interlinked-dylibs/OptimizerObjC.cpp

   1 /* -*- mode: C++; c-basic-offset: 4; indent-tabs-mode: nil -*-
   2  *
   3  * Copyright (c) 2014 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
  26 #include "mega-dylib-utils.h"
  27 #include "Logging.h"
  28 #include "MachOFileAbstraction.hpp"
  29
  30
  31 #include <dirent.h>
  32 #include <sys/errno.h>
  33 #include <sys/fcntl.h>
  34 #include <mach-o/loader.h>
  35 #include <mach-o/fat.h>
  36 #include <assert.h>
  37
  38
  39 // Scan a C++ or Swift length-mangled field.
  40 static bool scanMangledField(const char *&string, const char *end,
  41                              const char *&field, int& length)
  42 {
  43     // Leading zero not allowed.
  44     if (*string == '0') return false;
  45
  46     length = 0;
  47     field = string;
  48     while (field < end) {
  49         char c = *field;
  50         if (!isdigit(c)) break;
  51         field++;
  52         if (__builtin_smul_overflow(length, 10, &length)) return false;
  53         if (__builtin_sadd_overflow(length, c - '0', &length)) return false;
  54     }
  55
  56     string = field + length;
  57     return length > 0  &&  string <= end;
  58 }
  59
  60
  61 // copySwiftDemangledName
  62 // Returns the pretty form of the given Swift-mangled class or protocol name.
  63 // Returns nullptr if the string doesn't look like a mangled Swift name.
  64 // The result must be freed with free().
  65 static char *copySwiftDemangledName(const char *string, bool isProtocol = false)
  66 {
  67     if (!string) return nullptr;
  68
  69     // Swift mangling prefix.
  70     if (strncmp(string, isProtocol ? "_TtP" : "_TtC", 4) != 0) return nullptr;
  71     string += 4;
  72
  73     const char *end = string + strlen(string);
  74
  75     // Module name.
  76     const char *prefix;
  77     int prefixLength;
  78     if (string[0] == 's') {
  79         // "s" is the Swift module.
  80         prefix = "Swift";
  81         prefixLength = 5;
  82         string += 1;
  83     } else {
  84         if (! scanMangledField(string, end, prefix, prefixLength)) return nullptr;
  85     }
  86
  87     // Class or protocol name.
  88     const char *suffix;
  89     int suffixLength;
  90     if (! scanMangledField(string, end, suffix, suffixLength)) return nullptr;
  91
  92     if (isProtocol) {
  93         // Remainder must be "_".
  94         if (strcmp(string, "_") != 0) return nullptr;
  95     } else {
  96         // Remainder must be empty.
  97         if (string != end) return nullptr;
  98     }
  99
 100     char *result;
 101     asprintf(&result, "%.*s.%.*s", prefixLength,prefix, suffixLength,suffix);
 102     return result;
 103 }
 104
 105
 106 class ContentAccessor {
 107 public:
 108     ContentAccessor(SharedCache& cache) {
 109         cache.forEachRegion([&] (void* content, uint64_t vmAddr, uint64_t size, uint32_t permissions) {
 110             Info info = { (uint8_t*)content, (uint8_t*)content+size, vmAddr, vmAddr+size };
 111             _regions.push_back(info);
 112         });
 113     }
 114
 115     void* contentForVMAddr(uint64_t vmaddr) {
 116         for (Info& info : _regions) {
 117             if ( (info.startAddr <= vmaddr) && (vmaddr < info.endAddr) )
 118                 return (void*)(info.contentStart + vmaddr - info.startAddr);
 119         }
 120         if ( vmaddr == 0 )
 121             return nullptr;
 122         terminate("contentForVMAddr(0x%0llX) invalid vmaddr in ObjC data", vmaddr);
 123     }
 124
 125     uint64_t vmAddrForContent(const void* content) {
 126         for (Info& info : _regions) {
 127             if ( (info.contentStart <= content) && (content < info.contentEnd) )
 128                 return info.startAddr + ((uint8_t*)content - (uint8_t*)info.contentStart);
 129         }
 130         terminate("vmAddrForContent(%p) invalid content pointer in ObjC data", content);
 131     }
 132
 133 private:
 134     struct Info { uint8_t* contentStart; uint8_t* contentEnd; uint64_t startAddr; uint64_t endAddr; };
 135     std::vector<Info> _regions;
 136 };
 137
 138
 139 // Access a section containing a list of pointers
 140 template <typename P, typename T>
 141 class PointerSection
 142 {
 143     typedef typename P::uint_t   pint_t;
