1 //===- MachOObjectFile.cpp - Mach-O object file binding ---------*- C++ -*-===// 2 // 3 // The LLVM Compiler Infrastructure 4 // 5 // This file is distributed under the University of Illinois Open Source 6 // License. See LICENSE.TXT for details. 7 // 8 //===----------------------------------------------------------------------===// 9 // 10 // This file defines the MachOObjectFile class, which binds the MachOObject 11 // class to the generic ObjectFile wrapper. 12 // 13 //===----------------------------------------------------------------------===// 14 15 #include "llvm/Object/MachO.h" 16 #include "llvm/ADT/STLExtras.h" 17 #include "llvm/ADT/StringSwitch.h" 18 #include "llvm/ADT/Triple.h" 19 #include "llvm/Support/DataExtractor.h" 20 #include "llvm/Support/Debug.h" 21 #include "llvm/Support/Format.h" 22 #include "llvm/Support/Host.h" 23 #include "llvm/Support/LEB128.h" 24 #include "llvm/Support/MachO.h" 25 #include "llvm/Support/MemoryBuffer.h" 26 #include "llvm/Support/raw_ostream.h" 27 #include <cctype> 28 #include <cstring> 29 #include <limits> 30 #include <list> 31 32 using namespace llvm; 33 using namespace object; 34 35 namespace { 36 struct section_base { 37 char sectname[16]; 38 char segname[16]; 39 }; 40 } 41 42 static Error 43 malformedError(Twine Msg) { 44 std::string StringMsg = "truncated or malformed object (" + Msg.str() + ")"; 45 return make_error<GenericBinaryError>(std::move(StringMsg), 46 object_error::parse_failed); 47 } 48 49 // FIXME: Replace all uses of this function with getStructOrErr. 50 template <typename T> 51 static T getStruct(const MachOObjectFile &O, const char *P) { 52 // Don't read before the beginning or past the end of the file 53 if (P < O.getData().begin() || P + sizeof(T) > O.getData().end()) 54 report_fatal_error("Malformed MachO file."); 55 56 T Cmd; 57 memcpy(&Cmd, P, sizeof(T)); 58 if (O.isLittleEndian() != sys::IsLittleEndianHost) 59 MachO::swapStruct(Cmd); 60 return Cmd; 61 } 62 63 template <typename T> 64 static Expected<T> getStructOrErr(const MachOObjectFile &O, const char *P) { 65 // Don't read before the beginning or past the end of the file 66 if (P < O.getData().begin() || P + sizeof(T) > O.getData().end()) 67 return malformedError("Structure read out-of-range"); 68 69 T Cmd; 70 memcpy(&Cmd, P, sizeof(T)); 71 if (O.isLittleEndian() != sys::IsLittleEndianHost) 72 MachO::swapStruct(Cmd); 73 return Cmd; 74 } 75 76 static const char * 77 getSectionPtr(const MachOObjectFile &O, MachOObjectFile::LoadCommandInfo L, 78 unsigned Sec) { 79 uintptr_t CommandAddr = reinterpret_cast<uintptr_t>(L.Ptr); 80 81 bool Is64 = O.is64Bit(); 82 unsigned SegmentLoadSize = Is64 ? sizeof(MachO::segment_command_64) : 83 sizeof(MachO::segment_command); 84 unsigned SectionSize = Is64 ? sizeof(MachO::section_64) : 85 sizeof(MachO::section); 86 87 uintptr_t SectionAddr = CommandAddr + SegmentLoadSize + Sec * SectionSize; 88 return reinterpret_cast<const char*>(SectionAddr); 89 } 90 91 static const char *getPtr(const MachOObjectFile &O, size_t Offset) { 92 return O.getData().substr(Offset, 1).data(); 93 } 94 95 static MachO::nlist_base 96 getSymbolTableEntryBase(const MachOObjectFile &O, DataRefImpl DRI) { 97 const char *P = reinterpret_cast<const char *>(DRI.p); 98 return getStruct<MachO::nlist_base>(O, P); 99 } 100 101 static StringRef parseSegmentOrSectionName(const char *P) { 102 if (P[15] == 0) 103 // Null terminated. 104 return P; 105 // Not null terminated, so this is a 16 char string. 106 return StringRef(P, 16); 107 } 108 109 // Helper to advance a section or symbol iterator multiple increments at a time. 110 template<class T> 111 static void advance(T &it, size_t Val) { 112 while (Val--) 113 ++it; 114 } 115 116 static unsigned getCPUType(const MachOObjectFile &O) { 117 return O.getHeader().cputype; 118 } 119 120 static uint32_t 121 getPlainRelocationAddress(const MachO::any_relocation_info &RE) { 122 return RE.r_word0; 123 } 124 125 static unsigned 126 getScatteredRelocationAddress(const MachO::any_relocation_info &RE) { 127 return RE.r_word0 & 0xffffff; 128 } 129 130 static bool getPlainRelocationPCRel(const MachOObjectFile &O, 131 const MachO::any_relocation_info &RE) { 132 if (O.isLittleEndian()) 133 return (RE.r_word1 >> 24) & 1; 134 return (RE.r_word1 >> 7) & 1; 135 } 136 137 static bool 138 getScatteredRelocationPCRel(const MachO::any_relocation_info &RE) { 139 return (RE.r_word0 >> 30) & 1; 140 } 141 142 static unsigned getPlainRelocationLength(const MachOObjectFile &O, 143 const MachO::any_relocation_info &RE) { 144 if (O.isLittleEndian()) 145 return (RE.r_word1 >> 25) & 3; 146 return (RE.r_word1 >> 5) & 3; 147 } 148 149 static unsigned 150 getScatteredRelocationLength(const MachO::any_relocation_info &RE) { 151 return (RE.r_word0 >> 28) & 3; 152 } 153 154 static unsigned getPlainRelocationType(const MachOObjectFile &O, 155 const MachO::any_relocation_info &RE) { 156 if (O.isLittleEndian()) 157 return RE.r_word1 >> 28; 158 return RE.r_word1 & 0xf; 159 } 160 161 static uint32_t getSectionFlags(const MachOObjectFile &O, 162 DataRefImpl Sec) { 163 if (O.is64Bit()) { 164 MachO::section_64 Sect = O.getSection64(Sec); 165 return Sect.flags; 166 } 167 MachO::section Sect = O.getSection(Sec); 168 return Sect.flags; 169 } 170 171 static Expected<MachOObjectFile::LoadCommandInfo> 172 getLoadCommandInfo(const MachOObjectFile &Obj, const char *Ptr, 173 uint32_t LoadCommandIndex) { 174 if (auto CmdOrErr = getStructOrErr<MachO::load_command>(Obj, Ptr)) { 175 if (CmdOrErr->cmdsize < 8) 176 return malformedError("load command " + Twine(LoadCommandIndex) + 177 " with size less than 8 bytes"); 178 return MachOObjectFile::LoadCommandInfo({Ptr, *CmdOrErr}); 179 } else 180 return CmdOrErr.takeError(); 181 } 182 183 static Expected<MachOObjectFile::LoadCommandInfo> 184 getFirstLoadCommandInfo(const MachOObjectFile &Obj) { 185 unsigned HeaderSize = Obj.is64Bit() ? sizeof(MachO::mach_header_64) 186 : sizeof(MachO::mach_header); 187 if (sizeof(MachO::load_command) > Obj.getHeader().sizeofcmds) 188 return malformedError("load command 0 extends past the end all load " 189 "commands in the file"); 190 return getLoadCommandInfo(Obj, getPtr(Obj, HeaderSize), 0); 191 } 192 193 static Expected<MachOObjectFile::LoadCommandInfo> 194 getNextLoadCommandInfo(const MachOObjectFile &Obj, uint32_t LoadCommandIndex, 195 const MachOObjectFile::LoadCommandInfo &L) { 196 unsigned HeaderSize = Obj.is64Bit() ? sizeof(MachO::mach_header_64) 197 : sizeof(MachO::mach_header); 198 if (L.Ptr + L.C.cmdsize + sizeof(MachO::load_command) > 199 Obj.getData().data() + HeaderSize + Obj.getHeader().sizeofcmds) 200 return malformedError("load command " + Twine(LoadCommandIndex + 1) + 201 " extends past the end all load commands in the file"); 202 return getLoadCommandInfo(Obj, L.Ptr + L.C.cmdsize, LoadCommandIndex + 1); 203 } 204 205 template <typename T> 206 static void parseHeader(const MachOObjectFile &Obj, T &Header, 207 Error &Err) { 208 if (sizeof(T) > Obj.getData().size()) { 209 Err = malformedError("the mach header extends past the end of the " 210 "file"); 211 return; 212 } 213 if (auto HeaderOrErr = getStructOrErr<T>(Obj, getPtr(Obj, 0))) 214 Header = *HeaderOrErr; 215 else 216 Err = HeaderOrErr.takeError(); 217 } 218 219 // This is used to check for overlapping of Mach-O elements. 220 struct MachOElement { 221 uint64_t Offset; 222 uint64_t Size; 223 const char *Name; 224 }; 225 226 static Error checkOverlappingElement(std::list<MachOElement> &Elements, 227 uint64_t Offset, uint64_t Size, 228 const char *Name) { 229 if (Size == 0) 230 return Error::success(); 231 232 for (auto it=Elements.begin() ; it != Elements.end(); ++it) { 233 auto E = *it; 234 if ((Offset >= E.Offset && Offset < E.Offset + E.Size) || 235 (Offset + Size > E.Offset && Offset + Size < E.Offset + E.Size) || 236 (Offset <= E.Offset && Offset + Size >= E.Offset + E.Size)) 237 return malformedError(Twine(Name) + " at offset " + Twine(Offset) + 238 " with a size of " + Twine(Size) + ", overlaps " + 239 E.Name + " at offset " + Twine(E.Offset) + " with " 240 "a size of " + Twine(E.Size)); 241 auto nt = it; 242 nt++; 243 if (nt != Elements.end()) { 244 auto N = *nt; 245 if (Offset + Size <= N.Offset) { 246 Elements.insert(nt, {Offset, Size, Name}); 247 return Error::success(); 248 } 249 } 250 } 251 Elements.push_back({Offset, Size, Name}); 252 return Error::success(); 253 } 254 255 // Parses LC_SEGMENT or LC_SEGMENT_64 load command, adds addresses of all 256 // sections to \param Sections, and optionally sets 257 // \param IsPageZeroSegment to true. 258 template <typename Segment, typename Section> 259 static Error parseSegmentLoadCommand( 260 const MachOObjectFile &Obj, const MachOObjectFile::LoadCommandInfo &Load, 261 SmallVectorImpl<const char *> &Sections, bool &IsPageZeroSegment, 262 uint32_t LoadCommandIndex, const char *CmdName, uint64_t SizeOfHeaders, 263 std::list<MachOElement> &Elements) { 264 const unsigned SegmentLoadSize = sizeof(Segment); 265 if (Load.C.cmdsize < SegmentLoadSize) 266 return malformedError("load command " + Twine(LoadCommandIndex) + 267 " " + CmdName + " cmdsize too small"); 268 if (auto SegOrErr = getStructOrErr<Segment>(Obj, Load.Ptr)) { 269 Segment S = SegOrErr.get(); 270 const unsigned SectionSize = sizeof(Section); 271 uint64_t FileSize = Obj.getData().size(); 272 if (S.nsects > std::numeric_limits<uint32_t>::max() / SectionSize || 273 S.nsects * SectionSize > Load.C.cmdsize - SegmentLoadSize) 274 return malformedError("load command " + Twine(LoadCommandIndex) + 275 " inconsistent cmdsize in " + CmdName + 276 " for the number of sections"); 277 for (unsigned J = 0; J < S.nsects; ++J) { 278 const char *Sec = getSectionPtr(Obj, Load, J); 279 Sections.push_back(Sec); 280 Section s = getStruct<Section>(Obj, Sec); 281 if (Obj.getHeader().filetype != MachO::MH_DYLIB_STUB && 282 Obj.getHeader().filetype != MachO::MH_DSYM && 283 s.flags != MachO::S_ZEROFILL && 284 s.flags != MachO::S_THREAD_LOCAL_ZEROFILL && 285 s.offset > FileSize) 286 return malformedError("offset field of section " + Twine(J) + " in " + 287 CmdName + " command " + Twine(LoadCommandIndex) + 288 " extends past the end of the file"); 289 if (Obj.getHeader().filetype != MachO::MH_DYLIB_STUB && 290 Obj.getHeader().filetype != MachO::MH_DSYM && 291 s.flags != MachO::S_ZEROFILL && 292 s.flags != MachO::S_THREAD_LOCAL_ZEROFILL && S.fileoff == 0 && 293 s.offset < SizeOfHeaders && s.size != 0) 294 return malformedError("offset field of section " + Twine(J) + " in " + 295 CmdName + " command " + Twine(LoadCommandIndex) + 296 " not past the headers of the file"); 297 uint64_t BigSize = s.offset; 298 BigSize += s.size; 299 if (Obj.getHeader().filetype != MachO::MH_DYLIB_STUB && 300 Obj.getHeader().filetype != MachO::MH_DSYM && 301 s.flags != MachO::S_ZEROFILL && 302 s.flags != MachO::S_THREAD_LOCAL_ZEROFILL && 303 BigSize > FileSize) 304 return malformedError("offset field plus size field of section " + 305 Twine(J) + " in " + CmdName + " command " + 306 Twine(LoadCommandIndex) + 307 " extends past the end of the file"); 308 if (Obj.getHeader().filetype != MachO::MH_DYLIB_STUB && 309 Obj.getHeader().filetype != MachO::MH_DSYM && 310 s.flags != MachO::S_ZEROFILL && 311 s.flags != MachO::S_THREAD_LOCAL_ZEROFILL && 312 s.size > S.filesize) 313 return malformedError("size field of section " + 314 Twine(J) + " in " + CmdName + " command " + 315 Twine(LoadCommandIndex) + 316 " greater than the segment"); 317 if (Obj.getHeader().filetype != MachO::MH_DYLIB_STUB && 318 Obj.getHeader().filetype != MachO::MH_DSYM && s.size != 0 && 319 s.addr < S.vmaddr) 320 return malformedError("addr field of section " + Twine(J) + " in " + 321 CmdName + " command " + Twine(LoadCommandIndex) + 322 " less than the segment's vmaddr"); 323 BigSize = s.addr; 324 BigSize += s.size; 325 uint64_t BigEnd = S.vmaddr; 326 BigEnd += S.vmsize; 327 if (S.vmsize != 0 && s.size != 0 && BigSize > BigEnd) 328 return malformedError("addr field plus size of section " + Twine(J) + 329 " in " + CmdName + " command " + 330 Twine(LoadCommandIndex) + 331 " greater than than " 332 "the segment's vmaddr plus vmsize"); 333 if (Obj.getHeader().filetype != MachO::MH_DYLIB_STUB && 334 Obj.getHeader().filetype != MachO::MH_DSYM && 335 s.flags != MachO::S_ZEROFILL && 336 s.flags != MachO::S_THREAD_LOCAL_ZEROFILL) 337 if (Error Err = checkOverlappingElement(Elements, s.offset, s.size, 338 "section contents")) 339 return Err; 340 if (s.reloff > FileSize) 341 return malformedError("reloff field of section " + Twine(J) + " in " + 342 CmdName + " command " + Twine(LoadCommandIndex) + 343 " extends past the end of the file"); 344 BigSize = s.nreloc; 345 BigSize *= sizeof(struct MachO::relocation_info); 346 BigSize += s.reloff; 347 if (BigSize > FileSize) 348 return malformedError("reloff field plus nreloc field times sizeof(" 349 "struct relocation_info) of section " + 350 Twine(J) + " in " + CmdName + " command " + 351 Twine(LoadCommandIndex) + 352 " extends past the end of the file"); 353 if (Error Err = checkOverlappingElement(Elements, s.reloff, s.nreloc * 354 sizeof(struct 355 MachO::relocation_info), 356 "section relocation entries")) 357 return Err; 358 } 359 if (S.fileoff > FileSize) 360 return malformedError("load command " + Twine(LoadCommandIndex) + 361 " fileoff field in " + CmdName + 362 " extends past the end of the file"); 363 uint64_t BigSize = S.fileoff; 364 BigSize += S.filesize; 365 if (BigSize > FileSize) 366 return malformedError("load command " + Twine(LoadCommandIndex) + 367 " fileoff field plus filesize field in " + 368 CmdName + " extends past the end of the file"); 369 if (S.vmsize != 0 && S.filesize > S.vmsize) 370 return malformedError("load command " + Twine(LoadCommandIndex) + 371 " filesize field in " + CmdName + 372 " greater than vmsize field"); 373 IsPageZeroSegment |= StringRef("__PAGEZERO").equals(S.segname); 374 } else 375 return SegOrErr.takeError(); 376 377 return Error::success(); 378 } 379 380 static Error checkSymtabCommand(const MachOObjectFile &Obj, 381 const MachOObjectFile::LoadCommandInfo &Load, 382 uint32_t LoadCommandIndex, 383 const char **SymtabLoadCmd, 384 std::list<MachOElement> &Elements) { 385 if (Load.C.cmdsize < sizeof(MachO::symtab_command)) 386 return malformedError("load command " + Twine(LoadCommandIndex) + 387 " LC_SYMTAB cmdsize too small"); 388 if (*SymtabLoadCmd != nullptr) 389 return malformedError("more than one LC_SYMTAB command"); 390 MachO::symtab_command Symtab = 391 getStruct<MachO::symtab_command>(Obj, Load.Ptr); 392 if (Symtab.cmdsize != sizeof(MachO::symtab_command)) 393 return malformedError("LC_SYMTAB command " + Twine(LoadCommandIndex) + 394 " has incorrect cmdsize"); 395 uint64_t FileSize = Obj.getData().size(); 396 if (Symtab.symoff > FileSize) 397 return malformedError("symoff field of LC_SYMTAB command " + 398 Twine(LoadCommandIndex) + " extends past the end " 399 "of the file"); 400 uint64_t SymtabSize = Symtab.nsyms; 401 const char *struct_nlist_name; 402 if (Obj.is64Bit()) { 403 SymtabSize *= sizeof(MachO::nlist_64); 404 struct_nlist_name = "struct nlist_64"; 405 } else { 406 SymtabSize *= sizeof(MachO::nlist); 407 struct_nlist_name = "struct nlist"; 408 } 409 uint64_t BigSize = SymtabSize; 410 BigSize += Symtab.symoff; 411 if (BigSize > FileSize) 412 return malformedError("symoff field plus nsyms field times sizeof(" + 413 Twine(struct_nlist_name) + ") of LC_SYMTAB command " + 414 Twine(LoadCommandIndex) + " extends past the end " 415 "of the file"); 416 if (Error Err = checkOverlappingElement(Elements, Symtab.symoff, SymtabSize, 417 "symbol table")) 418 return Err; 419 if (Symtab.stroff > FileSize) 420 return malformedError("stroff field of LC_SYMTAB command " + 421 Twine(LoadCommandIndex) + " extends past the end " 422 "of the file"); 423 BigSize = Symtab.stroff; 424 BigSize += Symtab.strsize; 425 if (BigSize > FileSize) 426 return malformedError("stroff field plus strsize field of LC_SYMTAB " 427 "command " + Twine(LoadCommandIndex) + " extends " 428 "past the end of the file"); 429 if (Error Err = checkOverlappingElement(Elements, Symtab.stroff, 430 Symtab.strsize, "string table")) 431 return Err; 432 *SymtabLoadCmd = Load.Ptr; 433 return Error::success(); 434 } 435 436 static Error checkDysymtabCommand(const MachOObjectFile &Obj, 437 const MachOObjectFile::LoadCommandInfo &Load, 438 uint32_t LoadCommandIndex, 439 const char **DysymtabLoadCmd, 440 std::list<MachOElement> &Elements) { 441 if (Load.C.cmdsize < sizeof(MachO::dysymtab_command)) 442 return malformedError("load command " + Twine(LoadCommandIndex) + 443 " LC_DYSYMTAB cmdsize too small"); 444 if (*DysymtabLoadCmd != nullptr) 445 return malformedError("more than one LC_DYSYMTAB command"); 446 MachO::dysymtab_command Dysymtab = 447 getStruct<MachO::dysymtab_command>(Obj, Load.Ptr); 448 if (Dysymtab.cmdsize != sizeof(MachO::dysymtab_command)) 449 return malformedError("LC_DYSYMTAB command " + Twine(LoadCommandIndex) + 450 " has incorrect cmdsize"); 451 uint64_t FileSize = Obj.getData().size(); 452 if (Dysymtab.tocoff > FileSize) 453 return malformedError("tocoff field of LC_DYSYMTAB command " + 454 Twine(LoadCommandIndex) + " extends past the end of " 455 "the file"); 456 uint64_t BigSize = Dysymtab.ntoc; 457 BigSize *= sizeof(MachO::dylib_table_of_contents); 458 BigSize += Dysymtab.tocoff; 459 if (BigSize > FileSize) 460 return malformedError("tocoff field plus ntoc field times sizeof(struct " 461 "dylib_table_of_contents) of LC_DYSYMTAB command " + 462 Twine(LoadCommandIndex) + " extends past the end of " 463 "the file"); 464 if (Error Err = checkOverlappingElement(Elements, Dysymtab.tocoff, 465 Dysymtab.ntoc * sizeof(struct 466 MachO::dylib_table_of_contents), 467 "table of contents")) 468 return Err; 469 if (Dysymtab.modtaboff > FileSize) 470 return malformedError("modtaboff field of LC_DYSYMTAB command " + 471 Twine(LoadCommandIndex) + " extends past the end of " 472 "the file"); 473 BigSize = Dysymtab.nmodtab; 474 const char *struct_dylib_module_name; 475 uint64_t sizeof_modtab; 476 if (Obj.is64Bit()) { 477 sizeof_modtab = sizeof(MachO::dylib_module_64); 478 struct_dylib_module_name = "struct dylib_module_64"; 479 } else { 480 sizeof_modtab = sizeof(MachO::dylib_module); 481 struct_dylib_module_name = "struct dylib_module"; 482 } 483 BigSize *= sizeof_modtab; 484 BigSize += Dysymtab.modtaboff; 485 if (BigSize > FileSize) 486 return malformedError("modtaboff field plus nmodtab field times sizeof(" + 487 Twine(struct_dylib_module_name) + ") of LC_DYSYMTAB " 488 "command " + Twine(LoadCommandIndex) + " extends " 489 "past the end of the file"); 490 if (Error Err = checkOverlappingElement(Elements, Dysymtab.modtaboff, 491 Dysymtab.nmodtab * sizeof_modtab, 492 "module table")) 493 return Err; 494 if (Dysymtab.extrefsymoff > FileSize) 495 return malformedError("extrefsymoff field of LC_DYSYMTAB command " + 496 Twine(LoadCommandIndex) + " extends past the end of " 497 "the file"); 498 BigSize = Dysymtab.nextrefsyms; 499 BigSize *= sizeof(MachO::dylib_reference); 500 BigSize += Dysymtab.extrefsymoff; 501 if (BigSize > FileSize) 502 return malformedError("extrefsymoff field plus nextrefsyms field times " 503 "sizeof(struct dylib_reference) of LC_DYSYMTAB " 504 "command " + Twine(LoadCommandIndex) + " extends " 505 "past the end of the file"); 506 if (Error Err = checkOverlappingElement(Elements, Dysymtab.extrefsymoff, 507 Dysymtab.nextrefsyms * 508 sizeof(MachO::dylib_reference), 509 "reference table")) 510 return