP_PREC_24B "; 9432 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_53B) 9433 outs() << "FP_PREC_53B "; 9434 else if (fpu.fpu_fcw.pc == MachO::x86_FP_PREC_64B) 9435 outs() << "FP_PREC_64B "; 9436 else 9437 outs() << fpu.fpu_fcw.pc << " "; 9438 outs() << "rc "; 9439 if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_NEAR) 9440 outs() << "FP_RND_NEAR "; 9441 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_DOWN) 9442 outs() << "FP_RND_DOWN "; 9443 else if (fpu.fpu_fcw.rc == MachO::x86_FP_RND_UP) 9444 outs() << "FP_RND_UP "; 9445 else if (fpu.fpu_fcw.rc == MachO::x86_FP_CHOP) 9446 outs() << "FP_CHOP "; 9447 outs() << "\n"; 9448 outs() << "\t status: invalid " << fpu.fpu_fsw.invalid; 9449 outs() << " denorm " << fpu.fpu_fsw.denorm; 9450 outs() << " zdiv " << fpu.fpu_fsw.zdiv; 9451 outs() << " ovrfl " << fpu.fpu_fsw.ovrfl; 9452 outs() << " undfl " << fpu.fpu_fsw.undfl; 9453 outs() << " precis " << fpu.fpu_fsw.precis; 9454 outs() << " stkflt " << fpu.fpu_fsw.stkflt << "\n"; 9455 outs() << "\t errsumm " << fpu.fpu_fsw.errsumm; 9456 outs() << " c0 " << fpu.fpu_fsw.c0; 9457 outs() << " c1 " << fpu.fpu_fsw.c1; 9458 outs() << " c2 " << fpu.fpu_fsw.c2; 9459 outs() << " tos " << fpu.fpu_fsw.tos; 9460 outs() << " c3 " << fpu.fpu_fsw.c3; 9461 outs() << " busy " << fpu.fpu_fsw.busy << "\n"; 9462 outs() << "\t fpu_ftw " << format("0x%02" PRIx32, fpu.fpu_ftw); 9463 outs() << " fpu_rsrv1 " << format("0x%02" PRIx32, fpu.fpu_rsrv1); 9464 outs() << " fpu_fop " << format("0x%04" PRIx32, fpu.fpu_fop); 9465 outs() << " fpu_ip " << format("0x%08" PRIx32, fpu.fpu_ip) << "\n"; 9466 outs() << "\t fpu_cs " << format("0x%04" PRIx32, fpu.fpu_cs); 9467 outs() << " fpu_rsrv2 " << format("0x%04" PRIx32, fpu.fpu_rsrv2); 9468 outs() << " fpu_dp " << format("0x%08" PRIx32, fpu.fpu_dp); 9469 outs() << " fpu_ds " << format("0x%04" PRIx32, fpu.fpu_ds) << "\n"; 9470 outs() << "\t fpu_rsrv3 " << format("0x%04" PRIx32, fpu.fpu_rsrv3); 9471 outs() << " fpu_mxcsr " << format("0x%08" PRIx32, fpu.fpu_mxcsr); 9472 outs() << " fpu_mxcsrmask " << format("0x%08" PRIx32, fpu.fpu_mxcsrmask); 9473 outs() << "\n"; 9474 outs() << "\t fpu_stmm0:\n"; 9475 Print_mmst_reg(fpu.fpu_stmm0); 9476 outs() << "\t fpu_stmm1:\n"; 9477 Print_mmst_reg(fpu.fpu_stmm1); 9478 outs() << "\t fpu_stmm2:\n"; 9479 Print_mmst_reg(fpu.fpu_stmm2); 9480 outs() << "\t fpu_stmm3:\n"; 9481 Print_mmst_reg(fpu.fpu_stmm3); 9482 outs() << "\t fpu_stmm4:\n"; 9483 Print_mmst_reg(fpu.fpu_stmm4); 9484 outs() << "\t fpu_stmm5:\n"; 9485 Print_mmst_reg(fpu.fpu_stmm5); 9486 outs() << "\t fpu_stmm6:\n"; 9487 Print_mmst_reg(fpu.fpu_stmm6); 9488 outs() << "\t fpu_stmm7:\n"; 9489 Print_mmst_reg(fpu.fpu_stmm7); 9490 outs() << "\t fpu_xmm0:\n"; 9491 Print_xmm_reg(fpu.fpu_xmm0); 9492 outs() << "\t fpu_xmm1:\n"; 9493 Print_xmm_reg(fpu.fpu_xmm1); 9494 outs() << "\t fpu_xmm2:\n"; 9495 Print_xmm_reg(fpu.fpu_xmm2); 9496 outs() << "\t fpu_xmm3:\n"; 9497 Print_xmm_reg(fpu.fpu_xmm3); 9498 outs() << "\t fpu_xmm4:\n"; 9499 Print_xmm_reg(fpu.fpu_xmm4); 9500 outs() << "\t fpu_xmm5:\n"; 9501 Print_xmm_reg(fpu.fpu_xmm5); 9502 outs() << "\t fpu_xmm6:\n"; 9503 Print_xmm_reg(fpu.fpu_xmm6); 9504 outs() << "\t fpu_xmm7:\n"; 9505 Print_xmm_reg(fpu.fpu_xmm7); 9506 outs() << "\t fpu_xmm8:\n"; 9507 Print_xmm_reg(fpu.fpu_xmm8); 9508 outs() << "\t fpu_xmm9:\n"; 9509 Print_xmm_reg(fpu.fpu_xmm9); 9510 outs() << "\t fpu_xmm10:\n"; 9511 Print_xmm_reg(fpu.fpu_xmm10); 9512 outs() << "\t fpu_xmm11:\n"; 9513 Print_xmm_reg(fpu.fpu_xmm11); 9514 outs() << "\t fpu_xmm12:\n"; 9515 Print_xmm_reg(fpu.fpu_xmm12); 9516 outs() << "\t fpu_xmm13:\n"; 9517 Print_xmm_reg(fpu.fpu_xmm13); 9518 outs() << "\t fpu_xmm14:\n"; 9519 Print_xmm_reg(fpu.fpu_xmm14); 9520 outs() << "\t fpu_xmm15:\n"; 9521 Print_xmm_reg(fpu.fpu_xmm15); 9522 outs() << "\t fpu_rsrv4:\n"; 9523 for (uint32_t f = 0; f < 6; f++) { 9524 outs() << "\t "; 9525 for (uint32_t g = 0; g < 16; g++) 9526 outs() << format("%02" PRIx32, fpu.fpu_rsrv4[f * g]) << " "; 9527 outs() << "\n"; 9528 } 9529 outs() << "\t fpu_reserved1 " << format("0x%08" PRIx32, fpu.fpu_reserved1); 9530 outs() << "\n"; 9531 } 9532 9533 static void Print_x86_exception_state_t(MachO::x86_exception_state64_t &exc64) { 9534 outs() << "\t trapno " << format("0x%08" PRIx32, exc64.trapno); 9535 outs() << " err " << format("0x%08" PRIx32, exc64.err); 9536 outs() << " faultvaddr " << format("0x%016" PRIx64, exc64.faultvaddr) << "\n"; 9537 } 9538 9539 static void Print_arm_thread_state32_t(MachO::arm_thread_state32_t &cpu32) { 9540 outs() << "\t r0 " << format("0x%08" PRIx32, cpu32.r[0]); 9541 outs() << " r1 " << format("0x%08" PRIx32, cpu32.r[1]); 9542 outs() << " r2 " << format("0x%08" PRIx32, cpu32.r[2]); 9543 outs() << " r3 " << format("0x%08" PRIx32, cpu32.r[3]) << "\n"; 9544 outs() << "\t r4 " << format("0x%08" PRIx32, cpu32.r[4]); 9545 outs() << " r5 " << format("0x%08" PRIx32, cpu32.r[5]); 9546 outs() << " r6 " << format("0x%08" PRIx32, cpu32.r[6]); 9547 outs() << " r7 " << format("0x%08" PRIx32, cpu32.r[7]) << "\n"; 9548 outs() << "\t r8 " << format("0x%08" PRIx32, cpu32.r[8]); 9549 outs() << " r9 " << format("0x%08" PRIx32, cpu32.r[9]); 9550 outs() << " r10 " << format("0x%08" PRIx32, cpu32.r[10]); 9551 outs() << " r11 " << format("0x%08" PRIx32, cpu32.r[11]) << "\n"; 9552 outs() << "\t r12 " << format("0x%08" PRIx32, cpu32.r[12]); 9553 outs() << " sp " << format("0x%08" PRIx32, cpu32.sp); 9554 outs() << " lr " << format("0x%08" PRIx32, cpu32.lr); 9555 outs() << " pc " << format("0x%08" PRIx32, cpu32.pc) << "\n"; 9556 outs() << "\t cpsr " << format("0x%08" PRIx32, cpu32.cpsr) << "\n"; 9557 } 9558 9559 static void Print_arm_thread_state64_t(MachO::arm_thread_state64_t &cpu64) { 9560 outs() << "\t x0 " << format("0x%016" PRIx64, cpu64.x[0]); 9561 outs() << " x1 " << format("0x%016" PRIx64, cpu64.x[1]); 9562 outs() << " x2 " << format("0x%016" PRIx64, cpu64.x[2]) << "\n"; 9563 outs() << "\t x3 " << format("0x%016" PRIx64, cpu64.x[3]); 9564 outs() << " x4 " << format("0x%016" PRIx64, cpu64.x[4]); 9565 outs() << " x5 " << format("0x%016" PRIx64, cpu64.x[5]) << "\n"; 9566 outs() << "\t x6 " << format("0x%016" PRIx64, cpu64.x[6]); 9567 outs() << " x7 " << format("0x%016" PRIx64, cpu64.x[7]); 9568 outs() << " x8 " << format("0x%016" PRIx64, cpu64.x[8]) << "\n"; 9569 outs() << "\t x9 " << format("0x%016" PRIx64, cpu64.x[9]); 9570 outs() << " x10 " << format("0x%016" PRIx64, cpu64.x[10]); 9571 outs() << " x11 " << format("0x%016" PRIx64, cpu64.x[11]) << "\n"; 9572 outs() << "\t x12 " << format("0x%016" PRIx64, cpu64.x[12]); 9573 outs() << " x13 " << format("0x%016" PRIx64, cpu64.x[13]); 9574 outs() << " x14 " << format("0x%016" PRIx64, cpu64.x[14]) << "\n"; 9575 outs() << "\t x15 " << format("0x%016" PRIx64, cpu64.x[15]); 9576 outs() << " x16 " << format("0x%016" PRIx64, cpu64.x[16]); 9577 outs() << " x17 " << format("0x%016" PRIx64, cpu64.x[17]) << "\n"; 9578 outs() << "\t x18 " << format("0x%016" PRIx64, cpu64.x[18]); 9579 outs() << " x19 " << format("0x%016" PRIx64, cpu64.x[19]); 9580 outs() << " x20 " << format("0x%016" PRIx64, cpu64.x[20]) << "\n"; 9581 outs() << "\t x21 " << format("0x%016" PRIx64, cpu64.x[21]); 9582 outs() << " x22 " << format("0x%016" PRIx64, cpu64.x[22]); 9583 outs() << " x23 " << format("0x%016" PRIx64, cpu64.x[23]) << "\n"; 9584 outs() << "\t x24 " << format("0x%016" PRIx64, cpu64.x[24]); 9585 outs() << " x25 " << format("0x%016" PRIx64, cpu64.x[25]); 9586 outs() << " x26 " << format("0x%016" PRIx64, cpu64.x[26]) << "\n"; 9587 outs() << "\t x27 " << format("0x%016" PRIx64, cpu64.x[27]); 9588 outs() << " x28 " << format("0x%016" PRIx64, cpu64.x[28]); 9589 outs() << " fp " << format("0x%016" PRIx64, cpu64.fp) << "\n"; 9590 outs() << "\t lr " << format("0x%016" PRIx64, cpu64.lr); 9591 outs() << " sp " << format("0x%016" PRIx64, cpu64.sp); 9592 outs() << " pc " << format("0x%016" PRIx64, cpu64.pc) << "\n"; 9593 outs() << "\t cpsr " << format("0x%08" PRIx32, cpu64.cpsr) << "\n"; 9594 } 9595 9596 static void PrintThreadCommand(MachO::thread_command t, const char *Ptr, 9597 bool isLittleEndian, uint32_t cputype) { 9598 if (t.cmd == MachO::LC_THREAD) 9599 outs() << " cmd LC_THREAD\n"; 9600 else if (t.cmd == MachO::LC_UNIXTHREAD) 9601 outs() << " cmd LC_UNIXTHREAD\n"; 9602 else 9603 outs() << " cmd " << t.cmd << " (unknown)\n"; 9604 outs() << " cmdsize " << t.cmdsize; 9605 if (t.cmdsize < sizeof(struct MachO::thread_command) + 2 * sizeof(uint32_t)) 9606 outs() << " Incorrect size\n"; 9607 else 9608 outs() << "\n"; 9609 9610 const char *begin = Ptr + sizeof(struct MachO::thread_command); 9611 const char *end = Ptr + t.cmdsize; 9612 uint32_t flavor, count, left; 9613 if (cputype == MachO::CPU_TYPE_I386) { 9614 while (begin < end) { 9615 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9616 memcpy((char *)&flavor, begin, sizeof(uint32_t)); 9617 begin += sizeof(uint32_t); 9618 } else { 9619 flavor = 0; 9620 begin = end; 9621 } 9622 if (isLittleEndian != sys::IsLittleEndianHost) 9623 sys::swapByteOrder(flavor); 9624 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9625 memcpy((char *)&count, begin, sizeof(uint32_t)); 9626 begin += sizeof(uint32_t); 9627 } else { 9628 count = 0; 9629 begin = end; 9630 } 9631 if (isLittleEndian != sys::IsLittleEndianHost) 9632 sys::swapByteOrder(count); 9633 if (flavor == MachO::x86_THREAD_STATE32) { 9634 outs() << " flavor i386_THREAD_STATE\n"; 9635 if (count == MachO::x86_THREAD_STATE32_COUNT) 9636 outs() << " count i386_THREAD_STATE_COUNT\n"; 9637 else 9638 outs() << " count " << count 9639 << " (not x86_THREAD_STATE32_COUNT)\n"; 9640 MachO::x86_thread_state32_t cpu32; 9641 left = end - begin; 9642 if (left >= sizeof(MachO::x86_thread_state32_t)) { 9643 memcpy(&cpu32, begin, sizeof(MachO::x86_thread_state32_t)); 9644 begin += sizeof(MachO::x86_thread_state32_t); 9645 } else { 9646 memset(&cpu32, '\0', sizeof(MachO::x86_thread_state32_t)); 9647 memcpy(&cpu32, begin, left); 9648 begin += left; 9649 } 9650 if (isLittleEndian != sys::IsLittleEndianHost) 9651 swapStruct(cpu32); 9652 Print_x86_thread_state32_t(cpu32); 9653 } else if (flavor == MachO::x86_THREAD_STATE) { 9654 outs() << " flavor x86_THREAD_STATE\n"; 9655 if (count == MachO::x86_THREAD_STATE_COUNT) 9656 outs() << " count x86_THREAD_STATE_COUNT\n"; 9657 else 9658 outs() << " count " << count 9659 << " (not x86_THREAD_STATE_COUNT)\n"; 9660 struct MachO::x86_thread_state_t ts; 9661 left = end - begin; 9662 if (left >= sizeof(MachO::x86_thread_state_t)) { 9663 memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t)); 9664 begin += sizeof(MachO::x86_thread_state_t); 9665 } else { 9666 memset(&ts, '\0', sizeof(MachO::x86_thread_state_t)); 9667 memcpy(&ts, begin, left); 9668 begin += left; 9669 } 9670 if (isLittleEndian != sys::IsLittleEndianHost) 9671 swapStruct(ts); 9672 if (ts.tsh.flavor == MachO::x86_THREAD_STATE32) { 9673 outs() << "\t tsh.flavor x86_THREAD_STATE32 "; 9674 if (ts.tsh.count == MachO::x86_THREAD_STATE32_COUNT) 9675 outs() << "tsh.count x86_THREAD_STATE32_COUNT\n"; 9676 else 9677 outs() << "tsh.count " << ts.tsh.count 9678 << " (not