 144 public:
 145     PointerSection(ContentAccessor* cache, const macho_header<P>* mh,
 146                    const char* segname, const char* sectname)
 147         : _cache(cache),
 148           _section(mh->getSection(segname, sectname)),
 149           _base(_section ? (pint_t*)cache->contentForVMAddr(_section->addr()) : 0),
 150           _count(_section ? (pint_t)(_section->size() / sizeof(pint_t)) : 0) {
 151     }
 152
 153     pint_t count() const { return _count; }
 154
 155     pint_t getVMAddress(pint_t index) const {
 156         if ( index >= _count )
 157             terminate("index out of range in section %s", _section->sectname());
 158         return (pint_t)P::getP(_base[index]);
 159     }
 160
 161     T get(pint_t index) const {
 162         return (T)_cache->contentForVMAddr(getVMAddress(index));
 163     }
 164
 165     void setVMAddress(pint_t index, pint_t value) {
 166         if (index >= _count)
 167             terminate("index out of range in section %s", _section->sectname());
 168         P::setP(_base[index], value);
 169     }
 170
 171     void removeNulls() {
 172         pint_t shift = 0;
 173         for (pint_t i = 0; i < _count; i++) {
 174             pint_t value = _base[i];
 175             if (value) {
 176                 _base[i-shift] = value;
 177             } else {
 178                 shift++;
 179             }
 180         }
 181         _count -= shift;
 182         const_cast<macho_section<P>*>(_section)->set_size(_count * sizeof(pint_t));
 183     }
 184
 185 private:
 186     ContentAccessor* const         _cache;
 187     const macho_section<P>* const  _section;
 188     pint_t* const                  _base;
 189     pint_t const                   _count;
 190 };
 191
 192
 193 // Access a section containing an array of structures
 194 template <typename P, typename T>
 195 class ArraySection
 196 {
 197 public:
 198     ArraySection(ContentAccessor* cache, const macho_header<P>* mh,
 199                  const char *segname, const char *sectname)
 200         : _cache(cache),
 201           _section(mh->getSection(segname, sectname)),
 202           _base(_section ? (T *)cache->contentForVMAddr(_section->addr()) : 0),
 203           _count(_section ? _section->size() / sizeof(T) : 0) {
 204     }
 205
 206     uint64_t count() const { return _count; }
 207
 208     T& get(uint64_t index) const {
 209         if (index >= _count)
 210             terminate("index out of range in section %s", _section->sectname());
 211         return _base[index];
 212     }
 213
 214 private:
 215     ContentAccessor* const         _cache;
 216     const macho_section<P>* const  _section;
 217     T * const                      _base;
 218     uint64_t const                 _count;
 219 };
 220
 221
 222 #define SELOPT_WRITE
 223 #include "objc-shared-cache.h"
 224 #include "ObjC1Abstraction.hpp"
 225 #include "ObjC2Abstraction.hpp"
 226
 227
 228 namespace {
 229
 230
 231
 232 template <typename P>
 233 class ObjCSelectorUniquer
 234 {
 235 public:
 236     typedef typename P::uint_t  pint_t;
 237
 238     ObjCSelectorUniquer(ContentAccessor* cache) : _cache(cache) { }
 239
 240     pint_t visit(pint_t oldValue)
 241     {
 242         _count++;
 243         const char *s = (const char *)_cache->contentForVMAddr(oldValue);
 244         objc_opt::string_map::iterator element =
 245             _selectorStrings.insert(objc_opt::string_map::value_type(s, oldValue)).first;
 246         return (pint_t)element->second;
 247     }
 248
 249     objc_opt::string_map& strings() {
 250         return _selectorStrings;
 251     }
 252
 253     size_t count() const { return _count; }
 254
 255 private:
 256     objc_opt::string_map    _selectorStrings;
 257     ContentAccessor*        _cache;
 258     size_t                  _count = 0;
 259 };
 260
 261
 262 template <typename P>
 263 class ClassListBuilder
 264 {
 265 private:
 266     objc_opt::string_map    _classNames;
 267     objc_opt::class_map     _classes;
 268     size_t                  _count = 0;
 269     HeaderInfoOptimizer<P, objc_header_info_ro_t<P>>& _hInfos;
 270
 271 public:
 272
 273     ClassListBuilder(HeaderInfoOptimizer<P, objc_header_info_ro_t<P>>& hinfos) : _hInfos(hinfos) { }
