launch-cache/dsc_extractor.cpp

   1 /* -*- mode: C++; c-basic-offset: 4; tab-width: 4 -*-
   2  *
   3  * Copyright (c) 2011 Apple Inc. All rights reserved.
   4  *
   5  * @APPLE_LICENSE_HEADER_START@
   6  *
   7  * This file contains Original Code and/or Modifications of Original Code
   8  * as defined in and that are subject to the Apple Public Source License
   9  * Version 2.0 (the 'License'). You may not use this file except in
  10  * compliance with the License. Please obtain a copy of the License at
  11  * http://www.opensource.apple.com/apsl/ and read it before using this
  12  * file.
  13  *
  14  * The Original Code and all software distributed under the License are
  15  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  16  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  17  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  18  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  19  * Please see the License for the specific language governing rights and
  20  * limitations under the License.
  21  *
  22  * @APPLE_LICENSE_HEADER_END@
  23  */
  24
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27 #include <unistd.h>
  28 #include <sys/stat.h>
  29 #include <string.h>
  30 #include <fcntl.h>
  31 #include <stdlib.h>
  32 #include <errno.h>
  33 #include <sys/mman.h>
  34 #include <sys/syslimits.h>
  35 #include <libkern/OSByteOrder.h>
  36 #include <mach-o/fat.h>
  37 #include <mach-o/arch.h>
  38 #include <mach-o/loader.h>
  39 #include <Availability.h>
  40
  41 #define NO_ULEB
  42 #include "Architectures.hpp"
  43 #include "MachOFileAbstraction.hpp"
  44 #include "CacheFileAbstraction.hpp"
  45
  46 #include "dsc_iterator.h"
  47 #include "dsc_extractor.h"
  48 #include "MachOTrie.hpp"
  49
  50 #include <vector>
  51 #include <set>
  52 #include <map>
  53 #include <unordered_map>
  54 #include <algorithm>
  55 #include <dispatch/dispatch.h>
  56
  57 struct seg_info
  58 {
  59                                 seg_info(const char* n, uint64_t o, uint64_t s)
  60                                         : segName(n), offset(o), sizem(s) { }
  61         const char* segName;
  62         uint64_t        offset;
  63         uint64_t        sizem;
  64 };
  65
  66 class CStringHash {
  67 public:
  68         size_t operator()(const char* __s) const {
  69                 size_t __h = 0;
  70                 for ( ; *__s; ++__s)
  71                         __h = 5 * __h + *__s;
  72                 return __h;
  73         };
  74 };
  75 class CStringEquals {
  76 public:
  77         bool operator()(const char* left, const char* right) const { return (strcmp(left, right) == 0); }
  78 };
  79 typedef std::unordered_map<const char*, std::vector<seg_info>, CStringHash, CStringEquals> NameToSegments;
  80
  81 // Filter to find individual symbol re-exports in trie
  82 class NotReExportSymbol {
  83 public:
  84         NotReExportSymbol(const std::set<int> &rd) :_reexportDeps(rd) {}
  85         bool operator()(const mach_o::trie::Entry &entry) const {
  86                 bool result = isSymbolReExport(entry);
  87                 if (result) {
  88                         // <rdar://problem/17671438> Xcode 6 leaks in dyld_shared_cache_extract_dylibs
  89                         ::free((void*)entry.name);
  90                         const_cast<mach_o::trie::Entry*>(&entry)->name = NULL;
  91                 }
  92                 return result;
  93         }
  94 private:
  95         bool isSymbolReExport(const mach_o::trie::Entry &entry) const {
  96                 if ( (entry.flags & EXPORT_SYMBOL_FLAGS_KIND_MASK) != EXPORT_SYMBOL_FLAGS_KIND_REGULAR )
  97                         return true;
  98                 if ( (entry.flags & EXPORT_SYMBOL_FLAGS_REEXPORT) == 0 )
  99                         return true;
 100                 // If the symbol comes from a dylib that is re-exported, this is not an individual symbol re-export
 101                 if ( _reexportDeps.count((int)entry.other) != 0 )
 102                         return true;
 103                 return false;
 104         }
 105         const std::set<int> &_reexportDeps;
 106 };
 107
 108
 109 template <typename A>
 110 int optimize_linkedit(macho_header<typename A::P>* mh, uint64_t textOffsetInCache, const void* mapped_cache, uint64_t* newSize)
 111 {