x86_THREAD_STATE32_COUNT\n"; 9679 Print_x86_thread_state32_t(ts.uts.ts32); 9680 } else { 9681 outs() << "\t tsh.flavor " << ts.tsh.flavor << " tsh.count " 9682 << ts.tsh.count << "\n"; 9683 } 9684 } else { 9685 outs() << " flavor " << flavor << " (unknown)\n"; 9686 outs() << " count " << count << "\n"; 9687 outs() << " state (unknown)\n"; 9688 begin += count * sizeof(uint32_t); 9689 } 9690 } 9691 } else if (cputype == MachO::CPU_TYPE_X86_64) { 9692 while (begin < end) { 9693 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9694 memcpy((char *)&flavor, begin, sizeof(uint32_t)); 9695 begin += sizeof(uint32_t); 9696 } else { 9697 flavor = 0; 9698 begin = end; 9699 } 9700 if (isLittleEndian != sys::IsLittleEndianHost) 9701 sys::swapByteOrder(flavor); 9702 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9703 memcpy((char *)&count, begin, sizeof(uint32_t)); 9704 begin += sizeof(uint32_t); 9705 } else { 9706 count = 0; 9707 begin = end; 9708 } 9709 if (isLittleEndian != sys::IsLittleEndianHost) 9710 sys::swapByteOrder(count); 9711 if (flavor == MachO::x86_THREAD_STATE64) { 9712 outs() << " flavor x86_THREAD_STATE64\n"; 9713 if (count == MachO::x86_THREAD_STATE64_COUNT) 9714 outs() << " count x86_THREAD_STATE64_COUNT\n"; 9715 else 9716 outs() << " count " << count 9717 << " (not x86_THREAD_STATE64_COUNT)\n"; 9718 MachO::x86_thread_state64_t cpu64; 9719 left = end - begin; 9720 if (left >= sizeof(MachO::x86_thread_state64_t)) { 9721 memcpy(&cpu64, begin, sizeof(MachO::x86_thread_state64_t)); 9722 begin += sizeof(MachO::x86_thread_state64_t); 9723 } else { 9724 memset(&cpu64, '\0', sizeof(MachO::x86_thread_state64_t)); 9725 memcpy(&cpu64, begin, left); 9726 begin += left; 9727 } 9728 if (isLittleEndian != sys::IsLittleEndianHost) 9729 swapStruct(cpu64); 9730 Print_x86_thread_state64_t(cpu64); 9731 } else if (flavor == MachO::x86_THREAD_STATE) { 9732 outs() << " flavor x86_THREAD_STATE\n"; 9733 if (count == MachO::x86_THREAD_STATE_COUNT) 9734 outs() << " count x86_THREAD_STATE_COUNT\n"; 9735 else 9736 outs() << " count " << count 9737 << " (not x86_THREAD_STATE_COUNT)\n"; 9738 struct MachO::x86_thread_state_t ts; 9739 left = end - begin; 9740 if (left >= sizeof(MachO::x86_thread_state_t)) { 9741 memcpy(&ts, begin, sizeof(MachO::x86_thread_state_t)); 9742 begin += sizeof(MachO::x86_thread_state_t); 9743 } else { 9744 memset(&ts, '\0', sizeof(MachO::x86_thread_state_t)); 9745 memcpy(&ts, begin, left); 9746 begin += left; 9747 } 9748 if (isLittleEndian != sys::IsLittleEndianHost) 9749 swapStruct(ts); 9750 if (ts.tsh.flavor == MachO::x86_THREAD_STATE64) { 9751 outs() << "\t tsh.flavor x86_THREAD_STATE64 "; 9752 if (ts.tsh.count == MachO::x86_THREAD_STATE64_COUNT) 9753 outs() << "tsh.count x86_THREAD_STATE64_COUNT\n"; 9754 else 9755 outs() << "tsh.count " << ts.tsh.count 9756 << " (not x86_THREAD_STATE64_COUNT\n"; 9757 Print_x86_thread_state64_t(ts.uts.ts64); 9758 } else { 9759 outs() << "\t tsh.flavor " << ts.tsh.flavor << " tsh.count " 9760 << ts.tsh.count << "\n"; 9761 } 9762 } else if (flavor == MachO::x86_FLOAT_STATE) { 9763 outs() << " flavor x86_FLOAT_STATE\n"; 9764 if (count == MachO::x86_FLOAT_STATE_COUNT) 9765 outs() << " count x86_FLOAT_STATE_COUNT\n"; 9766 else 9767 outs() << " count " << count << " (not x86_FLOAT_STATE_COUNT)\n"; 9768 struct MachO::x86_float_state_t fs; 9769 left = end - begin; 9770 if (left >= sizeof(MachO::x86_float_state_t)) { 9771 memcpy(&fs, begin, sizeof(MachO::x86_float_state_t)); 9772 begin += sizeof(MachO::x86_float_state_t); 9773 } else { 9774 memset(&fs, '\0', sizeof(MachO::x86_float_state_t)); 9775 memcpy(&fs, begin, left); 9776 begin += left; 9777 } 9778 if (isLittleEndian != sys::IsLittleEndianHost) 9779 swapStruct(fs); 9780 if (fs.fsh.flavor == MachO::x86_FLOAT_STATE64) { 9781 outs() << "\t fsh.flavor x86_FLOAT_STATE64 "; 9782 if (fs.fsh.count == MachO::x86_FLOAT_STATE64_COUNT) 9783 outs() << "fsh.count x86_FLOAT_STATE64_COUNT\n"; 9784 else 9785 outs() << "fsh.count " << fs.fsh.count 9786 << " (not x86_FLOAT_STATE64_COUNT\n"; 9787 Print_x86_float_state_t(fs.ufs.fs64); 9788 } else { 9789 outs() << "\t fsh.flavor " << fs.fsh.flavor << " fsh.count " 9790 << fs.fsh.count << "\n"; 9791 } 9792 } else if (flavor == MachO::x86_EXCEPTION_STATE) { 9793 outs() << " flavor x86_EXCEPTION_STATE\n"; 9794 if (count == MachO::x86_EXCEPTION_STATE_COUNT) 9795 outs() << " count x86_EXCEPTION_STATE_COUNT\n"; 9796 else 9797 outs() << " count " << count 9798 << " (not x86_EXCEPTION_STATE_COUNT)\n"; 9799 struct MachO::x86_exception_state_t es; 9800 left = end - begin; 9801 if (left >= sizeof(MachO::x86_exception_state_t)) { 9802 memcpy(&es, begin, sizeof(MachO::x86_exception_state_t)); 9803 begin += sizeof(MachO::x86_exception_state_t); 9804 } else { 9805 memset(&es, '\0', sizeof(MachO::x86_exception_state_t)); 9806 memcpy(&es, begin, left); 9807 begin += left; 9808 } 9809 if (isLittleEndian != sys::IsLittleEndianHost) 9810 swapStruct(es); 9811 if (es.esh.flavor == MachO::x86_EXCEPTION_STATE64) { 9812 outs() << "\t esh.flavor x86_EXCEPTION_STATE64\n"; 9813 if (es.esh.count == MachO::x86_EXCEPTION_STATE64_COUNT) 9814 outs() << "\t esh.count x86_EXCEPTION_STATE64_COUNT\n"; 9815 else 9816 outs() << "\t esh.count " << es.esh.count 9817 << " (not x86_EXCEPTION_STATE64_COUNT\n"; 9818 Print_x86_exception_state_t(es.ues.es64); 9819 } else { 9820 outs() << "\t esh.flavor " << es.esh.flavor << " esh.count " 9821 << es.esh.count << "\n"; 9822 } 9823 } else if (flavor == MachO::x86_EXCEPTION_STATE64) { 9824 outs() << " flavor