 274
 275     void visitClass(ContentAccessor* cache,
 276                     const macho_header<P>* header,
 277                     objc_class_t<P>* cls)
 278     {
 279         if (cls->isMetaClass(cache)) return;
 280
 281         const char *name = cls->getName(cache);
 282         uint64_t name_vmaddr = cache->vmAddrForContent((void*)name);
 283         uint64_t cls_vmaddr = cache->vmAddrForContent(cls);
 284         uint64_t hinfo_vmaddr = cache->vmAddrForContent(_hInfos.hinfoForHeader(cache, header));
 285         _classNames.insert(objc_opt::string_map::value_type(name, name_vmaddr));
 286         _classes.insert(objc_opt::class_map::value_type(name, std::pair<uint64_t, uint64_t>(cls_vmaddr, hinfo_vmaddr)));
 287         _count++;
 288     }
 289
 290     objc_opt::string_map& classNames() {
 291         return _classNames;
 292     }
 293
 294     objc_opt::class_map& classes() {
 295         return _classes;
 296     }
 297
 298     size_t count() const { return _count; }
 299 };
 300
 301 template <typename P>
 302 class ProtocolOptimizer
 303 {
 304 private:
 305     typedef typename P::uint_t pint_t;
 306
 307     objc_opt::string_map    _protocolNames;
 308     objc_opt::protocol_map  _protocols;
 309     size_t                  _protocolCount;
 310     size_t                  _protocolReferenceCount;
 311
 312     friend class ProtocolReferenceWalker<P, ProtocolOptimizer<P>>;
 313
 314     pint_t visitProtocolReference(ContentAccessor* cache, pint_t oldValue)
 315     {
 316         objc_protocol_t<P>* proto = (objc_protocol_t<P>*)
 317             cache->contentForVMAddr(oldValue);
 318         pint_t newValue = (pint_t)_protocols[proto->getName(cache)];
 319         if (oldValue != newValue) _protocolReferenceCount++;
 320         return newValue;
 321     }
 322
 323 public:
 324
 325     ProtocolOptimizer()
 326         : _protocolNames()
 327         , _protocols()
 328         , _protocolCount(0)
 329         , _protocolReferenceCount(0)
 330     { }
 331
 332     void addProtocols(ContentAccessor* cache,
 333                       const macho_header<P>* header)
 334     {
 335         PointerSection<P, objc_protocol_t<P> *>
 336             protocols(cache, header, "__DATA", "__objc_protolist");
 337
 338         for (pint_t i = 0; i < protocols.count(); i++) {
 339             objc_protocol_t<P> *proto = protocols.get(i);
 340
 341             const char *name = proto->getName(cache);
 342             if (_protocolNames.count(name) == 0) {
 343                 if (proto->getSize() > sizeof(objc_protocol_t<P>)) {
 344                     terminate("objc protocol is too big");
 345                 }
 346
 347                 uint64_t name_vmaddr = cache->vmAddrForContent((void*)name);
 348                 uint64_t proto_vmaddr = cache->vmAddrForContent(proto);
 349                 _protocolNames.insert(objc_opt::string_map::value_type(name, name_vmaddr));
 350                 _protocols.insert(objc_opt::protocol_map::value_type(name, proto_vmaddr));
 351                 _protocolCount++;
 352             }
 353         }
 354     }
 355
 356     const char *writeProtocols(ContentAccessor* cache,
 357                                uint8_t *& rwdest, size_t& rwremaining,
 358                                uint8_t *& rodest, size_t& roremaining,
 359                                std::vector<void*>& pointersInData,
 360                                pint_t protocolClassVMAddr)
 361     {
 362         if (_protocolCount == 0) return NULL;
 363
 364         if (protocolClassVMAddr == 0) {
 365             return "libobjc's Protocol class symbol not found (metadata not optimized)";
 366         }
 367
 368         size_t rwrequired = _protocolCount * sizeof(objc_protocol_t<P>);
 369         if (rwremaining < rwrequired) {
 370             return "libobjc's read-write section is too small (metadata not optimized)";
 371         }
 372
 373         for (objc_opt::protocol_map::iterator iter = _protocols.begin();
 374              iter != _protocols.end();
 375              ++iter)
 376         {
 377             objc_protocol_t<P>* oldProto = (objc_protocol_t<P>*)
 378                 cache->contentForVMAddr(iter->second);
 379
 380             // Create a new protocol object.