 112         typedef typename A::P P;
 113         typedef typename A::P::E E;
 114     typedef typename A::P::uint_t pint_t;
 115
 116         // update header flags
 117         mh->set_flags(mh->flags() & 0x7FFFFFFF); // remove in-cache bit
 118
 119         // update load commands
 120         uint64_t cumulativeFileSize = 0;
 121         const unsigned origLoadCommandsSize = mh->sizeofcmds();
 122         unsigned bytesRemaining = origLoadCommandsSize;
 123         unsigned removedCount = 0;
 124         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)mh + sizeof(macho_header<P>));
 125         const uint32_t cmdCount = mh->ncmds();
 126         const macho_load_command<P>* cmd = cmds;
 127         macho_segment_command<P>* linkEditSegCmd = NULL;
 128         macho_symtab_command<P>* symtab = NULL;
 129         macho_dysymtab_command<P>*      dynamicSymTab = NULL;
 130         macho_linkedit_data_command<P>* functionStarts = NULL;
 131         macho_linkedit_data_command<P>* dataInCode = NULL;
 132         uint32_t exportsTrieOffset = 0;
 133         uint32_t exportsTrieSize = 0;
 134         std::set<int> reexportDeps;
 135         int depIndex = 0;
 136         for (uint32_t i = 0; i < cmdCount; ++i) {
 137             bool remove = false;
 138                 switch ( cmd->cmd() ) {
 139                 case macho_segment_command<P>::CMD:
 140                         {
 141                         // update segment/section file offsets
 142                         macho_segment_command<P>* segCmd = (macho_segment_command<P>*)cmd;
 143                         segCmd->set_fileoff(cumulativeFileSize);
 144                         macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
 145                         macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
 146                         for(macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
 147                                 if ( sect->offset() != 0 )
 148                                         sect->set_offset((uint32_t)(cumulativeFileSize+sect->addr()-segCmd->vmaddr()));
 149                         }
 150                         if ( strcmp(segCmd->segname(), "__LINKEDIT") == 0 ) {
 151                                 linkEditSegCmd = segCmd;
 152                         }
 153                         cumulativeFileSize += segCmd->filesize();
 154                         }
 155                         break;
 156                 case LC_DYLD_INFO_ONLY:
 157                         {
 158                         // zero out all dyld info
 159                         macho_dyld_info_command<P>* dyldInfo = (macho_dyld_info_command<P>*)cmd;
 160                         exportsTrieOffset = dyldInfo->export_off();
 161                         exportsTrieSize = dyldInfo->export_size();
 162                         dyldInfo->set_rebase_off(0);
 163                         dyldInfo->set_rebase_size(0);
 164                         dyldInfo->set_bind_off(0);
 165                         dyldInfo->set_bind_size(0);
 166                         dyldInfo->set_weak_bind_off(0);
 167                         dyldInfo->set_weak_bind_size(0);
 168                         dyldInfo->set_lazy_bind_off(0);
 169                         dyldInfo->set_lazy_bind_size(0);
 170                         dyldInfo->set_export_off(0);
 171                         dyldInfo->set_export_size(0);
 172                         }
 173                         break;
 174                 case LC_SYMTAB:
 175                         symtab = (macho_symtab_command<P>*)cmd;
 176                         break;
 177                 case LC_DYSYMTAB:
 178                         dynamicSymTab = (macho_dysymtab_command<P>*)cmd;
 179                         break;
 180                 case LC_FUNCTION_STARTS:
 181                         functionStarts = (macho_linkedit_data_command<P>*)cmd;
 182                         break;
 183                 case LC_DATA_IN_CODE:
 184                         dataInCode = (macho_linkedit_data_command<P>*)cmd;
 185                         break;
 186                 case LC_LOAD_DYLIB:
 187                 case LC_LOAD_WEAK_DYLIB:
 188                 case LC_REEXPORT_DYLIB:
 189                 case LC_LOAD_UPWARD_DYLIB:
 190                         ++depIndex;
 191                         if ( cmd->cmd() == LC_REEXPORT_DYLIB ) {
