launch-cache/MachOBinder.hpp

   1 /* -*- mode: C++; c-basic-offset: 4; tab-width: 4 -*-
   2  *
   3  * Copyright (c) 2006-2011 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 #ifndef __MACHO_BINDER__
  26 #define __MACHO_BINDER__
  27
  28 #include <sys/types.h>
  29 #include <sys/stat.h>
  30 #include <sys/mman.h>
  31 #include <mach/mach.h>
  32 #include <limits.h>
  33 #include <stdarg.h>
  34 #include <stdio.h>
  35 #include <fcntl.h>
  36 #include <errno.h>
  37 #include <unistd.h>
  38 #include <mach-o/loader.h>
  39 #include <mach-o/fat.h>
  40
  41 #include <vector>
  42 #include <set>
  43 #include <unordered_map>
  44 #include <unordered_set>
  45
  46 #include "MachOFileAbstraction.hpp"
  47 #include "Architectures.hpp"
  48 #include "MachOLayout.hpp"
  49 #include "MachORebaser.hpp"
  50 #include "MachOTrie.hpp"
  51
  52 #ifndef EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER
  53         #define EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER 0x10
  54 #endif
  55
  56 #ifndef EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE
  57         #define EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE 0x02
  58 #endif
  59
  60
  61 template <typename A>
  62 class Binder : public Rebaser<A>
  63 {
  64 public:
  65         class CStringHash {
  66         public:
  67                 size_t operator()(const char* __s) const {
  68                         size_t __h = 0;
  69                         for ( ; *__s; ++__s)
  70                                 __h = 5 * __h + *__s;
  71                         return __h;
  72                 };
  73         };
  74         struct CStringEquals {
  75                 bool operator()(const char* left, const char* right) const { return (strcmp(left, right) == 0); }
  76         };
  77         typedef std::unordered_map<const char*, class Binder<A>*, CStringHash, CStringEquals> Map;
  78
  79
  80                                                                                                 Binder(const MachOLayoutAbstraction&, uint64_t dyldBaseAddress);
  81         virtual                                                                         ~Binder() {}
  82
  83         const char*                                                                     getDylibID() const;
  84         void                                                                            setDependentBinders(const Map& map);
  85         void                                                                            bind(std::vector<void*>&);
  86         void                                                                            optimize();
  87     void                                        addResolverClient(Binder<A>* clientDylib, const char* symbolName);
  88 private:
  89         typedef typename A::P                                   P;
  90         typedef typename A::P::E                                E;
  91         typedef typename A::P::uint_t                   pint_t;
  92         struct BinderAndReExportFlag { Binder<A>* binder; bool reExport; };
  93         struct SymbolReExport { const char* exportName; int dylibOrdinal; const char* importName; };
  94         typedef std::unordered_map<const char*, pint_t, CStringHash, CStringEquals> NameToAddrMap;
  95         typedef std::unordered_set<const char*, CStringHash, CStringEquals> NameSet;
  96     struct ClientAndSymbol { Binder<A>* client; const char* symbolName; };
  97     struct SymbolAndLazyPointer { const char* symbolName; pint_t lpVMAddr; };
  98
  99         static bool                                                                     isPublicLocation(const char* pth);
 100         void                                                                            doBindExternalRelocations();
 101         void                                                                            doBindIndirectSymbols();
 102         void                                                                            doSetUpDyldSection();
 103         void                                                                            doSetPreboundUndefines();
 104         void                                                                            hoistPrivateRexports();
 105         int                                                                                     ordinalOfDependentBinder(Binder<A>* dep);
 106         void                                                                            doBindDyldInfo(std::vector<void*>& pointersInData);
 107         void                                                                            doBindDyldLazyInfo(std::vector<void*>& pointersInData);
 108         void                                                                            bindDyldInfoAt(uint8_t segmentIndex, uint64_t segmentOffset, uint8_t type,
 109                                                                                                                                 int libraryOrdinal, int64_t addend,
 110                                                                                                                                 const char* symbolName, bool lazyPointer, bool weakImport,
 111                                                                                                                                 std::vector<void*>& pointersInData);
 112         pint_t                                                                          resolveUndefined(const macho_nlist<P>* undefinedSymbol);
 113         bool                                                                            findExportedSymbolAddress(const char* name, pint_t* result, Binder<A>** foundIn,
 114                                                                                                                                                         bool* isResolverSymbol, bool* isAbsolute);
 115         void                                                                            bindStub(uint8_t elementSize, uint8_t* location, pint_t vmlocation, pint_t value);
 116         const char*                                                                     parentUmbrella();
 117         pint_t                                                                          runtimeAddressFromNList(const macho_nlist<P>* sym);
 118     void                                        optimizeStub(const char* symbolName, pint_t lpVMAddr);
 119     void                                        optimizeStub(uint8_t* stubMappedAddress, pint_t stubVMAddress, uint32_t stubSize, pint_t lpVMAddr);
 120     pint_t                                      findLazyPointerFor(const char* symbolName);
 121
 122
 123         static uint8_t                                                          pointerRelocSize();
 124         static uint8_t                                                          pointerRelocType();
 125
 126         std::vector<BinderAndReExportFlag>                      fDependentDylibs;
 127         NameToAddrMap                                                           fHashTable;
 128         NameSet                                                                         fSymbolResolvers;
 129         NameSet                                                                         fAbsoluteSymbols;
 130         std::vector<SymbolReExport>                                     fReExportedSymbols;
 131         uint64_t                                                                        fDyldBaseAddress;
 132         const macho_nlist<P>*                                           fSymbolTable;
 133         const char*                                                                     fStrings;
 134         const macho_dysymtab_command<P>*                        fDynamicInfo;
 135         const macho_segment_command<P>*                         fFristWritableSegment;
 136         const macho_dylib_command<P>*                           fDylibID;
 137         const macho_dylib_command<P>*                           fParentUmbrella;
 138         const macho_dyld_info_command<P>*                       fDyldInfo;
 139         bool                                                                            fOriginallyPrebound;
 140         bool                                                                            fReExportedSymbolsResolved;
 141     std::vector<ClientAndSymbol>                fClientAndSymbols;
 142         NameToAddrMap                                                           fResolverLazyPointers;
 143 };
 144
 145 template <>
 146 uint32_t Binder<arm>::runtimeAddressFromNList(const macho_nlist<Pointer32<LittleEndian> >* sym)
 147 {
 148         if (sym->n_desc() & N_ARM_THUMB_DEF)
 149                 return sym->n_value() + 1;
 150         else
 151                 return sym->n_value();
 152 }
 153
 154 template <typename A>
 155 typename A::P::uint_t   Binder<A>::runtimeAddressFromNList(const macho_nlist<P>* sym)
 156 {
 157         return sym->n_value();
 158 }
 159
 160
 161 template <typename A>
 162 Binder<A>::Binder(const MachOLayoutAbstraction& layout, uint64_t dyldBaseAddress)
 163         : Rebaser<A>(layout), fDyldBaseAddress(dyldBaseAddress),
 164           fSymbolTable(NULL), fStrings(NULL), fDynamicInfo(NULL),
 165           fFristWritableSegment(NULL), fDylibID(NULL), fDyldInfo(NULL),
 166           fParentUmbrella(NULL), fReExportedSymbolsResolved(false)
 167 {
 168         fOriginallyPrebound = ((this->fHeader->flags() & MH_PREBOUND) != 0);
 169         // update header flags so the cache looks prebound split-seg (0x80000000 is in-shared-cache bit)