 381             objc_protocol_t<P>* proto = (objc_protocol_t<P>*)rwdest;
 382             rwdest += sizeof(*proto);
 383             rwremaining -= sizeof(*proto);
 384
 385             // Initialize it.
 386             uint32_t oldSize = oldProto->getSize();
 387             memcpy(proto, oldProto, oldSize);
 388             if (!proto->getIsaVMAddr()) {
 389                 proto->setIsaVMAddr(protocolClassVMAddr);
 390             }
 391             if (oldSize < sizeof(*proto)) {
 392                 // Protocol object is old. Populate new fields.
 393                 proto->setSize(sizeof(objc_protocol_t<P>));
 394                 // missing extendedMethodTypes is already nil
 395             }
 396             // Some protocol objects are big enough to have the
 397             // demangledName field but don't initialize it.
 398             // Initialize it here if it is not already set.
 399             if (!proto->getDemangledName(cache)) {
 400                 const char *roName = proto->getName(cache);
 401                 char *demangledName = copySwiftDemangledName(roName, true);
 402                 if (demangledName) {
 403                     size_t length = 1 + strlen(demangledName);
 404                     if (roremaining < length) {
 405                         return "libobjc's read-only section is too small (metadata not optimized)";
 406                     }
 407
 408                     memmove(rodest, demangledName, length);
 409                     roName = (const char *)rodest;
 410                     rodest += length;
 411                     roremaining -= length;
 412
 413                     free(demangledName);
 414                 }
 415                 proto->setDemangledName(cache, roName);
 416             }
 417             proto->setFixedUp();
 418
 419             // Redirect the protocol table at our new object.
 420             iter->second = cache->vmAddrForContent(proto);
 421
 422             // Add new rebase entries.
 423             proto->addPointers(pointersInData);
 424         }
 425
 426         return NULL;
 427     }
 428
 429     void updateReferences(ContentAccessor* cache, const macho_header<P>* header)
 430     {
 431         ProtocolReferenceWalker<P, ProtocolOptimizer<P>> refs(*this);
 432         refs.walk(cache, header);
 433     }
 434
 435     objc_opt::string_map& protocolNames() {
 436         return _protocolNames;
 437     }
 438
 439     objc_opt::protocol_map& protocols() {
 440         return _protocols;
 441    }
 442
 443     size_t protocolCount() const { return _protocolCount; }
 444     size_t protocolReferenceCount() const { return _protocolReferenceCount; }
 445 };
 446
 447
 448 static int percent(size_t num, size_t denom) {
 449     if (denom)
 450         return (int)(num / (double)denom * 100);
 451     else
 452         return 100;
 453 }
 454
 455
 456 template <typename P>
 457 void optimizeObjC(SharedCache& cache, std::vector<void*>& pointersForASLR, bool forProduction)
 458 {
 459     typedef typename P::E           E;
 460     typedef typename P::uint_t      pint_t;
 461
 462     verboseLog("Optimizing objc metadata:");
 463     verboseLog("  cache type is %s",
 464                forProduction ? "production" : "development");
 465
 466     ContentAccessor cacheAccessor(cache);
 467
 468     size_t headerSize = P::round_up(sizeof(objc_opt::objc_opt_t));
 469     if (headerSize != sizeof(objc_opt::objc_opt_t)) {
 470         warning("libobjc's optimization structure size is wrong (metadata not optimized)");
 471     }
 472
 473     //
 474     // Find libobjc's empty sections and build list of images with objc metadata
 475     //
 476     const macho_section<P> *optROSection = nullptr;
 477     const macho_section<P> *optRWSection = nullptr;
 478     const macho_section<P> *optPointerListSection = nullptr;
 479     std::vector<const macho_header<P>*> objcDylibs;
 480     cache.forEachImage([&](const void* machHeader, const char* installName,
 481                         time_t, ino_t, const std::vector<MachOProxy::Segment>& segments) {
 482         const macho_header<P>* mh = (const macho_header<P>*)machHeader;
 483         if ( strstr(installName, "/libobjc.") != nullptr ) {
 484             optROSection = mh->getSection("__TEXT", "__objc_opt_ro");
 485             optRWSection = mh->getSection("__DATA", "__objc_opt_rw");