Err; 511 if (Dysymtab.indirectsymoff > FileSize) 512 return malformedError("indirectsymoff field of LC_DYSYMTAB command " + 513 Twine(LoadCommandIndex) + " extends past the end of " 514 "the file"); 515 BigSize = Dysymtab.nindirectsyms; 516 BigSize *= sizeof(uint32_t); 517 BigSize += Dysymtab.indirectsymoff; 518 if (BigSize > FileSize) 519 return malformedError("indirectsymoff field plus nindirectsyms field times " 520 "sizeof(uint32_t) of LC_DYSYMTAB command " + 521 Twine(LoadCommandIndex) + " extends past the end of " 522 "the file"); 523 if (Error Err = checkOverlappingElement(Elements, Dysymtab.indirectsymoff, 524 Dysymtab.nindirectsyms * 525 sizeof(uint32_t), 526 "indirect table")) 527 return Err; 528 if (Dysymtab.extreloff > FileSize) 529 return malformedError("extreloff field of LC_DYSYMTAB command " + 530 Twine(LoadCommandIndex) + " extends past the end of " 531 "the file"); 532 BigSize = Dysymtab.nextrel; 533 BigSize *= sizeof(MachO::relocation_info); 534 BigSize += Dysymtab.extreloff; 535 if (BigSize > FileSize) 536 return malformedError("extreloff field plus nextrel field times sizeof" 537 "(struct relocation_info) of LC_DYSYMTAB command " + 538 Twine(LoadCommandIndex) + " extends past the end of " 539 "the file"); 540 if (Error Err = checkOverlappingElement(Elements, Dysymtab.extreloff, 541 Dysymtab.nextrel * 542 sizeof(MachO::relocation_info), 543 "external relocation table")) 544 return Err; 545 if (Dysymtab.locreloff > FileSize) 546 return malformedError("locreloff field of LC_DYSYMTAB command " + 547 Twine(LoadCommandIndex) + " extends past the end of " 548 "the file"); 549 BigSize = Dysymtab.nlocrel; 550 BigSize *= sizeof(MachO::relocation_info); 551 BigSize += Dysymtab.locreloff; 552 if (BigSize > FileSize) 553 return malformedError("locreloff field plus nlocrel field times sizeof" 554 "(struct relocation_info) of LC_DYSYMTAB command " + 555 Twine(LoadCommandIndex) + " extends past the end of " 556 "the file"); 557 if (Error Err = checkOverlappingElement(Elements, Dysymtab.locreloff, 558 Dysymtab.nlocrel * 559 sizeof(MachO::relocation_info), 560 "local relocation table")) 561 return Err; 562 *DysymtabLoadCmd = Load.Ptr; 563 return Error::success(); 564 } 565 566 static Error checkLinkeditDataCommand(const MachOObjectFile &Obj, 567 const MachOObjectFile::LoadCommandInfo &Load, 568 uint32_t LoadCommandIndex, 569 const char **LoadCmd, const char *CmdName, 570 std::list<MachOElement> &Elements, 571 const char *ElementName) { 572 if (Load.C.cmdsize < sizeof(MachO::linkedit_data_command)) 573 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 574 CmdName + " cmdsize too small"); 575 if (*LoadCmd != nullptr) 576 return malformedError("more than one " + Twine(CmdName) + " command"); 577 MachO::linkedit_data_command LinkData = 578 getStruct<MachO::linkedit_data_command>(Obj, Load.Ptr); 579 if (LinkData.cmdsize != sizeof(MachO::linkedit_data_command)) 580 return malformedError(Twine(CmdName) + " command " + 581 Twine(LoadCommandIndex) + " has incorrect cmdsize"); 582 uint64_t FileSize = Obj.getData().size(); 583 if (LinkData.dataoff > FileSize) 584 return malformedError("dataoff field of " + Twine(CmdName) + " command " + 585 Twine(LoadCommandIndex) + " extends past the end of " 586 "the file"); 587 uint64_t BigSize = LinkData.dataoff; 588 BigSize += LinkData.datasize; 589 if (BigSize > FileSize) 590 return malformedError("dataoff field plus datasize field of " + 591 Twine(CmdName) + " command " + 592 Twine(LoadCommandIndex) + " extends past the end of " 593 "the file"); 594 if (Error Err = checkOverlappingElement(Elements, LinkData.dataoff, 595 LinkData.datasize, ElementName)) 596 return Err; 597 *LoadCmd = Load.Ptr; 598 return Error::success(); 599 } 600 601 static Error checkDyldInfoCommand(const MachOObjectFile &Obj, 602 const MachOObjectFile::LoadCommandInfo &Load, 603 uint32_t LoadCommandIndex, 604 const char **LoadCmd, const char *CmdName, 605 std::list<MachOElement> &Elements) { 606 if (Load.C.cmdsize < sizeof(MachO::dyld_info_command)) 607 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 608 CmdName + " cmdsize too small"); 609 if (*LoadCmd != nullptr) 610 return malformedError("more than one LC_DYLD_INFO and or LC_DYLD_INFO_ONLY " 611 "command"); 612 MachO::dyld_info_command DyldInfo = 613 getStruct<MachO::dyld_info_command>(Obj, Load.Ptr); 614 if (DyldInfo.cmdsize != sizeof(MachO::dyld_info_command)) 615 return malformedError(Twine(CmdName) + " command " + 616 Twine(LoadCommandIndex) + " has incorrect cmdsize"); 617 uint64_t FileSize = Obj.getData().size(); 618 if (DyldInfo.rebase_off > FileSize) 619 return malformedError("rebase_off field of " + Twine(CmdName) + 620 " command " + Twine(LoadCommandIndex) + " extends " 621 "past the end of the file"); 622 uint64_t BigSize = DyldInfo.rebase_off; 623 BigSize += DyldInfo.rebase_size; 624 if (BigSize > FileSize) 625 return malformedError("rebase_off field plus rebase_size field of " + 626 Twine(CmdName) + " command " + 627 Twine(LoadCommandIndex) + " extends past the end of " 628 "the file"); 629 if (Error Err = checkOverlappingElement(Elements, DyldInfo.rebase_off, 630 DyldInfo.rebase_size, 631 "dyld rebase info")) 632 return Err; 633 if (DyldInfo.bind_off > FileSize) 634 return malformedError("bind_off field of " + Twine(CmdName) + 635 " command " + Twine(LoadCommandIndex) + " extends " 636 "past the end of the file"); 637 BigSize = DyldInfo.bind_off; 638 BigSize += DyldInfo.bind_size; 639 if (BigSize > FileSize) 640 return malformedError("bind_off field plus bind_size field of " + 641 Twine(CmdName) + " command " + 642 Twine(LoadCommandIndex) + " extends past the end of " 643 "the file"); 644 if (Error Err = checkOverlappingElement(Elements, DyldInfo.bind_off, 645 DyldInfo.bind_size, 646 "dyld bind info")) 647 return Err; 648 if (DyldInfo.weak_bind_off > FileSize) 649 return malformedError("weak_bind_off field of " + Twine(CmdName) + 650 " command " + Twine(LoadCommandIndex) + " extends " 651 "past the end of the file"); 652 BigSize = DyldInfo.weak_bind_off; 653 BigSize += DyldInfo.weak_bind_size; 654 if (BigSize > FileSize) 655 return malformedError("weak_bind_off field plus weak_bind_size field of " + 656 Twine(CmdName) + " command " + 657 Twine(LoadCommandIndex) + " extends past the end of " 658 "the file"); 659 if (Error Err = checkOverlappingElement(Elements, DyldInfo.weak_bind_off, 660 DyldInfo.weak_bind_size, 661 "dyld weak bind info")) 662 return Err; 663 if (DyldInfo.lazy_bind_off > FileSize) 664 return malformedError("lazy_bind_off field of " + Twine(CmdName) + 665 " command " + Twine(LoadCommandIndex) + " extends " 666 "past the end of the file"); 667 BigSize = DyldInfo.lazy_bind_off; 668 BigSize += DyldInfo.lazy_bind_size; 669 if (BigSize > FileSize) 670 return malformedError("lazy_bind_off field plus lazy_bind_size field of " + 671 Twine(CmdName) + " command " + 672 Twine(LoadCommandIndex) + " extends past the end of " 673 "the file"); 674 if (Error Err = checkOverlappingElement(Elements, DyldInfo.lazy_bind_off, 675 DyldInfo.lazy_bind_size, 676 "dyld lazy bind info")) 677 return Err; 678 if (DyldInfo.export_off > FileSize) 679 return malformedError("export_off field of " + Twine(CmdName) + 680 " command " + Twine(LoadCommandIndex) + " extends " 681 "past the end of the file"); 682 BigSize = DyldInfo.export_off; 683 BigSize += DyldInfo.export_size; 684 if (BigSize > FileSize) 685 return malformedError("export_off field plus export_size field of " + 686 Twine(CmdName) + " command " + 687 Twine(LoadCommandIndex) + " extends past the end of " 688 "the file"); 689 if (Error Err = checkOverlappingElement(Elements, DyldInfo.export_off, 690 DyldInfo.export_size, 691 "dyld export info")) 692 return Err; 693 *LoadCmd = Load.Ptr; 694 return Error::success(); 695 } 696 697 static Error checkDylibCommand(const MachOObjectFile &Obj, 698 const MachOObjectFile::LoadCommandInfo &Load, 699 uint32_t LoadCommandIndex, const char *CmdName) { 700 if (Load.C.cmdsize < sizeof(MachO::dylib_command)) 701 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 702 CmdName + " cmdsize too small"); 703 MachO::dylib_command D = getStruct<MachO::dylib_command>(Obj, Load.Ptr); 704 if (D.dylib.name < sizeof(MachO::dylib_command)) 705 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 706 CmdName + " name.offset field too small, not past " 707 "the end of the dylib_command struct"); 708 if (D.dylib.name >= D.cmdsize) 709 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 710 CmdName + " name.offset field extends past the end " 711 "of the load command"); 712 // Make sure there is a null between the starting offset of the name and 713 // the end of the load command. 714 uint32_t i; 715 const char *P = (const char *)Load.Ptr; 716 for (i = D.dylib.name; i < D.cmdsize; i++) 717 if (P[i] == '\0') 718 break; 719 if (i >= D.cmdsize) 720 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 721 CmdName + " library name extends past the end of the " 722 "load command"); 723 return Error::success(); 724 } 725 726 static Error checkDylibIdCommand(const MachOObjectFile &Obj, 727 const MachOObjectFile::LoadCommandInfo &Load, 728 uint32_t LoadCommandIndex, 729 const char **LoadCmd) { 730 if (Error Err = checkDylibCommand(Obj, Load, LoadCommandIndex, 731 "LC_ID_DYLIB")) 732 return Err; 733 if (*LoadCmd != nullptr) 734 return malformedError("more than one LC_ID_DYLIB command"); 735 if (Obj.getHeader().filetype != MachO::MH_DYLIB && 736 Obj.getHeader().filetype != MachO::MH_DYLIB_STUB) 737 return malformedError("LC_ID_DYLIB load command in non-dynamic library " 738 "file type"); 739 *LoadCmd = Load.Ptr; 740 return Error::success(); 741 } 742 743 static Error checkDyldCommand(const MachOObjectFile &Obj, 744 const MachOObjectFile::LoadCommandInfo &Load, 745 uint32_t LoadCommandIndex, const char *CmdName) { 746 if (Load.C.cmdsize < sizeof(MachO::dylinker_command)) 747 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 748 CmdName + " cmdsize too small"); 749 MachO::dylinker_command D = getStruct<MachO::dylinker_command>(Obj, Load.Ptr); 750 if (D.name < sizeof(MachO::dylinker_command)) 751 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 752 CmdName + " name.offset field too small, not past " 753 "the end of the dylinker_command struct"); 754 if (D.name >= D.cmdsize) 755 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 756 CmdName + " name.offset field extends past the end " 757 "of the load command"); 758 // Make sure there is a null between the starting offset of the name and 759 // the end of the load command. 760 uint32_t i; 761 const char *P = (const char *)Load.Ptr; 762 for (i = D.name; i < D.cmdsize; i++) 763 if (P[i] == '\0') 764 break; 765 if (i >= D.cmdsize) 766 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 767 CmdName + " dyld name extends past the end of the " 768 "load command"); 769 return Error::success(); 770 } 771 772 static Error checkVersCommand(const MachOObjectFile &Obj, 773 const MachOObjectFile::LoadCommandInfo &Load, 774 uint32_t LoadCommandIndex, 775 const char **LoadCmd, const char *CmdName) { 776 if (Load.C.cmdsize != sizeof(MachO::version_min_command)) 777 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 778 CmdName + " has incorrect cmdsize"); 779 if (*LoadCmd != nullptr) 780 return malformedError("more than one LC_VERSION_MIN_MACOSX, " 781 "LC_VERSION_MIN_IPHONEOS, LC_VERSION_MIN_TVOS or " 782 "LC_VERSION_MIN_WATCHOS command"); 783 *LoadCmd = Load.Ptr; 784 return Error::success(); 785 } 786 787 static Error checkNoteCommand(const MachOObjectFile &Obj, 788 const MachOObjectFile::LoadCommandInfo &Load, 789 uint32_t LoadCommandIndex, 790 std::list<MachOElement> &Elements) { 791 if (Load.C.cmdsize != sizeof(MachO::note_command)) 792 return malformedError("load command " + Twine(LoadCommandIndex) + 793 " LC_NOTE has incorrect cmdsize"); 794 MachO::note_command Nt = getStruct<MachO::note_command>(Obj, Load.Ptr); 795 uint64_t FileSize = Obj.getData().size(); 796 if (Nt.offset > FileSize) 797 return malformedError("offset field of LC_NOTE command " + 798 Twine(LoadCommandIndex) + " extends " 799 "past the end of the file"); 800 uint64_t BigSize = Nt.offset; 801 BigSize += Nt.size; 802 if (BigSize > FileSize) 803 return malformedError("size field plus offset field of LC_NOTE command " + 804 Twine(LoadCommandIndex) + " extends past the end of " 805 "the file"); 806 if (Error Err = checkOverlappingElement(Elements, Nt.offset, Nt.size, 807 "LC_NOTE data")) 808 return Err; 809 return Error::success(); 810 } 811 812 static Error 813 parseBuildVersionCommand(const MachOObjectFile &Obj, 814 const MachOObjectFile::LoadCommandInfo &Load, 815 SmallVectorImpl<const char*> &BuildTools, 816 uint32_t LoadCommandIndex) { 817 MachO::build_version_command BVC = 818 getStruct<MachO::build_version_command>(Obj, Load.Ptr); 819 if (Load.C.cmdsize != 820 sizeof(MachO::build_version_command) + 821 BVC.ntools * sizeof(MachO::build_tool_version)) 822 return malformedError("load command " + Twine(LoadCommandIndex) + 823 " LC_BUILD_VERSION_COMMAND has incorrect cmdsize"); 824 825 auto Start = Load.Ptr + sizeof(MachO::build_version_command); 826 BuildTools.resize(BVC.ntools); 827 for (unsigned i = 0; i < BVC.ntools; ++i) 828 BuildTools[i] = Start + i * sizeof(MachO::build_tool_version); 829 830 return Error::success(); 831 } 832 833 static Error checkRpathCommand(const MachOObjectFile &Obj, 834 const MachOObjectFile::LoadCommandInfo &Load, 835 uint32_t LoadCommandIndex) { 836 if (Load.C.cmdsize < sizeof(MachO::rpath_command)) 837 return malformedError("load command " + Twine(LoadCommandIndex) + 838 " LC_RPATH cmdsize too small"); 839 MachO::rpath_command R = getStruct<MachO::rpath_command>(Obj, Load.Ptr); 840 if (R.path < sizeof(MachO::rpath_command)) 841 return malformedError("load command " + Twine(LoadCommandIndex) + 842 " LC_RPATH path.offset field too small, not past " 843 "the end of the rpath_command struct"); 844 if (R.path >= R.cmdsize) 845 return malformedError("load command " + Twine(LoadCommandIndex) + 846 " LC_RPATH path.offset field extends past the end " 847 "of the load command"); 848 // Make sure there is a null between the starting offset of the path and 849 // the end of the load command. 850 uint32_t i; 851 const char *P = (const char *)Load.Ptr; 852 for (i = R.path; i < R.cmdsize; i++) 853 if (P[i] == '\0') 854 break; 855 if (i >= R.cmdsize) 856 return malformedError("load command " + Twine(LoadCommandIndex) + 857 " LC_RPATH library name extends past the end of the " 858 "load command"); 859 return Error::success(); 860 } 861 862 static Error checkEncryptCommand(const MachOObjectFile &Obj, 863 const MachOObjectFile::LoadCommandInfo &Load, 864 uint32_t LoadCommandIndex, 865 uint64_t cryptoff, uint64_t cryptsize, 866 const char **LoadCmd, const char *CmdName) { 867 if (*LoadCmd != nullptr) 868 return malformedError("more than one LC_ENCRYPTION_INFO and or " 869 "LC_ENCRYPTION_INFO_64 command"); 870 uint64_t FileSize = Obj.getData().size(); 871 if (cryptoff > FileSize) 872 return malformedError("cryptoff field of " + Twine(CmdName) + 873 " command " + Twine(LoadCommandIndex) + " extends " 874 "past the end of the file"); 875 uint64_t BigSize = cryptoff; 876 BigSize += cryptsize; 877 if (BigSize > FileSize) 878 return malformedError("cryptoff field plus cryptsize field of " + 879 Twine(CmdName) + " command " + 880 Twine(LoadCommandIndex) + " extends past the end of " 881 "the file"); 882 *LoadCmd = Load.Ptr; 883 return Error::success(); 884 } 885 886 static Error checkLinkerOptCommand(const MachOObjectFile &Obj, 887 const MachOObjectFile::LoadCommandInfo &Load, 888 uint32_t LoadCommandIndex) { 889 if (Load.C.cmdsize < sizeof(MachO::linker_option_command)) 890 return malformedError("load command " + Twine(LoadCommandIndex) + 891 " LC_LINKER_OPTION cmdsize too small"); 892 MachO::linker_option_command L = 893 getStruct<MachO::linker_option_command>(Obj, Load.Ptr); 894 // Make sure the count of strings is correct. 895 const char *string = (const char *)Load.Ptr + 896 sizeof(struct MachO::linker_option_command); 897 uint32_t left = L.cmdsize - sizeof(struct MachO::linker_option_command); 898 uint32_t i = 0; 899 while (left > 0) { 900 while (*string == '\0' && left > 0) { 901 string++; 902 left--; 903 } 904 if (left > 0) { 905 i++; 906 uint32_t NullPos = StringRef(string, left).find('\0'); 907 uint32_t len = std::min(NullPos, left) + 1; 908 string += len; 909 left -= len; 910 } 911 } 912 if (L.count != i) 913 return malformedError("load command " + Twine(LoadCommandIndex) + 914 " LC_LINKER_OPTION string count " + Twine(L.count) + 915 " does not match number of strings"); 916 return Error::success(); 917 } 918 919 static Error checkSubCommand(const MachOObjectFile &Obj, 920 const MachOObjectFile::LoadCommandInfo &Load, 921 uint32_t LoadCommandIndex, const char *CmdName, 922 size_t SizeOfCmd, const char *CmdStructName, 923 uint32_t PathOffset, const char *PathFieldName) { 924 if (PathOffset < SizeOfCmd) 925 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 926 CmdName + " " + PathFieldName + ".offset field too " 927 "small, not past the end of the " + CmdStructName); 928 if (PathOffset >= Load.C.cmdsize) 929 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 930 CmdName + " " + PathFieldName + ".offset field " 931 "extends past the end of the load command"); 932 // Make sure there is a null between the starting offset of the path and 933 // the end of the load command. 