dyld3/shared-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/arch.h>
  37 #include <mach-o/loader.h>
  38 #include <Availability.h>
  39
  40 #include "CodeSigningTypes.h"
  41 #include <CommonCrypto/CommonHMAC.h>
  42 #include <CommonCrypto/CommonDigest.h>
  43 #include <CommonCrypto/CommonDigestSPI.h>
  44
  45 #define NO_ULEB
  46 #include "Architectures.hpp"
  47 #include "MachOFileAbstraction.hpp"
  48
  49 #include "dsc_iterator.h"
  50 #include "dsc_extractor.h"
  51 #include "DyldSharedCache.h"
  52 #include "MachOAnalyzer.h"
  53 #include "SupportedArchs.h"
  54 #include "Trie.hpp"
  55
  56 #include <vector>
  57 #include <set>
  58 #include <map>
  59 #include <unordered_map>
  60 #include <algorithm>
  61 #include <dispatch/dispatch.h>
  62
  63 struct seg_info
  64 {
  65     seg_info(const char* n, uint64_t o, uint64_t s)
  66     : segName(n), offset(o), sizem(s) { }
  67     const char* segName;
  68     uint64_t    offset;
  69     uint64_t    sizem;
  70 };
  71
  72 class CStringHash {
  73 public:
  74     size_t operator()(const char* __s) const {
  75         size_t __h = 0;
  76         for ( ; *__s; ++__s)
  77             __h = 5 * __h + *__s;
  78         return __h;
  79     };
  80 };
  81 class CStringEquals {
  82 public:
  83     bool operator()(const char* left, const char* right) const { return (strcmp(left, right) == 0); }
  84 };
  85 typedef std::unordered_map<const char*, std::vector<seg_info>, CStringHash, CStringEquals> NameToSegments;
  86
  87 // Filter to find individual symbol re-exports in trie
  88 class NotReExportSymbol {
  89 public:
  90     NotReExportSymbol(const std::set<int> &rd) :_reexportDeps(rd) {}
  91     bool operator()(const ExportInfoTrie::Entry &entry) const {
  92         return isSymbolReExport(entry);
  93     }
  94 private:
  95     bool isSymbolReExport(const ExportInfoTrie::Entry &entry) const {
  96         if ( (entry.info.flags & EXPORT_SYMBOL_FLAGS_KIND_MASK) != EXPORT_SYMBOL_FLAGS_KIND_REGULAR )
  97             return true;
  98         if ( (entry.info.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.info.other) != 0 )
 102             return true;
 103         return false;
 104     }
 105     const std::set<int> &_reexportDeps;
 106 };
 107
 108 template <typename P>
 109 struct LoadCommandInfo {
 110 };
 111
 112 template <typename A>
 113 class LinkeditOptimizer {
 114     typedef typename A::P P;
 115     typedef typename A::P::E E;
 116     typedef typename A::P::uint_t pint_t;
 117
 118 private:
 119     macho_segment_command<P>* linkEditSegCmd = nullptr;
 120     symtab_command* symtab = nullptr;
 121     dysymtab_command*    dynamicSymTab = nullptr;
 122     linkedit_data_command*    functionStarts = nullptr;
 123     linkedit_data_command*    dataInCode = nullptr;
 124     uint32_t exportsTrieOffset = 0;
 125     uint32_t exportsTrieSize = 0;
 126     std::set<int> reexportDeps;
 127
 128 public:
 129
 130     void optimize_loadcommands(dyld3::MachOAnalyzer* mh)
 131     {
 132         // update header flags
 133         mh->flags &= 0x7FFFFFFF; // remove in-cache bit
 134
 135         // update load commands
 136         __block uint64_t cumulativeFileSize = 0;
 137         __block int depIndex = 0;
 138         Diagnostics diag;
 139         mh->forEachLoadCommand(diag, ^(const load_command* cmd, bool &stop) {
 140             switch ( cmd->cmd ) {
 141                 case macho_segment_command<P>::CMD: {
 142                     auto segCmd = (macho_segment_command<P>*)cmd;
 143                     segCmd->set_fileoff(cumulativeFileSize);
 144                     segCmd->set_filesize(segCmd->vmsize());
 145
 146                     auto const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
 147                     auto const sectionsEnd = &sectionsStart[segCmd->nsects()];
 148                     for (auto sect = sectionsStart; sect < sectionsEnd; ++sect) {
 149                         if ( sect->offset() != 0 ) {
 150                             sect->set_offset((uint32_t)(cumulativeFileSize + sect->addr() - segCmd->vmaddr()));
 151                         }
 152                     }
 153                     if ( strcmp(segCmd->segname(), "__LINKEDIT") == 0 )
 154                         linkEditSegCmd = segCmd;
 155                     cumulativeFileSize += segCmd->filesize();
 156                 } break;
 157                 case LC_DYLD_INFO_ONLY: {
 158                     // zero out all dyld info. lldb only uses symbol table
 159                     auto dyldInfo = (dyld_info_command*)cmd;
 160                     exportsTrieOffset =  dyldInfo->export_off;
 161                     exportsTrieSize = dyldInfo->export_size;
 162                     dyldInfo->rebase_off = 0;
 163                     dyldInfo->rebase_size = 0;
 164                     dyldInfo->bind_off = 0;
 165                     dyldInfo->bind_size = 0;
 166                     dyldInfo->weak_bind_off = 0;