 170         ((macho_header<P>*)this->fHeader)->set_flags(this->fHeader->flags() | MH_PREBOUND | MH_SPLIT_SEGS | 0x80000000);
 171
 172         // calculate fDynamicInfo, fStrings, fSymbolTable
 173         const macho_symtab_command<P>* symtab;
 174         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
 175         const uint32_t cmd_count = this->fHeader->ncmds();
 176         const macho_load_command<P>* cmd = cmds;
 177         for (uint32_t i = 0; i < cmd_count; ++i) {
 178                 switch (cmd->cmd()) {
 179                         case LC_SYMTAB:
 180                                 symtab = (macho_symtab_command<P>*)cmd;
 181                                 fSymbolTable = (macho_nlist<P>*)(&this->fLinkEditBase[symtab->symoff()]);
 182                                 fStrings = (const char*)&this->fLinkEditBase[symtab->stroff()];
 183                                 break;
 184                         case LC_DYSYMTAB:
 185                                 fDynamicInfo = (macho_dysymtab_command<P>*)cmd;
 186                                 break;
 187                         case LC_ID_DYLIB:
 188                                 ((macho_dylib_command<P>*)cmd)->set_timestamp(0);
 189                                 fDylibID = (macho_dylib_command<P>*)cmd;
 190                                 break;
 191                         case LC_LOAD_DYLIB:
 192                         case LC_LOAD_WEAK_DYLIB:
 193                         case LC_REEXPORT_DYLIB:
 194                         case LC_LOAD_UPWARD_DYLIB:
 195                                 ((macho_dylib_command<P>*)cmd)->set_timestamp(0);
 196                                 break;
 197                         case LC_SUB_FRAMEWORK:
 198                                 fParentUmbrella = (macho_dylib_command<P>*)cmd;
 199                                 break;
 200                         case LC_DYLD_INFO:
 201                         case LC_DYLD_INFO_ONLY:
 202                                 fDyldInfo = (macho_dyld_info_command<P>*)cmd;
 203                                 break;
 204                         case LC_RPATH:
 205                                 throwf("dyld shared cache does not support LC_RPATH found in %s", layout.getFilePath());
 206                                 break;
 207                         default:
 208                                 if ( cmd->cmd() & LC_REQ_DYLD )
 209                                         throwf("unknown required load command 0x%08X", cmd->cmd());
 210                 }
 211                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 212         }
 213         if ( fDynamicInfo == NULL )
 214                 throw "no LC_DYSYMTAB";
 215         if ( fSymbolTable == NULL )
 216                 throw "no LC_SYMTAB";
 217         // build hash table
 218 //      fprintf(stderr, "exports for %s\n", layout.getFilePath());
 219         if ( fDyldInfo != NULL ) {
 220                 std::vector<mach_o::trie::Entry> exports;
 221                 const uint8_t* exportsStart = layout.getDyldInfoExports();
 222                 const uint8_t* exportsEnd = &exportsStart[fDyldInfo->export_size()];
 223                 mach_o::trie::parseTrie(exportsStart, exportsEnd, exports);
 224                 pint_t baseAddress = layout.getSegments()[0].newAddress();
 225                 for(std::vector<mach_o::trie::Entry>::iterator it = exports.begin(); it != exports.end(); ++it) {
 226                         switch ( it->flags & EXPORT_SYMBOL_FLAGS_KIND_MASK ) {
 227                                 case EXPORT_SYMBOL_FLAGS_KIND_REGULAR:
 228                                         if ( (it->flags & EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER) ) {
 229                                                 fSymbolResolvers.insert(it->name);
 230                                         }
 231                                         if ( it->flags & EXPORT_SYMBOL_FLAGS_REEXPORT ) {
 232                                                 //fprintf(stderr, "found re-export %s in %s\n", sym.exportName, this->getDylibID());
 233                                                 SymbolReExport sym;
 234                                                 sym.exportName = it->name;
 235                                                 sym.dylibOrdinal = it->other;
 236                                                 sym.importName = it->importName;
 237                                                 if ( (sym.importName == NULL) || (sym.importName[0] == '\0') )
 238                                                         sym.importName = sym.exportName;
 239                                                 fReExportedSymbols.push_back(sym);
 240                                                 // fHashTable entry will be added in first call to findExportedSymbolAddress()
 241                                         }
 242                                         else {
 243                                                 fHashTable[it->name] = it->address + baseAddress;
 244                                         }
 245                                         break;
 246                                 case EXPORT_SYMBOL_FLAGS_KIND_THREAD_LOCAL:
 247                                         fHashTable[it->name] = it->address + baseAddress;
 248                                         break;
 249                                 case EXPORT_SYMBOL_FLAGS_KIND_ABSOLUTE:
 250                                         fHashTable[it->name] = it->address;
 251                                         fAbsoluteSymbols.insert(it->name);
 252                                         break;
 253                                 default:
 254                                         throwf("non-regular symbol binding not supported for %s in %s", it->name, layout.getFilePath());
 255                                         break;
 256                         }
 257                         //fprintf(stderr, "0x%08llX %s\n", it->address + baseAddress, it->name);
 258                 }
 259         }
 260         else {
 261                 if ( fDynamicInfo->tocoff() == 0 ) {
 262                         const macho_nlist<P>* start = &fSymbolTable[fDynamicInfo->iextdefsym()];
 263                         const macho_nlist<P>* end = &start[fDynamicInfo->nextdefsym()];
 264                         fHashTable.reserve(fDynamicInfo->nextdefsym()); // set initial bucket count
 265                         for (const macho_nlist<P>* sym=start; sym < end; ++sym) {
 266                                 const char* name = &fStrings[sym->n_strx()];
 267                                 fHashTable[name] = runtimeAddressFromNList(sym);
 268                                 //fprintf(stderr, " 0x%08llX %s\n", sym->n_value(), name);
 269                         }
 270                 }
 271                 else {
 272                         int32_t count = fDynamicInfo->ntoc();
 273                         fHashTable.reserve(count); // set initial bucket count
 274                         const struct dylib_table_of_contents* toc = (dylib_table_of_contents*)&this->fLinkEditBase[fDynamicInfo->tocoff()];
 275                         for (int32_t i = 0; i < count; ++i) {
 276                                 const uint32_t index = E::get32(toc[i].symbol_index);
 277                                 const macho_nlist<P>* sym = &fSymbolTable[index];
 278                                 const char* name = &fStrings[sym->n_strx()];
 279                                 fHashTable[name] = runtimeAddressFromNList(sym);
 280                                 //fprintf(stderr, "- 0x%08llX %s\n", sym->n_value(), name);
 281                         }
 282                 }
 283         }
 284 }
 285
 286 template <> uint8_t     Binder<x86>::pointerRelocSize()   { return 2; }
 287 template <> uint8_t     Binder<x86_64>::pointerRelocSize() { return 3; }
 288 template <> uint8_t     Binder<arm>::pointerRelocSize() { return 2; }
 289 template <> uint8_t     Binder<arm64>::pointerRelocSize() { return 3; }
 290
 291 template <> uint8_t     Binder<x86>::pointerRelocType()   { return GENERIC_RELOC_VANILLA; }
 292 template <> uint8_t     Binder<x86_64>::pointerRelocType() { return X86_64_RELOC_UNSIGNED; }
 293 template <> uint8_t     Binder<arm>::pointerRelocType() { return ARM_RELOC_VANILLA; }
 294 template <> uint8_t     Binder<arm64>::pointerRelocType() { return ARM64_RELOC_UNSIGNED; }
 295
 296
 297 template <typename A>
 298 const char* Binder<A>::getDylibID() const
 299 {
 300         if ( fDylibID != NULL )
 301                 return fDylibID->name();
 302         else
 303                 return NULL;
 304 }
 305
 306 template <typename A>
 307 const char* Binder<A>::parentUmbrella()
 308 {
 309         if ( fParentUmbrella != NULL )
 310                 return fParentUmbrella->name();
 311         else
 312                 return NULL;
 313 }
 314
 315
 316 template <typename A>
 317 bool Binder<A>::isPublicLocation(const char* pth)
 318 {
 319         // /usr/lib is a public location
 320         if ( (strncmp(pth, "/usr/lib/", 9) == 0) && (strchr(&pth[9], '/') == NULL) )
 321                 return true;
 322
 323         // /System/Library/Frameworks/ is a public location
 324         if ( strncmp(pth, "/System/Library/Frameworks/", 27) == 0 ) {
 325                 const char* frameworkDot = strchr(&pth[27], '.');
 326                 // but only top level framework
 327                 // /System/Library/Frameworks/Foo.framework/Versions/A/Foo                 ==> true
