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
  45 #include "dsc_iterator.h"
  46 #include "dsc_extractor.h"
  47
  48 #include <vector>
  49 #include <map>
  50 #include <ext/hash_map>
  51 #include <algorithm>
  52 #include <dispatch/dispatch.h>
  53
  54 struct seg_info
  55 {
  56                                 seg_info(const char* n, uint64_t o, uint64_t s)
  57                                         : segName(n), offset(o), sizem(s) { }
  58         const char* segName;
  59         uint64_t        offset;
  60         uint64_t        sizem;
  61 };
  62
  63 class CStringEquals {
  64 public:
  65         bool operator()(const char* left, const char* right) const { return (strcmp(left, right) == 0); }
  66 };
  67 typedef __gnu_cxx::hash_map<const char*, std::vector<seg_info>, __gnu_cxx::hash<const char*>, CStringEquals> NameToSegments;
  68
  69
  70 template <typename A>
  71 int optimize_linkedit(macho_header<typename A::P>* mh, const void* mapped_cache, uint64_t* newSize)
  72 {
  73         typedef typename A::P P;
  74         typedef typename A::P::E E;
  75     typedef typename A::P::uint_t pint_t;
  76
  77         // update header flags
  78         mh->set_flags(mh->flags() & 0x7FFFFFFF); // remove in-cache bit
  79
  80         // update load commands
  81         uint64_t cumulativeFileSize = 0;
  82         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)mh + sizeof(macho_header<P>));
  83         const uint32_t cmd_count = mh->ncmds();
  84         const macho_load_command<P>* cmd = cmds;
  85         macho_segment_command<P>* linkEditSegCmd = NULL;
  86         macho_symtab_command<P>* symtab = NULL;
  87         macho_dysymtab_command<P>*      dynamicSymTab = NULL;
  88         macho_linkedit_data_command<P>* functionStarts = NULL;
  89         macho_linkedit_data_command<P>* dataInCode = NULL;
  90         for (uint32_t i = 0; i < cmd_count; ++i) {
  91                 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
  92                         // update segment/section file offsets
  93                         macho_segment_command<P>* segCmd = (macho_segment_command<P>*)cmd;
  94                         segCmd->set_fileoff(cumulativeFileSize);
  95                         macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)segCmd + sizeof(macho_segment_command<P>));
  96                         macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
  97                         for(macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
  98                                 if ( sect->offset() != 0 )
  99                                         sect->set_offset(cumulativeFileSize+sect->addr()-segCmd->vmaddr());
 100                         }
 101                         if ( strcmp(segCmd->segname(), "__LINKEDIT") == 0 ) {
 102                                 linkEditSegCmd = segCmd;
 103                         }
 104                         cumulativeFileSize += segCmd->filesize();
 105                 }
 106                 else if ( cmd->cmd() == LC_DYLD_INFO_ONLY ) {
 107                         // zero out all dyld info
 108                         macho_dyld_info_command<P>* dyldInfo = (macho_dyld_info_command<P>*)cmd;
 109                         dyldInfo->set_rebase_off(0);
 110                         dyldInfo->set_rebase_size(0);
 111                         dyldInfo->set_bind_off(0);
 112                         dyldInfo->set_bind_size(0);
 113                         dyldInfo->set_weak_bind_off(0);
 114                         dyldInfo->set_weak_bind_size(0);
 115                         dyldInfo->set_lazy_bind_off(0);
 116                         dyldInfo->set_lazy_bind_size(0);
 117                         dyldInfo->set_export_off(0);
 118                         dyldInfo->set_export_size(0);
 119                 }
 120                 else if ( cmd->cmd() == LC_SYMTAB ) {
 121                         symtab = (macho_symtab_command<P>*)cmd;
 122                 }
 123                 else if ( cmd->cmd() == LC_DYSYMTAB ) {
 124                         dynamicSymTab = (macho_dysymtab_command<P>*)cmd;
 125                 }
 126                 else if ( cmd->cmd() == LC_FUNCTION_STARTS ) {
 127                         functionStarts = (macho_linkedit_data_command<P>*)cmd;
 128                 }
 129                 else if ( cmd->cmd() == LC_DATA_IN_CODE ) {
 130                         dataInCode = (macho_linkedit_data_command<P>*)cmd;
 131                 }