 192                                 reexportDeps.insert(depIndex);
 193                         }
 194                         break;
 195                 case LC_SEGMENT_SPLIT_INFO:
 196                         // <rdar://problem/23212513> dylibs iOS 9 dyld caches have bogus LC_SEGMENT_SPLIT_INFO
 197                         remove = true;
 198                         break;
 199                 }
 200                 uint32_t cmdSize = cmd->cmdsize();
 201                 macho_load_command<P>* nextCmd = (macho_load_command<P>*)(((uint8_t*)cmd)+cmdSize);
 202                 if ( remove ) {
 203                         ::memmove((void*)cmd, (void*)nextCmd, bytesRemaining);
 204                         ++removedCount;
 205                 }
 206                 else {
 207                         bytesRemaining -= cmdSize;
 208                         cmd = nextCmd;
 209                 }
 210         }
 211         // zero out stuff removed
 212         ::bzero((void*)cmd, bytesRemaining);
 213         // update header
 214         mh->set_ncmds(cmdCount - removedCount);
 215         mh->set_sizeofcmds(origLoadCommandsSize - bytesRemaining);
 216
 217         // rebuild symbol table
 218         if ( linkEditSegCmd == NULL ) {
 219                 fprintf(stderr, "__LINKEDIT not found\n");
 220                 return -1;
 221         }
 222         if ( symtab == NULL ) {
 223                 fprintf(stderr, "LC_SYMTAB not found\n");
 224                 return -1;
 225         }
 226         if ( dynamicSymTab == NULL ) {
 227                 fprintf(stderr, "LC_DYSYMTAB not found\n");
 228                 return -1;
 229         }
 230
 231         const uint64_t newFunctionStartsOffset = linkEditSegCmd->fileoff();
 232         uint32_t functionStartsSize = 0;
 233         if ( functionStarts != NULL ) {
 234                 // copy function starts from original cache file to new mapped dylib file
 235                 functionStartsSize = functionStarts->datasize();
 236                 memcpy((char*)mh + newFunctionStartsOffset, (char*)mapped_cache + functionStarts->dataoff(), functionStartsSize);
 237         }
 238         const uint64_t newDataInCodeOffset = (newFunctionStartsOffset + functionStartsSize + sizeof(pint_t) - 1) & (-sizeof(pint_t)); // pointer align
 239         uint32_t dataInCodeSize = 0;
 240         if ( dataInCode != NULL ) {
 241                 // copy data-in-code info from original cache file to new mapped dylib file
 242                 dataInCodeSize = dataInCode->datasize();
 243                 memcpy((char*)mh + newDataInCodeOffset, (char*)mapped_cache + dataInCode->dataoff(), dataInCodeSize);
 244         }
 245
 246         std::vector<mach_o::trie::Entry> exports;
 247         if ( exportsTrieSize != 0 ) {
 248                 const uint8_t* exportsStart = ((uint8_t*)mapped_cache) + exportsTrieOffset;
 249                 const uint8_t* exportsEnd = &exportsStart[exportsTrieSize];
 250                 mach_o::trie::parseTrie(exportsStart, exportsEnd, exports);
 251                 exports.erase(std::remove_if(exports.begin(), exports.end(), NotReExportSymbol(reexportDeps)), exports.end());
 252         }
 253
 254         // look for local symbol info in unmapped part of shared cache
 255         dyldCacheHeader<E>* header = (dyldCacheHeader<E>*)mapped_cache;
 256         macho_nlist<P>* localNlists = NULL;
 257         uint32_t localNlistCount = 0;
 258         const char* localStrings = NULL;
 259         const char* localStringsEnd = NULL;
 260         if ( header->mappingOffset() > offsetof(dyld_cache_header,localSymbolsSize) ) {
 261                 dyldCacheLocalSymbolsInfo<E>* localInfo = (dyldCacheLocalSymbolsInfo<E>*)(((uint8_t*)mapped_cache) + header->localSymbolsOffset());
 262                 dyldCacheLocalSymbolEntry<E>* entries = (dyldCacheLocalSymbolEntry<E>*)(((uint8_t*)mapped_cache) + header->localSymbolsOffset() + localInfo->entriesOffset());
 263                 macho_nlist<P>* allLocalNlists = (macho_nlist<P>*)(((uint8_t*)localInfo) + localInfo->nlistOffset());
 264                 const uint32_t entriesCount = localInfo->entriesCount();
 265                 for (uint32_t i=0; i < entriesCount; ++i) {
 266                         if ( entries[i].dylibOffset() == textOffsetInCache ) {