934 uint32_t i; 935 const char *P = (const char *)Load.Ptr; 936 for (i = PathOffset; i < Load.C.cmdsize; i++) 937 if (P[i] == '\0') 938 break; 939 if (i >= Load.C.cmdsize) 940 return malformedError("load command " + Twine(LoadCommandIndex) + " " + 941 CmdName + " " + PathFieldName + " name extends past " 942 "the end of the load command"); 943 return Error::success(); 944 } 945 946 static Error checkThreadCommand(const MachOObjectFile &Obj, 947 const MachOObjectFile::LoadCommandInfo &Load, 948 uint32_t LoadCommandIndex, 949 const char *CmdName) { 950 if (Load.C.cmdsize < sizeof(MachO::thread_command)) 951 return malformedError("load command " + Twine(LoadCommandIndex) + 952 CmdName + " cmdsize too small"); 953 MachO::thread_command T = 954 getStruct<MachO::thread_command>(Obj, Load.Ptr); 955 const char *state = Load.Ptr + sizeof(MachO::thread_command); 956 const char *end = Load.Ptr + T.cmdsize; 957 uint32_t nflavor = 0; 958 uint32_t cputype = getCPUType(Obj); 959 while (state < end) { 960 if(state + sizeof(uint32_t) > end) 961 return malformedError("load command " + Twine(LoadCommandIndex) + 962 "flavor in " + CmdName + " extends past end of " 963 "command"); 964 uint32_t flavor; 965 memcpy(&flavor, state, sizeof(uint32_t)); 966 if (Obj.isLittleEndian() != sys::IsLittleEndianHost) 967 sys::swapByteOrder(flavor); 968 state += sizeof(uint32_t); 969 970 if(state + sizeof(uint32_t) > end) 971 return malformedError("load command " + Twine(LoadCommandIndex) + 972 " count in " + CmdName + " extends past end of " 973 "command"); 974 uint32_t count; 975 memcpy(&count, state, sizeof(uint32_t)); 976 if (Obj.isLittleEndian() != sys::IsLittleEndianHost) 977 sys::swapByteOrder(count); 978 state += sizeof(uint32_t); 979 980 if (cputype == MachO::CPU_TYPE_I386) { 981 if (flavor == MachO::x86_THREAD_STATE32) { 982 if (count != MachO::x86_THREAD_STATE32_COUNT) 983 return malformedError("load command " + Twine(LoadCommandIndex) + 984 " count not x86_THREAD_STATE32_COUNT for " 985 "flavor number " + Twine(nflavor) + " which is " 986 "a x86_THREAD_STATE32 flavor in " + CmdName + 987 " command"); 988 if (state + sizeof(MachO::x86_thread_state32_t) > end) 989 return malformedError("load command " + Twine(LoadCommandIndex) + 990 " x86_THREAD_STATE32 extends past end of " 991 "command in " + CmdName + " command"); 992 state += sizeof(MachO::x86_thread_state32_t); 993 } else { 994 return malformedError("load command " + Twine(LoadCommandIndex) + 995 " unknown flavor (" + Twine(flavor) + ") for " 996 "flavor number " + Twine(nflavor) + " in " + 997 CmdName + " command"); 998 } 999 } else if (cputype == MachO::CPU_TYPE_X86_64) { 1000 if (flavor == MachO::x86_THREAD_STATE64) { 1001 if (count != MachO::x86_THREAD_STATE64_COUNT) 1002 return malformedError("load command " + Twine(LoadCommandIndex) + 1003 " count not x86_THREAD_STATE64_COUNT for " 1004 "flavor number " + Twine(nflavor) + " which is " 1005 "a x86_THREAD_STATE64 flavor in " + CmdName + 1006 " command"); 1007 if (state + sizeof(MachO::x86_thread_state64_t) > end) 1008 return malformedError("load command " + Twine(LoadCommandIndex) + 1009 " x86_THREAD_STATE64 extends past end of " 1010 "command in " + CmdName + " command"); 1011 state += sizeof(MachO::x86_thread_state64_t); 1012 } else { 1013 return malformedError("load command " + Twine(LoadCommandIndex) + 1014 " unknown flavor (" + Twine(flavor) + ") for " 1015 "flavor number " + Twine(nflavor) + " in " + 1016 CmdName + " command"); 1017 } 1018 } else if (cputype == MachO::CPU_TYPE_ARM) { 1019 if (flavor == MachO::ARM_THREAD_STATE) { 1020 if (count != MachO::ARM_THREAD_STATE_COUNT) 1021 return malformedError("load command " + Twine(LoadCommandIndex) + 1022 " count not ARM_THREAD_STATE_COUNT for " 1023 "flavor number " + Twine(nflavor) + " which is " 1024 "a ARM_THREAD_STATE flavor in " + CmdName + 1025 " command"); 1026 if (state + sizeof(MachO::arm_thread_state32_t) > end) 1027 return malformedError("load command " + Twine(LoadCommandIndex) + 1028 " ARM_THREAD_STATE extends past end of " 1029 "command in " + CmdName + " command"); 1030 state += sizeof(MachO::arm_thread_state32_t); 1031 } else { 1032 return malformedError("load command " + Twine(LoadCommandIndex) + 1033 " unknown flavor (" + Twine(flavor) + ") for " 1034 "flavor number " + Twine(nflavor) + " in " + 1035 CmdName + " command"); 1036 } 1037 } else if (cputype == MachO::CPU_TYPE_ARM64) { 1038 if (flavor == MachO::ARM_THREAD_STATE64) { 1039 if (count != MachO::ARM_THREAD_STATE64_COUNT) 1040 return malformedError("load command " + Twine(LoadCommandIndex) + 1041 " count not ARM_THREAD_STATE64_COUNT for " 1042 "flavor number " + Twine(nflavor) + " which is " 1043 "a ARM_THREAD_STATE64 flavor in " + CmdName + 1044 " command"); 1045 if (state + sizeof(MachO::arm_thread_state64_t) > end) 1046 return malformedError("load command " + Twine(LoadCommandIndex) + 1047 " ARM_THREAD_STATE64 extends past end of " 1048 "command in " + CmdName + " command"); 1049 state += sizeof(MachO::arm_thread_state64_t); 1050 } else { 1051 return malformedError("load command " + Twine(LoadCommandIndex) + 1052 " unknown flavor (" + Twine(flavor) + ") for " 1053 "flavor number " + Twine(nflavor) + " in " + 1054 CmdName + " command"); 1055 } 1056 } else if (cputype == MachO::CPU_TYPE_POWERPC) { 1057 if (flavor == MachO::PPC_THREAD_STATE) { 1058 if (count != MachO::PPC_THREAD_STATE_COUNT) 1059 return malformedError("load command " + Twine(LoadCommandIndex) + 1060 " count not PPC_THREAD_STATE_COUNT for " 1061 "flavor number " + Twine(nflavor) + " which is " 1062 "a PPC_THREAD_STATE flavor in " + CmdName + 1063 " command"); 1064 if (state + sizeof(MachO::ppc_thread_state32_t) > end) 1065 return malformedError("load command " + Twine(LoadCommandIndex) + 1066 " PPC_THREAD_STATE extends past end of " 1067 "command in " + CmdName + " command"); 1068 state += sizeof(MachO::ppc_thread_state32_t); 1069 } else { 1070 return malformedError("load command " + Twine(LoadCommandIndex) + 1071 " unknown flavor (" + Twine(flavor) + ") for " 1072 "flavor number " + Twine(nflavor) + " in " + 1073 CmdName + " command"); 1074 } 1075 } else { 1076 return malformedError("unknown cputype (" + Twine(cputype) + ") load " 1077 "command " + Twine(LoadCommandIndex) + " for " + 1078 CmdName + " command can't be checked"); 1079 } 1080 nflavor++; 1081 } 1082 return Error::success(); 1083 } 1084 1085 static Error checkTwoLevelHintsCommand(const MachOObjectFile &Obj, 1086 const MachOObjectFile::LoadCommandInfo 1087 &Load, 1088 uint32_t LoadCommandIndex, 1089 const char **LoadCmd, 1090 std::list<MachOElement> &Elements) { 1091 if (Load.C.cmdsize != sizeof(MachO::twolevel_hints_command)) 1092 return malformedError("load command " + Twine(LoadCommandIndex) + 1093 " LC_TWOLEVEL_HINTS has incorrect cmdsize"); 1094 if (*LoadCmd != nullptr) 1095 return malformedError("more than one LC_TWOLEVEL_HINTS command"); 1096 MachO::twolevel_hints_command Hints = 1097 getStruct<MachO::twolevel_hints_command>(Obj, Load.Ptr); 1098 uint64_t FileSize = Obj.getData().size(); 1099 if (Hints.offset > FileSize) 1100 return malformedError("offset field of LC_TWOLEVEL_HINTS command " + 1101 Twine(LoadCommandIndex) + " extends past the end of " 1102 "the file"); 1103 uint64_t BigSize = Hints.nhints; 1104 BigSize *= Hints.nhints * sizeof(MachO::twolevel_hint); 1105 BigSize += Hints.offset; 1106 if (BigSize > FileSize) 1107 return malformedError("offset field plus nhints times sizeof(struct " 1108 "twolevel_hint) field of LC_TWOLEVEL_HINTS command " + 1109 Twine(LoadCommandIndex) + " extends past the end of " 1110 "the file"); 1111 if (Error Err = checkOverlappingElement(Elements, Hints.offset, Hints.nhints * 1112 sizeof(MachO::twolevel_hint), 1113 "two level hints")) 1114 return Err; 1115 *LoadCmd = Load.Ptr; 1116 return Error::success(); 1117 } 1118 1119 // Returns true if the libObject code does not support the load command and its 1120 // contents. The cmd value it is treated as an unknown load command but with 1121 // an error message that says the cmd value is obsolete. 1122 static bool isLoadCommandObsolete(uint32_t cmd) { 1123 if (cmd == MachO::LC_SYMSEG || 1124 cmd == MachO::LC_LOADFVMLIB || 1125 cmd == MachO::LC_IDFVMLIB || 1126 cmd == MachO::LC_IDENT || 1127 cmd == MachO::LC_FVMFILE || 1128 cmd == MachO::LC_PREPAGE || 1129 cmd == MachO::LC_PREBOUND_DYLIB || 1130 cmd == MachO::LC_TWOLEVEL_HINTS || 1131 cmd == MachO::LC_PREBIND_CKSUM) 1132 return true; 1133 return false; 1134 } 1135 1136 Expected<std::unique_ptr<MachOObjectFile>> 1137 MachOObjectFile::create(MemoryBufferRef Object, bool IsLittleEndian, 1138 bool Is64Bits, uint32_t UniversalCputype, 1139 uint32_t UniversalIndex) { 1140 Error Err = Error::success(); 1141 std::unique_ptr<MachOObjectFile> Obj( 1142 new MachOObjectFile(std::move(Object), IsLittleEndian, 1143 Is64Bits, Err, UniversalCputype, 1144 UniversalIndex)); 1145 if (Err) 1146 return std::move(Err); 1147 return std::move(Obj); 1148 } 1149 1150 MachOObjectFile::MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, 1151 bool Is64bits, Error &Err, 1152 uint32_t UniversalCputype, 1153 uint32_t UniversalIndex) 1154 : ObjectFile(getMachOType(IsLittleEndian, Is64bits), Object), 1155 SymtabLoadCmd(nullptr), DysymtabLoadCmd(nullptr), 1156 DataInCodeLoadCmd(nullptr), LinkOptHintsLoadCmd(nullptr), 1157 DyldInfoLoadCmd(nullptr), UuidLoadCmd(nullptr), 1158 HasPageZeroSegment(false) { 1159 ErrorAsOutParameter ErrAsOutParam(&Err); 1160 uint64_t SizeOfHeaders; 1161 uint32_t cputype; 1162 if (is64Bit()) { 1163 parseHeader(*this, Header64, Err); 1164 SizeOfHeaders = sizeof(MachO::mach_header_64); 1165 cputype = Header64.cputype; 1166 } else { 1167 parseHeader(*this, Header, Err); 1168 SizeOfHeaders = sizeof(MachO::mach_header); 1169 cputype = Header.cputype; 1170 } 1171 if (Err) 1172 return; 1173 SizeOfHeaders += getHeader().sizeofcmds; 1174 if (getData().data() + SizeOfHeaders > getData().end()) { 1175 Err = malformedError("load commands extend past the end of the file"); 1176 return; 1177 } 1178 if (UniversalCputype != 0 && cputype != UniversalCputype) { 1179 Err = malformedError("universal header architecture: " + 1180 Twine(UniversalIndex) + "'s cputype does not match " 1181 "object file's mach header"); 1182 return; 1183 } 1184 std::list<MachOElement> Elements; 1185 Elements.push_back({0, SizeOfHeaders, "Mach-O headers"}); 1186 1187 uint32_t LoadCommandCount = getHeader().ncmds; 1188 LoadCommandInfo Load; 1189 if (LoadCommandCount != 0) { 1190 if (auto LoadOrErr = getFirstLoadCommandInfo(*this)) 1191 Load = *LoadOrErr; 1192 else { 1193 Err = LoadOrErr.takeError(); 1194 return; 1195 } 1196 } 1197 1198 const char *DyldIdLoadCmd = nullptr; 1199 const char *FuncStartsLoadCmd = nullptr; 1200 const char *SplitInfoLoadCmd = nullptr; 1201 const char *CodeSignDrsLoadCmd = nullptr; 1202 const char *CodeSignLoadCmd = nullptr; 1203 const char *VersLoadCmd = nullptr; 1204 const char *SourceLoadCmd = nullptr; 1205 const char *EntryPointLoadCmd = nullptr; 1206 const char *EncryptLoadCmd = nullptr; 1207 const char *RoutinesLoadCmd = nullptr; 1208 const char *UnixThreadLoadCmd = nullptr; 1209 const char *TwoLevelHintsLoadCmd = nullptr; 1210 for (unsigned I = 0; I < LoadCommandCount; ++I) { 1211 if (is64Bit()) { 1212 if (Load.C.cmdsize % 8 != 0) { 1213 // We have a hack here to allow 64-bit Mach-O core files to have 1214 // LC_THREAD commands that are only a multiple of 4 and not 8 to be 1215 // allowed since the macOS kernel produces them. 1216 if (getHeader().filetype != MachO::MH_CORE || 1217 Load.C.cmd != MachO::LC_THREAD || Load.C.cmdsize % 4) { 1218 Err = malformedError("load command " + Twine(I) + " cmdsize not a " 1219 "multiple of 8"); 1220 return; 1221 } 1222 } 1223 } else { 1224 if (Load.C.cmdsize % 4 != 0) { 1225 Err = malformedError("load command " + Twine(I) + " cmdsize not a " 1226 "multiple of 4"); 1227 return; 1228 } 1229 } 1230 LoadCommands.push_back(Load); 1231 if (Load.C.cmd == MachO::LC_SYMTAB) { 1232 if ((Err = checkSymtabCommand(*this, Load, I, &SymtabLoadCmd, Elements))) 1233 return; 1234 } else if (Load.C.cmd == MachO::LC_DYSYMTAB) { 1235 if ((Err = checkDysymtabCommand(*this, Load, I, &DysymtabLoadCmd, 1236 Elements))) 1237 return; 1238 } else if (Load.C.cmd == MachO::LC_DATA_IN_CODE) { 1239 if ((Err = checkLinkeditDataCommand(*this, Load, I, &DataInCodeLoadCmd, 1240 "LC_DATA_IN_CODE", Elements, 1241 "data in code info"))) 1242 return; 1243 } else if (Load.C.cmd == MachO::LC_LINKER_OPTIMIZATION_HINT) { 1244 if ((Err = checkLinkeditDataCommand(*this, Load, I, &LinkOptHintsLoadCmd, 1245 "LC_LINKER_OPTIMIZATION_HINT", 1246 Elements, "linker optimization " 1247 "hints"))) 1248 return; 1249 } else if (Load.C.cmd == MachO::LC_FUNCTION_STARTS) { 1250 if ((Err = checkLinkeditDataCommand(*this, Load, I, &FuncStartsLoadCmd, 1251 "LC_FUNCTION_STARTS", Elements, 1252 "function starts data"))) 1253 return; 1254 } else if (Load.C.cmd == MachO::LC_SEGMENT_SPLIT_INFO) { 1255 if ((Err = checkLinkeditDataCommand(*this, Load, I, &SplitInfoLoadCmd, 1256 "LC_SEGMENT_SPLIT_INFO", Elements, 1257 "split info data"))) 1258 return; 1259 } else if (Load.C.cmd == MachO::LC_DYLIB_CODE_SIGN_DRS) { 1260 if ((Err = checkLinkeditDataCommand(*this, Load, I, &CodeSignDrsLoadCmd, 1261 "LC_DYLIB_CODE_SIGN_DRS", Elements, 1262 "code signing RDs data"))) 1263 return; 1264 } else if (Load.C.cmd == MachO::LC_CODE_SIGNATURE) { 1265 if ((Err = checkLinkeditDataCommand(*this, Load, I, &CodeSignLoadCmd, 1266 "LC_CODE_SIGNATURE", Elements, 1267 "code signature data"))) 1268 return; 1269 } else if (Load.C.cmd == MachO::LC_DYLD_INFO) { 1270 if ((Err = checkDyldInfoCommand(*this, Load, I, &DyldInfoLoadCmd, 1271 "LC_DYLD_INFO", Elements))) 1272 return; 1273 } else if (Load.C.cmd == MachO::LC_DYLD_INFO_ONLY) { 1274 if ((Err = checkDyldInfoCommand(*this, Load, I, &DyldInfoLoadCmd, 1275 "LC_DYLD_INFO_ONLY", Elements))) 1276 return; 1277 } else if (Load.C.cmd == MachO::LC_UUID) { 1278 if (Load.C.cmdsize != sizeof(MachO::uuid_command)) { 1279 Err = malformedError("LC_UUID command " + Twine(I) + " has incorrect " 1280 "cmdsize"); 1281 return; 1282 } 1283 if (UuidLoadCmd) { 1284 Err = malformedError("more than one LC_UUID command"); 1285 return; 1286 } 1287 UuidLoadCmd = Load.Ptr; 1288 } else if (Load.C.cmd == MachO::LC_SEGMENT_64) { 1289 if ((Err = parseSegmentLoadCommand<MachO::segment_command_64, 1290 MachO::section_64>( 1291 *this, Load, Sections, HasPageZeroSegment, I, 1292 "LC_SEGMENT_64", SizeOfHeaders, Elements))) 1293 return; 1294 } else if (Load.C.cmd == MachO::LC_SEGMENT) { 1295 if ((Err = parseSegmentLoadCommand<MachO::segment_command, 1296 MachO::section>( 1297 *this, Load, Sections, HasPageZeroSegment, I, 1298 "LC_SEGMENT", SizeOfHeaders, Elements))) 1299 return; 1300 } else if (Load.C.cmd == MachO::LC_ID_DYLIB) { 1301 if ((Err = checkDylibIdCommand(*this, Load, I, &DyldIdLoadCmd))) 1302 return; 1303 } else if (Load.C.cmd == MachO::LC_LOAD_DYLIB) { 1304 if ((Err = checkDylibCommand(*this, Load, I, "LC_LOAD_DYLIB"))) 1305 return; 1306 Libraries.push_back(Load.Ptr); 1307 } else if (Load.C.cmd == MachO::LC_LOAD_WEAK_DYLIB) { 1308 if ((Err = checkDylibCommand(*this, Load, I, "LC_LOAD_WEAK_DYLIB"))) 1309 return; 1310 Libraries.push_back(Load.Ptr); 1311 } else if (Load.C.cmd == MachO::LC_LAZY_LOAD_DYLIB) { 1312 if ((Err = checkDylibCommand(*this, Load, I, "LC_LAZY_LOAD_DYLIB"))) 1313 return; 1314 Libraries.push_back(Load.Ptr); 1315 } else if (Load.C.cmd == MachO::LC_REEXPORT_DYLIB) { 1316 if ((Err = checkDylibCommand(*this, Load, I, "LC_REEXPORT_DYLIB"))) 1317 return; 1318 Libraries.push_back(Load.Ptr); 1319 } else if (Load.C.cmd == MachO::LC_LOAD_UPWARD_DYLIB) { 1320 if ((Err = checkDylibCommand(*this, Load, I, "LC_LOAD_UPWARD_DYLIB"))) 1321 return; 1322 Libraries.push_back(Load.Ptr); 1323 } else if (Load.C.cmd == MachO::LC_ID_DYLINKER) { 1324 if ((Err = checkDyldCommand(*this, Load, I, "LC_ID_DYLINKER"))) 1325 return; 1326 } else if (Load.C.cmd == MachO::LC_LOAD_DYLINKER) { 1327 if ((Err = checkDyldCommand(*this, Load, I, "LC_LOAD_DYLINKER"))) 1328 return; 1329 } else if (Load.C.cmd == MachO::LC_DYLD_ENVIRONMENT) { 1330 if ((Err = checkDyldCommand(*this, Load, I, "LC_DYLD_ENVIRONMENT"))) 1331 return; 1332 } else if (Load.C.cmd == MachO::LC_VERSION_MIN_MACOSX) { 1333 if ((Err = checkVersCommand(*this, Load, I, &VersLoadCmd, 1334 "LC_VERSION_MIN_MACOSX"))) 1335 return; 1336 } else if (Load.C.cmd == MachO::LC_VERSION_MIN_IPHONEOS) { 1337 if ((Err = checkVersCommand(*this, Load, I, &VersLoadCmd, 1338 "LC_VERSION_MIN_IPHONEOS"))) 1339 return; 1340 } else if (Load.C.cmd == MachO::LC_VERSION_MIN_TVOS) { 1341 if ((Err = checkVersCommand(*this, Load, I, &VersLoadCmd, 1342 "LC_VERSION_MIN_TVOS"))) 1343 return; 1344 } else if (Load.C.cmd == MachO::LC_VERSION_MIN_WATCHOS) { 1345 if ((Err = checkVersCommand(*this, Load, I, &VersLoadCmd, 1346 "LC_VERSION_MIN_WATCHOS"))) 1347 return; 1348 } else if (Load.C.cmd == MachO::LC_NOTE) { 1349 if ((Err = checkNoteCommand(*this, Load, I, Elements))) 1350 return; 1351 } else if (Load.C.cmd == MachO::LC_BUILD_VERSION) { 1352 if ((Err = parseBuildVersionCommand(*this, Load, BuildTools, I))) 1353 return; 1354 } else if (Load.C.cmd == MachO::LC_RPATH) { 1355 if ((Err = checkRpathCommand(*this, Load, I))) 1356 return; 1357 } else if (Load.C.cmd == MachO::LC_SOURCE_VERSION) { 1358 if (Load.C.cmdsize != sizeof(MachO::source_version_command)) { 1359 Err = malformedError("LC_SOURCE_VERSION command " + Twine(I) + 1360 " has incorrect cmdsize"); 1361 return; 1362 } 1363 if (SourceLoadCmd) { 1364 Err = malformedError("more than one LC_SOURCE_VERSION command"); 1365 return; 1366 } 1367 SourceLoadCmd = Load.Ptr; 1368 } else if (Load.C.cmd == MachO::LC_MAIN) { 1369 if (Load.C.cmdsize != sizeof(MachO::entry_point_command)) { 1370 Err = malformedError("LC_MAIN command " + Twine(I) + 1371 " has incorrect cmdsize"); 1372 return; 1373 } 1374 if (EntryPointLoadCmd) { 1375 Err = malformedError("more than one LC_MAIN command"); 1376 return; 1377 } 1378 EntryPointLoadCmd = Load.Ptr; 1379 } else if (Load.C.cmd == MachO::LC_ENCRYPTION_INFO) { 1380 if (Load.C.cmdsize != sizeof(MachO::encryption_info_command)) { 1381 Err = malformedError("LC_ENCRYPTION_INFO command " + Twine(I) + 1382 " has incorrect cmdsize"); 1383 return; 1384 } 1385 MachO::encryption_info_command E = 1386 getStruct<MachO::encryption_info_command>(*this, Load.Ptr); 1387 if ((Err = checkEncryptCommand(*this, Load, I, E.cryptoff, E.cryptsize, 1388 &EncryptLoadCmd, "LC_ENCRYPTION_INFO"))) 1389 return; 1390 } else if (Load.C.cmd == MachO::LC_ENCRYPTION_INFO_64) { 1391 if (Load.C.cmdsize != sizeof(MachO::encryption_info_command_64)) { 1392 Err = malformedError("LC_ENCRYPTION_INFO_64 command " + Twine(I) + 1393 " has incorrect cmdsize"); 1394 return; 1395 } 1396 MachO::encryption_info_command_64 E = 1397 getStruct<MachO::encryption_info_command_64>(*this, Load.Ptr); 1398 if ((Err = checkEncryptCommand(*this, Load, I, E.cryptoff, E.cryptsize, 1399 &EncryptLoadCmd, "LC_ENCRYPTION_INFO_64"))) 1400 return; 1401 } else if (Load.C.cmd == MachO::LC_LINKER_OPTION) { 1402 if ((Err = checkLinkerOptCommand(*this, Load, I))) 1403 return; 1404 } else if (Load.C.cmd == MachO::LC_SUB_FRAMEWORK) { 1405 if (Load.C.cmdsize < sizeof(MachO::sub_framework_command)) { 1406 Err = malformedError("load command " + Twine(I) + 1407 " LC_SUB_FRAMEWORK cmdsize too small"); 1408 return; 1409 } 1410 MachO::sub_framework_command S = 1411 getStruct<MachO::sub_framework_command>(*this, Load.Ptr); 1412 if ((Err = checkSubCommand(*this, Load, I, "LC_SUB_FRAMEWORK", 1413 sizeof(MachO::sub_framework_command), 1414 "sub_framework_command", S.umbrella, 1415 "umbrella"))) 1416 return; 1417 } else if (Load.C.cmd == MachO::LC_SUB_UMBRELLA) { 1418 if (Load.C.cmdsize < sizeof(MachO::sub_umbrella_command)) { 1419 Err = malformedError("load command " + Twine(I) + 1420 " LC_SUB_UMBRELLA cmdsize too small"); 1421 return; 1422 } 1423 MachO::sub_umbrella_command S = 1424 getStruct<MachO::sub_umbrella_command>(*this, Load.Ptr); 1425 if ((Err = checkSubCommand(*this, Load, I, "LC_SUB_UMBRELLA", 1426 sizeof(MachO::sub_umbrella_command), 1427 "sub_umbrella_command", S.sub_umbrella, 1428 "sub_umbrella"))) 1429 return; 1430 } else if (Load.C.cmd == MachO::LC_SUB_LIBRARY) { 1431 if (Load.C.cmdsize < sizeof(MachO::sub_library_command)) { 1432 Err = malformedError("load command " + Twine(I) + 1433 " LC_SUB_LIBRARY cmdsize too small"); 1434 return; 1435 } 1436 MachO::sub_library_command S = 1437 getStruct<MachO::sub_library_command>(*this, Load.Ptr); 1438 if ((Err = checkSubCommand(*this, Load, I, "LC_SUB_LIBRARY", 1439 sizeof(MachO::sub_library_command), 1440 "sub_library_command", S.sub_library, 1441 "sub_library"))) 1442 return; 1443 } else if (Load.C.cmd == MachO::LC_SUB_CLIENT) { 1444 if (Load.C.cmdsize < sizeof(MachO::sub_client_command)) { 1445 Err = malformedError("load command " + Twine(I) + 1446 " LC_SUB_CLIENT cmdsize too small"); 1447 return; 1448 } 1449 MachO::sub_client_command S = 1450 getStruct<MachO::sub_client_command>(*this, Load.Ptr); 1451 if ((Err = checkSubCommand(*this, Load, I, "LC_SUB_CLIENT", 1452 sizeof(MachO::sub_client_command), 1453 "sub_client_command", S.client, "client"))) 1454 return; 1455 } else if (Load.C.cmd == MachO::LC_ROUTINES) { 1456 if (Load.C.cmdsize != sizeof(MachO::routines_command)) { 1457 Err = malformedError("LC_ROUTINES command " + Twine(I) + 1458 " has incorrect cmdsize"); 1459 return; 1460 } 1461 if (RoutinesLoadCmd) { 1462 Err = malformedError("more than one LC_ROUTINES and or LC_ROUTINES_64 " 1463 "command"); 1464 return; 1465 } 1466 RoutinesLoadCmd = Load.Ptr; 1467 } else if (Load.C.cmd == MachO::LC_ROUTINES_64) { 1468 if (Load.C.cmdsize != sizeof(MachO::routines_command_64)) { 1469 Err = malformedError("LC_ROUTINES_64 command " + Twine(I) + 1470 " has incorrect cmdsize"); 1471 return; 1472 } 1473 if (RoutinesLoadCmd) { 1474 Err = malformedError("more than one LC_ROUTINES_64 and or LC_ROUTINES " 1475 "command"); 1476 return; 1477 } 1478 RoutinesLoadCmd = Load.Ptr; 1479 } else if (Load.C.cmd == MachO::LC_UNIXTHREAD) { 1480 if ((Err = checkThreadCommand(*this, Load, I, "LC_UNIXTHREAD"))) 1481 return; 1482 if (UnixThreadLoadCmd) { 1483 Err = malformedError("more than one LC_UNIXTHREAD command"); 1484 return; 1485 } 1486 UnixThreadLoadCmd = Load.Ptr; 1487 } else if (Load.C.cmd == MachO::LC_THREAD) { 1488 if ((Err = checkThreadCommand(*this, Load, I, "LC_THREAD"))) 1489 return; 1490 // Note: LC_TWOLEVEL_HINTS is really obsolete and is not supported. 