 328                 // /System/Library/Frameworks/Foo.framework/Resources/libBar.dylib         ==> false
 329                 // /System/Library/Frameworks/Foo.framework/Frameworks/Bar.framework/Bar   ==> false
 330                 // /System/Library/Frameworks/Foo.framework/Frameworks/Xfoo.framework/XFoo ==> false
 331                 if ( frameworkDot != NULL ) {
 332                         int frameworkNameLen = frameworkDot - &pth[27];
 333                         if ( strncmp(&pth[strlen(pth)-frameworkNameLen-1], &pth[26], frameworkNameLen+1) == 0 )
 334                                 return true;
 335                 }
 336         }
 337
 338         return false;
 339 }
 340
 341 template <typename A>
 342 void Binder<A>::setDependentBinders(const Map& map)
 343 {
 344         // first pass to build vector of dylibs
 345         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
 346         const uint32_t cmd_count = this->fHeader->ncmds();
 347         const macho_load_command<P>* cmd = cmds;
 348         for (uint32_t i = 0; i < cmd_count; ++i) {
 349                 switch (cmd->cmd()) {
 350                         case LC_LOAD_DYLIB:
 351                         case LC_LOAD_WEAK_DYLIB:
 352                         case LC_REEXPORT_DYLIB:
 353                         case LC_LOAD_UPWARD_DYLIB:
 354                                 const char* path = ((struct macho_dylib_command<P>*)cmd)->name();
 355                                 typename Map::const_iterator pos = map.find(path);
 356                                 if ( pos != map.end() ) {
 357                                         BinderAndReExportFlag entry;
 358                                         entry.binder = pos->second;
 359                                         entry.reExport = ( cmd->cmd() == LC_REEXPORT_DYLIB );
 360                                         fDependentDylibs.push_back(entry);
 361                                 }
 362                                 else {
 363                                         // the load command string does not match the install name of any loaded dylib
 364                                         // this could happen if there was not a world build and some dylib changed its
 365                                         // install path to be some symlinked path
 366
 367                                         // use realpath() and walk map looking for a realpath match
 368                                         bool found = false;
 369                                         char targetPath[PATH_MAX];
 370                                         if ( realpath(path, targetPath) != NULL ) {
 371                                                 for(typename Map::const_iterator it=map.begin(); it != map.end(); ++it) {
 372                                                         char aPath[PATH_MAX];
 373                                                         if ( realpath(it->first, aPath) != NULL ) {
 374                                                                 if ( strcmp(targetPath, aPath) == 0 ) {
 375                                                                         BinderAndReExportFlag entry;
 376                                                                         entry.binder = it->second;
 377                                                                         entry.reExport = ( cmd->cmd() == LC_REEXPORT_DYLIB );
 378                                                                         fDependentDylibs.push_back(entry);
 379                                                                         found = true;
 380                                                                         fprintf(stderr, "update_dyld_shared_cache: warning mismatched install path in %s for %s\n",
 381                                                                                 this->getDylibID(), path);
 382                                                                         break;
 383                                                                 }
 384                                                         }
 385                                                 }
 386                                         }
 387                                         if ( ! found ) {
 388                                                 if ( cmd->cmd() == LC_LOAD_WEAK_DYLIB ) {
 389                                                         BinderAndReExportFlag entry;
 390                                                         entry.binder = NULL;
 391                                                         entry.reExport = false;
 392                                                         fDependentDylibs.push_back(entry);
 393                                                         break;
 394                                                 }
 395                                                 else {
 396                                                         throwf("in %s can't find dylib %s", this->getDylibID(), path);
 397                                                 }
 398                                         }
 399                                 }
 400                                 break;
 401                 }
 402                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 403         }
 404         // handle pre-10.5 re-exports
 405         if ( (this->fHeader->flags() & MH_NO_REEXPORTED_DYLIBS) == 0 ) {
 406                 cmd = cmds;
 407                 // LC_SUB_LIBRARY means re-export one with matching leaf name
 408                 const char* dylibBaseName;
 409                 const char* frameworkLeafName;
 410                 for (uint32_t i = 0; i < cmd_count; ++i) {
 411                         switch ( cmd->cmd() ) {
 412                                 case LC_SUB_LIBRARY:
 413                                         dylibBaseName = ((macho_sub_library_command<P>*)cmd)->sub_library();
 414                                         for (typename std::vector<BinderAndReExportFlag>::iterator it = fDependentDylibs.begin(); it != fDependentDylibs.end(); ++it) {
 415                                                 const char* dylibName = it->binder->getDylibID();
 416                                                 const char* lastSlash = strrchr(dylibName, '/');
 417                                                 const char* leafStart = &lastSlash[1];
 418                                                 if ( lastSlash == NULL )
 419                                                         leafStart = dylibName;
 420                                                 const char* firstDot = strchr(leafStart, '.');
 421                                                 int len = strlen(leafStart);
 422                                                 if ( firstDot != NULL )
 423                                                         len = firstDot - leafStart;
 424                                                 if ( strncmp(leafStart, dylibBaseName, len) == 0 )
 425                                                         it->reExport = true;
 426                                         }
 427                                         break;
 428                                 case LC_SUB_UMBRELLA:
 429                                         frameworkLeafName = ((macho_sub_umbrella_command<P>*)cmd)->sub_umbrella();
 430                                         for (typename std::vector<BinderAndReExportFlag>::iterator it = fDependentDylibs.begin(); it != fDependentDylibs.end(); ++it) {
 431                                                 const char* dylibName = it->binder->getDylibID();
 432                                                 const char* lastSlash = strrchr(dylibName, '/');
 433                                                 if ( (lastSlash != NULL) && (strcmp(&lastSlash[1], frameworkLeafName) == 0) )
 434                                                         it->reExport = true;
 435                                         }
 436                                         break;
 437                         }
 438                         cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 439                 }
 440                 // ask dependents if they re-export through me
 441                 const char* thisName = this->getDylibID();
 442                 if ( thisName != NULL ) {
 443                         const char* thisLeafName = strrchr(thisName, '/');
 444                         if ( thisLeafName != NULL )
 445                                 ++thisLeafName;
 446                         for (typename std::vector<BinderAndReExportFlag>::iterator it = fDependentDylibs.begin(); it != fDependentDylibs.end(); ++it) {
 447                                 if ( ! it->reExport ) {
 448                                         Binder<A>* dep = it->binder;
 449                                         if ( dep != NULL ) {
 450                                                 const char* parentUmbrellaName = dep->parentUmbrella();
 451                                                 if ( parentUmbrellaName != NULL ) {
 452                                                         if ( strcmp(parentUmbrellaName, thisLeafName) == 0 )
 453                                                                 it->reExport = true;
 454                                                 }
 455                                         }
 456                                 }
 457                         }
 458                 }
 459         }
 460
 461 }
 462
 463 template <typename A>
 464 int Binder<A>::ordinalOfDependentBinder(Binder<A>* dep)