x86_EXCEPTION_STATE64\n"; 9825 if (count == MachO::x86_EXCEPTION_STATE64_COUNT) 9826 outs() << " count x86_EXCEPTION_STATE64_COUNT\n"; 9827 else 9828 outs() << " count " << count 9829 << " (not x86_EXCEPTION_STATE64_COUNT)\n"; 9830 struct MachO::x86_exception_state64_t es64; 9831 left = end - begin; 9832 if (left >= sizeof(MachO::x86_exception_state64_t)) { 9833 memcpy(&es64, begin, sizeof(MachO::x86_exception_state64_t)); 9834 begin += sizeof(MachO::x86_exception_state64_t); 9835 } else { 9836 memset(&es64, '\0', sizeof(MachO::x86_exception_state64_t)); 9837 memcpy(&es64, begin, left); 9838 begin += left; 9839 } 9840 if (isLittleEndian != sys::IsLittleEndianHost) 9841 swapStruct(es64); 9842 Print_x86_exception_state_t(es64); 9843 } else { 9844 outs() << " flavor " << flavor << " (unknown)\n"; 9845 outs() << " count " << count << "\n"; 9846 outs() << " state (unknown)\n"; 9847 begin += count * sizeof(uint32_t); 9848 } 9849 } 9850 } else if (cputype == MachO::CPU_TYPE_ARM) { 9851 while (begin < end) { 9852 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9853 memcpy((char *)&flavor, begin, sizeof(uint32_t)); 9854 begin += sizeof(uint32_t); 9855 } else { 9856 flavor = 0; 9857 begin = end; 9858 } 9859 if (isLittleEndian != sys::IsLittleEndianHost) 9860 sys::swapByteOrder(flavor); 9861 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9862 memcpy((char *)&count, begin, sizeof(uint32_t)); 9863 begin += sizeof(uint32_t); 9864 } else { 9865 count = 0; 9866 begin = end; 9867 } 9868 if (isLittleEndian != sys::IsLittleEndianHost) 9869 sys::swapByteOrder(count); 9870 if (flavor == MachO::ARM_THREAD_STATE) { 9871 outs() << " flavor ARM_THREAD_STATE\n"; 9872 if (count == MachO::ARM_THREAD_STATE_COUNT) 9873 outs() << " count ARM_THREAD_STATE_COUNT\n"; 9874 else 9875 outs() << " count " << count 9876 << " (not ARM_THREAD_STATE_COUNT)\n"; 9877 MachO::arm_thread_state32_t cpu32; 9878 left = end - begin; 9879 if (left >= sizeof(MachO::arm_thread_state32_t)) { 9880 memcpy(&cpu32, begin, sizeof(MachO::arm_thread_state32_t)); 9881 begin += sizeof(MachO::arm_thread_state32_t); 9882 } else { 9883 memset(&cpu32, '\0', sizeof(MachO::arm_thread_state32_t)); 9884 memcpy(&cpu32, begin, left); 9885 begin += left; 9886 } 9887 if (isLittleEndian != sys::IsLittleEndianHost) 9888 swapStruct(cpu32); 9889 Print_arm_thread_state32_t(cpu32); 9890 } else { 9891 outs() << " flavor " << flavor << " (unknown)\n"; 9892 outs() << " count " << count << "\n"; 9893 outs() << " state (unknown)\n"; 9894 begin += count * sizeof(uint32_t); 9895 } 9896 } 9897 } else if (cputype == MachO::CPU_TYPE_ARM64 || 9898 cputype == MachO::CPU_TYPE_ARM64_32) { 9899 while (begin < end) { 9900 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9901 memcpy((char *)&flavor, begin, sizeof(uint32_t)); 9902 begin += sizeof(uint32_t); 9903 } else { 9904 flavor = 0; 9905 begin = end; 9906 } 9907 if (isLittleEndian != sys::IsLittleEndianHost) 9908 sys::swapByteOrder(flavor); 9909 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9910 memcpy((char *)&count, begin, sizeof(uint32_t)); 9911 begin += sizeof(uint32_t); 9912 } else { 9913 count = 0; 9914 begin = end; 9915 } 9916 if (isLittleEndian != sys::IsLittleEndianHost) 9917 sys::swapByteOrder(count); 9918 if (flavor == MachO::ARM_THREAD_STATE64) { 9919 outs() << " flavor ARM_THREAD_STATE64\n"; 9920 if (count == MachO::ARM_THREAD_STATE64_COUNT) 9921 outs() << " count ARM_THREAD_STATE64_COUNT\n"; 9922 else 9923 outs() << " count " << count 9924 << " (not ARM_THREAD_STATE64_COUNT)\n"; 9925 MachO::arm_thread_state64_t cpu64; 9926 left = end - begin; 9927 if (left >= sizeof(MachO::arm_thread_state64_t)) { 9928 memcpy(&cpu64, begin, sizeof(MachO::arm_thread_state64_t)); 9929 begin += sizeof(MachO::arm_thread_state64_t); 9930 } else { 9931 memset(&cpu64, '\0', sizeof(MachO::arm_thread_state64_t)); 9932 memcpy(&cpu64, begin, left); 9933 begin += left; 9934 } 9935 if (isLittleEndian != sys::IsLittleEndianHost) 9936 swapStruct(cpu64); 9937 Print_arm_thread_state64_t(cpu64); 9938 } else { 9939 outs() << " flavor " << flavor << " (unknown)\n"; 9940 outs() << " count " << count << "\n"; 9941 outs() << " state (unknown)\n"; 9942 begin += count * sizeof(uint32_t); 9943 } 9944 } 9945 } else { 9946 while (begin < end) { 9947 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9948 memcpy((char *)&flavor, begin, sizeof(uint32_t)); 9949 begin += sizeof(uint32_t); 9950 } else { 9951 flavor = 0; 9952 begin = end; 9953 } 9954 if (isLittleEndian != sys::IsLittleEndianHost) 9955 sys::swapByteOrder(flavor); 9956 if (end - begin > (ptrdiff_t)sizeof(uint32_t)) { 9957 memcpy((char *)&count, begin, sizeof(uint32_t)); 9958 begin += sizeof(uint32_t); 9959 } else { 9960 count = 0; 9961 begin = end; 9962 } 9963 if (isLittleEndian != sys::IsLittleEndianHost) 9964 sys::swapByteOrder(count); 9965 outs() << " flavor " << flavor << "\n"; 9966 outs() << " count " << count << "\n"; 9967 outs() << " state (Unknown cputype/cpusubtype)\n"; 9968 begin += count * sizeof(uint32_t); 9969 } 9970 } 9971 } 9972 9973 static void PrintDylibCommand(MachO::dylib_command dl, const char *Ptr) { 9974 if (dl.cmd == MachO::LC_ID_DYLIB) 9975 outs() << " cmd LC_ID_DYLIB\n"; 9976 else if (dl.cmd == MachO::LC_LOAD_DYLIB) 9977 outs() << " cmd LC_LOAD_DYLIB\n"; 9978 else if (dl.cmd == MachO::LC_LOAD_WEAK_DYLIB) 9979 outs() << " cmd LC_LOAD_WEAK_DYLIB\n"; 9980 else if (dl.cmd == MachO::LC_REEXPORT_DYLIB) 9981 outs() << " cmd LC_REEXPORT_DYLIB\n"; 9982 else if (dl.cmd == MachO::LC_LAZY_LOAD_DYLIB) 9983 outs() << " cmd LC_LAZY_LOAD_DYLIB\n"; 9984 else if (dl.cmd == MachO::LC_LOAD_UPWARD_DYLIB) 9985 outs() << " cmd LC_LOAD_UPWARD_DYLIB\n"; 9986 else 9987 outs() << " cmd " << dl.cmd << " (unknown)\n"; 9988 outs() << " cmdsize " << dl.cmdsize; 9989 if (dl.cmdsize < sizeof(struct MachO::dylib_command)) 9990 outs() << " Incorrect size\n"; 9991 else 9992 outs() << "\n"; 9993 if (dl.dylib.name < dl.cmdsize) { 9994 const char *P = (const char *)(Ptr) + dl.dylib.name; 9995 outs() << " name " << P << " (offset " << dl.dylib.name << ")\n"; 9996 } else { 9997 outs() << " name ?