 267                                 uint32_t localNlistStart = entries[i].nlistStartIndex();
 268                                 localNlistCount = entries[i].nlistCount();
 269                                 localNlists = &allLocalNlists[localNlistStart];
 270                                 localStrings = ((char*)localInfo) + localInfo->stringsOffset();
 271                                 localStringsEnd = &localStrings[localInfo->stringsSize()];
 272                                 break;
 273                         }
 274                 }
 275         }
 276         // compute number of symbols in new symbol table
 277         const macho_nlist<P>* const mergedSymTabStart = (macho_nlist<P>*)(((uint8_t*)mapped_cache) + symtab->symoff());
 278         const macho_nlist<P>* const mergedSymTabend = &mergedSymTabStart[symtab->nsyms()];
 279         uint32_t newSymCount = symtab->nsyms();
 280         if ( localNlists != NULL ) {
 281                 newSymCount = localNlistCount;
 282                 for (const macho_nlist<P>* s = mergedSymTabStart; s != mergedSymTabend; ++s) {
 283                         // skip any locals in cache
 284                         if ( (s->n_type() & (N_TYPE|N_EXT)) == N_SECT )
 285                                 continue;
 286                         ++newSymCount;
 287                 }
 288         }
 289
 290         // add room for N_INDR symbols for re-exported symbols
 291         newSymCount += exports.size();
 292
 293         // copy symbol entries and strings from original cache file to new mapped dylib file
 294         const uint64_t newSymTabOffset = (newDataInCodeOffset + dataInCodeSize + sizeof(pint_t) - 1) & (-sizeof(pint_t)); // pointer align
 295         const uint64_t newIndSymTabOffset = newSymTabOffset + newSymCount*sizeof(macho_nlist<P>);
 296         const uint64_t newStringPoolOffset = newIndSymTabOffset + dynamicSymTab->nindirectsyms()*sizeof(uint32_t);
 297         macho_nlist<P>* const newSymTabStart = (macho_nlist<P>*)(((uint8_t*)mh) + newSymTabOffset);
 298         char* const newStringPoolStart = (char*)mh + newStringPoolOffset;
 299         const uint32_t* mergedIndSymTab = (uint32_t*)((char*)mapped_cache + dynamicSymTab->indirectsymoff());
 300         const char* mergedStringPoolStart = (char*)mapped_cache + symtab->stroff();
 301         const char* mergedStringPoolEnd = &mergedStringPoolStart[symtab->strsize()];
 302         macho_nlist<P>* t = newSymTabStart;
 303         int poolOffset = 0;
 304         uint32_t symbolsCopied = 0;
 305         newStringPoolStart[poolOffset++] = '\0'; // first pool entry is always empty string
 306         for (const macho_nlist<P>* s = mergedSymTabStart; s != mergedSymTabend; ++s) {
 307                 // if we have better local symbol info, skip any locals here
 308                 if ( (localNlists != NULL) && ((s->n_type() & (N_TYPE|N_EXT)) == N_SECT) )
 309                         continue;
 310                 *t = *s;
 311                 t->set_n_strx(poolOffset);
 312                 const char* symName = &mergedStringPoolStart[s->n_strx()];
 313                 if ( symName > mergedStringPoolEnd )
 314                         symName = "<corrupt symbol name>";
 315                 strcpy(&newStringPoolStart[poolOffset], symName);
 316                 poolOffset += (strlen(symName) + 1);
 317                 ++t;
 318                 ++symbolsCopied;
 319         }
 320         // <rdar://problem/16529213> recreate N_INDR symbols in extracted dylibs for debugger
 321         for (std::vector<mach_o::trie::Entry>::iterator it = exports.begin(); it != exports.end(); ++it) {
 322                 strcpy(&newStringPoolStart[poolOffset], it->name);
 323                 t->set_n_strx(poolOffset);
 324                 poolOffset += (strlen(it->name) + 1);
 325                 t->set_n_type(N_INDR | N_EXT);
 326                 t->set_n_sect(0);
 327                 t->set_n_desc(0);
 328                 const char* importName = it->importName;
 329                 if ( *importName == '\0' )
 330                         importName = it->name;
 331                 strcpy(&newStringPoolStart[poolOffset], importName);
 332                 t->set_n_value(poolOffset);
 333                 poolOffset += (strlen(importName) + 1);
 334                 ++t;
 335                 ++symbolsCopied;
 336         }