 465 {
 466         for (int i=0; i < fDependentDylibs.size(); ++i) {
 467                 if ( fDependentDylibs[i].binder == dep )
 468                         return i+1;
 469         }
 470         throw "dependend dylib not found";
 471 }
 472
 473 template <typename A>
 474 void Binder<A>::hoistPrivateRexports()
 475 {
 476         std::vector<Binder<A>*> privateReExportedDylibs;
 477         for (typename std::vector<BinderAndReExportFlag>::iterator it = fDependentDylibs.begin(); it != fDependentDylibs.end(); ++it) {
 478                 if ( it->reExport && ! isPublicLocation(it->binder->getDylibID()) )
 479                         privateReExportedDylibs.push_back(it->binder);
 480         }
 481         if ( privateReExportedDylibs.size() != 0 ) {
 482                 // parse export info into vector of exports
 483                 const uint8_t* exportsStart = this->fLayout.getDyldInfoExports();
 484                 const uint8_t* exportsEnd = &exportsStart[fDyldInfo->export_size()];
 485                 std::vector<mach_o::trie::Entry> exports;
 486                 mach_o::trie::parseTrie(exportsStart, exportsEnd, exports);
 487                 //fprintf(stderr, "%s exports %lu symbols from trie of size %u \n", this->fLayout.getFilePath(), exports.size(), fDyldInfo->export_size());
 488
 489                 // add re-exports for each export from an re-exported dylib
 490                 for(typename std::vector<Binder<A>*>::iterator it = privateReExportedDylibs.begin(); it != privateReExportedDylibs.end(); ++it) {
 491                         Binder<A>* binder = *it;
 492                         int ordinal = ordinalOfDependentBinder(binder);
 493                         const uint8_t* aDylibsExportsStart = binder->fLayout.getDyldInfoExports();
 494                         const uint8_t* aDylibsExportsEnd = &aDylibsExportsStart[binder->fDyldInfo->export_size()];
 495                         std::vector<mach_o::trie::Entry> aDylibsExports;
 496                         mach_o::trie::parseTrie(aDylibsExportsStart, aDylibsExportsEnd, aDylibsExports);
 497                         //fprintf(stderr, "%s re-exports %lu symbols from %s\n", this->fLayout.getFilePath(), aDylibsExports.size(), binder->getDylibID());
 498                         for(std::vector<mach_o::trie::Entry>::iterator eit = aDylibsExports.begin(); eit != aDylibsExports.end(); ++eit) {
 499                                 mach_o::trie::Entry entry = *eit;
 500                                 entry.flags |= EXPORT_SYMBOL_FLAGS_REEXPORT;
 501                                 entry.other = ordinal;
 502                                 entry.importName = NULL;
 503                                 exports.push_back(entry);
 504                         }
 505                 }
 506                 // rebuild new combined trie
 507                 std::vector<uint8_t> newExportTrieBytes;
 508                 newExportTrieBytes.reserve(fDyldInfo->export_size());
 509                 mach_o::trie::makeTrie(exports, newExportTrieBytes);
 510                 //fprintf(stderr, "%s now exports %lu symbols from trie of size %lu\n", this->fLayout.getFilePath(), exports.size(), newExportTrieBytes.size());
 511
 512                 // allocate new buffer and set export_off to use new buffer instead
 513                 uint32_t newExportsSize = newExportTrieBytes.size();
 514                 uint8_t* sideTrie = new uint8_t[newExportsSize];
 515                 memcpy(sideTrie, &newExportTrieBytes[0], newExportsSize);
 516                 this->fLayout.setDyldInfoExports(sideTrie);
 517                 ((macho_dyld_info_command<P>*)fDyldInfo)->set_export_off(0); // invalidate old trie
 518                 ((macho_dyld_info_command<P>*)fDyldInfo)->set_export_size(newExportsSize);
 519         }
 520 }
 521
 522
 523 template <typename A>
 524 void Binder<A>::bind(std::vector<void*>& pointersInData)
 525 {
 526         this->doSetUpDyldSection();
 527         if ( fDyldInfo != NULL ) {
 528                 this->doBindDyldInfo(pointersInData);
 529                 this->doBindDyldLazyInfo(pointersInData);
 530                 this->hoistPrivateRexports();
 531                 // weak bind info is processed at launch time
 532         }
 533         else {
 534                 this->doBindExternalRelocations();
 535                 this->doBindIndirectSymbols();
 536                 this->doSetPreboundUndefines();
 537         }
 538 }
 539
 540
 541 template <typename A>
 542 void Binder<A>::doSetUpDyldSection()
 543 {
 544         // find __DATA __dyld section
 545         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
 546         const uint32_t cmd_count = this->fHeader->ncmds();
 547         const macho_load_command<P>* cmd = cmds;
 548         for (uint32_t i = 0; i < cmd_count; ++i) {
 549                 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
 550                         const macho_segment_command<P>* seg = (macho_segment_command<P>*)cmd;
 551                         if ( strcmp(seg->segname(), "__DATA") == 0 ) {
 552                                 const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)seg + sizeof(macho_segment_command<P>));
 553                                 const macho_section<P>* const sectionsEnd = &sectionsStart[seg->nsects()];
 554                                 for (const macho_section<P>* sect=sectionsStart; sect < sectionsEnd; ++sect) {
 555                                         if ( (strcmp(sect->sectname(), "__dyld") == 0) && (sect->size() >= 2*sizeof(pint_t)) ) {
 556                                                 // set two values in __dyld section to point into dyld
 557                                                 pint_t* lazyBinder = this->mappedAddressForNewAddress(sect->addr());
 558                                                 pint_t* dyldFuncLookup = this->mappedAddressForNewAddress(sect->addr()+sizeof(pint_t));
 559                                                 A::P::setP(*lazyBinder, fDyldBaseAddress + 0x1000);
 560                                                 A::P::setP(*dyldFuncLookup, fDyldBaseAddress + 0x1008);
 561                                         }
 562                                 }
 563                         }
 564                 }
 565                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 566         }
 567 }
 568
 569 template <typename A>
 570 void Binder<A>::bindDyldInfoAt(uint8_t segmentIndex, uint64_t segmentOffset, uint8_t type, int libraryOrdinal,
 571                                                         int64_t addend, const char* symbolName, bool lazyPointer, bool weakImport, std::vector<void*>& pointersInData)
 572 {
 573         //printf("%d 0x%08llX type=%d, lib=%d, addend=%lld, symbol=%s\n", segmentIndex, segmentOffset, type, libraryOrdinal, addend, symbolName);
 574         const std::vector<MachOLayoutAbstraction::Segment>& segments = this->fLayout.getSegments();
 575         if ( segmentIndex > segments.size() )
 576                 throw "bad segment index in rebase info";
 577
 578         if ( libraryOrdinal == BIND_SPECIAL_DYLIB_FLAT_LOOKUP )
 579                 throw "dynamic lookup linkage not allowed in dyld shared cache";
 580
 581         if ( libraryOrdinal == BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE )
 582                 throw "linkage to main executable not allowed in dyld shared cache";
 583
 584         if ( libraryOrdinal < 0 )
 585                 throw "bad mach-o binary, special library ordinal not allowd in dyld shared cache";
 586
 587         if ( (unsigned)libraryOrdinal > fDependentDylibs.size() )
 588                 throw "bad mach-o binary, library ordinal too big";
 589
 590         Binder<A>* binder;
 591         if ( libraryOrdinal == BIND_SPECIAL_DYLIB_SELF )
 592                 binder = this;
 593         else
 594                 binder = fDependentDylibs[libraryOrdinal-1].binder;
 595         pint_t targetSymbolAddress;
 596         bool isResolverSymbol = false;
 597         bool isAbsolute = false;
 598     Binder<A>* foundIn;
 599         if ( weakImport && (binder == NULL) ) {
 600                 targetSymbolAddress = 0;
 601                 foundIn = NULL;
 602         }
 603         else {
 604                 if ( ! binder->findExportedSymbolAddress(symbolName, &targetSymbolAddress, &foundIn, &isResolverSymbol, &isAbsolute) )
 605                         throwf("could not bind symbol %s in %s expected in %s", symbolName, this->getDylibID(), binder->getDylibID());
 606         }
 607
 608         // don't bind lazy pointers to resolver stubs in shared cache
 609         if ( lazyPointer && isResolverSymbol ) {
 610         if ( foundIn != this ) {
 611                         // record that this dylib has a lazy pointer to a resolver function
 612             foundIn->addResolverClient(this, symbolName);
 613            // fprintf(stderr, "have lazy pointer to resolver %s in %s\n", symbolName, this->getDylibID());
 614         }
 615                 return;
 616     }
 617
 618         // do actual update
 619         const MachOLayoutAbstraction::Segment& seg = segments[segmentIndex];
 620         uint8_t*  mappedAddr = (uint8_t*)seg.mappedAddress() + segmentOffset;
 621         pint_t*   mappedAddrP = (pint_t*)mappedAddr;
 622         uint32_t* mappedAddr32 = (uint32_t*)mappedAddr;
 623         int32_t svalue32new;
 624         switch ( type ) {
 625                 case BIND_TYPE_POINTER:
 626                         P::setP(*mappedAddrP, targetSymbolAddress + addend);