(bad offset " << dl.dylib.name << ")\n"; 9998 } 9999 outs() << " time stamp " << dl.dylib.timestamp << " "; 10000 time_t t = dl.dylib.timestamp; 10001 outs() << ctime(&t); 10002 outs() << " current version "; 10003 if (dl.dylib.current_version == 0xffffffff) 10004 outs() << "n/a\n"; 10005 else 10006 outs() << ((dl.dylib.current_version >> 16) & 0xffff) << "." 10007 << ((dl.dylib.current_version >> 8) & 0xff) << "." 10008 << (dl.dylib.current_version & 0xff) << "\n"; 10009 outs() << "compatibility version "; 10010 if (dl.dylib.compatibility_version == 0xffffffff) 10011 outs() << "n/a\n"; 10012 else 10013 outs() << ((dl.dylib.compatibility_version >> 16) & 0xffff) << "." 10014 << ((dl.dylib.compatibility_version >> 8) & 0xff) << "." 10015 << (dl.dylib.compatibility_version & 0xff) << "\n"; 10016 } 10017 10018 static void PrintLinkEditDataCommand(MachO::linkedit_data_command ld, 10019 uint32_t object_size) { 10020 if (ld.cmd == MachO::LC_CODE_SIGNATURE) 10021 outs() << " cmd LC_CODE_SIGNATURE\n"; 10022 else if (ld.cmd == MachO::LC_SEGMENT_SPLIT_INFO) 10023 outs() << " cmd LC_SEGMENT_SPLIT_INFO\n"; 10024 else if (ld.cmd == MachO::LC_FUNCTION_STARTS) 10025 outs() << " cmd LC_FUNCTION_STARTS\n"; 10026 else if (ld.cmd == MachO::LC_DATA_IN_CODE) 10027 outs() << " cmd LC_DATA_IN_CODE\n"; 10028 else if (ld.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS) 10029 outs() << " cmd LC_DYLIB_CODE_SIGN_DRS\n"; 10030 else if (ld.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) 10031 outs() << " cmd LC_LINKER_OPTIMIZATION_HINT\n"; 10032 else 10033 outs() << " cmd " << ld.cmd << " (?)\n"; 10034 outs() << " cmdsize " << ld.cmdsize; 10035 if (ld.cmdsize != sizeof(struct MachO::linkedit_data_command)) 10036 outs() << " Incorrect size\n"; 10037 else 10038 outs() << "\n"; 10039 outs() << " dataoff " << ld.dataoff; 10040 if (ld.dataoff > object_size) 10041 outs() << " (past end of file)\n"; 10042 else 10043 outs() << "\n"; 10044 outs() << " datasize " << ld.datasize; 10045 uint64_t big_size = ld.dataoff; 10046 big_size += ld.datasize; 10047 if (big_size > object_size) 10048 outs() << " (past end of file)\n"; 10049 else 10050 outs() << "\n"; 10051 } 10052 10053 static void PrintLoadCommands(const MachOObjectFile *Obj, uint32_t filetype, 10054 uint32_t cputype, bool verbose) { 10055 StringRef Buf = Obj->getData(); 10056 unsigned Index = 0; 10057 for (const auto &Command : Obj->load_commands()) { 10058 outs() << "Load command " << Index++ << "\n"; 10059 if (Command.C.cmd == MachO::LC_SEGMENT) { 10060 MachO::segment_command SLC = Obj->getSegmentLoadCommand(Command); 10061 const char *sg_segname = SLC.segname; 10062 PrintSegmentCommand(SLC.cmd, SLC.cmdsize, SLC.segname, SLC.vmaddr, 10063 SLC.vmsize, SLC.fileoff, SLC.filesize, SLC.maxprot, 10064 SLC.initprot, SLC.nsects, SLC.flags, Buf.size(), 10065 verbose); 10066 for (unsigned j = 0; j < SLC.nsects; j++) { 10067 MachO::section S = Obj->getSection(Command, j); 10068 PrintSection(S.sectname, S.segname, S.addr, S.size, S.offset, S.align, 10069 S.reloff, S.nreloc, S.flags, S.reserved1, S.reserved2, 10070 SLC.cmd, sg_segname, filetype, Buf.size(), verbose); 10071 } 10072 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) { 10073 MachO::segment_command_64 SLC_64 = Obj->getSegment64LoadCommand(Command); 10074 const char *sg_segname = SLC_64.segname; 10075 PrintSegmentCommand(SLC_64.cmd, SLC_64.cmdsize, SLC_64.segname, 10076 SLC_64.vmaddr, SLC_64.vmsize, SLC_64.fileoff, 10077 SLC_64.filesize, SLC_64.maxprot, SLC_64.initprot, 10078 SLC_64.nsects, SLC_64.flags, Buf.size(), verbose); 10079 for (unsigned j = 0; j < SLC_64.nsects; j++) { 10080 MachO::section_64 S_64 = Obj->getSection64(Command, j); 10081 PrintSection(S_64.sectname, S_64.segname, S_64.addr, S_64.size, 10082 S_64.offset, S_64.align, S_64.reloff, S_64.nreloc, 10083 S_64.flags, S_64.reserved1, S_64.reserved2, SLC_64.cmd, 10084 sg_segname, filetype, Buf.size(), verbose); 10085 } 10086 } else if (Command.C.cmd == MachO::LC_SYMTAB) { 10087 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand(); 10088 PrintSymtabLoadCommand(Symtab, Obj->is64Bit(), Buf.size()); 10089 } else if (Command.C.cmd == MachO::LC_DYSYMTAB) { 10090 MachO::dysymtab_command Dysymtab = Obj->getDysymtabLoadCommand(); 10091 MachO::symtab_command Symtab = Obj->getSymtabLoadCommand(); 10092 PrintDysymtabLoadCommand(Dysymtab, Symtab.nsyms, Buf.size(), 10093 Obj->is64Bit()); 10094 } else if (Command.C.cmd == MachO::LC_DYLD_INFO || 10095 Command.C.cmd == MachO::LC_DYLD_INFO_ONLY) { 10096 MachO::dyld_info_command DyldInfo = Obj->getDyldInfoLoadCommand(Command); 10097 PrintDyldInfoLoadCommand(DyldInfo, Buf.size()); 10098 } else if (Command.C.cmd == MachO::LC_LOAD_DYLINKER || 10099 Command.C.cmd == MachO::LC_ID_DYLINKER || 10100 Command.C.cmd == MachO::LC_DYLD_ENVIRONMENT) { 10101 MachO::dylinker_command Dyld = Obj->getDylinkerCommand(Command); 10102 PrintDyldLoadCommand(Dyld, Command.Ptr); 10103 } else if (Command.C.cmd == MachO::LC_UUID) { 10104 MachO::uuid_command Uuid = Obj->getUuidCommand(Command); 10105 PrintUuidLoadCommand(Uuid); 10106 } else if (Command.C.cmd == MachO::LC_RPATH) { 10107 MachO::rpath_command Rpath = Obj->getRpathCommand(Command); 10108 PrintRpathLoadCommand(Rpath, Command.Ptr); 10109 } else if (Command.C.cmd == MachO::LC_VERSION_MIN_MACOSX || 10110 Command.