 337         if ( localNlists != NULL ) {
 338                 // update load command to reflect new count of locals
 339                 dynamicSymTab->set_ilocalsym(symbolsCopied);
 340                 dynamicSymTab->set_nlocalsym(localNlistCount);
 341                 // copy local symbols
 342                 for (uint32_t i=0; i < localNlistCount; ++i) {
 343                         const char* localName = &localStrings[localNlists[i].n_strx()];
 344                         if ( localName > localStringsEnd )
 345                                 localName = "<corrupt local symbol name>";
 346                         *t = localNlists[i];
 347                         t->set_n_strx(poolOffset);
 348                         strcpy(&newStringPoolStart[poolOffset], localName);
 349                         poolOffset += (strlen(localName) + 1);
 350                         ++t;
 351                         ++symbolsCopied;
 352                 }
 353         }
 354
 355         if ( newSymCount != symbolsCopied ) {
 356                 fprintf(stderr, "symbol count miscalculation\n");
 357                 return -1;
 358         }
 359
 360         // pointer align string pool size
 361         while ( (poolOffset % sizeof(pint_t)) != 0 )
 362                 ++poolOffset;
 363         // copy indirect symbol table
 364         uint32_t* newIndSymTab = (uint32_t*)((char*)mh + newIndSymTabOffset);
 365         memcpy(newIndSymTab, mergedIndSymTab, dynamicSymTab->nindirectsyms()*sizeof(uint32_t));
 366
 367         // update load commands
 368         if ( functionStarts != NULL ) {
 369                 functionStarts->set_dataoff((uint32_t)newFunctionStartsOffset);
 370                 functionStarts->set_datasize(functionStartsSize);
 371         }
 372         if ( dataInCode != NULL ) {
 373                 dataInCode->set_dataoff((uint32_t)newDataInCodeOffset);
 374                 dataInCode->set_datasize(dataInCodeSize);
 375         }
 376         symtab->set_nsyms(symbolsCopied);
 377         symtab->set_symoff((uint32_t)newSymTabOffset);
 378         symtab->set_stroff((uint32_t)newStringPoolOffset);
 379         symtab->set_strsize(poolOffset);
 380         dynamicSymTab->set_extreloff(0);
 381         dynamicSymTab->set_nextrel(0);
 382         dynamicSymTab->set_locreloff(0);
 383         dynamicSymTab->set_nlocrel(0);
 384         dynamicSymTab->set_indirectsymoff((uint32_t)newIndSymTabOffset);
 385         linkEditSegCmd->set_filesize(symtab->stroff()+symtab->strsize() - linkEditSegCmd->fileoff());
 386         linkEditSegCmd->set_vmsize( (linkEditSegCmd->filesize()+4095) & (-4096) );
 387
 388         // return new size
 389         *newSize = (symtab->stroff()+symtab->strsize()+4095) & (-4096);
 390
 391         // <rdar://problem/17671438> Xcode 6 leaks in dyld_shared_cache_extract_dylibs
 392         for (std::vector<mach_o::trie::Entry>::iterator it = exports.begin(); it != exports.end(); ++it) {
 393                 ::free((void*)(it->name));
 394         }
 395
 396
 397         return 0;
 398 }
 399
 400
 401
 402 static void make_dirs(const char* file_path)
 403 {
 404         //printf("make_dirs(%s)\n", file_path);
 405         char dirs[strlen(file_path)+1];
 406         strcpy(dirs, file_path);
 407         char* lastSlash = strrchr(dirs, '/');
 408         if ( lastSlash == NULL )
 409                 return;
 410         lastSlash[1] = '\0';
 411         struct stat stat_buf;
 412         if ( stat(dirs, &stat_buf) != 0 ) {
 413                 char* afterSlash = &dirs[1];
 414                 char* slash;
 415                 while ( (slash = strchr(afterSlash, '/')) != NULL ) {
 416                         *slash = '\0';
 417                         ::mkdir(dirs, S_IRWXU | S_IRGRP|S_IXGRP | S_IROTH|S_IXOTH);
 418                         //printf("mkdir(%s)\n", dirs);
 419                         *slash = '/';
 420                         afterSlash = slash+1;
 421                 }
 422         }
 423 }
 424
 425
 426
 427 template <typename A>
 428 size_t dylib_maker(const void* mapped_cache, std::vector<uint8_t> &dylib_data, const std::vector<seg_info>& segments) {
 429         typedef typename A::P P;
 430
 431     size_t  additionalSize  = 0;
 432         for(std::vector<seg_info>::const_iterator it=segments.begin(); it != segments.end(); ++it) {