 486             optPointerListSection = mh->getSection("__DATA", "__objc_opt_ptrs");
 487         }
 488         if ( mh->getSection("__DATA", "__objc_imageinfo") || mh->getSection("__OBJC", "__image_info") ) {
 489             objcDylibs.push_back(mh);
 490         }
 491         // log("installName %s at mhdr 0x%016lx", installName, (uintptr_t)cacheAccessor.vmAddrForContent((void*)mh));
 492     });
 493     if ( optROSection == nullptr ) {
 494         warning("libobjc's read-only section missing (metadata not optimized)");
 495         return;
 496     }
 497     if ( optRWSection == nullptr ) {
 498         warning("libobjc's read/write section missing (metadata not optimized)");
 499         return;
 500     }
 501     if ( optPointerListSection == nullptr ) {
 502         warning("libobjc's pointer list section missing (metadata not optimized)");
 503         return;
 504     }
 505
 506     uint8_t* optROData = (uint8_t*)cacheAccessor.contentForVMAddr(optROSection->addr());
 507     size_t optRORemaining = optROSection->size();
 508     uint8_t* optRWData = (uint8_t*)cacheAccessor.contentForVMAddr(optRWSection->addr());
 509     size_t optRWRemaining = optRWSection->size();
 510     if (optRORemaining < headerSize) {
 511         warning("libobjc's read-only section is too small (metadata not optimized)");
 512         return;
 513     }
 514     objc_opt::objc_opt_t* optROHeader = (objc_opt::objc_opt_t *)optROData;
 515     optROData += headerSize;
 516     optRORemaining -= headerSize;
 517     if (E::get32(optROHeader->version) != objc_opt::VERSION) {
 518         warning("libobjc's read-only section version is unrecognized (metadata not optimized)");
 519         return;
 520     }
 521
 522     if (optPointerListSection->size() < sizeof(objc_opt::objc_opt_pointerlist_tt<pint_t>)) {
 523         warning("libobjc's pointer list section is too small (metadata not optimized)");
 524         return;
 525     }
 526     const objc_opt::objc_opt_pointerlist_tt<pint_t> *optPointerList = (const objc_opt::objc_opt_pointerlist_tt<pint_t> *)cacheAccessor.contentForVMAddr(optPointerListSection->addr());
 527
 528     // Write nothing to optROHeader until everything else is written.
 529     // If something fails below, libobjc will not use the section.
 530
 531
 532     //
 533     // Make copy of objcList and sort that list.
 534     //
 535     std::vector<const macho_header<P>*> addressSortedDylibs = objcDylibs;
 536     std::sort(addressSortedDylibs.begin(), addressSortedDylibs.end(), [](const macho_header<P>* lmh, const macho_header<P>* rmh) -> bool {
 537         return lmh < rmh;
 538     });
 539
 540     //
 541     // Build HeaderInfo list in cache
 542     //
 543     // First the RO header info
 544     // log("writing out %d RO dylibs at offset %d", (uint32_t)objcDylibs.size(), (uint32_t)(optROSection->size() - optRORemaining));
 545     uint64_t hinfoROVMAddr = optROSection->addr() + optROSection->size() - optRORemaining;
 546     HeaderInfoOptimizer<P, objc_header_info_ro_t<P>> hinfoROOptimizer;
 547     const char* err = hinfoROOptimizer.init((uint32_t)objcDylibs.size(), optROData, optRORemaining);
 548     if (err) {
 549         warning("%s", err);
 550         return;
 551     }
 552     else {
 553         for (const macho_header<P>* mh : addressSortedDylibs) {
 554             hinfoROOptimizer.update(&cacheAccessor, mh, pointersForASLR);
 555         }
 556     }
 557
 558     // Then the RW header info
 559     // log("writing out %d RW dylibs at offset %d", (uint32_t)objcDylibs.size(), (uint32_t)(optRWSection->size() - optRWRemaining));
 560     uint64_t hinfoRWVMAddr = (uint64_t)optRWSection->addr() + (uint64_t)optRWSection->size() - optRWRemaining;
 561     HeaderInfoOptimizer<P, objc_header_info_rw_t<P>> hinfoRWOptimizer;
 562     err = hinfoRWOptimizer.init((uint32_t)objcDylibs.size(), optRWData, optRWRemaining);
 563     if (err) {
 564         warning("%s", err);
 565         return;
 566     }
 567     else {
 568         for (const macho_header<P>* mh : addressSortedDylibs) {
 569             hinfoRWOptimizer.update(&cacheAccessor, mh, pointersForASLR);
 570         }
 571     }
 572
 573     //
 574     // Update selector references and build selector list
 575     //
 576     // This is SAFE: if we run out of room for the selector table,
 577     // the modified binaries are still usable.