 167                     dyldInfo->weak_bind_size = 0;
 168                     dyldInfo->lazy_bind_off = 0;
 169                     dyldInfo->lazy_bind_size = 0;
 170                     dyldInfo->export_off = 0;
 171                     dyldInfo->export_size = 0;
 172                 } break;
 173                 case LC_DYLD_EXPORTS_TRIE: {
 174                     // don't put export trie into extracted dylib.  lldb only uses symbol table
 175                     linkedit_data_command* exportsTrie = (linkedit_data_command*)cmd;
 176                     exportsTrieOffset =  exportsTrie->dataoff;
 177                     exportsTrieSize   = exportsTrie->datasize;
 178                     exportsTrie->dataoff = 0;
 179                     exportsTrie->datasize = 0;
 180                 } break;
 181                 case LC_SYMTAB:
 182                     symtab = (symtab_command*)cmd;
 183                     break;
 184                 case LC_DYSYMTAB:
 185                     dynamicSymTab = (dysymtab_command*)cmd;
 186                     break;
 187                 case LC_FUNCTION_STARTS:
 188                     functionStarts = (linkedit_data_command*)cmd;
 189                     break;
 190                 case LC_DATA_IN_CODE:
 191                     dataInCode = (linkedit_data_command*)cmd;
 192                     break;
 193                 case LC_LOAD_DYLIB:
 194                 case LC_LOAD_WEAK_DYLIB:
 195                 case LC_REEXPORT_DYLIB:
 196                 case LC_LOAD_UPWARD_DYLIB:
 197                     depIndex++;
 198                     if ( cmd->cmd == LC_REEXPORT_DYLIB ) {
 199                         reexportDeps.insert(depIndex);
 200                     }
 201                     break;
 202                 default:
 203                     break;
 204             }
 205         });
 206
 207         mh->removeLoadCommand(diag, ^(const load_command* cmd, bool& remove, bool &stop) {
 208             switch ( cmd->cmd ) {
 209                 case LC_SEGMENT_SPLIT_INFO:
 210                     // <rdar://problem/23212513> dylibs iOS 9 dyld caches have bogus LC_SEGMENT_SPLIT_INFO
 211                     remove = true;
 212                     stop = true;
 213                     break;
 214                 default:
 215                     break;
 216             }
 217         });
 218 }
 219
 220     int optimize_linkedit(std::vector<uint8_t> &new_linkedit_data, uint64_t textOffsetInCache, const void* mapped_cache)
 221     {
 222         // rebuild symbol table
 223         if ( linkEditSegCmd == nullptr ) {
 224             fprintf(stderr, "__LINKEDIT not found\n");
 225             return -1;
 226         }
 227         if ( symtab == nullptr ) {
 228             fprintf(stderr, "LC_SYMTAB not found\n");
 229             return -1;
 230         }
 231         if ( dynamicSymTab == nullptr ) {
 232             fprintf(stderr, "LC_DYSYMTAB not found\n");
 233             return -1;
 234         }
 235
 236         const uint64_t newFunctionStartsOffset = new_linkedit_data.size();
 237         uint32_t functionStartsSize = 0;
 238         if ( functionStarts != NULL ) {
 239             // copy function starts from original cache file to new mapped dylib file
 240             functionStartsSize = functionStarts->datasize;
 241             new_linkedit_data.insert(new_linkedit_data.end(),
 242                                      (char*)mapped_cache + functionStarts->dataoff,
 243                                      (char*)mapped_cache + functionStarts->dataoff + functionStartsSize);
 244         }
 245
 246         // pointer align
 247         while ((linkEditSegCmd->fileoff() + new_linkedit_data.size()) % sizeof(pint_t))
 248             new_linkedit_data.push_back(0);
 249
 250         const uint64_t newDataInCodeOffset = new_linkedit_data.size();
 251         uint32_t dataInCodeSize = 0;
 252         if ( dataInCode != NULL ) {
 253             // copy data-in-code info from original cache file to new mapped dylib file
 254             dataInCodeSize = dataInCode->datasize;
 255             new_linkedit_data.insert(new_linkedit_data.end(),
 256                                      (char*)mapped_cache + dataInCode->dataoff,
 257                                      (char*)mapped_cache + dataInCode->dataoff + dataInCodeSize);
 258         }
 259
 260         std::vector<ExportInfoTrie::Entry> exports;
 261         if ( exportsTrieSize != 0 ) {
 262             const uint8_t* exportsStart = ((uint8_t*)mapped_cache) + exportsTrieOffset;
 263             const uint8_t* exportsEnd = &exportsStart[exportsTrieSize];
 264             ExportInfoTrie::parseTrie(exportsStart, exportsEnd, exports);
 265             exports.erase(std::remove_if(exports.begin(), exports.end(), NotReExportSymbol(reexportDeps)), exports.end());
 266         }
 267
 268         const DyldSharedCache*  cache = (DyldSharedCache*)mapped_cache;