 132                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 133         }
 134
 135         // rebuild symbol table
 136         if ( linkEditSegCmd == NULL ) {
 137                 fprintf(stderr, "__LINKEDIT not found\n");
 138                 return -1;
 139         }
 140         if ( symtab == NULL ) {
 141                 fprintf(stderr, "LC_SYMTAB not found\n");
 142                 return -1;
 143         }
 144         if ( dynamicSymTab == NULL ) {
 145                 fprintf(stderr, "LC_DYSYMTAB not found\n");
 146                 return -1;
 147         }
 148
 149         const uint32_t newFunctionStartsOffset = linkEditSegCmd->fileoff();
 150         uint32_t functionStartsSize = 0;
 151         if ( functionStarts != NULL ) {
 152                 // copy function starts from original cache file to new mapped dylib file
 153                 functionStartsSize = functionStarts->datasize();
 154                 memcpy((char*)mh + newFunctionStartsOffset, (char*)mapped_cache + functionStarts->dataoff(), functionStartsSize);
 155         }
 156         const uint32_t newDataInCodeOffset = (newFunctionStartsOffset + functionStartsSize + sizeof(pint_t) - 1) & (-sizeof(pint_t)); // pointer align
 157         uint32_t dataInCodeSize = 0;
 158         if ( dataInCode != NULL ) {
 159                 // copy data-in-code info from original cache file to new mapped dylib file
 160                 dataInCodeSize = dataInCode->datasize();
 161                 memcpy((char*)mh + newDataInCodeOffset, (char*)mapped_cache + dataInCode->dataoff(), dataInCodeSize);
 162         }
 163         // copy symbol entries and strings from original cache file to new mapped dylib file
 164         const uint32_t newSymTabOffset = (newDataInCodeOffset + dataInCodeSize + sizeof(pint_t) - 1) & (-sizeof(pint_t)); // pointer align
 165         const uint32_t newIndSymTabOffset = newSymTabOffset + symtab->nsyms()*sizeof(macho_nlist<P>);
 166         const uint32_t newStringPoolOffset = newIndSymTabOffset + dynamicSymTab->nindirectsyms()*sizeof(uint32_t);
 167         macho_nlist<P>* const newSymTabStart = (macho_nlist<P>*)(((uint8_t*)mh) + newSymTabOffset);
 168         char* const newStringPoolStart = (char*)mh + newStringPoolOffset;
 169         uint32_t* newIndSymTab = (uint32_t*)((char*)mh + newIndSymTabOffset);
 170         const uint32_t* mergedIndSymTab = (uint32_t*)((char*)mapped_cache + dynamicSymTab->indirectsymoff());
 171         const macho_nlist<P>* const mergedSymTabStart = (macho_nlist<P>*)(((uint8_t*)mapped_cache) + symtab->symoff());
 172         const macho_nlist<P>* const mergedSymTabend = &mergedSymTabStart[symtab->nsyms()];
 173         const char* mergedStringPoolStart = (char*)mapped_cache + symtab->stroff();
 174         macho_nlist<P>* t = newSymTabStart;
 175         int poolOffset = 0;
 176         newStringPoolStart[poolOffset++] = '\0'; // first pool entry is always empty string
 177         for (const macho_nlist<P>* s = mergedSymTabStart; s != mergedSymTabend; ++s) {
 178                 *t = *s;
 179                 t->set_n_strx(poolOffset);
 180                 strcpy(&newStringPoolStart[poolOffset], &mergedStringPoolStart[s->n_strx()]);
 181                 poolOffset += (strlen(&newStringPoolStart[poolOffset]) + 1);
 182                 ++t;
 183         }
 184         // pointer align string pool size
 185         while ( (poolOffset % sizeof(pint_t)) != 0 )
 186                 ++poolOffset;
 187         // copy indirect symbol table
 188         memcpy(newIndSymTab, mergedIndSymTab, dynamicSymTab->nindirectsyms()*sizeof(uint32_t));
 189
 190         // update load commands
 191         if ( functionStarts != NULL ) {
 192                 functionStarts->set_dataoff(newFunctionStartsOffset);
 193                 functionStarts->set_datasize(functionStartsSize);
 194         }
 195         if ( dataInCode != NULL ) {
 196                 dataInCode->set_dataoff(newDataInCodeOffset);
 197                 dataInCode->set_datasize(dataInCodeSize);
 198         }
 199         symtab->set_symoff(newSymTabOffset);
 200         symtab->set_stroff(newStringPoolOffset);
 201         symtab->set_strsize(poolOffset);
 202         dynamicSymTab->set_extreloff(0);
 203         dynamicSymTab->set_nextrel(0);
 204         dynamicSymTab->set_locreloff(0);
 205         dynamicSymTab->set_nlocrel(0);
 206         dynamicSymTab->set_indirectsymoff(newIndSymTabOffset);
 207         linkEditSegCmd->set_filesize(symtab->stroff()+symtab->strsize() - linkEditSegCmd->fileoff());