 627                         break;
 628
 629                 case BIND_TYPE_TEXT_ABSOLUTE32:
 630                         E::set32(*mappedAddr32, targetSymbolAddress + addend);
 631                         break;
 632
 633                 case BIND_TYPE_TEXT_PCREL32:
 634                         svalue32new = seg.address() + segmentOffset + 4 - (targetSymbolAddress + addend);
 635                         E::set32(*mappedAddr32, svalue32new);
 636                         break;
 637
 638                 default:
 639                         throw "bad bind type";
 640         }
 641         if ( !isAbsolute )
 642                 pointersInData.push_back(mappedAddr);
 643 }
 644
 645
 646
 647 template <typename A>
 648 void Binder<A>::doBindDyldLazyInfo(std::vector<void*>& pointersInData)
 649 {
 650         const uint8_t* p = &this->fLinkEditBase[fDyldInfo->lazy_bind_off()];
 651         const uint8_t* end = &p[fDyldInfo->lazy_bind_size()];
 652
 653         uint8_t type = BIND_TYPE_POINTER;
 654         uint64_t segmentOffset = 0;
 655         uint8_t segmentIndex = 0;
 656         const char* symbolName = NULL;
 657         int libraryOrdinal = 0;
 658         int64_t addend = 0;
 659         bool weakImport = false;
 660         while ( p < end ) {
 661                 uint8_t immediate = *p & BIND_IMMEDIATE_MASK;
 662                 uint8_t opcode = *p & BIND_OPCODE_MASK;
 663                 ++p;
 664                 switch (opcode) {
 665                         case BIND_OPCODE_DONE:
 666                                 // this opcode marks the end of each lazy pointer binding
 667                                 break;
 668                         case BIND_OPCODE_SET_DYLIB_ORDINAL_IMM:
 669                                 libraryOrdinal = immediate;
 670                                 break;
 671                         case BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB:
 672                                 libraryOrdinal = read_uleb128(p, end);
 673                                 break;
 674                         case BIND_OPCODE_SET_DYLIB_SPECIAL_IMM:
 675                                 // the special ordinals are negative numbers
 676                                 if ( immediate == 0 )
 677                                         libraryOrdinal = 0;
 678                                 else {
 679                                         int8_t signExtended = BIND_OPCODE_MASK | immediate;
 680                                         libraryOrdinal = signExtended;
 681                                 }
 682                                 break;
 683                         case BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM:
 684                                 weakImport = ( (immediate & BIND_SYMBOL_FLAGS_WEAK_IMPORT) != 0 );
 685                                 symbolName = (char*)p;
 686                                 while (*p != '\0')
 687                                         ++p;
 688                                 ++p;
 689                                 break;
 690                         case BIND_OPCODE_SET_ADDEND_SLEB:
 691                                 addend = read_sleb128(p, end);
 692                                 break;
 693                         case BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB:
 694                                 segmentIndex = immediate;
 695                                 segmentOffset = read_uleb128(p, end);
 696                                 break;
 697                         case BIND_OPCODE_DO_BIND:
 698                                 bindDyldInfoAt(segmentIndex, segmentOffset, type, libraryOrdinal, addend, symbolName, true, weakImport, pointersInData);
 699                                 segmentOffset += sizeof(pint_t);
 700                                 break;
 701                         case BIND_OPCODE_SET_TYPE_IMM:
 702                         case BIND_OPCODE_ADD_ADDR_ULEB:
 703                         case BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB:
 704                         case BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED:
 705                         case BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB:
 706                         default:
 707                                 throwf("bad lazy bind opcode %d", *p);
 708                 }
 709         }
 710
 711
 712 }
 713
 714 template <typename A>
 715 void Binder<A>::doBindDyldInfo(std::vector<void*>& pointersInData)
 716 {
 717         const uint8_t* p = &this->fLinkEditBase[fDyldInfo->bind_off()];
 718         const uint8_t* end = &p[fDyldInfo->bind_size()];
 719
 720         uint8_t type = 0;
 721         uint64_t segmentOffset = 0;
 722         uint8_t segmentIndex = 0;
 723         const char* symbolName = NULL;
 724         int libraryOrdinal = 0;
 725         int64_t addend = 0;
 726         uint32_t count;
 727         uint32_t skip;
 728         bool weakImport = false;
 729         bool done = false;
 730         while ( !done && (p < end) ) {
 731                 uint8_t immediate = *p & BIND_IMMEDIATE_MASK;
 732                 uint8_t opcode = *p & BIND_OPCODE_MASK;
 733                 ++p;
 734                 switch (opcode) {
 735                         case BIND_OPCODE_DONE:
 736                                 done = true;
 737                                 break;
 738                         case BIND_OPCODE_SET_DYLIB_ORDINAL_IMM:
 739                                 libraryOrdinal = immediate;
 740                                 break;
 741                         case BIND_OPCODE_SET_DYLIB_ORDINAL_ULEB:
 742                                 libraryOrdinal = read_uleb128(p, end);
 743                                 break;
 744                         case BIND_OPCODE_SET_DYLIB_SPECIAL_IMM:
 745                                 // the special ordinals are negative numbers
 746                                 if ( immediate == 0 )
 747                                         libraryOrdinal = 0;
 748                                 else {
 749                                         int8_t signExtended = BIND_OPCODE_MASK | immediate;
 750                                         libraryOrdinal = signExtended;
 751                                 }
 752                                 break;
 753                         case BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM:
 754                                 weakImport = ( (immediate & BIND_SYMBOL_FLAGS_WEAK_IMPORT) != 0 );
 755                                 symbolName = (char*)p;
 756                                 while (*p != '\0')
 757                                         ++p;
 758                                 ++p;
 759                                 break;
 760                         case BIND_OPCODE_SET_TYPE_IMM:
 761                                 type = immediate;
 762                                 break;
 763                         case BIND_OPCODE_SET_ADDEND_SLEB:
 764                                 addend = read_sleb128(p, end);
 765                                 break;
 766                         case BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB:
 767                                 segmentIndex = immediate;
 768                                 segmentOffset = read_uleb128(p, end);
 769                                 break;
 770                         case BIND_OPCODE_ADD_ADDR_ULEB:
 771                                 segmentOffset += read_uleb128(p, end);
 772                                 break;
 773                         case BIND_OPCODE_DO_BIND:
 774                                 bindDyldInfoAt(segmentIndex, segmentOffset, type, libraryOrdinal, addend, symbolName, false, weakImport, pointersInData);
 775                                 segmentOffset += sizeof(pint_t);
 776                                 break;
 777                         case BIND_OPCODE_DO_BIND_ADD_ADDR_ULEB:
 778                                 bindDyldInfoAt(segmentIndex, segmentOffset, type, libraryOrdinal, addend, symbolName, false, weakImport, pointersInData);
 779                                 segmentOffset += read_uleb128(p, end) + sizeof(pint_t);
 780                                 break;
 781                         case BIND_OPCODE_DO_BIND_ADD_ADDR_IMM_SCALED:
 782                                 bindDyldInfoAt(segmentIndex, segmentOffset, type, libraryOrdinal, addend, symbolName, false, weakImport, pointersInData);
 783                                 segmentOffset += immediate*sizeof(pint_t) + sizeof(pint_t);
 784                                 break;
 785                         case BIND_OPCODE_DO_BIND_ULEB_TIMES_SKIPPING_ULEB:
 786                                 count = read_uleb128(p, end);
 787                                 skip = read_uleb128(p, end);
 788                                 for (uint32_t i=0; i < count; ++i) {
 789                                         bindDyldInfoAt(segmentIndex, segmentOffset, type, libraryOrdinal, addend, symbolName, false, weakImport, pointersInData);
 790                                         segmentOffset += skip + sizeof(pint_t);
 791                                 }
 792                                 break;
 793                         default:
 794                                 throwf("bad bind opcode %d", *p);
 795                 }
 796         }
 797
 798
 799
 800 }
 801
 802
 803 template <typename A>
 804 void Binder<A>::doSetPreboundUndefines()
 805 {
 806         const macho_dysymtab_command<P>* dysymtab = NULL;
 807         macho_nlist<P>* symbolTable = NULL;
 808
 809         // get symbol table info