1491 } else if (Load.C.cmd == MachO::LC_TWOLEVEL_HINTS) { 1492 if ((Err = checkTwoLevelHintsCommand(*this, Load, I, 1493 &TwoLevelHintsLoadCmd, Elements))) 1494 return; 1495 } else if (isLoadCommandObsolete(Load.C.cmd)) { 1496 Err = malformedError("load command " + Twine(I) + " for cmd value of: " + 1497 Twine(Load.C.cmd) + " is obsolete and not " 1498 "supported"); 1499 return; 1500 } 1501 // TODO: generate a error for unknown load commands by default. But still 1502 // need work out an approach to allow or not allow unknown values like this 1503 // as an option for some uses like lldb. 1504 if (I < LoadCommandCount - 1) { 1505 if (auto LoadOrErr = getNextLoadCommandInfo(*this, I, Load)) 1506 Load = *LoadOrErr; 1507 else { 1508 Err = LoadOrErr.takeError(); 1509 return; 1510 } 1511 } 1512 } 1513 if (!SymtabLoadCmd) { 1514 if (DysymtabLoadCmd) { 1515 Err = malformedError("contains LC_DYSYMTAB load command without a " 1516 "LC_SYMTAB load command"); 1517 return; 1518 } 1519 } else if (DysymtabLoadCmd) { 1520 MachO::symtab_command Symtab = 1521 getStruct<MachO::symtab_command>(*this, SymtabLoadCmd); 1522 MachO::dysymtab_command Dysymtab = 1523 getStruct<MachO::dysymtab_command>(*this, DysymtabLoadCmd); 1524 if (Dysymtab.nlocalsym != 0 && Dysymtab.ilocalsym > Symtab.nsyms) { 1525 Err = malformedError("ilocalsym in LC_DYSYMTAB load command " 1526 "extends past the end of the symbol table"); 1527 return; 1528 } 1529 uint64_t BigSize = Dysymtab.ilocalsym; 1530 BigSize += Dysymtab.nlocalsym; 1531 if (Dysymtab.nlocalsym != 0 && BigSize > Symtab.nsyms) { 1532 Err = malformedError("ilocalsym plus nlocalsym in LC_DYSYMTAB load " 1533 "command extends past the end of the symbol table"); 1534 return; 1535 } 1536 if (Dysymtab.nextdefsym != 0 && Dysymtab.ilocalsym > Symtab.nsyms) { 1537 Err = malformedError("nextdefsym in LC_DYSYMTAB load command " 1538 "extends past the end of the symbol table"); 1539 return; 1540 } 1541 BigSize = Dysymtab.iextdefsym; 1542 BigSize += Dysymtab.nextdefsym; 1543 if (Dysymtab.nextdefsym != 0 && BigSize > Symtab.nsyms) { 1544 Err = malformedError("iextdefsym plus nextdefsym in LC_DYSYMTAB " 1545 "load command extends past the end of the symbol " 1546 "table"); 1547 return; 1548 } 1549 if (Dysymtab.nundefsym != 0 && Dysymtab.iundefsym > Symtab.nsyms) { 1550 Err = malformedError("nundefsym in LC_DYSYMTAB load command " 1551 "extends past the end of the symbol table"); 1552 return; 1553 } 1554 BigSize = Dysymtab.iundefsym; 1555 BigSize += Dysymtab.nundefsym; 1556 if (Dysymtab.nundefsym != 0 && BigSize > Symtab.nsyms) { 1557 Err = malformedError("iundefsym plus nundefsym in LC_DYSYMTAB load " 1558 " command extends past the end of the symbol table"); 1559 return; 1560 } 1561 } 1562 if ((getHeader().filetype == MachO::MH_DYLIB || 1563 getHeader().filetype == MachO::MH_DYLIB_STUB) && 1564 DyldIdLoadCmd == nullptr) { 1565 Err = malformedError("no LC_ID_DYLIB load command in dynamic library " 1566 "filetype"); 1567 return; 1568 } 1569 assert(LoadCommands.size() == LoadCommandCount); 1570 1571 Err = Error::success(); 1572 } 1573 1574 Error MachOObjectFile::checkSymbolTable() const { 1575 uint32_t Flags = 0; 1576 if (is64Bit()) { 1577 MachO::mach_header_64 H_64 = MachOObjectFile::getHeader64(); 1578 Flags = H_64.flags; 1579 } else { 1580 MachO::mach_header H = MachOObjectFile::getHeader(); 1581 Flags = H.flags; 1582 } 1583 uint8_t NType = 0; 1584 uint8_t NSect = 0; 1585 uint16_t NDesc = 0; 1586 uint32_t NStrx = 0; 1587 uint64_t NValue = 0; 1588 uint32_t SymbolIndex = 0; 1589 MachO::symtab_command S = getSymtabLoadCommand(); 1590 for (const SymbolRef &Symbol : symbols()) { 1591 DataRefImpl SymDRI = Symbol.getRawDataRefImpl(); 1592 if (is64Bit()) { 1593 MachO::nlist_64 STE_64 = getSymbol64TableEntry(SymDRI); 1594 NType = STE_64.n_type; 1595 NSect = STE_64.n_sect; 1596 NDesc = STE_64.n_desc; 1597 NStrx = STE_64.n_strx; 1598 NValue = STE_64.n_value; 1599 } else { 1600 MachO::nlist STE = getSymbolTableEntry(SymDRI); 1601 NType = STE.n_type; 1602 NType = STE.n_type; 1603 NSect = STE.n_sect; 1604 NDesc = STE.n_desc; 1605 NStrx = STE.n_strx; 1606 NValue = STE.n_value; 1607 } 1608 if ((NType & MachO::N_STAB) == 0 && 1609 (NType & MachO::N_TYPE) == MachO::N_SECT) { 1610 if (NSect == 0 || NSect > Sections.size()) 1611 return malformedError("bad section index: " + Twine((int)NSect) + 1612 " for symbol at index " + Twine(SymbolIndex)); 1613 } 1614 if ((NType & MachO::N_STAB) == 0 && 1615 (NType & MachO::N_TYPE) == MachO::N_INDR) { 1616 if (NValue >= S.strsize) 1617 return malformedError("bad n_value: " + Twine((int)NValue) + " past " 1618 "the end of string table, for N_INDR symbol at " 1619 "index " + Twine(SymbolIndex)); 1620 } 1621 if ((Flags & MachO::MH_TWOLEVEL) == MachO::MH_TWOLEVEL && 1622 (((NType & MachO::N_TYPE) == MachO::N_UNDF && NValue == 0) || 1623 (NType & MachO::N_TYPE) == MachO::N_PBUD)) { 1624 uint32_t LibraryOrdinal = MachO::GET_LIBRARY_ORDINAL(NDesc); 1625 if (LibraryOrdinal != 0 && 1626 LibraryOrdinal != MachO::EXECUTABLE_ORDINAL && 1627 LibraryOrdinal != MachO::DYNAMIC_LOOKUP_ORDINAL && 1628 LibraryOrdinal - 1 >= Libraries.size() ) { 1629 return malformedError("bad library ordinal: " + Twine(LibraryOrdinal) + 1630 " for symbol at index " + Twine(SymbolIndex)); 1631 } 1632 } 1633 if (NStrx >= S.strsize) 1634 return malformedError("bad string table index: " + Twine((int)NStrx) + 1635 " past the end of string table, for symbol at " 1636 "index " + Twine(SymbolIndex)); 1637 SymbolIndex++; 1638 } 1639 return Error::success(); 1640 } 1641 1642 void MachOObjectFile::moveSymbolNext(DataRefImpl &Symb) const { 1643 unsigned SymbolTableEntrySize = is64Bit() ? 1644 sizeof(MachO::nlist_64) : 1645 sizeof(MachO::nlist); 1646 Symb.p += SymbolTableEntrySize; 1647 } 1648 1649 Expected<StringRef> MachOObjectFile::getSymbolName(DataRefImpl Symb) const { 1650 StringRef StringTable = getStringTableData(); 1651 MachO::nlist_base Entry = getSymbolTableEntryBase(*this, Symb); 1652 const char *Start = &StringTable.data()[Entry.n_strx]; 1653 if (Start < getData().begin() || Start >= getData().end()) { 1654 return malformedError("bad string index: " + Twine(Entry.n_strx) + 1655 " for symbol at index " + Twine(getSymbolIndex(Symb))); 1656 } 1657 return StringRef(Start); 1658 } 1659 1660 unsigned MachOObjectFile::getSectionType(SectionRef Sec) const { 1661 DataRefImpl DRI = Sec.getRawDataRefImpl(); 1662 uint32_t Flags = getSectionFlags(*this, DRI); 1663 return Flags & MachO::SECTION_TYPE; 1664 } 1665 1666 uint64_t MachOObjectFile::getNValue(DataRefImpl Sym) const { 1667 if (is64Bit()) { 1668 MachO::nlist_64 Entry = getSymbol64TableEntry(Sym); 1669 return Entry.n_value; 1670 } 1671 MachO::nlist Entry = getSymbolTableEntry(Sym); 1672 return Entry.n_value; 1673 } 1674 1675 // getIndirectName() returns the name of the alias'ed symbol who's string table 1676 // index is in the n_value field. 1677 std::error_code MachOObjectFile::getIndirectName(DataRefImpl Symb, 1678 StringRef &Res) const { 1679 StringRef StringTable = getStringTableData(); 1680 MachO::nlist_base Entry = getSymbolTableEntryBase(*this, Symb); 1681 if ((Entry.n_type & MachO::N_TYPE) != MachO::N_INDR) 1682 return object_error::parse_failed; 1683 uint64_t NValue = getNValue(Symb); 1684 if (NValue >= StringTable.size()) 1685 return object_error::parse_failed; 1686 const char *Start = &StringTable.data()[NValue]; 1687 Res = StringRef(Start); 1688 return std::error_code(); 1689 } 1690 1691 uint64_t MachOObjectFile::getSymbolValueImpl(DataRefImpl Sym) const { 1692 return getNValue(Sym); 1693 } 1694 1695 Expected<uint64_t> MachOObjectFile::getSymbolAddress(DataRefImpl Sym) const { 1696 return getSymbolValue(Sym); 1697 } 1698 1699 uint32_t MachOObjectFile::getSymbolAlignment(DataRefImpl DRI) const { 1700 uint32_t flags = getSymbolFlags(DRI); 1701 if (flags & SymbolRef::SF_Common) { 1702 MachO::nlist_base Entry = getSymbolTableEntryBase(*this, DRI); 1703 return 1 << MachO::GET_COMM_ALIGN(Entry.n_desc); 1704 } 1705 return 0; 1706 } 1707 1708 uint64_t MachOObjectFile::getCommonSymbolSizeImpl(DataRefImpl DRI) const { 1709 return getNValue(DRI); 1710 } 1711 1712 Expected<SymbolRef::Type> 1713 MachOObjectFile::getSymbolType(DataRefImpl Symb) const { 1714 MachO::nlist_base Entry = getSymbolTableEntryBase(*this, Symb); 1715 uint8_t n_type = Entry.n_type; 1716 1717 // If this is a STAB debugging symbol, we can do nothing more. 1718 if (n_type & MachO::N_STAB) 1719 return SymbolRef::ST_Debug; 1720 1721 switch (n_type & MachO::N_TYPE) { 1722 case MachO::N_UNDF : 1723 return SymbolRef::ST_Unknown; 1724 case MachO::N_SECT : 1725 Expected<section_iterator> SecOrError = getSymbolSection(Symb); 1726 if (!SecOrError) 1727 return SecOrError.takeError(); 1728 section_iterator Sec = *SecOrError; 1729 if (Sec->isData() || Sec->isBSS()) 1730 return SymbolRef::ST_Data; 1731 return SymbolRef::ST_Function; 1732 } 1733 return SymbolRef::ST_Other; 1734 } 1735 1736 uint32_t MachOObjectFile::getSymbolFlags(DataRefImpl DRI) const { 1737 MachO::nlist_base Entry = getSymbolTableEntryBase(*this, DRI); 1738 1739 uint8_t MachOType = Entry.n_type; 1740 uint16_t MachOFlags = Entry.n_desc; 1741 1742 uint32_t Result = SymbolRef::SF_None; 1743 1744 if ((MachOType & MachO::N_TYPE) == MachO::N_INDR) 1745 Result |= SymbolRef::SF_Indirect; 1746 1747 if (MachOType & MachO::N_STAB) 1748 Result |= SymbolRef::SF_FormatSpecific; 1749 1750 if (MachOType & MachO::N_EXT) { 1751 Result |= SymbolRef::SF_Global; 1752 if ((MachOType & MachO::N_TYPE) == MachO::N_UNDF) { 1753 if (getNValue(DRI)) 1754 Result |= SymbolRef::SF_Common; 1755 else 1756 Result |= SymbolRef::SF_Undefined; 1757 } 1758 1759 if (!(MachOType & MachO::N_PEXT)) 1760 Result |= SymbolRef::SF_Exported; 1761 } 1762 1763 if (MachOFlags & (MachO::N_WEAK_REF | MachO::N_WEAK_DEF)) 1764 Result |= SymbolRef::SF_Weak; 1765 1766 if (MachOFlags & (MachO::N_ARM_THUMB_DEF)) 1767 Result |= SymbolRef::SF_Thumb; 1768 1769 if ((MachOType & MachO::N_TYPE) == MachO::N_ABS) 1770 Result |= SymbolRef::SF_Absolute; 1771 1772 return Result; 1773 } 1774 1775 Expected<section_iterator> 1776 MachOObjectFile::getSymbolSection(DataRefImpl Symb) const { 1777 MachO::nlist_base Entry = getSymbolTableEntryBase(*this, Symb); 1778 uint8_t index = Entry.n_sect; 1779 1780 if (index == 0) 1781 return section_end(); 1782 DataRefImpl DRI; 1783 DRI.d.a = index - 1; 1784 if (DRI.d.a >= Sections.size()){ 1785 return malformedError("bad section index: " + Twine((int)index) + 1786 " for symbol at index " + Twine(getSymbolIndex(Symb))); 1787 } 1788 return section_iterator(SectionRef(DRI, this)); 1789 } 1790 1791 unsigned MachOObjectFile::getSymbolSectionID(SymbolRef Sym) const { 1792 MachO::nlist_base Entry = 1793 getSymbolTableEntryBase(*this, Sym.getRawDataRefImpl()); 1794 return Entry.n_sect - 1; 1795 } 1796 1797 void MachOObjectFile::moveSectionNext(DataRefImpl &Sec) const { 1798 Sec.d.a++; 1799 } 1800 1801 std::error_code MachOObjectFile::getSectionName(DataRefImpl Sec, 1802 StringRef &Result) const { 1803 ArrayRef<char> Raw = getSectionRawName(Sec); 1804 Result = parseSegmentOrSectionName(Raw.data()); 1805 return std::error_code(); 1806 } 1807 1808 uint64_t MachOObjectFile::getSectionAddress(DataRefImpl Sec) const { 1809 if (is64Bit()) 1810 return getSection64(Sec).addr; 1811 return getSection(Sec).addr; 1812 } 1813 1814 uint64_t MachOObjectFile::getSectionSize(DataRefImpl Sec) const { 1815 // In the case if a malformed Mach-O file where the section offset is past 1816 // the end of the file or some part of the section size is past the end of 1817 // the file return a size of zero or a size that covers the rest of the file 1818 // but does not extend past the end of the file. 1819 uint32_t SectOffset, SectType; 1820 uint64_t SectSize; 1821 1822 if (is64Bit()) { 1823 MachO::section_64 Sect = getSection64(Sec); 1824 SectOffset = Sect.offset; 1825 SectSize = Sect.size; 1826 SectType = Sect.flags & MachO::SECTION_TYPE; 1827 } else { 1828 MachO::section Sect = getSection(Sec); 1829 SectOffset = Sect.offset; 1830 SectSize = Sect.size; 1831 SectType = Sect.flags & MachO::SECTION_TYPE; 1832 } 1833 if (SectType == MachO::S_ZEROFILL || SectType == MachO::S_GB_ZEROFILL) 1834 return SectSize; 1835 uint64_t FileSize = getData().size(); 1836 if (SectOffset > FileSize) 1837 return 0; 1838 if (FileSize - SectOffset < SectSize) 1839 return FileSize - SectOffset; 1840 return SectSize; 1841 } 1842 1843 std::error_code MachOObjectFile::getSectionContents(DataRefImpl Sec, 1844 StringRef &Res) const { 1845 uint32_t Offset; 1846 uint64_t Size; 1847 1848 if (is64Bit()) { 1849 MachO::section_64 Sect = getSection64(Sec); 1850 Offset = Sect.offset; 1851 Size = Sect.size; 1852 } else { 1853 MachO::section Sect = getSection(Sec); 1854 Offset = Sect.offset; 1855 Size = Sect.size; 1856 } 1857 1858 Res = this->getData().substr(Offset, Size); 1859 return std::error_code(); 1860 } 1861 1862 uint64_t MachOObjectFile::getSectionAlignment(DataRefImpl Sec) const { 1863 uint32_t Align; 1864 if (is64Bit()) { 1865 MachO::section_64 Sect = getSection64(Sec); 1866 Align = Sect.align; 1867 } else { 1868 MachO::section Sect = getSection(Sec); 1869 Align = Sect.align; 1870 } 1871 1872 return uint64_t(1) << Align; 1873 } 1874 1875 bool MachOObjectFile::isSectionCompressed(DataRefImpl Sec) const { 1876 return false; 1877 } 1878 1879 bool MachOObjectFile::isSectionText(DataRefImpl Sec) const { 1880 uint32_t Flags = getSectionFlags(*this, Sec); 1881 return Flags & MachO::S_ATTR_PURE_INSTRUCTIONS; 1882 } 1883 1884 bool MachOObjectFile::isSectionData(DataRefImpl Sec) const { 1885 uint32_t Flags = getSectionFlags(*this, Sec); 1886 unsigned SectionType = Flags & MachO::SECTION_TYPE; 1887 return !(Flags & MachO::S_ATTR_PURE_INSTRUCTIONS) && 1888 !(SectionType == MachO::S_ZEROFILL || 1889 SectionType == MachO::S_GB_ZEROFILL); 1890 } 1891 1892 bool MachOObjectFile::isSectionBSS(DataRefImpl Sec) const { 1893 uint32_t Flags = getSectionFlags(*this, Sec); 1894 unsigned SectionType = Flags & MachO::SECTION_TYPE; 1895 return !(Flags & MachO::S_ATTR_PURE_INSTRUCTIONS) && 1896 (SectionType == MachO::S_ZEROFILL || 1897 SectionType == MachO::S_GB_ZEROFILL); 1898 } 1899 1900 unsigned MachOObjectFile::getSectionID(SectionRef Sec) const { 1901 return Sec.getRawDataRefImpl().d.a; 1902 } 1903 1904 bool MachOObjectFile::isSectionVirtual(DataRefImpl Sec) const { 1905 // FIXME: Unimplemented. 1906 return false; 1907 } 1908 1909 bool MachOObjectFile::isSectionBitcode(DataRefImpl Sec) const { 1910 StringRef SegmentName = getSectionFinalSegmentName(Sec); 1911 StringRef SectName; 1912 if (!getSectionName(Sec, SectName)) 1913 return (SegmentName == "__LLVM" && SectName == "__bitcode"); 1914 return false; 1915 } 1916 1917 relocation_iterator MachOObjectFile::section_rel_begin(DataRefImpl Sec) const { 1918 DataRefImpl Ret; 1919 Ret.d.a = Sec.d.a; 1920 Ret.d.b = 0; 1921 return relocation_iterator(RelocationRef(Ret, this)); 1922 } 1923 1924 relocation_iterator 1925 MachOObjectFile::section_rel_end(DataRefImpl Sec) const { 1926 uint32_t Num; 1927 if (is64Bit()) { 1928 MachO::section_64 Sect = getSection64(Sec); 1929 Num = Sect.nreloc; 1930 } else { 1931 MachO::section Sect = getSection(Sec); 1932 Num = Sect.nreloc; 1933 } 1934 1935 DataRefImpl Ret; 1936 Ret.d.a = Sec.d.a; 1937 Ret.d.b = Num; 1938 return relocation_iterator(RelocationRef(Ret, this)); 1939 } 1940 1941 void MachOObjectFile::moveRelocationNext(DataRefImpl &Rel) const { 1942 ++Rel.d.b; 1943 } 1944 1945 uint64_t MachOObjectFile::getRelocationOffset(DataRefImpl Rel) const { 1946 assert(getHeader().filetype == MachO::MH_OBJECT && 1947 "Only implemented for MH_OBJECT"); 1948 MachO::any_relocation_info RE = getRelocation(Rel); 1949 return getAnyRelocationAddress(RE); 1950 } 1951 1952 symbol_iterator 1953 MachOObjectFile::getRelocationSymbol(DataRefImpl Rel) const { 1954 MachO::any_relocation_info RE = getRelocation(Rel); 1955 if (isRelocationScattered(RE)) 1956 return symbol_end(); 1957 1958 uint32_t SymbolIdx = getPlainRelocationSymbolNum(RE); 1959 bool isExtern = getPlainRelocationExternal(RE); 1960 if (!isExtern) 1961 return symbol_end(); 1962 1963 MachO::symtab_command S = getSymtabLoadCommand(); 1964 unsigned SymbolTableEntrySize = is64Bit() ? 1965 sizeof(MachO::nlist_64) : 1966 sizeof(MachO::nlist); 1967 uint64_t Offset = S.symoff + SymbolIdx * SymbolTableEntrySize; 1968 DataRefImpl Sym; 1969 Sym.p = reinterpret_cast<uintptr_t>(getPtr(*this, Offset)); 1970 return symbol_iterator(SymbolRef(Sym, this)); 1971 } 1972 1973 section_iterator 1974 MachOObjectFile::getRelocationSection(DataRefImpl Rel) const { 1975 return section_iterator(getAnyRelocationSection(getRelocation(Rel))); 1976 } 1977 1978 uint64_t MachOObjectFile::getRelocationType(DataRefImpl Rel) const { 1979 MachO::any_relocation_info RE = getRelocation(Rel); 1980 return getAnyRelocationType(RE); 1981 } 1982 1983 void MachOObjectFile::getRelocationTypeName( 1984 DataRefImpl Rel, SmallVectorImpl<char> &Result) const { 1985 StringRef res; 1986 uint64_t RType = getRelocationType(Rel); 1987 1988 unsigned Arch = this->getArch(); 1989 1990 switch (Arch) { 1991 case Triple::x86: { 1992 static const char *const Table[] = { 1993 "GENERIC_RELOC_VANILLA", 1994 "GENERIC_RELOC_PAIR", 1995 "GENERIC_RELOC_SECTDIFF", 1996 "GENERIC_RELOC_PB_LA_PTR", 1997 "GENERIC_RELOC_LOCAL_SECTDIFF", 1998 "GENERIC_RELOC_TLV" }; 1999 2000 if (RType > 5) 2001 res = "Unknown"; 2002 else 2003 res = Table[RType]; 2004 break; 2005 } 2006 case Triple::x86_64: { 2007 static const char *const Table[] = { 2008 "X86_64_RELOC_UNSIGNED", 2009 "X86_64_RELOC_SIGNED", 2010 "X86_64_RELOC_BRANCH", 2011 "X86_64_RELOC_GOT_LOAD", 2012 "X86_64_RELOC_GOT", 2013 "X86_64_RELOC_SUBTRACTOR", 2014 "X86_64_RELOC_SIGNED_1", 2015 "X86_64_RELOC_SIGNED_2", 2016 "X86_64_RELOC_SIGNED_4", 2017 "X86_64_RELOC_TLV" }; 2018 2019 if (RType > 9) 2020 res = "Unknown"; 2021 else 2022 res = Table[RType]; 2023 break; 2024 } 2025 case Triple::arm: { 2026 static const char *const Table[] = { 2027 "ARM_RELOC_VANILLA", 2028 "ARM_RELOC_PAIR", 2029 "ARM_RELOC_SECTDIFF", 2030 "ARM_RELOC_LOCAL_SECTDIFF", 2031 "ARM_RELOC_PB_LA_PTR", 2032 "ARM_RELOC_BR24", 2033 "ARM_THUMB_RELOC_BR22", 2034 "ARM_THUMB_32BIT_BRANCH", 2035 "ARM_RELOC_HALF", 2036 "ARM_RELOC_HALF_SECTDIFF" }; 2037 2038 if (RType > 9) 2039 res = "Unknown"; 2040 else 2041 res = Table[RType]; 2042 break; 2043 } 2044 case Triple::aarch64: { 2045 static const char *const Table[] = { 2046 "ARM64_RELOC_UNSIGNED", "ARM64_RELOC_SUBTRACTOR", 2047 "ARM64_RELOC_BRANCH26", "ARM64_RELOC_PAGE21", 2048 "ARM64_RELOC_PAGEOFF12", "ARM64_RELOC_GOT_LOAD_PAGE21", 2049 "ARM64_RELOC_GOT_LOAD_PAGEOFF12", "ARM64_RELOC_POINTER_TO_GOT", 2050 "ARM64_RELOC_TLVP_LOAD_PAGE21", "ARM64_RELOC_TLVP_LOAD_PAGEOFF12", 2051 "ARM64_RELOC_ADDEND" 2052 }; 2053 2054 if (RType >= array_lengthof(Table)) 2055 res = "Unknown"; 2056 else 2057 res = Table[RType]; 2058 break; 2059 } 2060 case Triple::ppc: { 2061 static const char *const Table[] = { 2062 "PPC_RELOC_VANILLA", 2063 "PPC_RELOC_PAIR", 2064 "PPC_RELOC_BR14", 2065 "PPC_RELOC_BR24", 2066 "PPC_RELOC_HI16", 2067 "PPC_RELOC_LO16", 2068 "PPC_RELOC_HA16", 2069 "PPC_RELOC_LO14", 2070 "PPC_RELOC_SECTDIFF", 2071 "PPC_RELOC_PB_LA_PTR", 2072 "PPC_RELOC_HI16_SECTDIFF", 2073 "PPC_RELOC_LO16_SECTDIFF", 2074 "PPC_RELOC_HA16_SECTDIFF", 2075 "PPC_RELOC_JBSR", 2076 "PPC_RELOC_LO14_SECTDIFF", 2077 "PPC_RELOC_LOCAL_SECTDIFF" }; 2078 2079 if (RType > 15) 2080 res = "Unknown"; 2081 else 2082 res = Table[RType]; 2083 break; 2084 } 2085 case Triple::UnknownArch: 2086 res = "Unknown"; 2087 break; 2088 } 2089 Result.append(res.begin(), res.end()); 2090 } 2091 2092 uint8_t MachOObjectFile::getRelocationLength(DataRefImpl Rel) const { 2093 MachO::any_relocation_info RE = getRelocation(Rel); 2094 return getAnyRelocationLength(RE); 2095 } 2096 2097 // 2098 // guessLibraryShortName() is passed a name of a dynamic library and returns a 2099 // guess on what the short name is. Then name is returned as a substring of the 2100 // StringRef Name passed in. The name of the dynamic library is recognized as 2101 // a framework if it has one of the two following forms: 2102 // Foo.framework/Versions/A/Foo 2103 // Foo.framework/Foo 2104 // Where A and Foo can be any string. And may contain a trailing suffix 2105 // starting with an underbar. If the Name is recognized as a framework then 2106 // isFramework is set to true else it is set to false. If the Name has a 2107 // suffix then Suffix is set to the substring in Name that contains the suffix 2108 // else it is set to a NULL StringRef. 