 269         macho_nlist<P>*         allLocalNlists = (macho_nlist<P>*)cache->getLocalNlistEntries();
 270         __block macho_nlist<P>* localNlists = nullptr;
 271         __block uint32_t        localNlistCount = 0;
 272         cache->forEachLocalSymbolEntry(^(uint32_t dylibOffset, uint32_t nlistStartIndex, uint32_t nlistCount, bool& stop){
 273             if (dylibOffset == textOffsetInCache) {
 274                 localNlists     = &allLocalNlists[nlistStartIndex];
 275                 localNlistCount = nlistCount;
 276                 stop            = true;
 277             }
 278         });
 279         // compute number of symbols in new symbol table
 280         const macho_nlist<P>* const mergedSymTabStart = (macho_nlist<P>*)(((uint8_t*)mapped_cache) + symtab->symoff);
 281         const macho_nlist<P>* const mergedSymTabend = &mergedSymTabStart[symtab->nsyms];
 282         uint32_t newSymCount = symtab->nsyms;
 283         if ( localNlistCount != 0 ) {
 284             newSymCount = localNlistCount;
 285             for (const macho_nlist<P>* s = mergedSymTabStart; s != mergedSymTabend; ++s) {
 286                 // skip any locals in cache
 287                 if ( (s->n_type() & (N_TYPE|N_EXT)) == N_SECT )
 288                     continue;
 289                 ++newSymCount;
 290             }
 291         }
 292
 293         // add room for N_INDR symbols for re-exported symbols
 294         newSymCount += exports.size();
 295
 296         // copy symbol entries and strings from original cache file to new mapped dylib file
 297         const char* mergedStringPoolStart = (char*)mapped_cache + symtab->stroff;
 298         const char* mergedStringPoolEnd = &mergedStringPoolStart[symtab->strsize];
 299
 300         // First count how many entries we need
 301         std::vector<macho_nlist<P>> newSymTab;
 302         newSymTab.reserve(newSymCount);
 303         std::vector<char> newSymNames;
 304
 305         // first pool entry is always empty string
 306         newSymNames.push_back('\0');
 307
 308         for (const macho_nlist<P>* s = mergedSymTabStart; s != mergedSymTabend; ++s) {
 309             // if we have better local symbol info, skip any locals here
 310             if ( (localNlists != NULL) && ((s->n_type() & (N_TYPE|N_EXT)) == N_SECT) )
 311                 continue;
 312             macho_nlist<P> t = *s;
 313             t.set_n_strx((uint32_t)newSymNames.size());
 314             const char* symName = &mergedStringPoolStart[s->n_strx()];
 315             if ( symName > mergedStringPoolEnd )
 316                 symName = "<corrupt symbol name>";
 317             newSymNames.insert(newSymNames.end(),
 318                                symName,
 319                                symName + (strlen(symName) + 1));
 320             newSymTab.push_back(t);
 321         }
 322         // <rdar://problem/16529213> recreate N_INDR symbols in extracted dylibs for debugger
 323         for (std::vector<ExportInfoTrie::Entry>::iterator it = exports.begin(); it != exports.end(); ++it) {
 324             macho_nlist<P> t;
 325             memset(&t, 0, sizeof(t));
 326             t.set_n_strx((uint32_t)newSymNames.size());
 327             t.set_n_type(N_INDR | N_EXT);
 328             t.set_n_sect(0);
 329             t.set_n_desc(0);
 330             newSymNames.insert(newSymNames.end(),
 331                                it->name.c_str(),
 332                                it->name.c_str() + (it->name.size() + 1));
 333             const char* importName = it->info.importName.c_str();
 334             if ( *importName == '\0' )
 335                 importName = it->name.c_str();
 336             t.set_n_value(newSymNames.size());
 337             newSymNames.insert(newSymNames.end(),
 338                                importName,
 339                                importName + (strlen(importName) + 1));
 340             newSymTab.push_back(t);
 341         }
 342         if ( localNlistCount != 0 ) {
 343             const char* localStrings = cache->getLocalStrings();
 344             // update load command to reflect new count of locals
 345             dynamicSymTab->ilocalsym = (uint32_t)newSymTab.size();
 346             dynamicSymTab->nlocalsym = localNlistCount;
 347             // copy local symbols
 348             for (uint32_t i=0; i < localNlistCount; ++i) {
 349                 const char* localName = &localStrings[localNlists[i].n_strx()];
 350                 if ( localName > localStrings + cache->getLocalStringsSize() )
 351                     localName = "<corrupt local symbol name>";
 352                 macho_nlist<P> t = localNlists[i];
 353                 t.set_n_strx((uint32_t)newSymNames.size());
 354                 newSymNames.insert(newSymNames.end(),
 355                                    localName,
 356                                    localName + (strlen(localName) + 1));
 357                 newSymTab.push_back(t);