 208         linkEditSegCmd->set_vmsize( (linkEditSegCmd->filesize()+4095) & (-4096) );
 209
 210         // return new size
 211         *newSize = (symtab->stroff()+symtab->strsize()+4095) & (-4096);
 212
 213         return 0;
 214 }
 215
 216
 217
 218 static void make_dirs(const char* file_path)
 219 {
 220         //printf("make_dirs(%s)\n", file_path);
 221         char dirs[strlen(file_path)+1];
 222         strcpy(dirs, file_path);
 223         char* lastSlash = strrchr(dirs, '/');
 224         if ( lastSlash == NULL )
 225                 return;
 226         lastSlash[1] = '\0';
 227         struct stat stat_buf;
 228         if ( stat(dirs, &stat_buf) != 0 ) {
 229                 const char* afterSlash = &dirs[1];
 230                 char* slash;
 231                 while ( (slash = strchr(afterSlash, '/')) != NULL ) {
 232                         *slash = '\0';
 233                         ::mkdir(dirs, S_IRWXU | S_IRGRP|S_IXGRP | S_IROTH|S_IXOTH);
 234                         //printf("mkdir(%s)\n", dirs);
 235                         *slash = '/';
 236                         afterSlash = slash+1;
 237                 }
 238         }
 239 }
 240
 241
 242
 243 template <typename A>
 244 size_t dylib_maker(const void* mapped_cache, std::vector<uint8_t> &dylib_data, const std::vector<seg_info>& segments) {
 245         typedef typename A::P P;
 246
 247     size_t  additionalSize  = 0;
 248         for(std::vector<seg_info>::const_iterator it=segments.begin(); it != segments.end(); ++it) {
 249                 additionalSize                          += it->sizem;
 250         }
 251
 252     dylib_data.reserve(dylib_data.size() + additionalSize);
 253
 254     uint32_t                nfat_archs          = 0;
 255         uint32_t                offsetInFatFile     = 4096;
 256     uint8_t                 *base_ptr           = &dylib_data.front();
 257
 258 #define FH reinterpret_cast<fat_header*>(base_ptr)
 259 #define FA reinterpret_cast<fat_arch*>(base_ptr + (8 + (nfat_archs - 1) * sizeof(fat_arch)))
 260
 261     if(dylib_data.size() >= 4096 && OSSwapBigToHostInt32(FH->magic) == FAT_MAGIC) {
 262                 // have fat header, append new arch to end
 263         nfat_archs                              = OSSwapBigToHostInt32(FH->nfat_arch);
 264                 offsetInFatFile                         = OSSwapBigToHostInt32(FA->offset) + OSSwapBigToHostInt32(FA->size);
 265     }
 266
 267     dylib_data.resize(offsetInFatFile);
 268     base_ptr                                    = &dylib_data.front();
 269
 270     FH->magic                                   = OSSwapHostToBigInt32(FAT_MAGIC);
 271     FH->nfat_arch                               = OSSwapHostToBigInt32(++nfat_archs);
 272
 273     FA->cputype                                 = 0; // filled in later
 274     FA->cpusubtype                              = 0; // filled in later
 275     FA->offset                                  = OSSwapHostToBigInt32(offsetInFatFile);
 276     FA->size                                    = 0; // filled in later
 277     FA->align                                   = OSSwapHostToBigInt32(12);
 278
 279         // Write regular segments into the buffer
 280         uint32_t                totalSize           = 0;
 281
 282         for( std::vector<seg_info>::const_iterator it=segments.begin(); it != segments.end(); ++it) {
 283
 284         if(strcmp(it->segName, "__TEXT") == 0 ) {
 285             const macho_header<P>   *textMH     = reinterpret_cast<macho_header<P>*>((uint8_t*)mapped_cache+it->offset);
 286             FA->cputype                         = OSSwapHostToBigInt32(textMH->cputype());
 287             FA->cpusubtype                      = OSSwapHostToBigInt32(textMH->cpusubtype());
 288
 289             // if this cputype/subtype already exist in fat header, then return immediately
 290             for(uint32_t i=0; i < nfat_archs-1; ++i) {
 291                 fat_arch            *afa        = reinterpret_cast<fat_arch*>(base_ptr+8)+i;
 292
 293                 if(   afa->cputype == FA->cputype
 294                    && afa->cpusubtype == FA->cpusubtype) {
 295                     fprintf(stderr, "arch already exists in fat dylib\n");
 296                     dylib_data.resize(offsetInFatFile);
 297                     return offsetInFatFile;
 298                 }
 299             }
 300                 }
 301
 302                 //printf("segName=%s, offset=0x%llX, size=0x%0llX\n", it->segName, it->offset, it->sizem);