2109 // 2110 // The Name of the dynamic library is recognized as a library name if it has 2111 // one of the two following forms: 2112 // libFoo.A.dylib 2113 // libFoo.dylib 2114 // The library may have a suffix trailing the name Foo of the form: 2115 // libFoo_profile.A.dylib 2116 // libFoo_profile.dylib 2117 // 2118 // The Name of the dynamic library is also recognized as a library name if it 2119 // has the following form: 2120 // Foo.qtx 2121 // 2122 // If the Name of the dynamic library is none of the forms above then a NULL 2123 // StringRef is returned. 2124 // 2125 StringRef MachOObjectFile::guessLibraryShortName(StringRef Name, 2126 bool &isFramework, 2127 StringRef &Suffix) { 2128 StringRef Foo, F, DotFramework, V, Dylib, Lib, Dot, Qtx; 2129 size_t a, b, c, d, Idx; 2130 2131 isFramework = false; 2132 Suffix = StringRef(); 2133 2134 // Pull off the last component and make Foo point to it 2135 a = Name.rfind('/'); 2136 if (a == Name.npos || a == 0) 2137 goto guess_library; 2138 Foo = Name.slice(a+1, Name.npos); 2139 2140 // Look for a suffix starting with a '_' 2141 Idx = Foo.rfind('_'); 2142 if (Idx != Foo.npos && Foo.size() >= 2) { 2143 Suffix = Foo.slice(Idx, Foo.npos); 2144 Foo = Foo.slice(0, Idx); 2145 } 2146 2147 // First look for the form Foo.framework/Foo 2148 b = Name.rfind('/', a); 2149 if (b == Name.npos) 2150 Idx = 0; 2151 else 2152 Idx = b+1; 2153 F = Name.slice(Idx, Idx + Foo.size()); 2154 DotFramework = Name.slice(Idx + Foo.size(), 2155 Idx + Foo.size() + sizeof(".framework/")-1); 2156 if (F == Foo && DotFramework == ".framework/") { 2157 isFramework = true; 2158 return Foo; 2159 } 2160 2161 // Next look for the form Foo.framework/Versions/A/Foo 2162 if (b == Name.npos) 2163 goto guess_library; 2164 c = Name.rfind('/', b); 2165 if (c == Name.npos || c == 0) 2166 goto guess_library; 2167 V = Name.slice(c+1, Name.npos); 2168 if (!V.startswith("Versions/")) 2169 goto guess_library; 2170 d = Name.rfind('/', c); 2171 if (d == Name.npos) 2172 Idx = 0; 2173 else 2174 Idx = d+1; 2175 F = Name.slice(Idx, Idx + Foo.size()); 2176 DotFramework = Name.slice(Idx + Foo.size(), 2177 Idx + Foo.size() + sizeof(".framework/")-1); 2178 if (F == Foo && DotFramework == ".framework/") { 2179 isFramework = true; 2180 return Foo; 2181 } 2182 2183 guess_library: 2184 // pull off the suffix after the "." and make a point to it 2185 a = Name.rfind('.'); 2186 if (a == Name.npos || a == 0) 2187 return StringRef(); 2188 Dylib = Name.slice(a, Name.npos); 2189 if (Dylib != ".dylib") 2190 goto guess_qtx; 2191 2192 // First pull off the version letter for the form Foo.A.dylib if any. 2193 if (a >= 3) { 2194 Dot = Name.slice(a-2, a-1); 2195 if (Dot == ".") 2196 a = a - 2; 2197 } 2198 2199 b = Name.rfind('/', a); 2200 if (b == Name.npos) 2201 b = 0; 2202 else 2203 b = b+1; 2204 // ignore any suffix after an underbar like Foo_profile.A.dylib 2205 Idx = Name.find('_', b); 2206 if (Idx != Name.npos && Idx != b) { 2207 Lib = Name.slice(b, Idx); 2208 Suffix = Name.slice(Idx, a); 2209 } 2210 else 2211 Lib = Name.slice(b, a); 2212 // There are incorrect library names of the form: 2213 // libATS.A_profile.dylib so check for these. 2214 if (Lib.size() >= 3) { 2215 Dot = Lib.slice(Lib.size()-2, Lib.size()-1); 2216 if (Dot == ".") 2217 Lib = Lib.slice(0, Lib.size()-2); 2218 } 2219 return Lib; 2220 2221 guess_qtx: 2222 Qtx = Name.slice(a, Name.npos); 2223 if (Qtx != ".qtx") 2224 return StringRef(); 2225 b = Name.rfind('/', a); 2226 if (b == Name.npos) 2227 Lib = Name.slice(0, a); 2228 else 2229 Lib = Name.slice(b+1, a); 2230 // There are library names of the form: QT.A.qtx so check for these. 2231 if (Lib.size() >= 3) { 2232 Dot = Lib.slice(Lib.size()-2, Lib.size()-1); 2233 if (Dot == ".") 2234 Lib = Lib.slice(0, Lib.size()-2); 2235 } 2236 return Lib; 2237 } 2238 2239 // getLibraryShortNameByIndex() is used to get the short name of the library 2240 // for an undefined symbol in a linked Mach-O binary that was linked with the 2241 // normal two-level namespace default (that is MH_TWOLEVEL in the header). 2242 // It is passed the index (0 - based) of the library as translated from 2243 // GET_LIBRARY_ORDINAL (1 - based). 2244 std::error_code MachOObjectFile::getLibraryShortNameByIndex(unsigned Index, 2245 StringRef &Res) const { 2246 if (Index >= Libraries.size()) 2247 return object_error::parse_failed; 2248 2249 // If the cache of LibrariesShortNames is not built up do that first for 2250 // all the Libraries. 2251 if (LibrariesShortNames.size() == 0) { 2252 for (unsigned i = 0; i < Libraries.size(); i++) { 2253 MachO::dylib_command D = 2254 getStruct<MachO::dylib_command>(*this, Libraries[i]); 2255 if (D.dylib.name >= D.cmdsize) 2256 return object_error::parse_failed; 2257 const char *P = (const char *)(Libraries[i]) + D.dylib.name; 2258 StringRef Name = StringRef(P); 2259 if (D.dylib.name+Name.size() >= D.cmdsize) 2260 return object_error::parse_failed; 2261 StringRef Suffix; 2262 bool isFramework; 2263 StringRef shortName = guessLibraryShortName(Name, isFramework, Suffix); 2264 if (shortName.empty()) 2265 LibrariesShortNames.push_back(Name); 2266 else 2267 LibrariesShortNames.push_back(shortName); 2268 } 2269 } 2270 2271 Res = LibrariesShortNames[Index]; 2272 return std::error_code(); 2273 } 2274 2275 uint32_t MachOObjectFile::getLibraryCount() const { 2276 return Libraries.size(); 2277 } 2278 2279 section_iterator 2280 MachOObjectFile::getRelocationRelocatedSection(relocation_iterator Rel) const { 2281 DataRefImpl Sec; 2282 Sec.d.a = Rel->getRawDataRefImpl().d.a; 2283 return section_iterator(SectionRef(Sec, this)); 2284 } 2285 2286 basic_symbol_iterator MachOObjectFile::symbol_begin() const { 2287 DataRefImpl DRI; 2288 MachO::symtab_command Symtab = getSymtabLoadCommand(); 2289 if (!SymtabLoadCmd || Symtab.nsyms == 0) 2290 return basic_symbol_iterator(SymbolRef(DRI, this)); 2291 2292 return getSymbolByIndex(0); 2293 } 2294 2295 basic_symbol_iterator MachOObjectFile::symbol_end() const { 2296 DataRefImpl DRI; 2297 MachO::symtab_command Symtab = getSymtabLoadCommand(); 2298 if (!SymtabLoadCmd || Symtab.nsyms == 0) 2299 return basic_symbol_iterator(SymbolRef(DRI, this)); 2300 2301 unsigned SymbolTableEntrySize = is64Bit() ? 2302 sizeof(MachO::nlist_64) : 2303 sizeof(MachO::nlist); 2304 unsigned Offset = Symtab.symoff + 2305 Symtab.nsyms * SymbolTableEntrySize; 2306 DRI.p = reinterpret_cast<uintptr_t>(getPtr(*this, Offset)); 2307 return basic_symbol_iterator(SymbolRef(DRI, this)); 2308 } 2309 2310 basic_symbol_iterator MachOObjectFile::getSymbolByIndex(unsigned Index) const { 2311 MachO::symtab_command Symtab = getSymtabLoadCommand(); 2312 if (!SymtabLoadCmd || Index >= Symtab.nsyms) 2313 report_fatal_error("Requested symbol index is out of range."); 2314 unsigned SymbolTableEntrySize = 2315 is64Bit() ? sizeof(MachO::nlist_64) : sizeof(MachO::nlist); 2316 DataRefImpl DRI; 2317 DRI.p = reinterpret_cast<uintptr_t>(getPtr(*this, Symtab.symoff)); 2318 DRI.p += Index * SymbolTableEntrySize; 2319 return basic_symbol_iterator(SymbolRef(DRI, this)); 2320 } 2321 2322 uint64_t MachOObjectFile::getSymbolIndex(DataRefImpl Symb) const { 2323 MachO::symtab_command Symtab = getSymtabLoadCommand(); 2324 if (!SymtabLoadCmd) 2325 report_fatal_error("getSymbolIndex() called with no symbol table symbol"); 2326 unsigned SymbolTableEntrySize = 2327 is64Bit() ? sizeof(MachO::nlist_64) : sizeof(MachO::nlist); 2328 DataRefImpl DRIstart; 2329 DRIstart.p = reinterpret_cast<uintptr_t>(getPtr(*this, Symtab.symoff)); 2330 uint64_t Index = (Symb.p - DRIstart.p) / SymbolTableEntrySize; 2331 return Index; 2332 } 2333 2334 section_iterator MachOObjectFile::section_begin() const { 2335 DataRefImpl DRI; 2336 return section_iterator(SectionRef(DRI, this)); 2337 } 2338 2339 section_iterator MachOObjectFile::section_end() const { 2340 DataRefImpl DRI; 2341 DRI.d.a = Sections.size(); 2342 return section_iterator(SectionRef(DRI, this)); 2343 } 2344 2345 uint8_t MachOObjectFile::getBytesInAddress() const { 2346 return is64Bit() ? 8 : 4; 2347 } 2348 2349 StringRef MachOObjectFile::getFileFormatName() const { 2350 unsigned CPUType = getCPUType(*this); 2351 if (!is64Bit()) { 2352 switch (CPUType) { 2353 case llvm::MachO::CPU_TYPE_I386: 2354 return "Mach-O 32-bit i386"; 2355 case llvm::MachO::CPU_TYPE_ARM: 2356 return "Mach-O arm"; 2357 case llvm::MachO::CPU_TYPE_POWERPC: 2358 return "Mach-O 32-bit ppc"; 2359 default: 2360 return "Mach-O 32-bit unknown"; 2361 } 2362 } 2363 2364 switch (CPUType) { 2365 case llvm::MachO::CPU_TYPE_X86_64: 2366 return "Mach-O 64-bit x86-64"; 2367 case llvm::MachO::CPU_TYPE_ARM64: 2368 return "Mach-O arm64"; 2369 case llvm::MachO::CPU_TYPE_POWERPC64: 2370 return "Mach-O 64-bit ppc64"; 2371 default: 2372 return "Mach-O 64-bit unknown"; 2373 } 2374 } 2375 2376 Triple::ArchType MachOObjectFile::getArch(uint32_t CPUType) { 2377 switch (CPUType) { 2378 case llvm::MachO::CPU_TYPE_I386: 2379 return Triple::x86; 2380 case llvm::MachO::CPU_TYPE_X86_64: 2381 return Triple::x86_64; 2382 case llvm::MachO::CPU_TYPE_ARM: 2383 return Triple::arm; 2384 case llvm::MachO::CPU_TYPE_ARM64: 2385 return Triple::aarch64; 2386 case llvm::MachO::CPU_TYPE_POWERPC: 2387 return Triple::ppc; 2388 case llvm::MachO::CPU_TYPE_POWERPC64: 2389 return Triple::ppc64; 2390 default: 2391 return Triple::UnknownArch; 2392 } 2393 } 2394 2395 Triple MachOObjectFile::getArchTriple(uint32_t CPUType, uint32_t CPUSubType, 2396 const char **McpuDefault, 2397 const char **ArchFlag) { 2398 if (McpuDefault) 2399 *McpuDefault = nullptr; 2400 if (ArchFlag) 2401 *ArchFlag = nullptr; 2402 2403 switch (CPUType) { 2404 case MachO::CPU_TYPE_I386: 2405 switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) { 2406 case MachO::CPU_SUBTYPE_I386_ALL: 2407 if (ArchFlag) 2408 *ArchFlag = "i386"; 2409 return Triple("i386-apple-darwin"); 2410 default: 2411 return Triple(); 2412 } 2413 case MachO::CPU_TYPE_X86_64: 2414 switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) { 2415 case MachO::CPU_SUBTYPE_X86_64_ALL: 2416 if (ArchFlag) 2417 *ArchFlag = "x86_64"; 2418 return Triple("x86_64-apple-darwin"); 2419 case MachO::CPU_SUBTYPE_X86_64_H: 2420 if (ArchFlag) 2421 *ArchFlag = "x86_64h"; 2422 return Triple("x86_64h-apple-darwin"); 2423 default: 2424 return Triple(); 2425 } 2426 case MachO::CPU_TYPE_ARM: 2427 switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) { 2428 case MachO::CPU_SUBTYPE_ARM_V4T: 2429 if (ArchFlag) 2430 *ArchFlag = "armv4t"; 2431 return Triple("armv4t-apple-darwin"); 2432 case MachO::CPU_SUBTYPE_ARM_V5TEJ: 2433 if (ArchFlag) 2434 *ArchFlag = "armv5e"; 2435 return Triple("armv5e-apple-darwin"); 2436 case MachO::CPU_SUBTYPE_ARM_XSCALE: 2437 if (ArchFlag) 2438 *ArchFlag = "xscale"; 2439 return Triple("xscale-apple-darwin"); 2440 case MachO::CPU_SUBTYPE_ARM_V6: 2441 if (ArchFlag) 2442 *ArchFlag = "armv6"; 2443 return Triple("armv6-apple-darwin"); 2444 case MachO::CPU_SUBTYPE_ARM_V6M: 2445 if (McpuDefault) 2446 *McpuDefault = "cortex-m0"; 2447 if (ArchFlag) 2448 *ArchFlag = "armv6m"; 2449 return Triple("armv6m-apple-darwin"); 2450 case MachO::CPU_SUBTYPE_ARM_V7: 2451 if (ArchFlag) 2452 *ArchFlag = "armv7"; 2453 return Triple("armv7-apple-darwin"); 2454 case MachO::CPU_SUBTYPE_ARM_V7EM: 2455 if (McpuDefault) 2456 *McpuDefault = "cortex-m4"; 2457 if (ArchFlag) 2458 *ArchFlag = "armv7em"; 2459 return Triple("thumbv7em-apple-darwin"); 2460 case MachO::CPU_SUBTYPE_ARM_V7K: 2461 if (McpuDefault) 2462 *McpuDefault = "cortex-a7"; 2463 if (ArchFlag) 2464 *ArchFlag = "armv7k"; 2465 return Triple("armv7k-apple-darwin"); 2466 case MachO::CPU_SUBTYPE_ARM_V7M: 2467 if (McpuDefault) 2468 *McpuDefault = "cortex-m3"; 2469 if (ArchFlag) 2470 *ArchFlag = "armv7m"; 2471 return Triple("thumbv7m-apple-darwin"); 2472 case MachO::CPU_SUBTYPE_ARM_V7S: 2473 if (McpuDefault) 2474 *McpuDefault = "cortex-a7"; 2475 if (ArchFlag) 2476 *ArchFlag = "armv7s"; 2477 return Triple("armv7s-apple-darwin"); 2478 default: 2479 return Triple(); 2480 } 2481 case MachO::CPU_TYPE_ARM64: 2482 switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) { 2483 case MachO::CPU_SUBTYPE_ARM64_ALL: 2484 if (McpuDefault) 2485 *McpuDefault = "cyclone"; 2486 if (ArchFlag) 2487 *ArchFlag = "arm64"; 2488 return Triple("arm64-apple-darwin"); 2489 default: 2490 return Triple(); 2491 } 2492 case MachO::CPU_TYPE_POWERPC: 2493 switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) { 2494 case MachO::CPU_SUBTYPE_POWERPC_ALL: 2495 if (ArchFlag) 2496 *ArchFlag = "ppc"; 2497 return Triple("ppc-apple-darwin"); 2498 default: 2499 return Triple(); 2500 } 2501 case MachO::CPU_TYPE_POWERPC64: 2502 switch (CPUSubType & ~MachO::CPU_SUBTYPE_MASK) { 2503 case MachO::CPU_SUBTYPE_POWERPC_ALL: 2504 if (ArchFlag) 2505 *ArchFlag = "ppc64"; 2506 return Triple("ppc64-apple-darwin"); 2507 default: 2508 return Triple(); 2509 } 2510 default: 2511 return Triple(); 2512 } 2513 } 2514 2515 Triple MachOObjectFile::getHostArch() { 2516 return Triple(sys::getDefaultTargetTriple()); 2517 } 2518 2519 bool MachOObjectFile::isValidArch(StringRef ArchFlag) { 2520 return StringSwitch<bool>(ArchFlag) 2521 .Case("i386", true) 2522 .Case("x86_64", true) 2523 .Case("x86_64h", true) 2524 .Case("armv4t", true) 2525 .Case("arm", true) 2526 .Case("armv5e", true) 2527 .Case("armv6", true) 2528 .Case("armv6m", true) 2529 .Case("armv7", true) 2530 .Case("armv7em", true) 2531 .Case("armv7k", true) 2532 .Case("armv7m", true) 2533 .Case("armv7s", true) 2534 .Case("arm64", true) 2535 .Case("ppc", true) 2536 .Case("ppc64", true) 2537 .Default(false); 2538 } 2539 2540 unsigned MachOObjectFile::getArch() const { 2541 return getArch(getCPUType(*this)); 2542 } 2543 2544 Triple MachOObjectFile::getArchTriple(const char **McpuDefault) const { 2545 return getArchTriple(Header.cputype, Header.cpusubtype, McpuDefault); 2546 } 2547 2548 relocation_iterator MachOObjectFile::section_rel_begin(unsigned Index) const { 2549 DataRefImpl DRI; 2550 DRI.d.a = Index; 2551 return section_rel_begin(DRI); 2552 } 2553 2554 relocation_iterator MachOObjectFile::section_rel_end(unsigned Index) const { 2555 DataRefImpl DRI; 2556 DRI.d.a = Index; 2557 return section_rel_end(DRI); 2558 } 2559 2560 dice_iterator MachOObjectFile::begin_dices() const { 2561 DataRefImpl DRI; 2562 if (!DataInCodeLoadCmd) 2563 return dice_iterator(DiceRef(DRI, this)); 2564 2565 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand(); 2566 DRI.p = reinterpret_cast<uintptr_t>(getPtr(*this, DicLC.dataoff)); 2567 return dice_iterator(DiceRef(DRI, this)); 2568 } 2569 2570 dice_iterator MachOObjectFile::end_dices() const { 2571 DataRefImpl DRI; 2572 if (!DataInCodeLoadCmd) 2573 return dice_iterator(DiceRef(DRI, this)); 2574 2575 MachO::linkedit_data_command DicLC = getDataInCodeLoadCommand(); 2576 unsigned Offset = DicLC.dataoff + DicLC.datasize; 2577 DRI.p = reinterpret_cast<uintptr_t>(getPtr(*this, Offset)); 2578 return dice_iterator(DiceRef(DRI, this)); 2579 } 2580 2581 ExportEntry::ExportEntry(ArrayRef<uint8_t> T) 2582 : Trie(T), Malformed(false), Done(false) {} 2583 2584 void ExportEntry::moveToFirst() { 2585 pushNode(0); 2586 pushDownUntilBottom(); 2587 } 2588 2589 void ExportEntry::moveToEnd() { 2590 Stack.clear(); 2591 Done = true; 2592 } 2593 2594 bool ExportEntry::operator==(const ExportEntry &Other) const { 2595 // Common case, one at end, other iterating from begin. 2596 if (Done || Other.Done) 2597 return (Done == Other.Done); 2598 // Not equal if different stack sizes. 2599 if (Stack.size() != Other.Stack.size()) 2600 return false; 2601 // Not equal if different cumulative strings. 2602 if (!CumulativeString.equals(Other.CumulativeString)) 2603 return false; 2604 // Equal if all nodes in both stacks match. 2605 for (unsigned i=0; i < Stack.size(); ++i) { 2606 if (Stack[i].Start != Other.Stack[i].Start) 2607 return false; 2608 } 2609 return true; 2610 } 2611 2612 uint64_t ExportEntry::readULEB128(const uint8_t *&Ptr) { 2613 unsigned Count; 2614 uint64_t Result = decodeULEB128(Ptr, &Count); 2615 Ptr += Count; 2616 if (Ptr > Trie.end()) { 2617 Ptr = Trie.end(); 2618 Malformed = true; 2619 } 2620 return Result; 2621 } 2622 2623 StringRef ExportEntry::name() const { 2624 return CumulativeString; 2625 } 2626 2627 uint64_t ExportEntry::flags() const { 2628 return Stack.back().Flags; 2629 } 2630 2631 uint64_t ExportEntry::address() const { 2632 return Stack.back().Address; 2633 } 2634 2635 uint64_t ExportEntry::other() const { 2636 return Stack.back().Other; 2637 } 2638 2639 StringRef ExportEntry::otherName() const { 2640 const char* ImportName = Stack.back().ImportName; 2641 if (ImportName) 2642 return StringRef(ImportName); 2643 return StringRef(); 2644 } 2645 2646 uint32_t ExportEntry::nodeOffset() const { 2647 return Stack.back().Start - Trie.begin(); 2648 } 2649 2650 ExportEntry::NodeState::NodeState(const uint8_t *Ptr) 2651 : Start(Ptr), Current(Ptr), Flags(0), Address(0), Other(0), 2652 ImportName(nullptr), ChildCount(0), NextChildIndex(0), 2653 ParentStringLength(0), IsExportNode(false) {} 2654 2655 void ExportEntry::pushNode(uint64_t offset) { 2656 const uint8_t *Ptr = Trie.begin() + offset; 2657 NodeState State(Ptr); 2658 uint64_t ExportInfoSize = readULEB128(State.Current); 2659 State.IsExportNode = (ExportInfoSize != 0); 2660 const uint8_t* Children = State.Current + ExportInfoSize; 2661 if (State.IsExportNode) { 2662 State.Flags = readULEB128(State.Current); 2663 if (State.Flags & MachO::EXPORT_SYMBOL_FLAGS_REEXPORT) { 2664 State.Address = 0; 2665 State.Other = readULEB128(State.Current); // dylib ordinal 2666 State.ImportName = reinterpret_cast<const char*>(State.Current); 2667 } else { 2668 State.Address = readULEB128(State.Current); 2669 if (State.Flags & MachO::EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER) 2670 State.Other = readULEB128(State.Current); 2671 } 2672 } 2673 State.ChildCount = *Children; 2674 State.Current = Children + 1; 2675 State.NextChildIndex = 0; 2676 State.ParentStringLength = CumulativeString.size(); 2677 Stack.push_back(State); 2678 } 2679 2680 void ExportEntry::pushDownUntilBottom() { 2681 while (Stack.back().NextChildIndex < Stack.back().ChildCount) { 2682 NodeState &Top = Stack.back(); 2683 CumulativeString.resize(Top.ParentStringLength); 2684 for (;*Top.Current != 0; Top.Current++) { 2685 char C = *Top.Current; 2686 CumulativeString.push_back(C); 2687 } 2688 Top.Current += 1; 2689 uint64_t childNodeIndex = readULEB128(Top.Current); 2690 Top.NextChildIndex += 1; 2691 pushNode(childNodeIndex); 2692 } 2693 if (!Stack.back().IsExportNode) { 2694 Malformed = true; 2695 moveToEnd(); 2696 } 2697 } 2698 2699 // We have a trie data structure and need a way to walk it that is compatible 2700 // with the C++ iterator model. The solution is a non-recursive depth first 2701 // traversal where the iterator contains a stack of parent nodes along with a 2702 // string that is the accumulation of all edge strings along the parent chain 2703 // to this point. 