 358             }
 359         }
 360
 361         if ( newSymCount != newSymTab.size() ) {
 362             fprintf(stderr, "symbol count miscalculation\n");
 363             return -1;
 364         }
 365
 366         //const uint64_t newStringPoolOffset = newIndSymTabOffset + dynamicSymTab->nindirectsyms()*sizeof(uint32_t);
 367         //macho_nlist<P>* const newSymTabStart = (macho_nlist<P>*)(((uint8_t*)mh) + newSymTabOffset);
 368         //char* const newStringPoolStart = (char*)mh + newStringPoolOffset;
 369
 370         // pointer align
 371         while ((linkEditSegCmd->fileoff() + new_linkedit_data.size()) % sizeof(pint_t))
 372             new_linkedit_data.push_back(0);
 373
 374         const uint64_t newSymTabOffset = new_linkedit_data.size();
 375
 376         // Copy sym tab
 377         for (macho_nlist<P>& sym : newSymTab) {
 378             uint8_t symData[sizeof(macho_nlist<P>)];
 379             memcpy(&symData, &sym, sizeof(sym));
 380             new_linkedit_data.insert(new_linkedit_data.end(), &symData[0], &symData[sizeof(macho_nlist<P>)]);
 381         }
 382
 383         const uint64_t newIndSymTabOffset = new_linkedit_data.size();
 384
 385         // Copy indirect symbol table
 386         const uint32_t* mergedIndSymTab = (uint32_t*)((char*)mapped_cache + dynamicSymTab->indirectsymoff);
 387         new_linkedit_data.insert(new_linkedit_data.end(),
 388                                  (char*)mergedIndSymTab,
 389                                  (char*)(mergedIndSymTab + dynamicSymTab->nindirectsyms));
 390
 391         const uint64_t newStringPoolOffset = new_linkedit_data.size();
 392
 393         // pointer align string pool size
 394         while (newSymNames.size() % sizeof(pint_t))
 395             newSymNames.push_back('\0');
 396
 397         new_linkedit_data.insert(new_linkedit_data.end(), newSymNames.begin(), newSymNames.end());
 398
 399         // update load commands
 400         if ( functionStarts != NULL ) {
 401             functionStarts->dataoff = (uint32_t)(newFunctionStartsOffset + linkEditSegCmd->fileoff());
 402             functionStarts->datasize = functionStartsSize;
 403         }
 404         if ( dataInCode != NULL ) {
 405             dataInCode->dataoff = (uint32_t)(newDataInCodeOffset + linkEditSegCmd->fileoff());
 406             dataInCode->datasize = dataInCodeSize;
 407         }
 408
 409         symtab->nsyms = newSymCount;
 410         symtab->symoff = (uint32_t)(newSymTabOffset + linkEditSegCmd->fileoff());
 411         symtab->stroff = (uint32_t)(newStringPoolOffset + linkEditSegCmd->fileoff());
 412         symtab->strsize = (uint32_t)newSymNames.size();
 413         dynamicSymTab->extreloff = 0;
 414         dynamicSymTab->nextrel = 0;
 415         dynamicSymTab->locreloff = 0;
 416         dynamicSymTab->nlocrel = 0;
 417         dynamicSymTab->indirectsymoff = (uint32_t)(newIndSymTabOffset + linkEditSegCmd->fileoff());
 418         linkEditSegCmd->set_filesize(symtab->stroff + symtab->strsize - linkEditSegCmd->fileoff());
 419         linkEditSegCmd->set_vmsize((linkEditSegCmd->filesize() + 4095) & (-4096));
 420
 421         return 0;
 422     }
 423
 424 };
 425
 426 static void make_dirs(const char* file_path)
 427 {
 428     //printf("make_dirs(%s)\n", file_path);
 429     char dirs[strlen(file_path)+1];
 430     strcpy(dirs, file_path);
 431     char* lastSlash = strrchr(dirs, '/');
 432     if ( lastSlash == NULL )
 433         return;
 434     lastSlash[1] = '\0';
 435     struct stat stat_buf;
 436     if ( stat(dirs, &stat_buf) != 0 ) {
 437         char* afterSlash = &dirs[1];
 438         char* slash;
 439         while ( (slash = strchr(afterSlash, '/')) != NULL ) {
 440             *slash = '\0';
 441             ::mkdir(dirs, S_IRWXU | S_IRGRP|S_IXGRP | S_IROTH|S_IXOTH);
 442             //printf("mkdir(%s)\n", dirs);
 443             *slash = '/';
 444             afterSlash = slash+1;
 445         }
 446     }
 447 }
 448
 449
 450
 451 template <typename A>
 452 void dylib_maker(const void* mapped_cache, std::vector<uint8_t> &dylib_data, const std::vector<seg_info>& segments) {
 453     typedef typename A::P P;
 454
 455     size_t  additionalSize  = 0;
 456     for(std::vector<seg_info>::const_iterator it=segments.begin(); it != segments.end(); ++it) {
 457         if ( strcmp(it->segName, "__LINKEDIT") != 0 )
 458             additionalSize += it->sizem;
 459     }
 460
 461     std::vector<uint8_t> new_dylib_data;
 462     new_dylib_data.reserve(additionalSize);
 463
 464     // Write regular segments into the buffer
 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
 471         //printf("segName=%s, offset=0x%llX, size=0x%0llX\n", it->segName, it->offset, it->sizem);
 472         // Copy all but the __LINKEDIT.  It will be copied later during the optimizer in to a temporary buffer but it would
 473         // not be efficient to copy it all now for each dylib.