 433                 additionalSize                          += it->sizem;
 434         }
 435
 436     dylib_data.reserve(dylib_data.size() + additionalSize);
 437
 438     uint32_t                nfat_archs          = 0;
 439         uint32_t                offsetInFatFile     = 4096;
 440     uint8_t                 *base_ptr           = &dylib_data.front();
 441
 442 #define FH reinterpret_cast<fat_header*>(base_ptr)
 443 #define FA reinterpret_cast<fat_arch*>(base_ptr + (8 + (nfat_archs - 1) * sizeof(fat_arch)))
 444
 445     if(dylib_data.size() >= 4096 && OSSwapBigToHostInt32(FH->magic) == FAT_MAGIC) {
 446                 // have fat header, append new arch to end
 447         nfat_archs                              = OSSwapBigToHostInt32(FH->nfat_arch);
 448                 offsetInFatFile                         = OSSwapBigToHostInt32(FA->offset) + OSSwapBigToHostInt32(FA->size);
 449     }
 450
 451     dylib_data.resize(offsetInFatFile);
 452     base_ptr                                    = &dylib_data.front();
 453
 454     FH->magic                                   = OSSwapHostToBigInt32(FAT_MAGIC);
 455     FH->nfat_arch                               = OSSwapHostToBigInt32(++nfat_archs);
 456
 457     FA->cputype                                 = 0; // filled in later
 458     FA->cpusubtype                              = 0; // filled in later
 459     FA->offset                                  = OSSwapHostToBigInt32(offsetInFatFile);
 460     FA->size                                    = 0; // filled in later
 461     FA->align                                   = OSSwapHostToBigInt32(12);
 462
 463         // Write regular segments into the buffer
 464         uint64_t                totalSize           = 0;
 465     uint64_t                            textOffsetInCache       = 0;
 466         for( std::vector<seg_info>::const_iterator it=segments.begin(); it != segments.end(); ++it) {
 467
 468         if(strcmp(it->segName, "__TEXT") == 0 ) {
 469                         textOffsetInCache                                       = it->offset;
 470             const macho_header<P>   *textMH     = reinterpret_cast<macho_header<P>*>((uint8_t*)mapped_cache+textOffsetInCache);
 471             FA->cputype                         = OSSwapHostToBigInt32(textMH->cputype());
 472             FA->cpusubtype                      = OSSwapHostToBigInt32(textMH->cpusubtype());
 473
 474             // if this cputype/subtype already exist in fat header, then return immediately
 475             for(uint32_t i=0; i < nfat_archs-1; ++i) {
 476                 fat_arch            *afa        = reinterpret_cast<fat_arch*>(base_ptr+8)+i;
 477
 478                 if(   afa->cputype == FA->cputype
 479                    && afa->cpusubtype == FA->cpusubtype) {
 480                     //fprintf(stderr, "arch already exists in fat dylib\n");
 481                     dylib_data.resize(offsetInFatFile);
 482                     return offsetInFatFile;
 483                 }
 484             }
 485                 }
 486
 487                 //printf("segName=%s, offset=0x%llX, size=0x%0llX\n", it->segName, it->offset, it->sizem);
 488         std::copy(((uint8_t*)mapped_cache)+it->offset, ((uint8_t*)mapped_cache)+it->offset+it->sizem, std::back_inserter(dylib_data));
 489         base_ptr                                = &dylib_data.front();
 490         totalSize                               += it->sizem;
 491         }
 492
 493         FA->size                                    = OSSwapHostToBigInt32(totalSize);
 494
 495         // optimize linkedit
 496         uint64_t                newSize             = dylib_data.size();
 497         optimize_linkedit<A>(((macho_header<P>*)(base_ptr+offsetInFatFile)), textOffsetInCache, mapped_cache, &newSize);
 498
 499         // update fat header with new file size
 500     dylib_data.resize((size_t)(offsetInFatFile+newSize));
 501     base_ptr                                    = &dylib_data.front();
 502         FA->size                                    = OSSwapHostToBigInt32(newSize);
 503 #undef FH
 504 #undef FA
 505         return offsetInFatFile;
 506 }
 507
 508
 509 int dyld_shared_cache_extract_dylibs_progress(const char* shared_cache_file_path, const char* extraction_root_path,
 510                                                                                                         void (^progress)(unsigned current, unsigned total))