2704 // 2705 // There is one "export" node for each exported symbol. But because some 2706 // symbols may be a prefix of another symbol (e.g. _dup and _dup2), an export 2707 // node may have child nodes too. 2708 // 2709 // The algorithm for moveNext() is to keep moving down the leftmost unvisited 2710 // child until hitting a node with no children (which is an export node or 2711 // else the trie is malformed). On the way down, each node is pushed on the 2712 // stack ivar. If there is no more ways down, it pops up one and tries to go 2713 // down a sibling path until a childless node is reached. 2714 void ExportEntry::moveNext() { 2715 if (Stack.empty() || !Stack.back().IsExportNode) { 2716 Malformed = true; 2717 moveToEnd(); 2718 return; 2719 } 2720 2721 Stack.pop_back(); 2722 while (!Stack.empty()) { 2723 NodeState &Top = Stack.back(); 2724 if (Top.NextChildIndex < Top.ChildCount) { 2725 pushDownUntilBottom(); 2726 // Now at the next export node. 2727 return; 2728 } else { 2729 if (Top.IsExportNode) { 2730 // This node has no children but is itself an export node. 2731 CumulativeString.resize(Top.ParentStringLength); 2732 return; 2733 } 2734 Stack.pop_back(); 2735 } 2736 } 2737 Done = true; 2738 } 2739 2740 iterator_range<export_iterator> 2741 MachOObjectFile::exports(ArrayRef<uint8_t> Trie) { 2742 ExportEntry Start(Trie); 2743 if (Trie.size() == 0) 2744 Start.moveToEnd(); 2745 else 2746 Start.moveToFirst(); 2747 2748 ExportEntry Finish(Trie); 2749 Finish.moveToEnd(); 2750 2751 return make_range(export_iterator(Start), export_iterator(Finish)); 2752 } 2753 2754 iterator_range<export_iterator> MachOObjectFile::exports() const { 2755 return exports(getDyldInfoExportsTrie()); 2756 } 2757 2758 MachORebaseEntry::MachORebaseEntry(Error *E, const MachOObjectFile *O, 2759 ArrayRef<uint8_t> Bytes, bool is64Bit) 2760 : E(E), O(O), Opcodes(Bytes), Ptr(Bytes.begin()), SegmentOffset(0), 2761 SegmentIndex(-1), RemainingLoopCount(0), AdvanceAmount(0), RebaseType(0), 2762 PointerSize(is64Bit ? 8 : 4), Done(false) {} 2763 2764 void MachORebaseEntry::moveToFirst() { 2765 Ptr = Opcodes.begin(); 2766 moveNext(); 2767 } 2768 2769 void MachORebaseEntry::moveToEnd() { 2770 Ptr = Opcodes.end(); 2771 RemainingLoopCount = 0; 2772 Done = true; 2773 } 2774 2775 void MachORebaseEntry::moveNext() { 2776 ErrorAsOutParameter ErrAsOutParam(E); 2777 // If in the middle of some loop, move to next rebasing in loop. 2778 SegmentOffset += AdvanceAmount; 2779 if (RemainingLoopCount) { 2780 --RemainingLoopCount; 2781 return; 2782 } 2783 if (Ptr >= Opcodes.end()) { 2784 if (Opcodes.begin() != Opcodes.end() && Done != true) { 2785 *E = malformedError("missing REBASE_OPCODE_DONE at end of opcodes"); 2786 moveToEnd(); 2787 return; 2788 } 2789 Done = true; 2790 return; 2791 } 2792 bool More = true; 2793 while (More) { 2794 // Parse next opcode and set up next loop. 2795 const uint8_t *OpcodeStart = Ptr; 2796 uint8_t Byte = *Ptr++; 2797 uint8_t ImmValue = Byte & MachO::REBASE_IMMEDIATE_MASK; 2798 uint8_t Opcode = Byte & MachO::REBASE_OPCODE_MASK; 2799 uint32_t Count, Skip; 2800 const char *error = nullptr; 2801 switch (Opcode) { 2802 case MachO::REBASE_OPCODE_DONE: 2803 More = false; 2804 Done = true; 2805 moveToEnd(); 2806 DEBUG_WITH_TYPE("mach-o-rebase", llvm::dbgs() << "REBASE_OPCODE_DONE\n"); 2807 break; 2808 case MachO::REBASE_OPCODE_SET_TYPE_IMM: 2809 RebaseType = ImmValue; 2810 if (RebaseType > MachO::REBASE_TYPE_TEXT_PCREL32) { 2811 *E = malformedError("for REBASE_OPCODE_SET_TYPE_IMM bad bind type: " + 2812 Twine((int)RebaseType) + " for opcode at: 0x" + 2813 utohexstr(OpcodeStart - Opcodes.begin())); 2814 moveToEnd(); 2815 return; 2816 } 2817 DEBUG_WITH_TYPE( 2818 "mach-o-rebase", 2819 llvm::dbgs() << "REBASE_OPCODE_SET_TYPE_IMM: " 2820 << "RebaseType=" << (int) RebaseType << "\n"); 2821 break; 2822 case MachO::REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB: 2823 SegmentIndex = ImmValue; 2824 SegmentOffset = readULEB128(&error); 2825 if (error) { 2826 *E = malformedError("for REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB " + 2827 Twine(error) + " for opcode at: 0x" + 2828 utohexstr(OpcodeStart - Opcodes.begin())); 2829 moveToEnd(); 2830 return; 2831 } 2832 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2833 true); 2834 if (error) { 2835 *E = malformedError("for REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB " + 2836 Twine(error) + " for opcode at: 0x" + 2837 utohexstr(OpcodeStart - Opcodes.begin())); 2838 moveToEnd(); 2839 return; 2840 } 2841 DEBUG_WITH_TYPE( 2842 "mach-o-rebase", 2843 llvm::dbgs() << "REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB: " 2844 << "SegmentIndex=" << SegmentIndex << ", " 2845 << format("SegmentOffset=0x%06X", SegmentOffset) 2846 << "\n"); 2847 break; 2848 case MachO::REBASE_OPCODE_ADD_ADDR_ULEB: 2849 SegmentOffset += readULEB128(&error); 2850 if (error) { 2851 *E = malformedError("for REBASE_OPCODE_ADD_ADDR_ULEB " + 2852 Twine(error) + " for opcode at: 0x" + 2853 utohexstr(OpcodeStart - Opcodes.begin())); 2854 moveToEnd(); 2855 return; 2856 } 2857 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2858 true); 2859 if (error) { 2860 *E = malformedError("for REBASE_OPCODE_ADD_ADDR_ULEB " + 2861 Twine(error) + " for opcode at: 0x" + 2862 utohexstr(OpcodeStart - Opcodes.begin())); 2863 moveToEnd(); 2864 return; 2865 } 2866 DEBUG_WITH_TYPE("mach-o-rebase", 2867 llvm::dbgs() << "REBASE_OPCODE_ADD_ADDR_ULEB: " 2868 << format("SegmentOffset=0x%06X", 2869 SegmentOffset) << "\n"); 2870 break; 2871 case MachO::REBASE_OPCODE_ADD_ADDR_IMM_SCALED: 2872 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2873 true); 2874 if (error) { 2875 *E = malformedError("for REBASE_OPCODE_ADD_ADDR_IMM_SCALED " + 2876 Twine(error) + " for opcode at: 0x" + 2877 utohexstr(OpcodeStart - Opcodes.begin())); 2878 moveToEnd(); 2879 return; 2880 } 2881 SegmentOffset += ImmValue * PointerSize; 2882 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2883 false); 2884 if (error) { 2885 *E = malformedError("for REBASE_OPCODE_ADD_ADDR_IMM_SCALED " 2886 " (after adding immediate times the pointer size) " + 2887 Twine(error) + " for opcode at: 0x" + 2888 utohexstr(OpcodeStart - Opcodes.begin())); 2889 moveToEnd(); 2890 return; 2891 } 2892 DEBUG_WITH_TYPE("mach-o-rebase", 2893 llvm::dbgs() << "REBASE_OPCODE_ADD_ADDR_IMM_SCALED: " 2894 << format("SegmentOffset=0x%06X", 2895 SegmentOffset) << "\n"); 2896 break; 2897 case MachO::REBASE_OPCODE_DO_REBASE_IMM_TIMES: 2898 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2899 true); 2900 if (error) { 2901 *E = malformedError("for REBASE_OPCODE_DO_REBASE_IMM_TIMES " + 2902 Twine(error) + " for opcode at: 0x" + 2903 utohexstr(OpcodeStart - Opcodes.begin())); 2904 moveToEnd(); 2905 return; 2906 } 2907 AdvanceAmount = PointerSize; 2908 Skip = 0; 2909 Count = ImmValue; 2910 if (ImmValue != 0) 2911 RemainingLoopCount = ImmValue - 1; 2912 else 2913 RemainingLoopCount = 0; 2914 error = O->RebaseEntryCheckCountAndSkip(Count, Skip, PointerSize, 2915 SegmentIndex, SegmentOffset); 2916 if (error) { 2917 *E = malformedError("for REBASE_OPCODE_DO_REBASE_IMM_TIMES " 2918 + Twine(error) + " for opcode at: 0x" + 2919 utohexstr(OpcodeStart - Opcodes.begin())); 2920 moveToEnd(); 2921 return; 2922 } 2923 DEBUG_WITH_TYPE( 2924 "mach-o-rebase", 2925 llvm::dbgs() << "REBASE_OPCODE_DO_REBASE_IMM_TIMES: " 2926 << format("SegmentOffset=0x%06X", SegmentOffset) 2927 << ", AdvanceAmount=" << AdvanceAmount 2928 << ", RemainingLoopCount=" << RemainingLoopCount 2929 << "\n"); 2930 return; 2931 case MachO::REBASE_OPCODE_DO_REBASE_ULEB_TIMES: 2932 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2933 true); 2934 if (error) { 2935 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES " + 2936 Twine(error) + " for opcode at: 0x" + 2937 utohexstr(OpcodeStart - Opcodes.begin())); 2938 moveToEnd(); 2939 return; 2940 } 2941 AdvanceAmount = PointerSize; 2942 Skip = 0; 2943 Count = readULEB128(&error); 2944 if (error) { 2945 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES " + 2946 Twine(error) + " for opcode at: 0x" + 2947 utohexstr(OpcodeStart - Opcodes.begin())); 2948 moveToEnd(); 2949 return; 2950 } 2951 if (Count != 0) 2952 RemainingLoopCount = Count - 1; 2953 else 2954 RemainingLoopCount = 0; 2955 error = O->RebaseEntryCheckCountAndSkip(Count, Skip, PointerSize, 2956 SegmentIndex, SegmentOffset); 2957 if (error) { 2958 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES " 2959 + Twine(error) + " for opcode at: 0x" + 2960 utohexstr(OpcodeStart - Opcodes.begin())); 2961 moveToEnd(); 2962 return; 2963 } 2964 DEBUG_WITH_TYPE( 2965 "mach-o-rebase", 2966 llvm::dbgs() << "REBASE_OPCODE_DO_REBASE_ULEB_TIMES: " 2967 << format("SegmentOffset=0x%06X", SegmentOffset) 2968 << ", AdvanceAmount=" << AdvanceAmount 2969 << ", RemainingLoopCount=" << RemainingLoopCount 2970 << "\n"); 2971 return; 2972 case MachO::REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB: 2973 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 2974 true); 2975 if (error) { 2976 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB " + 2977 Twine(error) + " for opcode at: 0x" + 2978 utohexstr(OpcodeStart - Opcodes.begin())); 2979 moveToEnd(); 2980 return; 2981 } 2982 Skip = readULEB128(&error); 2983 if (error) { 2984 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB " + 2985 Twine(error) + " for opcode at: 0x" + 2986 utohexstr(OpcodeStart - Opcodes.begin())); 2987 moveToEnd(); 2988 return; 2989 } 2990 AdvanceAmount = Skip + PointerSize; 2991 Count = 1; 2992 RemainingLoopCount = 0; 2993 error = O->RebaseEntryCheckCountAndSkip(Count, Skip, PointerSize, 2994 SegmentIndex, SegmentOffset); 2995 if (error) { 2996 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB " 2997 + Twine(error) + " for opcode at: 0x" + 2998 utohexstr(OpcodeStart - Opcodes.begin())); 2999 moveToEnd(); 3000 return; 3001 } 3002 DEBUG_WITH_TYPE( 3003 "mach-o-rebase", 3004 llvm::dbgs() << "REBASE_OPCODE_DO_REBASE_ADD_ADDR_ULEB: " 3005 << format("SegmentOffset=0x%06X", SegmentOffset) 3006 << ", AdvanceAmount=" << AdvanceAmount 3007 << ", RemainingLoopCount=" << RemainingLoopCount 3008 << "\n"); 3009 return; 3010 case MachO::REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_ULEB: 3011 error = O->RebaseEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, 3012 true); 3013 if (error) { 3014 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_" 3015 "ULEB " + Twine(error) + " for opcode at: 0x" + 3016 utohexstr(OpcodeStart - Opcodes.begin())); 3017 moveToEnd(); 3018 return; 3019 } 3020 Count = readULEB128(&error); 3021 if (error) { 3022 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_" 3023 "ULEB " + Twine(error) + " for opcode at: 0x" + 3024 utohexstr(OpcodeStart - Opcodes.begin())); 3025 moveToEnd(); 3026 return; 3027 } 3028 if (Count != 0) 3029 RemainingLoopCount = Count - 1; 3030 else 3031 RemainingLoopCount = 0; 3032 Skip = readULEB128(&error); 3033 if (error) { 3034 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_" 3035 "ULEB " + Twine(error) + " for opcode at: 0x" + 3036 utohexstr(OpcodeStart - Opcodes.begin())); 3037 moveToEnd(); 3038 return; 3039 } 3040 AdvanceAmount = Skip + PointerSize; 3041 3042 error = O->RebaseEntryCheckCountAndSkip(Count, Skip, PointerSize, 3043 SegmentIndex, SegmentOffset); 3044 if (error) { 3045 *E = malformedError("for REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_" 3046 "ULEB " + Twine(error) + " for opcode at: 0x" + 3047 utohexstr(OpcodeStart - Opcodes.begin())); 3048 moveToEnd(); 3049 return; 3050 } 3051 DEBUG_WITH_TYPE( 3052 "mach-o-rebase", 3053 llvm::dbgs() << "REBASE_OPCODE_DO_REBASE_ULEB_TIMES_SKIPPING_ULEB: " 3054 << format("SegmentOffset=0x%06X", SegmentOffset) 3055 << ", AdvanceAmount=" << AdvanceAmount 3056 << ", RemainingLoopCount=" << RemainingLoopCount 3057 << "\n"); 3058 return; 3059 default: 3060 *E = malformedError("bad rebase info (bad opcode value 0x" + 3061 utohexstr(Opcode) + " for opcode at: 0x" + 3062 utohexstr(OpcodeStart - Opcodes.begin())); 3063 moveToEnd(); 3064 return; 3065 } 3066 } 3067 } 3068 3069 uint64_t MachORebaseEntry::readULEB128(const char **error) { 3070 unsigned Count; 3071 uint64_t Result = decodeULEB128(Ptr, &Count, Opcodes.end(), error); 3072 Ptr += Count; 3073 if (Ptr > Opcodes.end()) 3074 Ptr = Opcodes.end(); 3075 return Result; 3076 } 3077 3078 int32_t MachORebaseEntry::segmentIndex() const { return SegmentIndex; } 3079 3080 uint64_t MachORebaseEntry::segmentOffset() const { return SegmentOffset; } 3081 3082 StringRef MachORebaseEntry::typeName() const { 3083 switch (RebaseType) { 3084 case MachO::REBASE_TYPE_POINTER: 3085 return "pointer"; 3086 case MachO::REBASE_TYPE_TEXT_ABSOLUTE32: 3087 return "text abs32"; 3088 case MachO::REBASE_TYPE_TEXT_PCREL32: 3089 return "text rel32"; 3090 } 3091 return "unknown"; 3092 } 3093 3094 // For use with the SegIndex of a checked Mach-O Rebase entry 3095 // to get the segment name. 3096 StringRef MachORebaseEntry::segmentName() const { 3097 return O->BindRebaseSegmentName(SegmentIndex); 3098 } 3099 3100 // For use with a SegIndex,SegOffset pair from a checked Mach-O Rebase entry 3101 // to get the section name. 3102 StringRef MachORebaseEntry::sectionName() const { 3103 return O->BindRebaseSectionName(SegmentIndex, SegmentOffset); 3104 } 3105 3106 // For use with a SegIndex,SegOffset pair from a checked Mach-O Rebase entry 3107 // to get the address. 3108 uint64_t MachORebaseEntry::address() const { 3109 return O->BindRebaseAddress(SegmentIndex, SegmentOffset); 3110 } 3111 3112 bool MachORebaseEntry::operator==(const MachORebaseEntry &Other) const { 3113 #ifdef EXPENSIVE_CHECKS 3114 assert(Opcodes == Other.Opcodes && "compare iterators of different files"); 3115 #else 3116 assert(Opcodes.data() == Other.Opcodes.data() && "compare iterators of different files"); 3117 #endif 3118 return (Ptr == Other.Ptr) && 3119 (RemainingLoopCount == Other.RemainingLoopCount) && 3120 (Done == Other.Done); 3121 } 3122 3123 iterator_range<rebase_iterator> 3124 MachOObjectFile::rebaseTable(Error &Err, MachOObjectFile *O, 3125 ArrayRef<uint8_t> Opcodes, bool is64) { 3126 if (O->BindRebaseSectionTable == nullptr) 3127 O->BindRebaseSectionTable = llvm::make_unique<BindRebaseSegInfo>(O); 3128 MachORebaseEntry Start(&Err, O, Opcodes, is64); 3129 Start.moveToFirst(); 3130 3131 MachORebaseEntry Finish(&Err, O, Opcodes, is64); 3132 Finish.moveToEnd(); 3133 3134 return make_range(rebase_iterator(Start), rebase_iterator(Finish)); 3135 } 3136 3137 iterator_range<rebase_iterator> MachOObjectFile::rebaseTable(Error &Err) { 3138 return rebaseTable(Err, this, getDyldInfoRebaseOpcodes(), is64Bit()); 3139 } 3140 3141 MachOBindEntry::MachOBindEntry(Error *E, const MachOObjectFile *O, 3142 ArrayRef<uint8_t> Bytes, bool is64Bit, Kind BK) 3143 : E(E), O(O), Opcodes(Bytes), Ptr(Bytes.begin()), SegmentOffset(0), 3144 SegmentIndex(-1), LibraryOrdinalSet(false), Ordinal(0), Flags(0), 3145 Addend(0), RemainingLoopCount(0), AdvanceAmount(0), BindType(0), 3146 PointerSize(is64Bit ? 8 : 4), TableKind(BK), Done(false) {} 3147 3148 void MachOBindEntry::moveToFirst() { 3149 Ptr = Opcodes.begin(); 3150 moveNext(); 3151 } 3152 3153 void MachOBindEntry::moveToEnd() { 3154 Ptr = Opcodes.end(); 3155 RemainingLoopCount = 0; 3156 Done = true; 3157 } 3158 3159 void MachOBindEntry::moveNext() { 3160 ErrorAsOutParameter ErrAsOutParam(E); 3161 // If in the middle of some loop, move to next binding in loop. 3162 SegmentOffset += AdvanceAmount; 3163 if (RemainingLoopCount) { 3164 --RemainingLoopCount; 3165 return; 3166 } 3167 if (Ptr >= Opcodes.end()) { 3168 if (Opcodes.begin() != Opcodes.end() && Done != true) { 3169 *E = malformedError("missing BIND_OPCODE_DONE at end of opcodes"); 3170 moveToEnd(); 3171 return; 3172 } 3173 Done = true; 3174 return; 3175 } 3176 bool More = true; 3177 while (More) { 3178 // Parse next opcode and set up next loop. 3179 const uint8_t *OpcodeStart = Ptr; 3180 uint8_t Byte = *Ptr++; 3181 uint8_t ImmValue = Byte & MachO::BIND_IMMEDIATE_MASK; 3182 uint8_t Opcode = Byte & MachO::BIND_OPCODE_MASK; 3183 int8_t SignExtended; 3184 const uint8_t *SymStart; 3185 uint32_t Count, Skip; 3186 const char *error = nullptr; 3187 switch (Opcode) { 3188 case MachO::BIND_OPCODE_DONE: 3189 if (TableKind == Kind::Lazy) { 3190 // Lazying bindings have a DONE opcode between entries. Need to ignore 3191 // it to advance to next entry. But need not if this is last entry. 