 810         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
 811         const uint32_t cmd_count = this->fHeader->ncmds();
 812         const macho_load_command<P>* cmd = cmds;
 813         for (uint32_t i = 0; i < cmd_count; ++i) {
 814                 switch (cmd->cmd()) {
 815                         case LC_SYMTAB:
 816                                 {
 817                                         const macho_symtab_command<P>* symtab = (macho_symtab_command<P>*)cmd;
 818                                         symbolTable = (macho_nlist<P>*)(&this->fLinkEditBase[symtab->symoff()]);
 819                                 }
 820                                 break;
 821                         case LC_DYSYMTAB:
 822                                 dysymtab = (macho_dysymtab_command<P>*)cmd;
 823                                 break;
 824                 }
 825                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 826         }
 827
 828         // walk all undefines and set their prebound n_value
 829         macho_nlist<P>* const lastUndefine = &symbolTable[dysymtab->iundefsym()+dysymtab->nundefsym()];
 830         for (macho_nlist<P>* entry = &symbolTable[dysymtab->iundefsym()]; entry < lastUndefine; ++entry) {
 831                 if ( entry->n_type() & N_EXT ) {
 832                         //fprintf(stderr, "doSetPreboundUndefines: r_sym=%s, pbaddr=0x%08X, in %s\n",
 833                         //      &fStrings[entry->n_strx()], pbaddr, this->getDylibID());
 834                         pint_t pbaddr = this->resolveUndefined(entry);
 835                         entry->set_n_value(pbaddr);
 836                 }
 837         }
 838 }
 839
 840
 841 template <typename A>
 842 void Binder<A>::doBindExternalRelocations()
 843 {
 844         // get where reloc addresses start
 845         // these address are always relative to first writable segment because they are in cache which always
 846         // has writable segments far from read-only segments
 847         pint_t firstWritableSegmentBaseAddress = 0;
 848         const std::vector<MachOLayoutAbstraction::Segment>& segments = this->fLayout.getSegments();
 849         for(std::vector<MachOLayoutAbstraction::Segment>::const_iterator it = segments.begin(); it != segments.end(); ++it) {
 850                 const MachOLayoutAbstraction::Segment& seg = *it;
 851                 if ( seg.writable() ) {
 852                         firstWritableSegmentBaseAddress = seg.newAddress();
 853                         break;
 854                 }
 855         }
 856
 857         // loop through all external relocation records and bind each
 858         const macho_relocation_info<P>* const relocsStart = (macho_relocation_info<P>*)(&this->fLinkEditBase[fDynamicInfo->extreloff()]);
 859         const macho_relocation_info<P>* const relocsEnd = &relocsStart[fDynamicInfo->nextrel()];
 860         for (const macho_relocation_info<P>* reloc=relocsStart; reloc < relocsEnd; ++reloc) {
 861                 if ( reloc->r_length() != pointerRelocSize() )
 862                         throw "bad external relocation length";
 863                 if ( reloc->r_type() != pointerRelocType() )
 864                         throw "unknown external relocation type";
 865                 if ( reloc->r_pcrel() )
 866                         throw "r_pcrel external relocaiton not supported";
 867
 868                 const macho_nlist<P>* undefinedSymbol = &fSymbolTable[reloc->r_symbolnum()];
 869                 pint_t* location;
 870                 try {
 871                         location = this->mappedAddressForNewAddress(reloc->r_address() + firstWritableSegmentBaseAddress);
 872                 }
 873                 catch (const char* msg) {
 874                         throwf("%s processesing external relocation r_address 0x%08X", msg, reloc->r_address());
 875                 }
 876                 pint_t addend =  P::getP(*location);
 877                 if ( fOriginallyPrebound ) {
 878                         // in a prebound binary, the n_value field of an undefined symbol is set to the address where the symbol was found when prebound
 879                         // so, subtracting that gives the initial displacement which we need to add to the newly found symbol address
 880                         // if mach-o relocation structs had an "addend" field this complication would not be necessary.
 881                         addend -= undefinedSymbol->n_value();
 882                         // To further complicate things, if this is defined symbol, then its n_value has already been adjust to the
 883                         // new base address, so we need to back off the slide too..
 884                         if ( (undefinedSymbol->n_type() & N_TYPE) == N_SECT ) {
 885                                 addend += this->getSlideForNewAddress(undefinedSymbol->n_value());
 886                         }
 887                 }
 888                 pint_t symbolAddr = this->resolveUndefined(undefinedSymbol);
 889                 //fprintf(stderr, "external reloc: r_address=0x%08X, r_sym=%s, symAddr=0x%08llX, addend=0x%08llX in %s\n",
 890                 //              reloc->r_address(), &fStrings[undefinedSymbol->n_strx()], (uint64_t)symbolAddr, (uint64_t)addend, this->getDylibID());
 891                 P::setP(*location, symbolAddr + addend);
 892         }
 893 }
 894
 895
 896 // most architectures use pure code, unmodifiable stubs
 897 template <typename A>
 898 void Binder<A>::bindStub(uint8_t elementSize, uint8_t* location, pint_t vmlocation, pint_t value)
 899 {
 900         // do nothing
 901 }
 902
 903 // x86 supports fast stubs
 904 template <>
 905 void Binder<x86>::bindStub(uint8_t elementSize, uint8_t* location, pint_t vmlocation, pint_t value)
 906 {
 907         // if the stub is not 5-bytes, it is an old slow stub
 908         if ( elementSize == 5 ) {
 909                 uint32_t rel32 = value - (vmlocation + 5);
 910                 location[0] = 0xE9; // JMP rel32
 911                 location[1] = rel32 & 0xFF;
 912                 location[2] = (rel32 >> 8) & 0xFF;
 913                 location[3] = (rel32 >> 16) & 0xFF;
 914                 location[4] = (rel32 >> 24) & 0xFF;
 915         }
 916 }
 917
 918 template <typename A>
 919 void Binder<A>::doBindIndirectSymbols()
 920 {
 921         const uint32_t* const indirectTable = (uint32_t*)&this->fLinkEditBase[fDynamicInfo->indirectsymoff()];
 922         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
 923         const uint32_t cmd_count = this->fHeader->ncmds();
 924         const macho_load_command<P>* cmd = cmds;
 925         //fprintf(stderr, "doBindIndirectSymbols() %s\n", this->fLayout.getFilePath());
 926         for (uint32_t i = 0; i < cmd_count; ++i) {
 927                 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
 928                         const macho_segment_command<P>* seg = (macho_segment_command<P>*)cmd;
 929                         const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)seg + sizeof(macho_segment_command<P>));
 930                         const macho_section<P>* const sectionsEnd = &sectionsStart[seg->nsects()];
 931                         for (const macho_section<P>* sect=sectionsStart; sect < sectionsEnd; ++sect) {
 932                                 uint8_t elementSize = 0;
 933                                 uint8_t sectionType = sect->flags() & SECTION_TYPE;
 934                                 switch ( sectionType ) {
 935                                         case S_SYMBOL_STUBS:
 936                                                 elementSize = sect->reserved2();
 937                                                 break;
 938                                         case S_NON_LAZY_SYMBOL_POINTERS:
 939                                         case S_LAZY_SYMBOL_POINTERS:
 940                                                 elementSize = sizeof(pint_t);
 941                                                 break;
 942                                 }
 943                                 if ( elementSize != 0 ) {
 944                                         uint32_t elementCount = sect->size() / elementSize;
 945                                         const uint32_t indirectTableOffset = sect->reserved1();
 946                                         uint8_t* location = NULL;
 947                                         if ( sect->size() != 0 )
 948                                                 location = (uint8_t*)this->mappedAddressForNewAddress(sect->addr());
 949                                         pint_t vmlocation = sect->addr();
 950                                         for (uint32_t j=0; j < elementCount; ++j, location += elementSize, vmlocation += elementSize) {
 951                                                 uint32_t symbolIndex = E::get32(indirectTable[indirectTableOffset + j]);
 952                                                 switch ( symbolIndex ) {
 953                                                         case INDIRECT_SYMBOL_ABS:
 954                                                         case INDIRECT_SYMBOL_LOCAL:
 955                                                                 break;
 956                                                         default:
 957                                                                 const macho_nlist<P>* undefinedSymbol = &fSymbolTable[symbolIndex];
 958                                                                 //fprintf(stderr, " sect=%s, index=%d, symbolIndex=%d, sym=%s\n", sect->sectname(), j, symbolIndex, &fStrings[undefinedSymbol->n_strx()]);