 474         if (strcmp(it->segName, "__LINKEDIT") == 0 )
 475             continue;
 476         std::copy(((uint8_t*)mapped_cache)+it->offset, ((uint8_t*)mapped_cache)+it->offset+it->sizem, std::back_inserter(new_dylib_data));
 477     }
 478
 479     // optimize linkedit
 480     std::vector<uint8_t> new_linkedit_data;
 481     new_linkedit_data.reserve(1 << 20);
 482
 483     LinkeditOptimizer<A> linkeditOptimizer;
 484     dyld3::MachOAnalyzer* mh = (dyld3::MachOAnalyzer*)&new_dylib_data.front();
 485     linkeditOptimizer.optimize_loadcommands(mh);
 486     linkeditOptimizer.optimize_linkedit(new_linkedit_data, textOffsetInCache, mapped_cache);
 487
 488     new_dylib_data.insert(new_dylib_data.end(), new_linkedit_data.begin(), new_linkedit_data.end());
 489
 490     // Page align file
 491     while (new_dylib_data.size() % 4096)
 492         new_dylib_data.push_back(0);
 493
 494     dylib_data.insert(dylib_data.end(), new_dylib_data.begin(), new_dylib_data.end());
 495 }
 496
 497 typedef __typeof(dylib_maker<x86>) dylib_maker_func;
 498 typedef void (^progress_block)(unsigned current, unsigned total);
 499
 500 class SharedCacheExtractor;
 501 struct SharedCacheDylibExtractor {
 502     SharedCacheDylibExtractor(const char* name, std::vector<seg_info> segInfo)
 503         : name(name), segInfo(segInfo) { }
 504
 505     void extractCache(SharedCacheExtractor& context);
 506
 507     const char*                     name;
 508     const std::vector<seg_info>     segInfo;
 509     int                             result = 0;
 510 };
 511
 512 struct SharedCacheExtractor {
 513     SharedCacheExtractor(const NameToSegments& map,
 514                          const char* extraction_root_path,
 515                          dylib_maker_func* dylib_create_func,
 516                          void* mapped_cache,
 517                          progress_block progress)
 518         : map(map), extraction_root_path(extraction_root_path),
 519           dylib_create_func(dylib_create_func), mapped_cache(mapped_cache),
 520           progress(progress) {
 521
 522       extractors.reserve(map.size());
 523       for (auto it : map)
 524           extractors.emplace_back(it.first, it.second);
 525
 526         // Limit the number of open files.  16 seems to give better performance than higher numbers.
 527         sema = dispatch_semaphore_create(16);
 528     }
 529     int extractCaches();
 530
 531     static void extractCache(void *ctx, size_t i);
 532
 533     const NameToSegments&                   map;
 534     std::vector<SharedCacheDylibExtractor>  extractors;
 535     dispatch_semaphore_t                    sema;
 536     const char*                             extraction_root_path;
 537     dylib_maker_func*                       dylib_create_func;
 538     void*                                   mapped_cache;
 539     progress_block                          progress;
 540     std::atomic_int                         count = { 0 };
 541 };
 542
 543 int SharedCacheExtractor::extractCaches() {
 544     dispatch_queue_t process_queue = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_LOW, 0);
 545     dispatch_apply_f(map.size(), process_queue,
 546                      this, extractCache);
 547
 548     int result = 0;
 549     for (const SharedCacheDylibExtractor& extractor : extractors) {
 550         if (extractor.result != 0) {
 551             result = extractor.result;
 552             break;
 553         }
 554     }
 555     return result;
 556 }
 557
 558 void SharedCacheExtractor::extractCache(void *ctx, size_t i) {
 559     SharedCacheExtractor& context = *(SharedCacheExtractor*)ctx;
 560     dispatch_semaphore_wait(context.sema, DISPATCH_TIME_FOREVER);
 561     context.extractors[i].extractCache(context);
 562     dispatch_semaphore_signal(context.sema);
 563 }
 564
 565 void SharedCacheDylibExtractor::extractCache(SharedCacheExtractor &context) {
 566
 567     char    dylib_path[PATH_MAX];
 568     strcpy(dylib_path, context.extraction_root_path);
 569     strcat(dylib_path, "/");
 570     strcat(dylib_path, name);
 571
 572     //printf("%s with %lu segments\n", dylib_path, it->second.size());
 573     // make sure all directories in this path exist
 574     make_dirs(dylib_path);
 575
 576     // open file, create if does not already exist
 577     int fd = ::open(dylib_path, O_CREAT | O_TRUNC | O_EXLOCK | O_RDWR, 0644);
 578     if ( fd == -1 ) {
 579         fprintf(stderr, "can't open or create dylib file %s, errnor=%d\n", dylib_path, errno);
 580         result = -1;
 581         return;
 582     }
 583
 584     std::vector<uint8_t> vec;
 585     context.dylib_create_func(context.mapped_cache, vec, segInfo);
 586     context.progress(context.count++, (unsigned)context.map.size());
 587
 588     // Write file data
 589     if( write(fd, &vec.front(), vec.size()) == -1) {
 590         fprintf(stderr, "error writing, errnor=%d\n", errno);
 591         result = -1;
 592     }
 593
 594     close(fd);
 595 }
 596
 597 static int sharedCacheIsValid(const void* mapped_cache, uint64_t size) {
 598     // First check that the size is good.