 511 {
 512         struct stat statbuf;
 513         if (stat(shared_cache_file_path, &statbuf)) {
 514                 fprintf(stderr, "Error: stat failed for dyld shared cache at %s\n", shared_cache_file_path);
 515                 return -1;
 516         }
 517
 518         int cache_fd = open(shared_cache_file_path, O_RDONLY);
 519         if (cache_fd < 0) {
 520                 fprintf(stderr, "Error: failed to open shared cache file at %s\n", shared_cache_file_path);
 521                 return -1;
 522         }
 523
 524         void* mapped_cache = mmap(NULL, (size_t)statbuf.st_size, PROT_READ, MAP_PRIVATE, cache_fd, 0);
 525         if (mapped_cache == MAP_FAILED) {
 526                 fprintf(stderr, "Error: mmap() for shared cache at %s failed, errno=%d\n", shared_cache_file_path, errno);
 527                 return -1;
 528         }
 529
 530     close(cache_fd);
 531
 532         // instantiate arch specific dylib maker
 533     size_t (*dylib_create_func)(const void*, std::vector<uint8_t>&, const std::vector<seg_info>&) = NULL;
 534              if ( strcmp((char*)mapped_cache, "dyld_v1    i386") == 0 )
 535                 dylib_create_func = dylib_maker<x86>;
 536         else if ( strcmp((char*)mapped_cache, "dyld_v1  x86_64") == 0 )
 537                 dylib_create_func = dylib_maker<x86_64>;
 538         else if ( strcmp((char*)mapped_cache, "dyld_v1 x86_64h") == 0 )
 539                 dylib_create_func = dylib_maker<x86_64>;
 540         else if ( strcmp((char*)mapped_cache, "dyld_v1   armv5") == 0 )
 541                 dylib_create_func = dylib_maker<arm>;
 542         else if ( strcmp((char*)mapped_cache, "dyld_v1   armv6") == 0 )
 543                 dylib_create_func = dylib_maker<arm>;
 544         else if ( strcmp((char*)mapped_cache, "dyld_v1   armv7") == 0 )
 545                 dylib_create_func = dylib_maker<arm>;
 546         else if ( strncmp((char*)mapped_cache, "dyld_v1  armv7", 14) == 0 )
 547                 dylib_create_func = dylib_maker<arm>;
 548         else if ( strcmp((char*)mapped_cache, "dyld_v1   arm64") == 0 )
 549                 dylib_create_func = dylib_maker<arm64>;
 550         else if ( strcmp((char*)mapped_cache, "dyld_v1  arm64e") == 0 )
 551                 dylib_create_func = dylib_maker<arm64>;
 552         else {
 553                 fprintf(stderr, "Error: unrecognized dyld shared cache magic.\n");
 554         munmap(mapped_cache, (size_t)statbuf.st_size);
 555                 return -1;
 556         }
 557
 558         // iterate through all images in cache and build map of dylibs and segments
 559         __block NameToSegments  map;
 560         __block int                             result              = dyld_shared_cache_iterate(mapped_cache, (uint32_t)statbuf.st_size, ^(const dyld_shared_cache_dylib_info* dylibInfo, const dyld_shared_cache_segment_info* segInfo) {
 561         map[dylibInfo->path].push_back(seg_info(segInfo->name, segInfo->fileOffset, segInfo->fileSize));
 562     });
 563
 564     if(result != 0) {
 565                 fprintf(stderr, "Error: dyld_shared_cache_iterate_segments_with_slide failed.\n");
 566         munmap(mapped_cache, (size_t)statbuf.st_size);
 567                 return result;
 568     }
 569
 570         // for each dylib instantiate a dylib file
 571     dispatch_group_t        group               = dispatch_group_create();
 572     dispatch_semaphore_t    sema                = dispatch_semaphore_create(2);
 573     dispatch_queue_t        process_queue       = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_LOW, 0);
 574     dispatch_queue_t        writer_queue        = dispatch_queue_create("dyld writer queue", 0);
 575
 576         __block unsigned        count               = 0;
 577
 578         for ( NameToSegments::iterator it = map.begin(); it != map.end(); ++it) {
 579                 dispatch_semaphore_wait(sema, DISPATCH_TIME_FOREVER);
 580         dispatch_group_async(group, process_queue, ^{
 581
 582             char    dylib_path[PATH_MAX];
 583             strcpy(dylib_path, extraction_root_path);
 584             strcat(dylib_path, "/");
 585             strcat(dylib_path, it->first);
 586
 587             //printf("%s with %lu segments\n", dylib_path, it->second.size());