 578     //
 579     // Heuristic: choose selectors from libraries with more selector cstring data first.
 580     // This tries to localize selector cstring memory.
 581     //
 582     ObjCSelectorUniquer<P> uniq(&cacheAccessor);
 583     std::vector<const macho_header<P>*> sizeSortedDylibs = objcDylibs;
 584     std::sort(sizeSortedDylibs.begin(), sizeSortedDylibs.end(),  [](const macho_header<P>* lmh, const macho_header<P>* rmh) -> bool {
 585         const macho_section<P>* lSection = lmh->getSection("__TEXT", "__objc_methname");
 586         const macho_section<P>* rSection = rmh->getSection("__TEXT", "__objc_methname");
 587         uint64_t lSelectorSize = (lSection ? lSection->size() : 0);
 588         uint64_t rSelectorSize = (rSection ? rSection->size() : 0);
 589         return lSelectorSize > rSelectorSize;
 590     });
 591
 592     SelectorOptimizer<P, ObjCSelectorUniquer<P> > selOptimizer(uniq);
 593     for (const macho_header<P>* mh : sizeSortedDylibs) {
 594         LegacySelectorUpdater<P, ObjCSelectorUniquer<P>>::update(&cacheAccessor, mh, uniq);
 595         selOptimizer.optimize(&cacheAccessor, mh);
 596     }
 597
 598     verboseLog("  uniqued  % 6ld selectors",
 599                uniq.strings().size());
 600     verboseLog("  updated  % 6ld selector references",
 601                uniq.count());
 602
 603     uint64_t seloptVMAddr = optROSection->addr() + optROSection->size() - optRORemaining;
 604     objc_opt::objc_selopt_t *selopt = new(optROData) objc_opt::objc_selopt_t;
 605     err = selopt->write(seloptVMAddr, optRORemaining, uniq.strings());
 606     if (err) {
 607         warning("%s", err);
 608         return;
 609     }
 610     optROData += selopt->size();
 611     optRORemaining -= selopt->size();
 612     uint32_t seloptCapacity = selopt->capacity;
 613     uint32_t seloptOccupied = selopt->occupied;
 614     selopt->byteswap(E::little_endian), selopt = nullptr;
 615
 616     verboseLog("  selector table occupancy %u/%u (%u%%)",
 617                seloptOccupied, seloptCapacity,
 618                (unsigned)(seloptOccupied/(double)seloptCapacity*100));
 619
 620
 621     //
 622     // Detect classes that have missing weak-import superclasses.
 623     //
 624     // Production only. Development cache does not do this: a replacement
 625     // library could omit a class at runtime that was present during
 626     // cache construction.
 627     //
 628     // This is SAFE: the binaries themselves are unmodified.
 629     bool noMissingWeakSuperclasses = false; // dev cache can't promise otherwise
 630     if (forProduction) {
 631         WeakClassDetector<P> weakopt;
 632         noMissingWeakSuperclasses =
 633             weakopt.noMissingWeakSuperclasses(&cacheAccessor, sizeSortedDylibs);
 634
 635         // Shared cache does not currently support unbound weak references.
 636         // Here we assert that there are none. If support is added later then
 637         // this assertion needs to be removed and this path needs to be tested.