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS || 10111 Command.C.cmd == MachO::LC_VERSION_MIN_TVOS || 10112 Command.C.cmd == MachO::LC_VERSION_MIN_WATCHOS) { 10113 MachO::version_min_command Vd = Obj->getVersionMinLoadCommand(Command); 10114 PrintVersionMinLoadCommand(Vd); 10115 } else if (Command.C.cmd == MachO::LC_NOTE) { 10116 MachO::note_command Nt = Obj->getNoteLoadCommand(Command); 10117 PrintNoteLoadCommand(Nt); 10118 } else if (Command.C.cmd == MachO::LC_BUILD_VERSION) { 10119 MachO::build_version_command Bv = 10120 Obj->getBuildVersionLoadCommand(Command); 10121 PrintBuildVersionLoadCommand(Obj, Bv); 10122 } else if (Command.C.cmd == MachO::LC_SOURCE_VERSION) { 10123 MachO::source_version_command Sd = Obj->getSourceVersionCommand(Command); 10124 PrintSourceVersionCommand(Sd); 10125 } else if (Command.C.cmd == MachO::LC_MAIN) { 10126 MachO::entry_point_command Ep = Obj->getEntryPointCommand(Command); 10127 PrintEntryPointCommand(Ep); 10128 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO) { 10129 MachO::encryption_info_command Ei = 10130 Obj->getEncryptionInfoCommand(Command); 10131 PrintEncryptionInfoCommand(Ei, Buf.size()); 10132 } else if (Command.C.cmd == MachO::LC_ENCRYPTION_INFO_64) { 10133 MachO::encryption_info_command_64 Ei = 10134 Obj->getEncryptionInfoCommand64(Command); 10135 PrintEncryptionInfoCommand64(Ei, Buf.size()); 10136 } else if (Command.C.cmd == MachO::LC_LINKER_OPTION) { 10137 MachO::linker_option_command Lo = 10138 Obj->getLinkerOptionLoadCommand(Command); 10139 PrintLinkerOptionCommand(Lo, Command.Ptr); 10140 } else if (Command.C.cmd == MachO::LC_SUB_FRAMEWORK) { 10141 MachO::sub_framework_command Sf = Obj->getSubFrameworkCommand(Command); 10142 PrintSubFrameworkCommand(Sf, Command.Ptr); 10143 } else if (Command.C.cmd == MachO::LC_SUB_UMBRELLA) { 10144 MachO::sub_umbrella_command Sf = Obj->getSubUmbrellaCommand(Command); 10145 PrintSubUmbrellaCommand(Sf, Command.Ptr); 10146 } else if (Command.C.cmd == MachO::LC_SUB_LIBRARY) { 10147 MachO::sub_library_command Sl = Obj->getSubLibraryCommand(Command); 10148 PrintSubLibraryCommand(Sl, Command.Ptr); 10149 } else if (Command.C.cmd == MachO::LC_SUB_CLIENT) { 10150 MachO::sub_client_command Sc = Obj->getSubClientCommand(Command); 10151 PrintSubClientCommand(Sc, Command.Ptr); 10152 } else if (Command.C.cmd == MachO::LC_ROUTINES) { 10153 MachO::routines_command Rc = Obj->getRoutinesCommand(Command); 10154 PrintRoutinesCommand(Rc); 10155 } else if (Command.C.cmd == MachO::LC_ROUTINES_64) { 10156 MachO::routines_command_64 Rc = Obj->getRoutinesCommand64(Command); 10157 PrintRoutinesCommand64(Rc); 10158 } else if (Command.C.cmd == MachO::LC_THREAD || 10159 Command.C.cmd == MachO::LC_UNIXTHREAD) { 10160 MachO::thread_command Tc = Obj->getThreadCommand(Command); 10161 PrintThreadCommand(Tc, Command.Ptr, Obj->isLittleEndian(), cputype); 10162 } else if (Command.C.cmd == MachO::LC_LOAD_DYLIB || 10163 Command.C.cmd == MachO::LC_ID_DYLIB || 10164 Command.C.cmd == MachO::LC_LOAD_WEAK_DYLIB || 10165 Command.C.cmd == MachO::LC_REEXPORT_DYLIB || 10166 Command.C.cmd == MachO::LC_LAZY_LOAD_DYLIB || 10167 Command.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) { 10168 MachO::dylib_command Dl = Obj->getDylibIDLoadCommand(Command); 10169 PrintDylibCommand(Dl, Command.Ptr); 10170 } else if (Command.C.cmd == MachO::LC_CODE_SIGNATURE || 10171 Command.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO || 10172 Command.C.cmd == MachO::LC_FUNCTION_STARTS || 10173 Command.C.cmd == MachO::LC_DATA_IN_CODE || 10174 Command.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS || 10175 Command.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) { 10176 MachO::linkedit_data_command Ld = 10177 Obj->getLinkeditDataLoadCommand(Command); 10178 PrintLinkEditDataCommand(Ld, Buf.size()); 10179 } else { 10180 outs() << " cmd ?(" << format("0x%08" PRIx32, Command.C.cmd) 10181 << ")\n"; 10182 outs() << " cmdsize " << Command.C.cmdsize << "\n"; 10183 // TODO: get and print the raw bytes of the load command. 10184 } 10185 // TODO: print all the other kinds of load commands. 10186 } 10187 } 10188 10189 static void PrintMachHeader(const MachOObjectFile *Obj, bool verbose) { 10190 if (Obj->is64Bit()) { 10191 MachO::mach_header_64 H_64; 10192 H_64 = Obj->getHeader64(); 10193 PrintMachHeader(H_64.magic, H_64.cputype, H_64.cpusubtype, H_64.filetype, 10194 H_64.ncmds, H_64.sizeofcmds, H_64.flags, verbose); 10195 } else { 10196 MachO::mach_header H; 10197 H = Obj->getHeader(); 10198 PrintMachHeader(H.magic, H.cputype, H.cpusubtype, H.filetype, H.ncmds, 10199 H.sizeofcmds, H.flags, verbose); 10200 } 10201 } 10202 10203 void objdump::printMachOFileHeader(const object::ObjectFile *Obj) { 10204 const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj); 10205 PrintMachHeader(file, !NonVerbose); 10206 } 10207 10208 void objdump::printMachOLoadCommands(const object::ObjectFile *Obj) { 10209 const MachOObjectFile *file = dyn_cast<const MachOObjectFile>(Obj); 10210 uint32_t filetype = 0; 10211 uint32_t cputype = 0; 10212 if (file->is64Bit()) { 10213 MachO::mach_header_64 H_64; 10214 H_64 = file->getHeader64(); 10215 filetype = H_64.filetype; 