 588             // make sure all directories in this path exist
 589             make_dirs(dylib_path);
 590
 591             // open file, create if does not already exist
 592             int fd = ::open(dylib_path, O_CREAT | O_EXLOCK | O_RDWR, 0644);
 593             if ( fd == -1 ) {
 594                 fprintf(stderr, "can't open or create dylib file %s, errnor=%d\n", dylib_path, errno);
 595                 result    = -1;
 596                 return;
 597             }
 598
 599             struct stat statbuf;
 600             if (fstat(fd, &statbuf)) {
 601                 fprintf(stderr, "Error: stat failed for dyld file %s, errnor=%d\n", dylib_path, errno);
 602                 close(fd);
 603                 result    = -1;
 604                 return;
 605             }
 606
 607             std::vector<uint8_t> *vec   = new std::vector<uint8_t>((size_t)statbuf.st_size);
 608             if(pread(fd, &vec->front(), vec->size(), 0) != (long)vec->size()) {
 609                 fprintf(stderr, "can't read dylib file %s, errnor=%d\n", dylib_path, errno);
 610                 close(fd);
 611                 result    = -1;
 612                 return;
 613             }
 614
 615             const size_t    offset  = dylib_create_func(mapped_cache, *vec, it->second);
 616
 617             dispatch_group_async(group, writer_queue, ^{
 618                 progress(count++, (unsigned)map.size());
 619
 620                 if(offset != vec->size()) {
 621                     //Write out the first page, and everything after offset
 622                     if(   pwrite(fd, &vec->front(), 4096, 0) == -1
 623                        || pwrite(fd, &vec->front() + offset, vec->size() - offset, offset) == -1) {
 624                         fprintf(stderr, "error writing, errnor=%d\n", errno);
 625                         result    = -1;
 626                     }
 627                 }
 628
 629                 delete vec;
 630                 close(fd);
 631                 dispatch_semaphore_signal(sema);
 632             });
 633         });
 634         }
 635
 636     dispatch_group_wait(group, DISPATCH_TIME_FOREVER);
 637     dispatch_release(group);
 638     dispatch_release(writer_queue);
 639
 640     munmap(mapped_cache, (size_t)statbuf.st_size);
 641         return result;
 642 }
 643
 644
 645
 646 int dyld_shared_cache_extract_dylibs(const char* shared_cache_file_path, const char* extraction_root_path)
 647 {
 648         return dyld_shared_cache_extract_dylibs_progress(shared_cache_file_path, extraction_root_path,
 649                                                                                                         ^(unsigned , unsigned) {} );
 650 }
 651
 652
 653 #if 0
 654 // test program
 655 #include <stdio.h>
 656 #include <stddef.h>
 657 #include <dlfcn.h>
 658
 659
 660 typedef int (*extractor_proc)(const char* shared_cache_file_path, const char* extraction_root_path,
 661                                                                                                         void (^progress)(unsigned current, unsigned total));
 662
 663 int main(int argc, const char* argv[])
 664 {
 665         if ( argc != 3 ) {
 666                 fprintf(stderr, "usage: dsc_extractor <path-to-cache-file> <path-to-device-dir>\n");
 667                 return 1;
 668         }
 669
 670         //void* handle = dlopen("/Volumes/my/src/dyld/build/Debug/dsc_extractor.bundle", RTLD_LAZY);
 671         void* handle = dlopen("/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/usr/lib/dsc_extractor.bundle", RTLD_LAZY);
 672         if ( handle == NULL ) {
 673                 fprintf(stderr, "dsc_extractor.bundle could not be loaded\n");
 674                 return 1;
 675         }
 676
 677         extractor_proc proc = (extractor_proc)dlsym(handle, "dyld_shared_cache_extract_dylibs_progress");
 678         if ( proc == NULL ) {
 679                 fprintf(stderr, "dsc_extractor.bundle did not have dyld_shared_cache_extract_dylibs_progress symbol\n");
 680                 return 1;
 681         }
 682
 683         int result = (*proc)(argv[1], argv[2], ^(unsigned c, unsigned total) { printf("%d/%d\n", c, total); } );
 684         fprintf(stderr, "dyld_shared_cache_extract_dylibs_progress() => %d\n", result);
 685         return 0;
 686 }
 687
 688
 689 #endif
 690
 691
 692
 693