 599     // Note the shared cache may not have a codeSignatureSize value set so we need to first make
 600     // sure we have space for the CS_SuperBlob, then later crack that to check for the size of the rest.
 601     const DyldSharedCache* dyldSharedCache = (DyldSharedCache*)mapped_cache;
 602     uint64_t requiredSizeForCSSuperBlob = dyldSharedCache->header.codeSignatureOffset + sizeof(CS_SuperBlob);
 603     const dyld_cache_mapping_info* mappings = (dyld_cache_mapping_info*)((uint8_t*)mapped_cache + dyldSharedCache->header.mappingOffset);
 604     if ( requiredSizeForCSSuperBlob > size ) {
 605         fprintf(stderr, "Error: dyld shared cache size 0x%08llx is less than required size of 0x%08llx.\n", size, requiredSizeForCSSuperBlob);
 606         return -1;
 607     }
 608
 609     // Now see if the code signatures are valid as that tells us the pages aren't corrupt.
 610     // First find all of the regions of the shared cache we computed cd hashes
 611     std::vector<std::pair<uint64_t, uint64_t>> sharedCacheRegions;
 612     for (uint32_t i = 0; i != dyldSharedCache->header.mappingCount; ++i) {
 613         sharedCacheRegions.emplace_back(std::make_pair(mappings[i].fileOffset, mappings[i].fileOffset + mappings[i].size));
 614     }
 615     if (dyldSharedCache->header.localSymbolsSize)
 616         sharedCacheRegions.emplace_back(std::make_pair(dyldSharedCache->header.localSymbolsOffset, dyldSharedCache->header.localSymbolsOffset + dyldSharedCache->header.localSymbolsSize));
 617     size_t inBbufferSize = 0;
 618     for (auto& sharedCacheRegion : sharedCacheRegions)
 619         inBbufferSize += (sharedCacheRegion.second - sharedCacheRegion.first);
 620
 621     // Now take the cd hash from the cache itself and validate the regions we found.
 622     uint8_t* codeSignatureRegion = (uint8_t*)mapped_cache + dyldSharedCache->header.codeSignatureOffset;
 623     CS_SuperBlob* sb = reinterpret_cast<CS_SuperBlob*>(codeSignatureRegion);
 624     if (sb->magic != htonl(CSMAGIC_EMBEDDED_SIGNATURE)) {
 625         fprintf(stderr, "Error: dyld shared cache code signature magic is incorrect.\n");
 626         return -1;
 627     }
 628
 629     size_t sbSize = ntohl(sb->length);
 630     uint64_t requiredSizeForCS = dyldSharedCache->header.codeSignatureOffset + sbSize;
 631     if ( requiredSizeForCS > size ) {
 632         fprintf(stderr, "Error: dyld shared cache size 0x%08llx is less than required size of 0x%08llx.\n", size, requiredSizeForCS);
 633         return -1;
 634     }
 635
 636     // Find the offset to the code directory.