 959                                                                 pint_t symbolAddr = this->resolveUndefined(undefinedSymbol);
 960                                                                 switch ( sectionType ) {
 961                                                                         case S_NON_LAZY_SYMBOL_POINTERS:
 962                                                                         case S_LAZY_SYMBOL_POINTERS:
 963                                                                                 P::setP(*((pint_t*)location), symbolAddr);
 964                                                                                 break;
 965                                                                         case S_SYMBOL_STUBS:
 966                                                                                 this->bindStub(elementSize, location, vmlocation, symbolAddr);
 967                                                                                         break;
 968                                                                 }
 969                                                                 break;
 970                                                 }
 971                                         }
 972                                 }
 973                         }
 974                 }
 975                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 976         }
 977 }
 978
 979
 980
 981
 982 template <typename A>
 983 typename A::P::uint_t Binder<A>::resolveUndefined(const macho_nlist<P>* undefinedSymbol)
 984 {
 985         if ( (undefinedSymbol->n_type() & N_TYPE) == N_SECT ) {
 986                 if ( (undefinedSymbol->n_type() & N_PEXT) != 0 ) {
 987                         // is a multi-module private_extern internal reference that the linker did not optimize away
 988                         return runtimeAddressFromNList(undefinedSymbol);
 989                 }
 990                 if ( (undefinedSymbol->n_desc() & N_WEAK_DEF) != 0 ) {
 991                         // is a weak definition, we should prebind to this one in the same linkage unit
 992                         return runtimeAddressFromNList(undefinedSymbol);
 993                 }
 994         }
 995         const char* symbolName = &fStrings[undefinedSymbol->n_strx()];
 996         if ( (this->fHeader->flags() & MH_TWOLEVEL) == 0 ) {
 997                 // flat namespace binding
 998                 throw "flat namespace not supported";
 999         }
1000         else {
1001                 uint8_t ordinal = GET_LIBRARY_ORDINAL(undefinedSymbol->n_desc());
1002                 Binder<A>* binder = NULL;
1003                 switch ( ordinal ) {
1004                         case EXECUTABLE_ORDINAL:
1005                         case DYNAMIC_LOOKUP_ORDINAL:
1006                                 throw "magic ordineal not supported";
1007                         case SELF_LIBRARY_ORDINAL:
1008                                 binder = this;
1009                                 break;
1010                         default:
1011                                 if ( ordinal > fDependentDylibs.size() )
1012                                         throw "two-level ordinal out of range";
1013                                 binder = fDependentDylibs[ordinal-1].binder;
1014                 }
1015                 pint_t addr;
1016                 bool isResolver;
1017                 bool isAbsolute;
1018         Binder<A>* foundIn;
1019                 if ( ! binder->findExportedSymbolAddress(symbolName, &addr, &foundIn, &isResolver, &isAbsolute) )
1020                         throwf("could not resolve undefined symbol %s in %s expected in %s", symbolName, this->getDylibID(), binder->getDylibID());
1021                 return addr;
1022         }
1023 }
1024
1025 template <typename A>
1026 bool Binder<A>::findExportedSymbolAddress(const char* name, pint_t* result, Binder<A>** foundIn, bool* isResolverSymbol, bool* isAbsolute)
1027 {
1028     *foundIn = NULL;
1029         // since re-export chains can be any length, re-exports cannot be resolved in setDependencies()
1030         // instead we lazily, recursively update
1031         if ( !fReExportedSymbolsResolved ) {
1032
1033                 // update fHashTable with any individual symbol re-exports
1034                 for (typename std::vector<SymbolReExport>::iterator it=fReExportedSymbols.begin(); it != fReExportedSymbols.end(); ++it) {
1035                         pint_t targetSymbolAddress;
1036                         bool isResolver;
1037                         bool isAb;
1038
1039                         if ( it->dylibOrdinal <= 0 )
1040                                 throw "bad mach-o binary, special library ordinal not allowed in re-exported symbols in dyld shared cache";
1041
1042                         Binder<A>* binder = fDependentDylibs[it->dylibOrdinal-1].binder;
1043
1044                         if ( ! binder->findExportedSymbolAddress(it->importName, &targetSymbolAddress, foundIn, &isResolver, &isAb) )
1045                                 throwf("could not bind symbol %s in %s expected in %s", it->importName, this->getDylibID(), binder->getDylibID());
1046
1047                         if ( isResolver )
1048                                 fSymbolResolvers.insert(name);
1049
1050                         fHashTable[it->exportName] = targetSymbolAddress;
1051                 }
1052                 // mark as done
1053                 fReExportedSymbolsResolved = true;
1054         }
1055
1056         *isResolverSymbol = false;
1057         if ( !fSymbolResolvers.empty() && fSymbolResolvers.count(name) ) {
1058                 // lazy pointers should be left unbound, rather than bind to resolver stub
1059                 *isResolverSymbol = true;
1060         }
1061
1062         // search this dylib
1063         typename NameToAddrMap::iterator pos = fHashTable.find(name);
1064         if ( pos != fHashTable.end() ) {
1065                 *result = pos->second;
1066                 //fprintf(stderr, "findExportedSymbolAddress(%s) => 0x%08llX in %s\n", name, (uint64_t)*result, this->getDylibID());
1067         *foundIn = this;
1068                 *isAbsolute = (fAbsoluteSymbols.count(name) != 0);
1069                 return true;
1070         }
1071
1072         // search re-exported dylibs
1073         for (typename std::vector<BinderAndReExportFlag>::iterator it = fDependentDylibs.begin(); it != fDependentDylibs.end(); ++it) {
1074                 if ( it->reExport ) {
1075                         if ( it->binder->findExportedSymbolAddress(name, result, foundIn, isResolverSymbol, isAbsolute) )
1076                                 return true;
1077                 }
1078         }
1079         //fprintf(stderr, "findExportedSymbolAddress(%s) => not found in %s\n", name, this->getDylibID());
1080         return false;
1081 }
1082
1083 // record which dylibs will be using this dylibs lazy pointer
1084 template <typename A>
1085 void Binder<A>::addResolverClient(Binder<A>* clientDylib, const char* symbolName)
1086 {
1087     ClientAndSymbol x;
1088     x.client = clientDylib;
1089     x.symbolName = symbolName;
1090     fClientAndSymbols.push_back(x);
1091 }
1092
1093
1094 template <typename A>
1095 typename A::P::uint_t Binder<A>::findLazyPointerFor(const char* symbolName)
1096 {
1097         static const bool log = false;
1098
1099         // first check cache
1100         typename NameToAddrMap::iterator pos = fResolverLazyPointers.find(symbolName);
1101         if ( pos != fResolverLazyPointers.end() ) {
1102                 if ( log ) fprintf(stderr, "found cached shared lazy pointer at 0x%llX for %s in %s\n", (uint64_t)(pos->second), symbolName, this->getDylibID());
1103                 return pos->second;
1104         }
1105
1106         // do slow lookup in lazy pointer section
1107         const uint32_t* const indirectTable = (uint32_t*)&this->fLinkEditBase[fDynamicInfo->indirectsymoff()];
1108         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
1109         const uint32_t cmd_count = this->fHeader->ncmds();
1110         const macho_load_command<P>* cmd = cmds;
1111         for (uint32_t i = 0; i < cmd_count; ++i) {
1112                 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
1113                         const macho_segment_command<P>* seg = (macho_segment_command<P>*)cmd;
1114                         const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)seg + sizeof(macho_segment_command<P>));
1115                         const macho_section<P>* const sectionsEnd = &sectionsStart[seg->nsects()];
1116                         for (const macho_section<P>* sect=sectionsStart; sect < sectionsEnd; ++sect) {
1117                                 uint8_t sectionType = sect->flags() & SECTION_TYPE;
1118                                 if ( sectionType == S_LAZY_SYMBOL_POINTERS) {
1119                                         uint32_t elementCount = sect->size() / sizeof(pint_t);
1120                                         const uint32_t indirectTableOffset = sect->reserved1();
1121                                         pint_t vmlocation = sect->addr();
1122                                         for (uint32_t j=0; j < elementCount; ++j, vmlocation += sizeof(pint_t)) {
1123                                                 uint32_t symbolIndex = E::get32(indirectTable[indirectTableOffset + j]);
1124                                                 switch ( symbolIndex ) {