10216 cputype = H_64.cputype; 10217 } else { 10218 MachO::mach_header H; 10219 H = file->getHeader(); 10220 filetype = H.filetype; 10221 cputype = H.cputype; 10222 } 10223 PrintLoadCommands(file, filetype, cputype, !NonVerbose); 10224 } 10225 10226 //===----------------------------------------------------------------------===// 10227 // export trie dumping 10228 //===----------------------------------------------------------------------===// 10229 10230 static void printMachOExportsTrie(const object::MachOObjectFile *Obj) { 10231 uint64_t BaseSegmentAddress = 0; 10232 for (const auto &Command : Obj->load_commands()) { 10233 if (Command.C.cmd == MachO::LC_SEGMENT) { 10234 MachO::segment_command Seg = Obj->getSegmentLoadCommand(Command); 10235 if (Seg.fileoff == 0 && Seg.filesize != 0) { 10236 BaseSegmentAddress = Seg.vmaddr; 10237 break; 10238 } 10239 } else if (Command.C.cmd == MachO::LC_SEGMENT_64) { 10240 MachO::segment_command_64 Seg = Obj->getSegment64LoadCommand(Command); 10241 if (Seg.fileoff == 0 && Seg.filesize != 0) { 10242 BaseSegmentAddress = Seg.vmaddr; 10243 break; 10244 } 10245 } 10246 } 10247 Error Err = Error::success(); 10248 for (const object::ExportEntry &Entry : Obj->exports(Err)) { 10249 uint64_t Flags = Entry.flags(); 10250 bool ReExport = (Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT); 10251 bool WeakDef = (Flags & MachO::EXPORT_SYMBOL_FLAGS_WEAK_DEFINITION); 10252 bool ThreadLocal = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) == 10253 MachO::EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL); 10254 bool Abs = ((Flags & MachO::EXPORT_SYMBOL_FLAGS_KIND_MASK) == 10255 MachO::EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE); 10256 bool Resolver = (Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER); 10257 if (ReExport) 10258 outs() << "[re-export] "; 10259 else 10260 outs() << format("0x%08llX ", 10261 Entry.address() + BaseSegmentAddress); 10262 outs() << Entry.name(); 10263 if (WeakDef || ThreadLocal || Resolver || Abs) { 10264 bool NeedsComma = false; 10265 outs() << " ["; 10266 if (WeakDef) { 10267 outs() << "weak_def"; 10268 NeedsComma = true; 10269 } 10270 if (ThreadLocal) { 10271 if (NeedsComma) 10272 outs() << ", "; 10273 outs() << "per-thread"; 10274 NeedsComma = true; 10275 } 10276 if (Abs) { 10277 if (NeedsComma) 10278 outs() << ", "; 10279 outs() << "absolute"; 10280 NeedsComma = true; 10281 } 10282 if (Resolver) { 10283 if (NeedsComma) 10284 outs() << ", "; 10285 outs() << format("resolver=0x%08llX", Entry.other()); 10286 NeedsComma = true; 10287 } 10288 outs() << "]"; 10289 } 10290 if (ReExport) { 10291 StringRef DylibName = "unknown"; 10292 int Ordinal = Entry.other() - 1; 10293 Obj->getLibraryShortNameByIndex(Ordinal, DylibName); 10294 if (Entry.otherName().empty()) 10295 outs() << " (from " << DylibName << ")"; 10296 else 10297 outs() << " (" << Entry.otherName() << " from " << DylibName << ")"; 10298 } 10299 outs() << "\n"; 10300 } 10301 if (Err) 10302 reportError(std::move(Err), Obj->getFileName()); 10303 } 10304 10305 //===----------------------------------------------------------------------===// 10306 // rebase table dumping 10307 //===----------------------------------------------------------------------===// 10308 10309 static void printMachORebaseTable(object::MachOObjectFile *Obj) { 10310 outs() << "segment section address type\n"; 10311 Error Err = Error::success(); 10312 for (const object::MachORebaseEntry &Entry : Obj->rebaseTable(Err)) { 10313 StringRef SegmentName = Entry.segmentName(); 10314 StringRef SectionName = Entry.sectionName(); 10315 uint64_t Address = Entry.address(); 10316 10317 // Table lines look like: __DATA __nl_symbol_ptr 0x0000F00C pointer 10318 outs() << format("%-8s %-18s 0x%08" PRIX64 " %s\n", 10319 SegmentName.str().c_str(), SectionName.str().c_str(), 10320 Address, Entry.typeName().str().c_str()); 10321 } 10322 if (Err) 10323 reportError(std::move(Err), Obj->getFileName()); 10324 } 10325 10326 static StringRef ordinalName(const object::MachOObjectFile *Obj, int Ordinal) { 10327 StringRef DylibName; 10328 switch (Ordinal) { 10329 case MachO::BIND_SPECIAL_DYLIB_SELF: 10330 return "this-image"; 10331 case MachO::BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE: 10332 return "main-executable"; 10333 case MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP: 10334 return "flat-namespace"; 10335 default: 10336 if (Ordinal > 0) { 10337 std::error_code EC = 10338 Obj->getLibraryShortNameByIndex(Ordinal - 1, DylibName); 10339 if (EC) 10340 return "<<bad library ordinal>>"; 10341 return DylibName; 10342 } 10343 } 10344 return "<<unknown special ordinal>>"; 10345 } 10346 10347 //===----------------------------------------------------------------------===// 10348 // bind table dumping 10349 //===----------------------------------------------------------------------===// 10350 10351 static void printMachOBindTable(object::MachOObjectFile *Obj) { 10352 // Build table of sections so names can used in final output. 