 303         std::copy(((uint8_t*)mapped_cache)+it->offset, ((uint8_t*)mapped_cache)+it->offset+it->sizem, std::back_inserter(dylib_data));
 304         base_ptr                                = &dylib_data.front();
 305         totalSize                               += it->sizem;
 306         }
 307
 308         FA->size                                    = OSSwapHostToBigInt32(totalSize);
 309
 310         // optimize linkedit
 311         uint64_t                newSize             = dylib_data.size();
 312         optimize_linkedit<A>(((macho_header<P>*)(base_ptr+offsetInFatFile)), mapped_cache, &newSize);
 313
 314         // update fat header with new file size
 315     dylib_data.resize(offsetInFatFile+newSize);
 316     base_ptr                                    = &dylib_data.front();
 317         FA->size                                    = OSSwapHostToBigInt32(newSize);
 318 #undef FH
 319 #undef FA
 320         return offsetInFatFile;
 321 }
 322
 323
 324 int dyld_shared_cache_extract_dylibs_progress(const char* shared_cache_file_path, const char* extraction_root_path,
 325                                                                                                         void (^progress)(unsigned current, unsigned total))
 326 {
 327         struct stat statbuf;
 328         if (stat(shared_cache_file_path, &statbuf)) {
 329                 fprintf(stderr, "Error: stat failed for dyld shared cache at %s\n", shared_cache_file_path);
 330                 return -1;
 331         }
 332
 333         int cache_fd = open(shared_cache_file_path, O_RDONLY);
 334         if (cache_fd < 0) {
 335                 fprintf(stderr, "Error: failed to open shared cache file at %s\n", shared_cache_file_path);
 336                 return -1;
 337         }
 338
 339         void* mapped_cache = mmap(NULL, statbuf.st_size, PROT_READ, MAP_PRIVATE, cache_fd, 0);
 340         if (mapped_cache == MAP_FAILED) {
 341                 fprintf(stderr, "Error: mmap() for shared cache at %s failed, errno=%d\n", shared_cache_file_path, errno);
 342                 return -1;
 343         }
 344
 345     close(cache_fd);
 346
 347         // instantiate arch specific dylib maker
 348     size_t (*dylib_create_func)(const void*, std::vector<uint8_t>&, const std::vector<seg_info>&) = NULL;
 349              if ( strcmp((char*)mapped_cache, "dyld_v1    i386") == 0 )
 350                 dylib_create_func = dylib_maker<x86>;
 351         else if ( strcmp((char*)mapped_cache, "dyld_v1  x86_64") == 0 )
 352                 dylib_create_func = dylib_maker<x86_64>;
 353         else if ( strcmp((char*)mapped_cache, "dyld_v1   armv5") == 0 )
 354                 dylib_create_func = dylib_maker<arm>;
 355         else if ( strcmp((char*)mapped_cache, "dyld_v1   armv6") == 0 )
 356                 dylib_create_func = dylib_maker<arm>;
 357         else if ( strcmp((char*)mapped_cache, "dyld_v1   armv7") == 0 )
 358                 dylib_create_func = dylib_maker<arm>;
 359         else if ( strncmp((char*)mapped_cache, "dyld_v1  armv7", 14) == 0 )
 360                 dylib_create_func = dylib_maker<arm>;
 361         else {
 362                 fprintf(stderr, "Error: unrecognized dyld shared cache magic.\n");
 363         munmap(mapped_cache, statbuf.st_size);
 364                 return -1;
 365         }
 366
 367         // iterate through all images in cache and build map of dylibs and segments
 368         __block NameToSegments map;
 369         dyld_shared_cache_iterate_segments_with_slide(mapped_cache,
 370                 ^(const char* dylib, const char* segName, uint64_t offset, uint64_t sizem,
 371                                                         uint64_t mappedddress, uint64_t slide) {
 372                         map[dylib].push_back(seg_info(segName, offset, sizem));
 373                 });
 374
 375         // for each dylib instantiate a dylib file
 376     dispatch_group_t        group               = dispatch_group_create();
 377     dispatch_semaphore_t    sema                = dispatch_semaphore_create(2);
 378     dispatch_queue_t        process_queue       = dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_LOW, 0);
 379     dispatch_queue_t        writer_queue        = dispatch_queue_create("dyld writer queue", 0);
 380
 381         __block int             cumulativeResult    = 0;
 382         __block unsigned        count               = 0;
 383
 384         for ( NameToSegments::iterator it = map.begin(); it != map.end(); ++it) {