1125                                                         case INDIRECT_SYMBOL_ABS:
1126                                                         case INDIRECT_SYMBOL_LOCAL:
1127                                                                 break;
1128                                                         default:
1129                                                                 const macho_nlist<P>* aSymbol = &fSymbolTable[symbolIndex];
1130                                                                 const char* aName = &fStrings[aSymbol->n_strx()];
1131                                                                 //fprintf(stderr, " sect=%s, index=%d, symbolIndex=%d, sym=%s\n", sect->sectname(), j, symbolIndex, &fStrings[undefinedSymbol->n_strx()]);
1132                                                                 if ( strcmp(aName, symbolName) == 0 ) {
1133                                                                         fResolverLazyPointers[symbolName] = vmlocation;
1134                                                                         if ( log ) fprintf(stderr, "found slow-path shared lazy pointer at 0x%llX for %s in %s\n", (uint64_t)vmlocation, symbolName, this->getDylibID());
1135                                                                         return vmlocation;
1136                                                                 }
1137                                                                 break;
1138                                                 }
1139                                         }
1140                                 }
1141                         }
1142                 }
1143                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
1144         }
1145
1146         if ( log ) fprintf(stderr, "not found shared lazy pointer for %s in %s, checking for re-export symbol\n", symbolName, this->getDylibID());
1147         for (typename std::vector<SymbolReExport>::iterator it=fReExportedSymbols.begin(); it != fReExportedSymbols.end(); ++it) {
1148                 if ( strcmp(it->exportName, symbolName) != 0 )
1149                         continue;
1150
1151                 if ( it->dylibOrdinal <= 0 )
1152                         throw "bad mach-o binary, special library ordinal not allowed in re-exported symbols in dyld shared cache";
1153
1154                 Binder<A>* binder = fDependentDylibs[it->dylibOrdinal-1].binder;
1155                 return binder->findLazyPointerFor(it->importName);
1156         }
1157
1158         if ( log ) fprintf(stderr, "not found shared lazy pointer for %s in %s, checking re-export dylibs\n", symbolName, this->getDylibID());
1159         for (typename std::vector<BinderAndReExportFlag>::iterator it = fDependentDylibs.begin(); it != fDependentDylibs.end(); ++it) {
1160                 if ( it->reExport ) {
1161                         pint_t result = it->binder->findLazyPointerFor(symbolName);
1162                         if ( result != 0 )
1163                                 return result;
1164                 }
1165         }
1166
1167         if ( log ) fprintf(stderr, "NOT found shared lazy pointer for %s in %s\n", symbolName, this->getDylibID());
1168     return 0;
1169 }
1170
1171 // called after all binding is done to optimize lazy pointers
1172 template <typename A>
1173 void Binder<A>::optimize()
1174 {
1175     for (typename std::vector<ClientAndSymbol>::iterator it = fClientAndSymbols.begin(); it != fClientAndSymbols.end(); ++it) {
1176         pint_t lpVMAddr = findLazyPointerFor(it->symbolName);
1177         if ( lpVMAddr != 0 ) {
1178             it->client->optimizeStub(it->symbolName, lpVMAddr);
1179         }
1180         else {
1181                         fprintf(stderr, "not able to optimize lazy pointer for %s in %s\n", it->symbolName, it->client->getDylibID());
1182         }
1183
1184     }
1185 }
1186
1187 template <>
1188 void Binder<arm>::optimizeStub(uint8_t* stubMappedAddress, pint_t stubVMAddress, uint32_t stubSize, pint_t lpVMAddr)
1189 {
1190     if ( stubSize != 16 ) {
1191         fprintf(stderr, "could not optimize ARM stub to resolver function in %s because it is wrong size\n", this->getDylibID());
1192         return;
1193     }
1194     uint32_t* instructions = (uint32_t*)stubMappedAddress;
1195     if (   (E::get32(instructions[0]) != 0xe59fc004)
1196         || (E::get32(instructions[1]) != 0xe08fc00c)
1197         || (E::get32(instructions[2]) != 0xe59cf000)
1198         ) {
1199         fprintf(stderr, "could not optimize ARM stub to resolver function in %s because instructions are not as expected\n", this->getDylibID());
1200         return;
1201     }
1202     // last .long in stub is:  lazyPtr - (stub+8)
1203     // alter to point to more optimal lazy pointer
1204     uint32_t betterOffset = lpVMAddr  - (stubVMAddress + 12);
1205     E::set32(instructions[3], betterOffset);
1206 }
1207
1208
1209 template <>
1210 void Binder<x86_64>::optimizeStub(uint8_t* stubMappedAddress, pint_t stubVMAddress, uint32_t stubSize, pint_t lpVMAddr)
1211 {
1212     if ( stubSize != 6 ) {
1213         fprintf(stderr, "could not optimize x86_64 stub to resolver function in %s because it is wrong size\n", this->getDylibID());
1214         return;
1215     }
1216     if ( (stubMappedAddress[0] != 0xFF) || (stubMappedAddress[1] != 0x25) ) {
1217         fprintf(stderr, "could not optimize stub to resolver function in %s because instructions are not as expected\n", this->getDylibID());
1218         return;
1219     }
1220     // last four bytes in stub is RIP relative offset to lazy pointer
1221     // alter to point to more optimal lazy pointer
1222     uint32_t betterOffset = lpVMAddr  - (stubVMAddress + 6);
1223     E::set32(*((uint32_t*)(&stubMappedAddress[2])), betterOffset);
1224 }
1225
1226 template <typename A>
1227 void Binder<A>::optimizeStub(uint8_t* stubMappedAddress, pint_t stubVMAddress, uint32_t stubSize, pint_t lpVMAddress)
1228 {
1229     // Remaining architectures are not optimized
1230     //fprintf(stderr, "optimize stub at %p in %s to use lazyPointer at 0x%llX\n", stubMappedAddress, this->getDylibID(), (uint64_t)lpVMAddress);
1231 }
1232
1233 // search for stub in this image that call target symbol name and then optimize its lazy pointer
1234 template <typename A>
1235 void Binder<A>::optimizeStub(const char* stubName, pint_t lpVMAddr)
1236 {
1237         // find named stub
1238         const uint32_t* const indirectTable = (uint32_t*)&this->fLinkEditBase[fDynamicInfo->indirectsymoff()];
1239         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)this->fHeader + sizeof(macho_header<P>));
1240         const uint32_t cmd_count = this->fHeader->ncmds();
1241         const macho_load_command<P>* cmd = cmds;
1242         for (uint32_t i = 0; i < cmd_count; ++i) {
1243                 if ( cmd->cmd() == macho_segment_command<P>::CMD ) {
1244                         macho_segment_command<P>* seg = (macho_segment_command<P>*)cmd;
1245                         macho_section<P>* const sectionsStart = (macho_section<P>*)((char*)seg + sizeof(macho_segment_command<P>));
1246                         macho_section<P>* const sectionsEnd = &sectionsStart[seg->nsects()];
1247                         for(macho_section<P>* sect = sectionsStart; sect < sectionsEnd; ++sect) {
1248                                 if ( ((sect->flags() & SECTION_TYPE) == S_SYMBOL_STUBS) && (sect->size() != 0) ) {
1249                                         pint_t stubsVMStart = sect->addr();
1250                                         uint8_t* stubsMappingStart = (uint8_t*)this->mappedAddressForNewAddress(stubsVMStart);
1251                                         const uint32_t indirectTableOffset = sect->reserved1();
1252                                         const uint32_t stubSize = sect->reserved2();
1253                                         uint32_t elementCount = sect->size() / stubSize;
1254                     pint_t stubVMAddr = stubsVMStart;
1255                     uint8_t* stubMappedAddr = stubsMappingStart;
1256                                         for (uint32_t j=0; j < elementCount; ++j, stubMappedAddr += stubSize, stubVMAddr += stubSize) {
1257                                                 uint32_t symbolIndex = E::get32(indirectTable[indirectTableOffset + j]);
1258                                                 switch ( symbolIndex ) {
1259                                                         case INDIRECT_SYMBOL_ABS:
1260                                                         case INDIRECT_SYMBOL_LOCAL:
1261                                                                 break;
1262                                                         default:
1263                                 {
1264                                    const macho_nlist<P>* sym = &this->fSymbolTable[symbolIndex];
1265                                    const char* symName = &fStrings[sym->n_strx()];
1266                                     if ( strcmp(symName, stubName) == 0 )
1267                                         this->optimizeStub(stubMappedAddr, stubVMAddr, stubSize, lpVMAddr);
1268                                 }
1269                                                                 break;
1270                                                 }
1271                                         }
1272                                 }
1273                         }
1274                 }
1275                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
1276         }
1277 }
1278
1279
1280 #endif // __MACHO_BINDER__
1281
1282
1283
1284