10353 outs() << "segment section address type " 10354 "addend dylib symbol\n"; 10355 Error Err = Error::success(); 10356 for (const object::MachOBindEntry &Entry : Obj->bindTable(Err)) { 10357 StringRef SegmentName = Entry.segmentName(); 10358 StringRef SectionName = Entry.sectionName(); 10359 uint64_t Address = Entry.address(); 10360 10361 // Table lines look like: 10362 // __DATA __got 0x00012010 pointer 0 libSystem ___stack_chk_guard 10363 StringRef Attr; 10364 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_WEAK_IMPORT) 10365 Attr = " (weak_import)"; 10366 outs() << left_justify(SegmentName, 8) << " " 10367 << left_justify(SectionName, 18) << " " 10368 << format_hex(Address, 10, true) << " " 10369 << left_justify(Entry.typeName(), 8) << " " 10370 << format_decimal(Entry.addend(), 8) << " " 10371 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " " 10372 << Entry.symbolName() << Attr << "\n"; 10373 } 10374 if (Err) 10375 reportError(std::move(Err), Obj->getFileName()); 10376 } 10377 10378 //===----------------------------------------------------------------------===// 10379 // lazy bind table dumping 10380 //===----------------------------------------------------------------------===// 10381 10382 static void printMachOLazyBindTable(object::MachOObjectFile *Obj) { 10383 outs() << "segment section address " 10384 "dylib symbol\n"; 10385 Error Err = Error::success(); 10386 for (const object::MachOBindEntry &Entry : Obj->lazyBindTable(Err)) { 10387 StringRef SegmentName = Entry.segmentName(); 10388 StringRef SectionName = Entry.sectionName(); 10389 uint64_t Address = Entry.address(); 10390 10391 // Table lines look like: 10392 // __DATA __got 0x00012010 libSystem ___stack_chk_guard 10393 outs() << left_justify(SegmentName, 8) << " " 10394 << left_justify(SectionName, 18) << " " 10395 << format_hex(Address, 10, true) << " " 10396 << left_justify(ordinalName(Obj, Entry.ordinal()), 16) << " " 10397 << Entry.symbolName() << "\n"; 10398 } 10399 if (Err) 10400 reportError(std::move(Err), Obj->getFileName()); 10401 } 10402 10403 //===----------------------------------------------------------------------===// 10404 // weak bind table dumping 10405 //===----------------------------------------------------------------------===// 10406 10407 static void printMachOWeakBindTable(object::MachOObjectFile *Obj) { 10408 outs() << "segment section address " 10409 "type addend symbol\n"; 10410 Error Err = Error::success(); 10411 for (const object::MachOBindEntry &Entry : Obj->weakBindTable(Err)) { 10412 // Strong symbols don't have a location to update. 10413 if (Entry.flags() & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) { 10414 outs() << " strong " 10415 << Entry.symbolName() << "\n"; 10416 continue; 10417 } 10418 StringRef SegmentName = Entry.segmentName(); 10419 StringRef SectionName = Entry.sectionName(); 10420 uint64_t Address = Entry.address(); 10421 10422 // Table lines look like: 10423 // __DATA __data 0x00001000 pointer 0 _foo 10424 outs() << left_justify(SegmentName, 8) << " " 10425 << left_justify(SectionName, 18) << " " 10426 << format_hex(Address, 10, true) << " " 10427 << left_justify(Entry.typeName(), 8) << " " 10428 << format_decimal(Entry.addend(), 8) << " " << Entry.symbolName() 10429 << "\n"; 10430 } 10431 if (Err) 10432 reportError(std::move(Err), Obj->getFileName()); 10433 } 10434 10435 // get_dyld_bind_info_symbolname() is used for disassembly and passed an 10436 // address, ReferenceValue, in the Mach-O file and looks in the dyld bind 10437 // information for that address. If the address is found its binding symbol 10438 // name is returned. If not nullptr is returned. 10439 static const char *get_dyld_bind_info_symbolname(uint64_t ReferenceValue, 10440 struct DisassembleInfo *info) { 10441 if (info->bindtable == nullptr) { 10442 info->bindtable = std::make_unique<SymbolAddressMap>(); 10443 Error Err = Error::success(); 10444 for (const object::MachOBindEntry &Entry : info->O->bindTable(Err)) { 10445 uint64_t Address = Entry.address(); 10446 StringRef name = Entry.symbolName(); 10447 if (!name.empty()) 10448 (*info->bindtable)[Address] = name; 10449 } 10450 if (Err) 10451 reportError(std::move(Err), info->O->getFileName()); 10452 } 10453 auto name = info->bindtable->lookup(ReferenceValue); 10454 return !name.empty() ? name.data() : nullptr; 10455 } 10456 10457 void objdump::printLazyBindTable(ObjectFile *o) { 10458 outs() << "Lazy bind table:\n"; 10459 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o)) 10460 printMachOLazyBindTable(MachO); 10461 else 10462 WithColor::error() 10463 << "This operation is only currently supported " 10464 "for Mach-O executable files.\n"; 10465 } 10466 10467 void objdump::printWeakBindTable(ObjectFile *o) { 10468 outs() << "Weak bind table:\n"; 10469 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o)) 10470 printMachOWeakBindTable(MachO); 10471 else 10472 WithColor::error() 10473 << "This operation is only currently supported " 10474 "for Mach-O executable files.\n"; 10475 } 10476 10477 void objdump::printExportsTrie(const ObjectFile *o) { 10478 outs() << "Exports trie:\n"; 10479 if (const MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o)) 10480 printMachOExportsTrie(MachO); 10481 else 10482 WithColor::error() 10483 << "This operation is only currently supported " 10484 "for Mach-O executable files.\n"; 10485 } 10486 10487 void objdump::printRebaseTable(ObjectFile *o) { 10488 outs() << "Rebase table:\n"; 10489 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o)) 10490 printMachORebaseTable(MachO); 10491 else 10492 WithColor::error() 10493 << "This operation is only currently supported " 10494 "for Mach-O executable files.\n"; 10495 } 10496 10497 void objdump::printBindTable(ObjectFile *o) { 10498 outs() << "Bind table:\n"; 10499 if (MachOObjectFile *MachO = dyn_cast<MachOObjectFile>(o)) 10500 printMachOBindTable(MachO); 10501 else 10502 WithColor::error() 10503 << "This operation is only currently supported " 10504 "for Mach-O executable files.\n"; 10505 } 10506