3192 bool NotLastEntry = false; 3193 for (const uint8_t *P = Ptr; P < Opcodes.end(); ++P) { 3194 if (*P) { 3195 NotLastEntry = true; 3196 } 3197 } 3198 if (NotLastEntry) 3199 break; 3200 } 3201 More = false; 3202 moveToEnd(); 3203 DEBUG_WITH_TYPE("mach-o-bind", llvm::dbgs() << "BIND_OPCODE_DONE\n"); 3204 break; 3205 case MachO::BIND_OPCODE_SET_DYLIB_ORDINAL_IMM: 3206 if (TableKind == Kind::Weak) { 3207 *E = malformedError("BIND_OPCODE_SET_DYLIB_ORDINAL_IMM not allowed in " 3208 "weak bind table for opcode at: 0x" + 3209 utohexstr(OpcodeStart - Opcodes.begin())); 3210 moveToEnd(); 3211 return; 3212 } 3213 Ordinal = ImmValue; 3214 LibraryOrdinalSet = true; 3215 if (ImmValue > O->getLibraryCount()) { 3216 *E = malformedError("for BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB bad " 3217 "library ordinal: " + Twine((int)ImmValue) + " (max " + 3218 Twine((int)O->getLibraryCount()) + ") for opcode at: 0x" + 3219 utohexstr(OpcodeStart - Opcodes.begin())); 3220 moveToEnd(); 3221 return; 3222 } 3223 DEBUG_WITH_TYPE( 3224 "mach-o-bind", 3225 llvm::dbgs() << "BIND_OPCODE_SET_DYLIB_ORDINAL_IMM: " 3226 << "Ordinal=" << Ordinal << "\n"); 3227 break; 3228 case MachO::BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB: 3229 if (TableKind == Kind::Weak) { 3230 *E = malformedError("BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB not allowed in " 3231 "weak bind table for opcode at: 0x" + 3232 utohexstr(OpcodeStart - Opcodes.begin())); 3233 moveToEnd(); 3234 return; 3235 } 3236 Ordinal = readULEB128(&error); 3237 LibraryOrdinalSet = true; 3238 if (error) { 3239 *E = malformedError("for BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB " + 3240 Twine(error) + " for opcode at: 0x" + 3241 utohexstr(OpcodeStart - Opcodes.begin())); 3242 moveToEnd(); 3243 return; 3244 } 3245 if (Ordinal > (int)O->getLibraryCount()) { 3246 *E = malformedError("for BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB bad " 3247 "library ordinal: " + Twine((int)Ordinal) + " (max " + 3248 Twine((int)O->getLibraryCount()) + ") for opcode at: 0x" + 3249 utohexstr(OpcodeStart - Opcodes.begin())); 3250 moveToEnd(); 3251 return; 3252 } 3253 DEBUG_WITH_TYPE( 3254 "mach-o-bind", 3255 llvm::dbgs() << "BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB: " 3256 << "Ordinal=" << Ordinal << "\n"); 3257 break; 3258 case MachO::BIND_OPCODE_SET_DYLIB_SPECIAL_IMM: 3259 if (TableKind == Kind::Weak) { 3260 *E = malformedError("BIND_OPCODE_SET_DYLIB_SPECIAL_IMM not allowed in " 3261 "weak bind table for opcode at: 0x" + 3262 utohexstr(OpcodeStart - Opcodes.begin())); 3263 moveToEnd(); 3264 return; 3265 } 3266 if (ImmValue) { 3267 SignExtended = MachO::BIND_OPCODE_MASK | ImmValue; 3268 Ordinal = SignExtended; 3269 LibraryOrdinalSet = true; 3270 if (Ordinal < MachO::BIND_SPECIAL_DYLIB_FLAT_LOOKUP) { 3271 *E = malformedError("for BIND_OPCODE_SET_DYLIB_SPECIAL_IMM unknown " 3272 "special ordinal: " + Twine((int)Ordinal) + " for opcode at: " 3273 "0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3274 moveToEnd(); 3275 return; 3276 } 3277 } else 3278 Ordinal = 0; 3279 DEBUG_WITH_TYPE( 3280 "mach-o-bind", 3281 llvm::dbgs() << "BIND_OPCODE_SET_DYLIB_SPECIAL_IMM: " 3282 << "Ordinal=" << Ordinal << "\n"); 3283 break; 3284 case MachO::BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM: 3285 Flags = ImmValue; 3286 SymStart = Ptr; 3287 while (*Ptr && (Ptr < Opcodes.end())) { 3288 ++Ptr; 3289 } 3290 if (Ptr == Opcodes.end()) { 3291 *E = malformedError("for BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM " 3292 "symbol name extends past opcodes for opcode at: 0x" + 3293 utohexstr(OpcodeStart - Opcodes.begin())); 3294 moveToEnd(); 3295 return; 3296 } 3297 SymbolName = StringRef(reinterpret_cast<const char*>(SymStart), 3298 Ptr-SymStart); 3299 ++Ptr; 3300 DEBUG_WITH_TYPE( 3301 "mach-o-bind", 3302 llvm::dbgs() << "BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM: " 3303 << "SymbolName=" << SymbolName << "\n"); 3304 if (TableKind == Kind::Weak) { 3305 if (ImmValue & MachO::BIND_SYMBOL_FLAGS_NON_WEAK_DEFINITION) 3306 return; 3307 } 3308 break; 3309 case MachO::BIND_OPCODE_SET_TYPE_IMM: 3310 BindType = ImmValue; 3311 if (ImmValue > MachO::BIND_TYPE_TEXT_PCREL32) { 3312 *E = malformedError("for BIND_OPCODE_SET_TYPE_IMM bad bind type: " + 3313 Twine((int)ImmValue) + " for opcode at: 0x" + 3314 utohexstr(OpcodeStart - Opcodes.begin())); 3315 moveToEnd(); 3316 return; 3317 } 3318 DEBUG_WITH_TYPE( 3319 "mach-o-bind", 3320 llvm::dbgs() << "BIND_OPCODE_SET_TYPE_IMM: " 3321 << "BindType=" << (int)BindType << "\n"); 3322 break; 3323 case MachO::BIND_OPCODE_SET_ADDEND_SLEB: 3324 Addend = readSLEB128(&error); 3325 if (error) { 3326 *E = malformedError("for BIND_OPCODE_SET_ADDEND_SLEB " + 3327 Twine(error) + " for opcode at: 0x" + 3328 utohexstr(OpcodeStart - Opcodes.begin())); 3329 moveToEnd(); 3330 return; 3331 } 3332 DEBUG_WITH_TYPE( 3333 "mach-o-bind", 3334 llvm::dbgs() << "BIND_OPCODE_SET_ADDEND_SLEB: " 3335 << "Addend=" << Addend << "\n"); 3336 break; 3337 case MachO::BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB: 3338 SegmentIndex = ImmValue; 3339 SegmentOffset = readULEB128(&error); 3340 if (error) { 3341 *E = malformedError("for BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB " + 3342 Twine(error) + " for opcode at: 0x" + 3343 utohexstr(OpcodeStart - Opcodes.begin())); 3344 moveToEnd(); 3345 return; 3346 } 3347 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, true); 3348 if (error) { 3349 *E = malformedError("for BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB " + 3350 Twine(error) + " for opcode at: 0x" + 3351 utohexstr(OpcodeStart - Opcodes.begin())); 3352 moveToEnd(); 3353 return; 3354 } 3355 DEBUG_WITH_TYPE( 3356 "mach-o-bind", 3357 llvm::dbgs() << "BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB: " 3358 << "SegmentIndex=" << SegmentIndex << ", " 3359 << format("SegmentOffset=0x%06X", SegmentOffset) 3360 << "\n"); 3361 break; 3362 case MachO::BIND_OPCODE_ADD_ADDR_ULEB: 3363 SegmentOffset += readULEB128(&error); 3364 if (error) { 3365 *E = malformedError("for BIND_OPCODE_ADD_ADDR_ULEB " + 3366 Twine(error) + " for opcode at: 0x" + 3367 utohexstr(OpcodeStart - Opcodes.begin())); 3368 moveToEnd(); 3369 return; 3370 } 3371 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, true); 3372 if (error) { 3373 *E = malformedError("for BIND_OPCODE_ADD_ADDR_ULEB " + 3374 Twine(error) + " for opcode at: 0x" + 3375 utohexstr(OpcodeStart - Opcodes.begin())); 3376 moveToEnd(); 3377 return; 3378 } 3379 DEBUG_WITH_TYPE("mach-o-bind", 3380 llvm::dbgs() << "BIND_OPCODE_ADD_ADDR_ULEB: " 3381 << format("SegmentOffset=0x%06X", 3382 SegmentOffset) << "\n"); 3383 break; 3384 case MachO::BIND_OPCODE_DO_BIND: 3385 AdvanceAmount = PointerSize; 3386 RemainingLoopCount = 0; 3387 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, true); 3388 if (error) { 3389 *E = malformedError("for BIND_OPCODE_DO_BIND " + Twine(error) + 3390 " for opcode at: 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3391 moveToEnd(); 3392 return; 3393 } 3394 if (SymbolName == StringRef()) { 3395 *E = malformedError("for BIND_OPCODE_DO_BIND missing preceding " 3396 "BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM for opcode at: 0x" + 3397 utohexstr(OpcodeStart - Opcodes.begin())); 3398 moveToEnd(); 3399 return; 3400 } 3401 if (!LibraryOrdinalSet && TableKind != Kind::Weak) { 3402 *E = malformedError("for BIND_OPCODE_DO_BIND missing preceding " 3403 "BIND_OPCODE_SET_DYLIB_ORDINAL_* for opcode at: 0x" + 3404 utohexstr(OpcodeStart - Opcodes.begin())); 3405 moveToEnd(); 3406 return; 3407 } 3408 DEBUG_WITH_TYPE("mach-o-bind", 3409 llvm::dbgs() << "BIND_OPCODE_DO_BIND: " 3410 << format("SegmentOffset=0x%06X", 3411 SegmentOffset) << "\n"); 3412 return; 3413 case MachO::BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB: 3414 if (TableKind == Kind::Lazy) { 3415 *E = malformedError("BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB not allowed in " 3416 "lazy bind table for opcode at: 0x" + 3417 utohexstr(OpcodeStart - Opcodes.begin())); 3418 moveToEnd(); 3419 return; 3420 } 3421 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, true); 3422 if (error) { 3423 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB " + 3424 Twine(error) + " for opcode at: 0x" + 3425 utohexstr(OpcodeStart - Opcodes.begin())); 3426 moveToEnd(); 3427 return; 3428 } 3429 if (SymbolName == StringRef()) { 3430 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB missing " 3431 "preceding BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM for opcode " 3432 "at: 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3433 moveToEnd(); 3434 return; 3435 } 3436 if (!LibraryOrdinalSet && TableKind != Kind::Weak) { 3437 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB missing " 3438 "preceding BIND_OPCODE_SET_DYLIB_ORDINAL_* for opcode at: 0x" + 3439 utohexstr(OpcodeStart - Opcodes.begin())); 3440 moveToEnd(); 3441 return; 3442 } 3443 AdvanceAmount = readULEB128(&error) + PointerSize; 3444 if (error) { 3445 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB " + 3446 Twine(error) + " for opcode at: 0x" + 3447 utohexstr(OpcodeStart - Opcodes.begin())); 3448 moveToEnd(); 3449 return; 3450 } 3451 // Note, this is not really an error until the next bind but make no sense 3452 // for a BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB to not be followed by another 3453 // bind operation. 3454 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset + 3455 AdvanceAmount, false); 3456 if (error) { 3457 *E = malformedError("for BIND_OPCODE_ADD_ADDR_ULEB (after adding " 3458 "ULEB) " + Twine(error) + " for opcode at: 0x" + 3459 utohexstr(OpcodeStart - Opcodes.begin())); 3460 moveToEnd(); 3461 return; 3462 } 3463 RemainingLoopCount = 0; 3464 DEBUG_WITH_TYPE( 3465 "mach-o-bind", 3466 llvm::dbgs() << "BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB: " 3467 << format("SegmentOffset=0x%06X", SegmentOffset) 3468 << ", AdvanceAmount=" << AdvanceAmount 3469 << ", RemainingLoopCount=" << RemainingLoopCount 3470 << "\n"); 3471 return; 3472 case MachO::BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED: 3473 if (TableKind == Kind::Lazy) { 3474 *E = malformedError("BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED not " 3475 "allowed in lazy bind table for opcode at: 0x" + 3476 utohexstr(OpcodeStart - Opcodes.begin())); 3477 moveToEnd(); 3478 return; 3479 } 3480 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, true); 3481 if (error) { 3482 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED " + 3483 Twine(error) + " for opcode at: 0x" + 3484 utohexstr(OpcodeStart - Opcodes.begin())); 3485 moveToEnd(); 3486 return; 3487 } 3488 if (SymbolName == StringRef()) { 3489 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED " 3490 "missing preceding BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM for " 3491 "opcode at: 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3492 moveToEnd(); 3493 return; 3494 } 3495 if (!LibraryOrdinalSet && TableKind != Kind::Weak) { 3496 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED " 3497 "missing preceding BIND_OPCODE_SET_DYLIB_ORDINAL_* for opcode " 3498 "at: 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3499 moveToEnd(); 3500 return; 3501 } 3502 AdvanceAmount = ImmValue * PointerSize + PointerSize; 3503 RemainingLoopCount = 0; 3504 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset + 3505 AdvanceAmount, false); 3506 if (error) { 3507 *E = malformedError("for BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED " 3508 " (after adding immediate times the pointer size) " + 3509 Twine(error) + " for opcode at: 0x" + 3510 utohexstr(OpcodeStart - Opcodes.begin())); 3511 moveToEnd(); 3512 return; 3513 } 3514 DEBUG_WITH_TYPE("mach-o-bind", 3515 llvm::dbgs() 3516 << "BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED: " 3517 << format("SegmentOffset=0x%06X", 3518 SegmentOffset) << "\n"); 3519 return; 3520 case MachO::BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB: 3521 if (TableKind == Kind::Lazy) { 3522 *E = malformedError("BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB not " 3523 "allowed in lazy bind table for opcode at: 0x" + 3524 utohexstr(OpcodeStart - Opcodes.begin())); 3525 moveToEnd(); 3526 return; 3527 } 3528 Count = readULEB128(&error); 3529 if (Count != 0) 3530 RemainingLoopCount = Count - 1; 3531 else 3532 RemainingLoopCount = 0; 3533 if (error) { 3534 *E = malformedError("for BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB " 3535 " (count value) " + Twine(error) + " for opcode at" 3536 ": 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3537 moveToEnd(); 3538 return; 3539 } 3540 Skip = readULEB128(&error); 3541 AdvanceAmount = Skip + PointerSize; 3542 if (error) { 3543 *E = malformedError("for BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB " 3544 " (skip value) " + Twine(error) + " for opcode at" 3545 ": 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3546 moveToEnd(); 3547 return; 3548 } 3549 error = O->BindEntryCheckSegAndOffset(SegmentIndex, SegmentOffset, true); 3550 if (error) { 3551 *E = malformedError("for BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB " 3552 + Twine(error) + " for opcode at: 0x" + 3553 utohexstr(OpcodeStart - Opcodes.begin())); 3554 moveToEnd(); 3555 return; 3556 } 3557 if (SymbolName == StringRef()) { 3558 *E = malformedError("for BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB " 3559 "missing preceding BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM for " 3560 "opcode at: 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3561 moveToEnd(); 3562 return; 3563 } 3564 if (!LibraryOrdinalSet && TableKind != Kind::Weak) { 3565 *E = malformedError("for BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB " 3566 "missing preceding BIND_OPCODE_SET_DYLIB_ORDINAL_* for opcode " 3567 "at: 0x" + utohexstr(OpcodeStart - Opcodes.begin())); 3568 moveToEnd(); 3569 return; 3570 } 3571 error = O->BindEntryCheckCountAndSkip(Count, Skip, PointerSize, 3572 SegmentIndex, SegmentOffset); 3573 if (error) { 3574 *E = malformedError("for BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB " 3575 + Twine(error) + " for opcode at: 0x" + 3576 utohexstr(OpcodeStart - Opcodes.begin())); 3577 moveToEnd(); 3578 return; 3579 } 3580 DEBUG_WITH_TYPE( 3581 "mach-o-bind", 3582 llvm::dbgs() << "BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB: " 3583 << format("SegmentOffset=0x%06X", SegmentOffset) 3584 << ", AdvanceAmount=" << AdvanceAmount 3585 << ", RemainingLoopCount=" << RemainingLoopCount 3586 << "\n"); 3587 return; 3588 default: 3589 *E = malformedError("bad bind info (bad opcode value 0x" + 3590 utohexstr(Opcode) + " for opcode at: 0x" + 3591 utohexstr(OpcodeStart - Opcodes.begin())); 3592 moveToEnd(); 3593 return; 3594 } 3595 } 3596 } 3597 3598 uint64_t MachOBindEntry::readULEB128(const char **error) { 3599 unsigned Count; 3600 uint64_t Result = decodeULEB128(Ptr, &Count, Opcodes.end(), error); 3601 Ptr += Count; 3602 if (Ptr > Opcodes.end()) 3603 Ptr = Opcodes.end(); 3604 return Result; 3605 } 3606 3607 int64_t MachOBindEntry::readSLEB128(const char **error) { 3608 unsigned Count; 3609 int64_t Result = decodeSLEB128(Ptr, &Count, Opcodes.end(), error); 3610 Ptr += Count; 3611 if (Ptr > Opcodes.end()) 3612 Ptr = Opcodes.end(); 3613 return Result; 3614 } 3615 3616 int32_t MachOBindEntry::segmentIndex() const { return SegmentIndex; } 3617 3618 uint64_t MachOBindEntry::segmentOffset() const { return SegmentOffset; } 3619 3620 StringRef MachOBindEntry::typeName() const { 3621 switch (BindType) { 3622 case MachO::BIND_TYPE_POINTER: 3623 return "pointer"; 3624 case MachO::BIND_TYPE_TEXT_ABSOLUTE32: 3625 return "text abs32"; 3626 case MachO::BIND_TYPE_TEXT_PCREL32: 3627 return "text rel32"; 3628 } 3629 return "unknown"; 3630 } 3631 3632 StringRef MachOBindEntry::symbolName() const { return SymbolName; } 3633 3634 int64_t MachOBindEntry::addend() const { return Addend; } 3635 3636 uint32_t MachOBindEntry::flags() const { return Flags; } 3637 3638 int MachOBindEntry::ordinal() const { return Ordinal; } 3639 3640 // For use with the SegIndex of a checked Mach-O Bind entry 3641 // to get the segment name. 3642 StringRef MachOBindEntry::segmentName() const { 3643 return O->BindRebaseSegmentName(SegmentIndex); 3644 } 3645 3646 // For use with a SegIndex,SegOffset pair from a checked Mach-O Bind entry 3647 // to get the section name. 3648 StringRef MachOBindEntry::sectionName() const { 3649 return O->BindRebaseSectionName(SegmentIndex, SegmentOffset); 3650 } 3651 3652 // For use with a SegIndex,SegOffset pair from a checked Mach-O Bind entry 3653 // to get the address. 3654 uint64_t MachOBindEntry::address() const { 3655 return O->BindRebaseAddress(SegmentIndex, SegmentOffset); 3656 } 3657 3658 bool MachOBindEntry::operator==(const MachOBindEntry &Other) const { 3659 #ifdef EXPENSIVE_CHECKS 3660 assert(Opcodes == Other.Opcodes && "compare iterators of different files"); 3661 #else 3662 assert(Opcodes.data() == Other.Opcodes.data() && "compare iterators of different files"); 3663 #endif 3664 return (Ptr == Other.Ptr) && 3665 (RemainingLoopCount == Other.RemainingLoopCount) && 3666 (Done == Other.Done); 3667 } 3668 3669 // Build table of sections so SegIndex/SegOffset pairs can be translated. 3670 BindRebaseSegInfo::BindRebaseSegInfo(const object::MachOObjectFile *Obj) { 3671 uint32_t CurSegIndex = Obj->hasPageZeroSegment() ? 1 : 0; 3672 StringRef CurSegName; 3673 uint64_t CurSegAddress; 3674 for (const SectionRef &Section : Obj->sections()) { 3675 SectionInfo Info; 3676 Section.getName(Info.SectionName); 3677 Info.Address = Section.getAddress(); 3678 Info.Size = Section.getSize(); 3679 Info.SegmentName = 3680 Obj->getSectionFinalSegmentName(Section.getRawDataRefImpl()); 3681 if (!Info.SegmentName.equals(CurSegName)) { 3682 ++CurSegIndex; 3683 CurSegName = Info.SegmentName; 3684 CurSegAddress = Info.Address; 3685 } 3686 Info.SegmentIndex = CurSegIndex - 1; 3687 Info.OffsetInSegment = Info.Address - CurSegAddress; 3688 Info.SegmentStartAddress = CurSegAddress; 3689 Sections.push_back(Info); 3690 } 3691 MaxSegIndex = CurSegIndex; 3692 } 3693 3694 // For use with a SegIndex,SegOffset pair in MachOBindEntry::moveNext() to 3695 // validate a MachOBindEntry or MachORebaseEntry. 3696 const char * BindRebaseSegInfo::checkSegAndOffset(int32_t SegIndex, 3697 uint64_t SegOffset, 3698 bool endInvalid) { 3699 if (SegIndex == -1) 3700 return "missing preceding *_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB"; 3701 if (SegIndex >= MaxSegIndex) 3702 return "bad segIndex (too large)"; 3703 for (const SectionInfo &SI : Sections) { 3704 if (SI.SegmentIndex != SegIndex) 3705 continue; 3706 if (SI.OffsetInSegment > SegOffset) 3707 continue; 3708 if (SegOffset > (SI.OffsetInSegment + SI.Size)) 3709 continue; 3710 if (endInvalid && SegOffset >= (SI.OffsetInSegment + SI.Size)) 3711 continue; 3712 return nullptr; 3713 } 3714 return "bad segOffset, too large"; 3715 } 3716 3717 // For use in MachOBindEntry::moveNext() to validate a MachOBindEntry for 3718 // the BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB opcode and for use in 3719 // MachORebaseEntry::moveNext() to validate a MachORebaseEntry for 3720 // REBASE_OPCODE_DO_*_TIMES* opcodes. The SegIndex and SegOffset must have 3721 // been already checked. 3722 const char * BindRebaseSegInfo::checkCountAndSkip(uint32_t Count, uint32_t Skip, 3723 uint8_t PointerSize, 3724 int32_t SegIndex, 3725 uint64_t SegOffset) { 3726 const SectionInfo &SI = findSection(SegIndex, SegOffset); 3727 uint64_t addr = SI.SegmentStartAddress + SegOffset; 3728 if (addr >= SI.Address + SI.Size) 3729 return "bad segOffset, too large"; 3730 uint64_t i = 0; 3731 if (Count > 1) 3732 i = (Skip + PointerSize) * (Count - 1); 3733 else if (Count == 1) 3734 i = Skip + PointerSize; 3735 if (addr + i >= SI.Address + SI.Size) { 3736 // For rebase opcodes they can step from one section to another. 3737 uint64_t TrailingSegOffset = (addr + i) - SI.SegmentStartAddress; 3738 const char *error = checkSegAndOffset(SegIndex, TrailingSegOffset, false); 3739 if (error) 3740 return "bad count and skip, too large"; 3741 } 3742 return nullptr; 3743 } 3744 3745 // For use with the SegIndex of a checked Mach-O Bind or Rebase entry 3746 // to get the segment name. 3747 StringRef BindRebaseSegInfo::segmentName(int32_t SegIndex) { 3748 for (const SectionInfo &SI : Sections) { 3749 if (SI.SegmentIndex == SegIndex) 3750 return SI.SegmentName; 3751 } 3752 llvm_unreachable("invalid SegIndex"); 3753 } 3754 3755 // For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or Rebase 3756 // to get the SectionInfo. 