 637     CS_CodeDirectory* cd = nullptr;
 638     for (unsigned i =0; i != sb->count; ++i) {
 639         if (ntohl(sb->index[i].type) == CSSLOT_CODEDIRECTORY) {
 640             cd = (CS_CodeDirectory*)(codeSignatureRegion + ntohl(sb->index[i].offset));
 641             break;
 642         }
 643     }
 644
 645     if (!cd) {
 646         fprintf(stderr, "Error: dyld shared cache code signature directory is missing.\n");
 647         return -1;
 648     }
 649
 650     if ( (uint8_t*)cd > (codeSignatureRegion + sbSize) ) {
 651         fprintf(stderr, "Error: dyld shared cache code signature directory is out of bounds.\n");
 652         return -1;
 653     }
 654
 655     if ( cd->magic != htonl(CSMAGIC_CODEDIRECTORY) ) {
 656         fprintf(stderr, "Error: dyld shared cache code signature directory magic is incorrect.\n");
 657         return -1;
 658     }
 659
 660     uint32_t pageSize = 1 << cd->pageSize;
 661     uint32_t slotCountFromRegions = (uint32_t)((inBbufferSize + pageSize - 1) / pageSize);
 662     if ( ntohl(cd->nCodeSlots) < slotCountFromRegions ) {
 663         fprintf(stderr, "Error: dyld shared cache code signature directory num slots is incorrect.\n");
 664         return -1;
 665     }
 666
 667     uint32_t dscDigestFormat = kCCDigestNone;
 668     switch (cd->hashType) {
 669         case CS_HASHTYPE_SHA1:
 670 #pragma clang diagnostic push
 671 #pragma clang diagnostic ignored "-Wdeprecated-declarations"
 672             dscDigestFormat = kCCDigestSHA1;
 673 #pragma clang diagnostic pop
 674             break;
 675         case CS_HASHTYPE_SHA256:
 676             dscDigestFormat = kCCDigestSHA256;
 677             break;
 678         default:
 679             break;
 680     }
 681
 682     if (dscDigestFormat != kCCDigestNone) {
 683         const uint64_t csPageSize = 1 << cd->pageSize;
 684         size_t   hashOffset = ntohl(cd->hashOffset);
 685         uint8_t* hashSlot = (uint8_t*)cd + hashOffset;
 686         uint8_t cdHashBuffer[cd->hashSize];
 687
 688         // Skip local symbols for now as those aren't being codesign correctly right now.
 689         size_t inBbufferSize = 0;
 690         for (auto& sharedCacheRegion : sharedCacheRegions) {
 691             if (sharedCacheRegion.first == dyldSharedCache->header.localSymbolsOffset)
 692                 continue;
 693             inBbufferSize += (sharedCacheRegion.second - sharedCacheRegion.first);
 694         }
 695         uint32_t slotCountToProcess = (uint32_t)((inBbufferSize + pageSize - 1) / pageSize);
 696
 697         for (unsigned i = 0; i != slotCountToProcess; ++i) {
 698             // Skip data pages as those may have been slid by ASLR in the extracted file
 699             uint64_t fileOffset = i * csPageSize;
 700             bool isDataPage = false;
 701             for (unsigned mappingIndex = 1; mappingIndex != (dyldSharedCache->header.mappingCount - 1); ++mappingIndex) {
 702                 if ( (fileOffset >= mappings[mappingIndex].fileOffset) && (fileOffset < (mappings[mappingIndex].fileOffset + mappings[mappingIndex].size)) ) {
 703                     isDataPage = true;
 704                     break;
 705                 }
 706             }
 707             if ( isDataPage )
 708                 continue;
 709
 710             CCDigest(dscDigestFormat, (uint8_t*)mapped_cache + fileOffset, (size_t)csPageSize, cdHashBuffer);
 711             uint8_t* cacheCdHashBuffer = hashSlot + (i * cd->hashSize);
 712             if (memcmp(cdHashBuffer, cacheCdHashBuffer, cd->hashSize) != 0)  {
 713                 fprintf(stderr, "Error: dyld shared cache code signature for page %d is incorrect.\n", i);
 714                 return -1;
 715             }
 716         }
 717     }
 718     return 0;
 719 }
 720
 721 int dyld_shared_cache_extract_dylibs_progress(const char* shared_cache_file_path, const char* extraction_root_path,
 722                                               progress_block progress)
 723 {
 724     struct stat statbuf;
 725     if (stat(shared_cache_file_path, &statbuf)) {
 726         fprintf(stderr, "Error: stat failed for dyld shared cache at %s\n", shared_cache_file_path);
 727         return -1;
 728     }
 729
 730     int cache_fd = open(shared_cache_file_path, O_RDONLY);
 731     if (cache_fd < 0) {
 732         fprintf(stderr, "Error: failed to open shared cache file at %s\n", shared_cache_file_path);
 733         return -1;
 734     }
 735
 736     void* mapped_cache = mmap(NULL, (size_t)statbuf.st_size, PROT_READ, MAP_PRIVATE, cache_fd, 0);
 737     if (mapped_cache == MAP_FAILED) {
 738         fprintf(stderr, "Error: mmap() for shared cache at %s failed, errno=%d\n", shared_cache_file_path, errno);