 638         if (!noMissingWeakSuperclasses) {
 639             terminate("Some Objective-C class has a superclass that is "
 640                       "weak-import and missing from the cache.");
 641         }
 642     }
 643
 644
 645     //
 646     // Build class table.
 647     //
 648     // This is SAFE: the binaries themselves are unmodified.
 649     ClassListBuilder<P> classes(hinfoROOptimizer);
 650     ClassWalker<P, ClassListBuilder<P>> classWalker(classes);
 651     for (const macho_header<P>* mh : sizeSortedDylibs) {
 652         classWalker.walk(&cacheAccessor, mh);
 653     }
 654
 655     verboseLog("  recorded % 6ld classes",
 656                classes.classNames().size());
 657
 658     uint64_t clsoptVMAddr = optROSection->addr() + optROSection->size() - optRORemaining;
 659     objc_opt::objc_clsopt_t *clsopt = new(optROData) objc_opt::objc_clsopt_t;
 660     err = clsopt->write(clsoptVMAddr, optRORemaining,
 661                         classes.classNames(), classes.classes(), false);
 662     if (err) {
 663         warning("%s", err);
 664         return;
 665     }
 666     optROData += clsopt->size();
 667     optRORemaining -= clsopt->size();
 668     size_t duplicateCount = clsopt->duplicateCount();
 669     uint32_t clsoptCapacity = clsopt->capacity;
 670     uint32_t clsoptOccupied = clsopt->occupied;
 671     clsopt->byteswap(E::little_endian);
 672     clsopt = nullptr;
 673
 674     verboseLog("  found    % 6ld duplicate classes",
 675                duplicateCount);
 676     verboseLog("  class table occupancy %u/%u (%u%%)",
 677                clsoptOccupied, clsoptCapacity,
 678                (unsigned)(clsoptOccupied/(double)clsoptCapacity*100));
 679
 680
 681     //
 682     // Sort method lists.
 683     //
 684     // This is SAFE: modified binaries are still usable as unsorted lists.
 685     // This must be done AFTER uniquing selectors.
 686     MethodListSorter<P> methodSorter;
 687     for (const macho_header<P>* mh : sizeSortedDylibs) {
 688         methodSorter.optimize(&cacheAccessor, mh);
 689     }
 690
 691     verboseLog("  sorted   % 6ld method lists",
 692                methodSorter.optimized());
 693
 694
 695     // Unique protocols and build protocol table.
 696
 697     // This is SAFE: no protocol references are updated yet
 698     // This must be done AFTER updating method lists.
 699
 700     ProtocolOptimizer<P> protocolOptimizer;
 701     for (const macho_header<P>* mh : sizeSortedDylibs) {
 702         protocolOptimizer.addProtocols(&cacheAccessor, mh);
 703     }
 704
 705     verboseLog("  uniqued  % 6ld protocols",
 706                protocolOptimizer.protocolCount());
 707
 708     pint_t protocolClassVMAddr = (pint_t)P::getP(optPointerList->protocolClass);
 709     err = protocolOptimizer.writeProtocols(&cacheAccessor,
 710                                            optRWData, optRWRemaining,
 711                                            optROData, optRORemaining,
 712                                            pointersForASLR, protocolClassVMAddr);
 713     if (err) {
 714         warning("%s", err);
 715         return;
 716     }
 717
 718     uint64_t protocoloptVMAddr = optROSection->addr() + optROSection->size() - optRORemaining;
 719     objc_opt::objc_protocolopt_t *protocolopt = new (optROData) objc_opt::objc_protocolopt_t;
 720     err = protocolopt->write(protocoloptVMAddr, optRORemaining,
 721                              protocolOptimizer.protocolNames(),
 722                              protocolOptimizer.protocols(), true);
 723     if (err) {
 724         warning("%s", err);
 725         return;
 726     }
 727     optROData += protocolopt->size();
 728     optRORemaining -= protocolopt->size();
 729     uint32_t protocoloptCapacity = protocolopt->capacity;
 730     uint32_t protocoloptOccupied = protocolopt->occupied;
 731     protocolopt->byteswap(E::little_endian), protocolopt = NULL;
 732
 733     verboseLog("  protocol table occupancy %u/%u (%u%%)",
 734                protocoloptOccupied, protocoloptCapacity,
 735                (unsigned)(protocoloptOccupied/(double)protocoloptCapacity*100));
 736
 737
 738     // Redirect protocol references to the uniqued protocols.