 385         dispatch_semaphore_wait(sema, DISPATCH_TIME_FOREVER);
 386         dispatch_group_async(group, process_queue, ^{
 387
 388             char    dylib_path[PATH_MAX];
 389             strcpy(dylib_path, extraction_root_path);
 390             strcat(dylib_path, "/");
 391             strcat(dylib_path, it->first);
 392
 393             //printf("%s with %lu segments\n", dylib_path, segments.size());
 394             // make sure all directories in this path exist
 395             make_dirs(dylib_path);
 396
 397             // open file, create if does not already exist
 398             int fd = ::open(dylib_path, O_CREAT | O_EXLOCK | O_RDWR, 0644);
 399             if ( fd == -1 ) {
 400                 fprintf(stderr, "can't open or create dylib file %s, errnor=%d\n", dylib_path, errno);
 401                 cumulativeResult    = -1;
 402                 return;
 403             }
 404
 405             struct stat statbuf;
 406             if (fstat(fd, &statbuf)) {
 407                 fprintf(stderr, "Error: stat failed for dyld file %s, errnor=%d\n", dylib_path, errno);
 408                 close(fd);
 409                 cumulativeResult    = -1;
 410                 return;
 411             }
 412
 413             std::vector<uint8_t> *vec   = new std::vector<uint8_t>(statbuf.st_size);
 414             if(pread(fd, &vec->front(), vec->size(), 0) != (long)vec->size()) {
 415                 fprintf(stderr, "can't read dylib file %s, errnor=%d\n", dylib_path, errno);
 416                 close(fd);
 417                 cumulativeResult    = -1;
 418                 return;
 419             }
 420
 421             const size_t    offset  = dylib_create_func(mapped_cache, *vec, it->second);
 422
 423             dispatch_group_async(group, writer_queue, ^{
 424                 progress(count++, map.size());
 425
 426                 if(offset != vec->size()) {
 427                     //Write out the first page, and everything after offset
 428                     if(   pwrite(fd, &vec->front(), 4096, 0) == -1
 429                        || pwrite(fd, &vec->front() + offset, vec->size() - offset, offset) == -1) {
 430                         fprintf(stderr, "error writing, errnor=%d\n", errno);
 431                         cumulativeResult    = -1;
 432                     }
 433                 }
 434
 435                 delete vec;
 436                 close(fd);
 437                 dispatch_semaphore_signal(sema);
 438             });
 439         });
 440         }
 441
 442     dispatch_group_wait(group, DISPATCH_TIME_FOREVER);
 443     dispatch_release(group);
 444     dispatch_release(writer_queue);
 445
 446     munmap(mapped_cache, statbuf.st_size);
 447         return cumulativeResult;
 448 }
 449
 450
 451
 452 int dyld_shared_cache_extract_dylibs(const char* shared_cache_file_path, const char* extraction_root_path)
 453 {
 454         return dyld_shared_cache_extract_dylibs_progress(shared_cache_file_path, extraction_root_path,
 455                                                                                                         ^(unsigned , unsigned) {} );
 456 }
 457
 458
 459 #if 0
 460
 461 typedef int (*extractor_proc)(const char* shared_cache_file_path, const char* extraction_root_path,
 462                                                                                                         void (^progress)(unsigned current, unsigned total));
 463
 464 int main(int argc, const char* argv[])
 465 {
 466         if ( argc != 3 ) {
 467                 fprintf(stderr, "usage: dsc_extractor <path-to-cache-file> <path-to-device-dir>\n");
 468                 return 1;
 469         }
 470
 471         void* handle = dlopen("/Developer/Platforms/iPhoneOS.platform/usr/lib/dsc_extractor.bundle", RTLD_LAZY);
 472         if ( handle == NULL ) {
 473                 fprintf(stderr, "dsc_extractor.bundle could not be loaded\n");
 474                 return 1;
 475         }
 476
 477         extractor_proc proc = (extractor_proc)dlsym(handle, "dyld_shared_cache_extract_dylibs_progress");
 478         if ( proc == NULL ) {
 479                 fprintf(stderr, "dsc_extractor.bundle did not have dyld_shared_cache_extract_dylibs_progress symbol\n");
 480                 return 1;
 481         }
 482
 483         int result = (*proc)(argv[1], argv[2], ^(unsigned c, unsigned total) { printf("%d/%d\n", c, total); } );
 484         fprintf(stderr, "dyld_shared_cache_extract_dylibs_progress() => %d\n", result);
 485         return 0;
 486 }
 487 #endif
 488
 489
 490
 491