 739         return -1;
 740     }
 741
 742     close(cache_fd);
 743
 744     // instantiate arch specific dylib maker
 745     dylib_maker_func* dylib_create_func = nullptr;
 746     if ( strcmp((char*)mapped_cache, "dyld_v1    i386") == 0 )
 747         dylib_create_func = dylib_maker<x86>;
 748     else if ( strcmp((char*)mapped_cache, "dyld_v1  x86_64") == 0 )
 749         dylib_create_func = dylib_maker<x86_64>;
 750     else if ( strcmp((char*)mapped_cache, "dyld_v1 x86_64h") == 0 )
 751         dylib_create_func = dylib_maker<x86_64>;
 752     else if ( strcmp((char*)mapped_cache, "dyld_v1   armv5") == 0 )
 753         dylib_create_func = dylib_maker<arm>;
 754     else if ( strcmp((char*)mapped_cache, "dyld_v1   armv6") == 0 )
 755         dylib_create_func = dylib_maker<arm>;
 756     else if ( strcmp((char*)mapped_cache, "dyld_v1   armv7") == 0 )
 757         dylib_create_func = dylib_maker<arm>;
 758     else if ( strncmp((char*)mapped_cache, "dyld_v1  armv7", 14) == 0 )
 759         dylib_create_func = dylib_maker<arm>;
 760     else if ( strcmp((char*)mapped_cache, "dyld_v1   arm64") == 0 )
 761         dylib_create_func = dylib_maker<arm64>;
 762 #if SUPPORT_ARCH_arm64e
 763     else if ( strcmp((char*)mapped_cache, "dyld_v1  arm64e") == 0 )
 764         dylib_create_func = dylib_maker<arm64>;
 765 #endif
 766 #if SUPPORT_ARCH_arm64_32
 767     else if ( strcmp((char*)mapped_cache, "dyld_v1arm64_32") == 0 )
 768         dylib_create_func = dylib_maker<arm64_32>;
 769 #endif
 770     else {
 771         fprintf(stderr, "Error: unrecognized dyld shared cache magic.\n");
 772         munmap(mapped_cache, (size_t)statbuf.st_size);
 773         return -1;
 774     }
 775
 776     // Verify that the cache isn't corrupt.
 777     if (int result = sharedCacheIsValid(mapped_cache, (uint64_t)statbuf.st_size)) {
 778         munmap(mapped_cache, (size_t)statbuf.st_size);
 779         return result;
 780     }
 781
 782     // iterate through all images in cache and build map of dylibs and segments
 783     __block NameToSegments  map;
 784     int                     result = 0;
 785
 786     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) {
 787         map[dylibInfo->path].push_back(seg_info(segInfo->name, segInfo->fileOffset, segInfo->fileSize));
 788     });
 789
 790     if(result != 0) {
 791         fprintf(stderr, "Error: dyld_shared_cache_iterate_segments_with_slide failed.\n");
 792         munmap(mapped_cache, (size_t)statbuf.st_size);
 793         return result;
 794     }
 795
 796     // for each dylib instantiate a dylib file
 797     SharedCacheExtractor extractor(map, extraction_root_path, dylib_create_func, mapped_cache, progress);
 798     result = extractor.extractCaches();
 799
 800     munmap(mapped_cache, (size_t)statbuf.st_size);
 801     return result;
 802 }
 803
 804
 805
 806 int dyld_shared_cache_extract_dylibs(const char* shared_cache_file_path, const char* extraction_root_path)
 807 {
 808     return dyld_shared_cache_extract_dylibs_progress(shared_cache_file_path, extraction_root_path,
 809                                                      ^(unsigned , unsigned) {} );
 810 }
 811
 812
 813 #if 0
 814 // test program
 815 #include <stdio.h>
 816 #include <stddef.h>
 817 #include <dlfcn.h>
 818
 819
 820 typedef int (*extractor_proc)(const char* shared_cache_file_path, const char* extraction_root_path,
 821                               void (^progress)(unsigned current, unsigned total));
 822
 823 int main(int argc, const char* argv[])
 824 {
 825     if ( argc != 3 ) {
 826         fprintf(stderr, "usage: dsc_extractor <path-to-cache-file> <path-to-device-dir>\n");
 827         return 1;
 828     }
 829
 830     //void* handle = dlopen("/Volumes/my/src/dyld/build/Debug/dsc_extractor.bundle", RTLD_LAZY);
 831     void* handle = dlopen("/Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/usr/lib/dsc_extractor.bundle", RTLD_LAZY);
 832     if ( handle == NULL ) {
 833         fprintf(stderr, "dsc_extractor.bundle could not be loaded\n");
 834         return 1;
 835     }
 836
 837     extractor_proc proc = (extractor_proc)dlsym(handle, "dyld_shared_cache_extract_dylibs_progress");
 838     if ( proc == NULL ) {
 839         fprintf(stderr, "dsc_extractor.bundle did not have dyld_shared_cache_extract_dylibs_progress symbol\n");
 840         return 1;
 841     }
 842
 843     int result = (*proc)(argv[1], argv[2], ^(unsigned c, unsigned total) { printf("%d/%d\n", c, total); } );
 844     fprintf(stderr, "dyld_shared_cache_extract_dylibs_progress() => %d\n", result);
 845     return 0;
 846 }
 847
 848
 849 #endif
 850
 851
 852
 853