3757 const BindRebaseSegInfo::SectionInfo &BindRebaseSegInfo::findSection( 3758 int32_t SegIndex, uint64_t SegOffset) { 3759 for (const SectionInfo &SI : Sections) { 3760 if (SI.SegmentIndex != SegIndex) 3761 continue; 3762 if (SI.OffsetInSegment > SegOffset) 3763 continue; 3764 if (SegOffset >= (SI.OffsetInSegment + SI.Size)) 3765 continue; 3766 return SI; 3767 } 3768 llvm_unreachable("SegIndex and SegOffset not in any section"); 3769 } 3770 3771 // For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or Rebase 3772 // entry to get the section name. 3773 StringRef BindRebaseSegInfo::sectionName(int32_t SegIndex, 3774 uint64_t SegOffset) { 3775 return findSection(SegIndex, SegOffset).SectionName; 3776 } 3777 3778 // For use with a SegIndex,SegOffset pair from a checked Mach-O Bind or Rebase 3779 // entry to get the address. 3780 uint64_t BindRebaseSegInfo::address(uint32_t SegIndex, uint64_t OffsetInSeg) { 3781 const SectionInfo &SI = findSection(SegIndex, OffsetInSeg); 3782 return SI.SegmentStartAddress + OffsetInSeg; 3783 } 3784 3785 iterator_range<bind_iterator> 3786 MachOObjectFile::bindTable(Error &Err, MachOObjectFile *O, 3787 ArrayRef<uint8_t> Opcodes, bool is64, 3788 MachOBindEntry::Kind BKind) { 3789 if (O->BindRebaseSectionTable == nullptr) 3790 O->BindRebaseSectionTable = llvm::make_unique<BindRebaseSegInfo>(O); 3791 MachOBindEntry Start(&Err, O, Opcodes, is64, BKind); 3792 Start.moveToFirst(); 3793 3794 MachOBindEntry Finish(&Err, O, Opcodes, is64, BKind); 3795 Finish.moveToEnd(); 3796 3797 return make_range(bind_iterator(Start), bind_iterator(Finish)); 3798 } 3799 3800 iterator_range<bind_iterator> MachOObjectFile::bindTable(Error &Err) { 3801 return bindTable(Err, this, getDyldInfoBindOpcodes(), is64Bit(), 3802 MachOBindEntry::Kind::Regular); 3803 } 3804 3805 iterator_range<bind_iterator> MachOObjectFile::lazyBindTable(Error &Err) { 3806 return bindTable(Err, this, getDyldInfoLazyBindOpcodes(), is64Bit(), 3807 MachOBindEntry::Kind::Lazy); 3808 } 3809 3810 iterator_range<bind_iterator> MachOObjectFile::weakBindTable(Error &Err) { 3811 return bindTable(Err, this, getDyldInfoWeakBindOpcodes(), is64Bit(), 3812 MachOBindEntry::Kind::Weak); 3813 } 3814 3815 MachOObjectFile::load_command_iterator 3816 MachOObjectFile::begin_load_commands() const { 3817 return LoadCommands.begin(); 3818 } 3819 3820 MachOObjectFile::load_command_iterator 3821 MachOObjectFile::end_load_commands() const { 3822 return LoadCommands.end(); 3823 } 3824 3825 iterator_range<MachOObjectFile::load_command_iterator> 3826 MachOObjectFile::load_commands() const { 3827 return make_range(begin_load_commands(), end_load_commands()); 3828 } 3829 3830 StringRef 3831 MachOObjectFile::getSectionFinalSegmentName(DataRefImpl Sec) const { 3832 ArrayRef<char> Raw = getSectionRawFinalSegmentName(Sec); 3833 return parseSegmentOrSectionName(Raw.data()); 3834 } 3835 3836 ArrayRef<char> 3837 MachOObjectFile::getSectionRawName(DataRefImpl Sec) const { 3838 assert(Sec.d.a < Sections.size() && "Should have detected this earlier"); 3839 const section_base *Base = 3840 reinterpret_cast<const section_base *>(Sections[Sec.d.a]); 3841 return makeArrayRef(Base->sectname); 3842 } 3843 3844 ArrayRef<char> 3845 MachOObjectFile::getSectionRawFinalSegmentName(DataRefImpl Sec) const { 3846 assert(Sec.d.a < Sections.size() && "Should have detected this earlier"); 3847 const section_base *Base = 3848 reinterpret_cast<const section_base *>(Sections[Sec.d.a]); 3849 return makeArrayRef(Base->segname); 3850 } 3851 3852 bool 3853 MachOObjectFile::isRelocationScattered(const MachO::any_relocation_info &RE) 3854 const { 3855 if (getCPUType(*this) == MachO::CPU_TYPE_X86_64) 3856 return false; 3857 return getPlainRelocationAddress(RE) & MachO::R_SCATTERED; 3858 } 3859 3860 unsigned MachOObjectFile::getPlainRelocationSymbolNum( 3861 const MachO::any_relocation_info &RE) const { 3862 if (isLittleEndian()) 3863 return RE.r_word1 & 0xffffff; 3864 return RE.r_word1 >> 8; 3865 } 3866 3867 bool MachOObjectFile::getPlainRelocationExternal( 3868 const MachO::any_relocation_info &RE) const { 3869 if (isLittleEndian()) 3870 return (RE.r_word1 >> 27) & 1; 3871 return (RE.r_word1 >> 4) & 1; 3872 } 3873 3874 bool MachOObjectFile::getScatteredRelocationScattered( 3875 const MachO::any_relocation_info &RE) const { 3876 return RE.r_word0 >> 31; 3877 } 3878 3879 uint32_t MachOObjectFile::getScatteredRelocationValue( 3880 const MachO::any_relocation_info &RE) const { 3881 return RE.r_word1; 3882 } 3883 3884 uint32_t MachOObjectFile::getScatteredRelocationType( 3885 const MachO::any_relocation_info &RE) const { 3886 return (RE.r_word0 >> 24) & 0xf; 3887 } 3888 3889 unsigned MachOObjectFile::getAnyRelocationAddress( 3890 const MachO::any_relocation_info &RE) const { 3891 if (isRelocationScattered(RE)) 3892 return getScatteredRelocationAddress(RE); 3893 return getPlainRelocationAddress(RE); 3894 } 3895 3896 unsigned MachOObjectFile::getAnyRelocationPCRel( 3897 const MachO::any_relocation_info &RE) const { 3898 if (isRelocationScattered(RE)) 3899 return getScatteredRelocationPCRel(RE); 3900 return getPlainRelocationPCRel(*this, RE); 3901 } 3902 3903 unsigned MachOObjectFile::getAnyRelocationLength( 3904 const MachO::any_relocation_info &RE) const { 3905 if (isRelocationScattered(RE)) 3906 return getScatteredRelocationLength(RE); 3907 return getPlainRelocationLength(*this, RE); 3908 } 3909 3910 unsigned 3911 MachOObjectFile::getAnyRelocationType( 3912 const MachO::any_relocation_info &RE) const { 3913 if (isRelocationScattered(RE)) 3914 return getScatteredRelocationType(RE); 3915 return getPlainRelocationType(*this, RE); 3916 } 3917 3918 SectionRef 3919 MachOObjectFile::getAnyRelocationSection( 3920 const MachO::any_relocation_info &RE) const { 3921 if (isRelocationScattered(RE) || getPlainRelocationExternal(RE)) 3922 return *section_end(); 3923 unsigned SecNum = getPlainRelocationSymbolNum(RE); 3924 if (SecNum == MachO::R_ABS || SecNum > Sections.size()) 3925 return *section_end(); 3926 DataRefImpl DRI; 3927 DRI.d.a = SecNum - 1; 3928 return SectionRef(DRI, this); 3929 } 3930 3931 MachO::section MachOObjectFile::getSection(DataRefImpl DRI) const { 3932 assert(DRI.d.a < Sections.size() && "Should have detected this earlier"); 3933 return getStruct<MachO::section>(*this, Sections[DRI.d.a]); 3934 } 3935 3936 MachO::section_64 MachOObjectFile::getSection64(DataRefImpl DRI) const { 3937 assert(DRI.d.a < Sections.size() && "Should have detected this earlier"); 3938 return getStruct<MachO::section_64>(*this, Sections[DRI.d.a]); 3939 } 3940 3941 MachO::section MachOObjectFile::getSection(const LoadCommandInfo &L, 3942 unsigned Index) const { 3943 const char *Sec = getSectionPtr(*this, L, Index); 3944 return getStruct<MachO::section>(*this, Sec); 3945 } 3946 3947 MachO::section_64 MachOObjectFile::getSection64(const LoadCommandInfo &L, 3948 unsigned Index) const { 3949 const char *Sec = getSectionPtr(*this, L, Index); 3950 return getStruct<MachO::section_64>(*this, Sec); 3951 } 3952 3953 MachO::nlist 3954 MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI) const { 3955 const char *P = reinterpret_cast<const char *>(DRI.p); 3956 return getStruct<MachO::nlist>(*this, P); 3957 } 3958 3959 MachO::nlist_64 3960 MachOObjectFile::getSymbol64TableEntry(DataRefImpl DRI) const { 3961 const char *P = reinterpret_cast<const char *>(DRI.p); 3962 return getStruct<MachO::nlist_64>(*this, P); 3963 } 3964 3965 MachO::linkedit_data_command 3966 MachOObjectFile::getLinkeditDataLoadCommand(const LoadCommandInfo &L) const { 3967 return getStruct<MachO::linkedit_data_command>(*this, L.Ptr); 3968 } 3969 3970 MachO::segment_command 3971 MachOObjectFile::getSegmentLoadCommand(const LoadCommandInfo &L) const { 3972 return getStruct<MachO::segment_command>(*this, L.Ptr); 3973 } 3974 3975 MachO::segment_command_64 3976 MachOObjectFile::getSegment64LoadCommand(const LoadCommandInfo &L) const { 3977 return getStruct<MachO::segment_command_64>(*this, L.Ptr); 3978 } 3979 3980 MachO::linker_option_command 3981 MachOObjectFile::getLinkerOptionLoadCommand(const LoadCommandInfo &L) const { 3982 return getStruct<MachO::linker_option_command>(*this, L.Ptr); 3983 } 3984 3985 MachO::version_min_command 3986 MachOObjectFile::getVersionMinLoadCommand(const LoadCommandInfo &L) const { 3987 return getStruct<MachO::version_min_command>(*this, L.Ptr); 3988 } 3989 3990 MachO::note_command 3991 MachOObjectFile::getNoteLoadCommand(const LoadCommandInfo &L) const { 3992 return getStruct<MachO::note_command>(*this, L.Ptr); 3993 } 3994 3995 MachO::build_version_command 3996 MachOObjectFile::getBuildVersionLoadCommand(const LoadCommandInfo &L) const { 3997 return getStruct<MachO::build_version_command>(*this, L.Ptr); 3998 } 3999 4000 MachO::build_tool_version 4001 MachOObjectFile::getBuildToolVersion(unsigned index) const { 4002 return getStruct<MachO::build_tool_version>(*this, BuildTools[index]); 4003 } 4004 4005 MachO::dylib_command 4006 MachOObjectFile::getDylibIDLoadCommand(const LoadCommandInfo &L) const { 4007 return getStruct<MachO::dylib_command>(*this, L.Ptr); 4008 } 4009 4010 MachO::dyld_info_command 4011 MachOObjectFile::getDyldInfoLoadCommand(const LoadCommandInfo &L) const { 4012 return getStruct<MachO::dyld_info_command>(*this, L.Ptr); 4013 } 4014 4015 MachO::dylinker_command 4016 MachOObjectFile::getDylinkerCommand(const LoadCommandInfo &L) const { 4017 return getStruct<MachO::dylinker_command>(*this, L.Ptr); 4018 } 4019 4020 MachO::uuid_command 4021 MachOObjectFile::getUuidCommand(const LoadCommandInfo &L) const { 4022 return getStruct<MachO::uuid_command>(*this, L.Ptr); 4023 } 4024 4025 MachO::rpath_command 4026 MachOObjectFile::getRpathCommand(const LoadCommandInfo &L) const { 4027 return getStruct<MachO::rpath_command>(*this, L.Ptr); 4028 } 4029 4030 MachO::source_version_command 4031 MachOObjectFile::getSourceVersionCommand(const LoadCommandInfo &L) const { 4032 return getStruct<MachO::source_version_command>(*this, L.Ptr); 4033 } 4034 4035 MachO::entry_point_command 4036 MachOObjectFile::getEntryPointCommand(const LoadCommandInfo &L) const { 4037 return getStruct<MachO::entry_point_command>(*this, L.Ptr); 4038 } 4039 4040 MachO::encryption_info_command 4041 MachOObjectFile::getEncryptionInfoCommand(const LoadCommandInfo &L) const { 4042 return getStruct<MachO::encryption_info_command>(*this, L.Ptr); 4043 } 4044 4045 MachO::encryption_info_command_64 4046 MachOObjectFile::getEncryptionInfoCommand64(const LoadCommandInfo &L) const { 4047 return getStruct<MachO::encryption_info_command_64>(*this, L.Ptr); 4048 } 4049 4050 MachO::sub_framework_command 4051 MachOObjectFile::getSubFrameworkCommand(const LoadCommandInfo &L) const { 4052 return getStruct<MachO::sub_framework_command>(*this, L.Ptr); 4053 } 4054 4055 MachO::sub_umbrella_command 4056 MachOObjectFile::getSubUmbrellaCommand(const LoadCommandInfo &L) const { 4057 return getStruct<MachO::sub_umbrella_command>(*this, L.Ptr); 4058 } 4059 4060 MachO::sub_library_command 4061 MachOObjectFile::getSubLibraryCommand(const LoadCommandInfo &L) const { 4062 return getStruct<MachO::sub_library_command>(*this, L.Ptr); 4063 } 4064 4065 MachO::sub_client_command 4066 MachOObjectFile::getSubClientCommand(const LoadCommandInfo &L) const { 4067 return getStruct<MachO::sub_client_command>(*this, L.Ptr); 4068 } 4069 4070 MachO::routines_command 4071 MachOObjectFile::getRoutinesCommand(const LoadCommandInfo &L) const { 4072 return getStruct<MachO::routines_command>(*this, L.Ptr); 4073 } 4074 4075 MachO::routines_command_64 4076 MachOObjectFile::getRoutinesCommand64(const LoadCommandInfo &L) const { 4077 return getStruct<MachO::routines_command_64>(*this, L.Ptr); 4078 } 4079 4080 MachO::thread_command 4081 MachOObjectFile::getThreadCommand(const LoadCommandInfo &L) const { 4082 return getStruct<MachO::thread_command>(*this, L.Ptr); 4083 } 4084 4085 MachO::any_relocation_info 4086 MachOObjectFile::getRelocation(DataRefImpl Rel) const { 4087 DataRefImpl Sec; 4088 Sec.d.a = Rel.d.a; 4089 uint32_t Offset; 4090 if (is64Bit()) { 4091 MachO::section_64 Sect = getSection64(Sec); 4092 Offset = Sect.reloff; 4093 } else { 4094 MachO::section Sect = getSection(Sec); 4095 Offset = Sect.reloff; 4096 } 4097 4098 auto P = reinterpret_cast<const MachO::any_relocation_info *>( 4099 getPtr(*this, Offset)) + Rel.d.b; 4100 return getStruct<MachO::any_relocation_info>( 4101 *this, reinterpret_cast<const char *>(P)); 4102 } 4103 4104 MachO::data_in_code_entry 4105 MachOObjectFile::getDice(DataRefImpl Rel) const { 4106 const char *P = reinterpret_cast<const char *>(Rel.p); 4107 return getStruct<MachO::data_in_code_entry>(*this, P); 4108 } 4109 4110 const MachO::mach_header &MachOObjectFile::getHeader() const { 4111 return Header; 4112 } 4113 4114 const MachO::mach_header_64 &MachOObjectFile::getHeader64() const { 4115 assert(is64Bit()); 4116 return Header64; 4117 } 4118 4119 uint32_t MachOObjectFile::getIndirectSymbolTableEntry( 4120 const MachO::dysymtab_command &DLC, 4121 unsigned Index) const { 4122 uint64_t Offset = DLC.indirectsymoff + Index * sizeof(uint32_t); 4123 return getStruct<uint32_t>(*this, getPtr(*this, Offset)); 4124 } 4125 4126 MachO::data_in_code_entry 4127 MachOObjectFile::getDataInCodeTableEntry(uint32_t DataOffset, 4128 unsigned Index) const { 4129 uint64_t Offset = DataOffset + Index * sizeof(MachO::data_in_code_entry); 4130 return getStruct<MachO::data_in_code_entry>(*this, getPtr(*this, Offset)); 4131 } 4132 4133 MachO::symtab_command MachOObjectFile::getSymtabLoadCommand() const { 4134 if (SymtabLoadCmd) 4135 return getStruct<MachO::symtab_command>(*this, SymtabLoadCmd); 4136 4137 // If there is no SymtabLoadCmd return a load command with zero'ed fields. 4138 MachO::symtab_command Cmd; 4139 Cmd.cmd = MachO::LC_SYMTAB; 4140 Cmd.cmdsize = sizeof(MachO::symtab_command); 4141 Cmd.symoff = 0; 4142 Cmd.nsyms = 0; 4143 Cmd.stroff = 0; 4144 Cmd.strsize = 0; 4145 return Cmd; 4146 } 4147 4148 MachO::dysymtab_command MachOObjectFile::getDysymtabLoadCommand() const { 4149 if (DysymtabLoadCmd) 4150 return getStruct<MachO::dysymtab_command>(*this, DysymtabLoadCmd); 4151 4152 // If there is no DysymtabLoadCmd return a load command with zero'ed fields. 4153 MachO::dysymtab_command Cmd; 4154 Cmd.cmd = MachO::LC_DYSYMTAB; 4155 Cmd.cmdsize = sizeof(MachO::dysymtab_command); 4156 Cmd.ilocalsym = 0; 4157 Cmd.nlocalsym = 0; 4158 Cmd.iextdefsym = 0; 4159 Cmd.nextdefsym = 0; 4160 Cmd.iundefsym = 0; 4161 Cmd.nundefsym = 0; 4162 Cmd.tocoff = 0; 4163 Cmd.ntoc = 0; 4164 Cmd.modtaboff = 0; 4165 Cmd.nmodtab = 0; 4166 Cmd.extrefsymoff = 0; 4167 Cmd.nextrefsyms = 0; 4168 Cmd.indirectsymoff = 0; 4169 Cmd.nindirectsyms = 0; 4170 Cmd.extreloff = 0; 4171 Cmd.nextrel = 0; 4172 Cmd.locreloff = 0; 4173 Cmd.nlocrel = 0; 4174 return Cmd; 4175 } 4176 4177 MachO::linkedit_data_command 4178 MachOObjectFile::getDataInCodeLoadCommand() const { 4179 if (DataInCodeLoadCmd) 4180 return getStruct<MachO::linkedit_data_command>(*this, DataInCodeLoadCmd); 4181 4182 // If there is no DataInCodeLoadCmd return a load command with zero'ed fields. 4183 MachO::linkedit_data_command Cmd; 4184 Cmd.cmd = MachO::LC_DATA_IN_CODE; 4185 Cmd.cmdsize = sizeof(MachO::linkedit_data_command); 4186 Cmd.dataoff = 0; 4187 Cmd.datasize = 0; 4188 return Cmd; 4189 } 4190 4191 MachO::linkedit_data_command 4192 MachOObjectFile::getLinkOptHintsLoadCommand() const { 4193 if (LinkOptHintsLoadCmd) 4194 return getStruct<MachO::linkedit_data_command>(*this, LinkOptHintsLoadCmd); 4195 4196 // If there is no LinkOptHintsLoadCmd return a load command with zero'ed 4197 // fields. 4198 MachO::linkedit_data_command Cmd; 4199 Cmd.cmd = MachO::LC_LINKER_OPTIMIZATION_HINT; 4200 Cmd.cmdsize = sizeof(MachO::linkedit_data_command); 4201 Cmd.dataoff = 0; 4202 Cmd.datasize = 0; 4203 return Cmd; 4204 } 4205 4206 ArrayRef<uint8_t> MachOObjectFile::getDyldInfoRebaseOpcodes() const { 4207 if (!DyldInfoLoadCmd) 4208 return None; 4209 4210 MachO::dyld_info_command DyldInfo = 4211 getStruct<MachO::dyld_info_command>(*this, DyldInfoLoadCmd); 4212 const uint8_t *Ptr = 4213 reinterpret_cast<const uint8_t *>(getPtr(*this, DyldInfo.rebase_off)); 4214 return makeArrayRef(Ptr, DyldInfo.rebase_size); 4215 } 4216 4217 ArrayRef<uint8_t> MachOObjectFile::getDyldInfoBindOpcodes() const { 4218 if (!DyldInfoLoadCmd) 4219 return None; 4220 4221 MachO::dyld_info_command DyldInfo = 4222 getStruct<MachO::dyld_info_command>(*this, DyldInfoLoadCmd); 4223 const uint8_t *Ptr = 4224 reinterpret_cast<const uint8_t *>(getPtr(*this, DyldInfo.bind_off)); 4225 return makeArrayRef(Ptr, DyldInfo.bind_size); 4226 } 4227 4228 ArrayRef<uint8_t> MachOObjectFile::getDyldInfoWeakBindOpcodes() const { 4229 if (!DyldInfoLoadCmd) 4230 return None; 4231 4232 MachO::dyld_info_command DyldInfo = 4233 getStruct<MachO::dyld_info_command>(*this, DyldInfoLoadCmd); 4234 const uint8_t *Ptr = 4235 reinterpret_cast<const uint8_t *>(getPtr(*this, DyldInfo.weak_bind_off)); 4236 return makeArrayRef(Ptr, DyldInfo.weak_bind_size); 4237 } 4238 4239 ArrayRef<uint8_t> MachOObjectFile::getDyldInfoLazyBindOpcodes() const { 4240 if (!DyldInfoLoadCmd) 4241 return None; 4242 4243 MachO::dyld_info_command DyldInfo = 4244 getStruct<MachO::dyld_info_command>(*this, DyldInfoLoadCmd); 4245 const uint8_t *Ptr = 4246 reinterpret_cast<const uint8_t *>(getPtr(*this, DyldInfo.lazy_bind_off)); 4247 return makeArrayRef(Ptr, DyldInfo.lazy_bind_size); 4248 } 4249 4250 ArrayRef<uint8_t> MachOObjectFile::getDyldInfoExportsTrie() const { 4251 if (!DyldInfoLoadCmd) 4252 return None; 4253 4254 MachO::dyld_info_command DyldInfo = 4255 getStruct<MachO::dyld_info_command>(*this, DyldInfoLoadCmd); 4256 const uint8_t *Ptr = 4257 reinterpret_cast<const uint8_t *>(getPtr(*this, DyldInfo.export_off)); 4258 return makeArrayRef(Ptr, DyldInfo.export_size); 4259 } 4260 4261 ArrayRef<uint8_t> MachOObjectFile::getUuid() const { 4262 if (!UuidLoadCmd) 4263 return None; 4264 // Returning a pointer is fine as uuid doesn't need endian swapping. 4265 const char *Ptr = UuidLoadCmd + offsetof(MachO::uuid_command, uuid); 4266 return makeArrayRef(reinterpret_cast<const uint8_t *>(Ptr), 16); 4267 } 4268 4269 StringRef MachOObjectFile::getStringTableData() const { 4270 MachO::symtab_command S = getSymtabLoadCommand(); 4271 return getData().substr(S.stroff, S.strsize); 4272 } 4273 4274 bool MachOObjectFile::is64Bit() const { 4275 return getType() == getMachOType(false, true) || 4276 getType() == getMachOType(true, true); 4277 } 4278 4279 void MachOObjectFile::ReadULEB128s(uint64_t Index, 4280 SmallVectorImpl<uint64_t> &Out) const { 4281 DataExtractor extractor(ObjectFile::getData(), true, 0); 4282 4283 uint32_t offset = Index; 4284 uint64_t data = 0; 4285 while (uint64_t delta = extractor.getULEB128(&offset)) { 4286 data += delta; 4287 Out.push_back(data); 4288 } 4289 } 4290 4291 bool MachOObjectFile::isRelocatableObject() const { 4292 return getHeader().filetype == MachO::MH_OBJECT; 4293 } 4294 4295 Expected<std::unique_ptr<MachOObjectFile>> 4296 ObjectFile::createMachOObjectFile(MemoryBufferRef Buffer, 4297 uint32_t UniversalCputype, 4298 uint32_t UniversalIndex) { 4299 StringRef Magic = Buffer.getBuffer().slice(0, 4); 4300 if (Magic == "\xFE\xED\xFA\xCE") 4301 return MachOObjectFile::create(Buffer, false, false, 4302 UniversalCputype, UniversalIndex); 4303 if (Magic == "\xCE\xFA\xED\xFE") 4304 return MachOObjectFile::create(Buffer, true, false, 4305 UniversalCputype, UniversalIndex); 4306 if (Magic == "\xFE\xED\xFA\xCF") 4307 return MachOObjectFile::create(Buffer, false, true, 4308 UniversalCputype, UniversalIndex); 4309 if (Magic == "\xCF\xFA\xED\xFE") 4310 return MachOObjectFile::create(Buffer, true, true, 4311 UniversalCputype, UniversalIndex); 4312 return make_error<GenericBinaryError>("Unrecognized MachO magic number", 4313 object_error::invalid_file_type); 4314 } 4315