 739
 740     // This is SAFE: the new protocol objects are still usable as-is.
 741     for (const macho_header<P>* mh : sizeSortedDylibs) {
 742         protocolOptimizer.updateReferences(&cacheAccessor, mh);
 743     }
 744
 745     verboseLog("  updated  % 6ld protocol references",
 746                protocolOptimizer.protocolReferenceCount());
 747
 748
 749     //
 750     // Repair ivar offsets.
 751     //
 752     // This is SAFE: the runtime always validates ivar offsets at runtime.
 753     IvarOffsetOptimizer<P> ivarOffsetOptimizer;
 754     for (const macho_header<P>* mh : sizeSortedDylibs) {
 755         ivarOffsetOptimizer.optimize(&cacheAccessor, mh);
 756     }
 757
 758     verboseLog("  updated  % 6ld ivar offsets",
 759                ivarOffsetOptimizer.optimized());
 760
 761
 762     // Collect flags.
 763     uint32_t headerFlags = 0;
 764     if (forProduction) {
 765         headerFlags |= objc_opt::IsProduction;
 766     }
 767     if (noMissingWeakSuperclasses) {
 768         headerFlags |= objc_opt::NoMissingWeakSuperclasses;
 769     }
 770
 771
 772     // Success. Mark dylibs as optimized.
 773     for (const macho_header<P>* mh : sizeSortedDylibs) {
 774         const macho_section<P>* imageInfoSection = mh->getSection("__DATA", "__objc_imageinfo");
 775         if (!imageInfoSection) {
 776             imageInfoSection = mh->getSection("__OBJC", "__image_info");
 777         }
 778         if (imageInfoSection) {
 779             objc_image_info<P>* info = (objc_image_info<P>*)cacheAccessor.contentForVMAddr(imageInfoSection->addr());
 780             info->setOptimizedByDyld();
 781         }
 782     }
 783
 784
 785     // Success. Update RO header last.
 786     E::set32(optROHeader->flags, headerFlags);
 787     E::set32(optROHeader->selopt_offset, (uint32_t)(seloptVMAddr - optROSection->addr()));
 788     E::set32(optROHeader->clsopt_offset, (uint32_t)(clsoptVMAddr - optROSection->addr()));
 789     E::set32(optROHeader->protocolopt_offset, (uint32_t)(protocoloptVMAddr - optROSection->addr()));
 790     E::set32(optROHeader->headeropt_ro_offset, (uint32_t)(hinfoROVMAddr - optROSection->addr()));
 791     E::set32(optROHeader->headeropt_rw_offset, (uint32_t)(hinfoRWVMAddr - optROSection->addr()));
 792
 793     // Log statistics.
 794     size_t roSize = optROSection->size() - optRORemaining;
 795     size_t rwSize = optRWSection->size() - optRWRemaining;
 796     verboseLog("  %zu/%llu bytes "
 797                "(%d%%) used in libobjc read-only optimization section",
 798                 roSize, optROSection->size(),
 799                 percent(roSize, optROSection->size()));
 800     verboseLog("  %zu/%llu bytes "
 801                "(%d%%) used in libobjc read/write optimization section",
 802                 rwSize, optRWSection->size(),
 803                 percent(rwSize, optRWSection->size()));
 804     verboseLog("  wrote objc metadata optimization version %d",
 805                 objc_opt::VERSION);
 806 }
 807
 808
 809 } // anon namespace
 810
 811
 812 void SharedCache::optimizeObjC(bool forProduction)
 813 {
 814      switch ( _arch.arch ) {
 815         case CPU_TYPE_ARM:
 816         case CPU_TYPE_I386:
 817             ::optimizeObjC<Pointer32<LittleEndian>>(*this, _pointersForASLR, forProduction);
 818             break;
 819         case CPU_TYPE_X86_64:
 820         case CPU_TYPE_ARM64:
 821             ::optimizeObjC<Pointer64<LittleEndian>>(*this, _pointersForASLR, forProduction);
 822             break;
 823         default:
 824             terminate("unsupported arch 0x%08X", _arch.arch);
 825     }
 826 }