dyld3/ClosureBuilder.cpp

   1 /*
   2  * Copyright (c) 2017 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. Please obtain a copy of the License at
  10  * http://www.opensource.apple.com/apsl/ and read it before using this
  11  * file.
  12  *
  13  * The Original Code and all software distributed under the License are
  14  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  15  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  16  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  18  * Please see the License for the specific language governing rights and
  19  * limitations under the License.
  20  *
  21  * @APPLE_LICENSE_HEADER_END@
  22  */
  23
  24 #include <sys/types.h>
  25 #include <sys/stat.h>
  26 #include <sys/errno.h>
  27 #include <sys/mman.h>
  28 #include <sys/mman.h>
  29 #include <sys/param.h>
  30 #include <ext/__hash>
  31 #include <fcntl.h>
  32 #include <unistd.h>
  33 #include <string.h>
  34 #include <sys/types.h>
  35 #include <sys/sysctl.h>
  36
  37 #include <mach-o/dyld_priv.h>
  38
  39 #include "ClosureWriter.h"
  40 #include "ClosureBuilder.h"
  41 #include "MachOAnalyzer.h"
  42 #include "libdyldEntryVector.h"
  43 #include "Tracing.h"
  44
  45 #define CLOSURE_SELOPT_WRITE
  46 #include "objc-shared-cache.h"
  47
  48 namespace dyld3 {
  49 namespace closure {
  50
  51
  52 const DlopenClosure* ClosureBuilder::sRetryDlopenClosure = (const DlopenClosure*)(-1);
  53
  54 ClosureBuilder::ClosureBuilder(uint32_t startImageNum, const FileSystem& fileSystem, const DyldSharedCache* dyldCache, bool dyldCacheIsLive,
  55                                const GradedArchs& archs, const PathOverrides& pathOverrides, AtPath atPathHandling, bool allowRelativePaths,
  56                                LaunchErrorInfo* errorInfo, Platform platform, const CacheDylibsBindingHandlers* handlers)
  57     : _fileSystem(fileSystem), _dyldCache(dyldCache), _pathOverrides(pathOverrides), _archs(archs), _platform(platform), _startImageNum(startImageNum),
  58       _handlers(handlers), _atPathHandling(atPathHandling), _launchErrorInfo(errorInfo), _dyldCacheIsLive(dyldCacheIsLive), _allowRelativePaths(allowRelativePaths)
  59 {
  60     if ( dyldCache != nullptr ) {
  61         _dyldImageArray = dyldCache->cachedDylibsImageArray();
  62         if ( (dyldCache->header.otherImageArrayAddr != 0) && (dyldCache->header.progClosuresSize == 0) )
  63             _makingClosuresInCache = true;
  64     }
  65 }
  66
  67
  68 ClosureBuilder::~ClosureBuilder() {
  69     if ( _tempPaths != nullptr )
  70         PathPool::deallocate(_tempPaths);
  71     if ( _mustBeMissingPaths != nullptr )
  72         PathPool::deallocate(_mustBeMissingPaths);
  73     if ( _objcDuplicateClassWarnings != nullptr )
  74         PathPool::deallocate(_objcDuplicateClassWarnings);
  75 }
  76
  77 bool ClosureBuilder::findImage(const char* loadPath, const LoadedImageChain& forImageChain, BuilderLoadedImage*& foundImage, LinkageType linkageType,
  78                                 uint32_t compatVersion, bool canUseSharedCacheClosure)
  79 {
  80     // There shouldn't be an error here as the callers should stop trying to find more images if they get an error for an image
  81     _diag.assertNoError();
  82
  83     __block bool result = false;
  84
  85     // record if this is a non-overridable path
  86     bool pathIsInDyldCacheWhichCannotBeOverridden = false;
  87     bool dylibsExpectedOnDisk = true;
  88     if ( _dyldCache != nullptr ) {
  89         pathIsInDyldCacheWhichCannotBeOverridden = _dyldCache->hasNonOverridablePath(loadPath);
  90         dylibsExpectedOnDisk = _dyldCache->header.dylibsExpectedOnDisk;
  91     }
  92
  93     _pathOverrides.forEachPathVariant(loadPath, pathIsInDyldCacheWhichCannotBeOverridden, ^(const char* possibleVariantPath, bool isFallbackPath, bool& stopPathVariant) {
  94
  95         // This check is within forEachPathVariant() to let DYLD_LIBRARY_PATH override LC_RPATH
  96         bool isRPath = (strncmp(possibleVariantPath, "@rpath/", 7) == 0);
  97
  98         // passing a leaf name to dlopen() allows rpath searching for it
  99         // FIXME: Does this apply to DYLD_INSERT_LIBRARIES too?
 100         bool implictRPath = (linkageType == LinkageType::kDynamic) && (loadPath[0] != '/') && (loadPath == possibleVariantPath) && (_atPathHandling != AtPath::none);
 101
 102         // expand @ paths
 103         forEachResolvedPathVar(possibleVariantPath, forImageChain, implictRPath, linkageType,
 104                                ^(const char* possiblePath, bool& stop) {
 105             if ( possibleVariantPath != possiblePath )
 106                 _atPathUsed = true;
 107
 108             // look at already loaded images
 109             const char* leafName = strrchr(possiblePath, '/');
 110             for (BuilderLoadedImage& li: _loadedImages) {
 111                 if ( strcmp(li.path(), possiblePath) == 0 ) {
 112                     foundImage = &li;
 113                     result = true;
 114                     stop = true;
 115                     return;
 116                 }
 117                 else if ( isRPath ) {
 118                     // Special case @rpath/ because name in li.fileInfo.path is full path.
 119                     // Getting installName is expensive, so first see if an already loaded image
 120                     // has same leaf name and if so see if its installName matches request @rpath
 121                     if (const char* aLeaf = strrchr(li.path(), '/')) {
 122                         if ( strcmp(aLeaf, leafName) == 0 ) {
 123                             if ( li.loadAddress()->isDylib() && (strcmp(loadPath, li.loadAddress()->installName()) == 0) ) {
 124                                 foundImage = &li;
 125                                 result = true;
 126                                 stop = true;
 127                                 return;
 128                             }
 129                         }
 130                     }
 131                 }
 132             }
 133
 134             // look to see if image already loaded via a different symlink
 135             bool                  fileFound             = false;
 136             uint64_t              fileFoundINode        = 0;
 137             uint64_t              fileFoundMTime        = 0;
 138             bool                  inodesMatchRuntime    = false;
 139             // Note, we only do this check if we even expect to find this on-disk
 140             // We can also use the pathIsInDyldCacheWhichCannotBeOverridden result if we are still trying the same path
 141             // it was computed from
 142             if ( dylibsExpectedOnDisk || !pathIsInDyldCacheWhichCannotBeOverridden || (loadPath != possiblePath) ) {
 143                 if ( _fileSystem.fileExists(possiblePath, &fileFoundINode, &fileFoundMTime, nullptr, &inodesMatchRuntime) ) {
 144                     fileFound = true;
 145                     for (BuilderLoadedImage& li: _loadedImages) {
 146                         if ( (li.loadedFileInfo.inode == fileFoundINode) && (li.loadedFileInfo.mtime == fileFoundMTime) )  {
 147                             foundImage = &li;
 148                             result = true;
 149                             stop = true;
 150                             return;
 151                         }
 152                     }
 153                 }
 154             }
 155
 156             bool                  unmapWhenDone     = false;
 157             bool                  contentRebased    = false;
 158             bool                  hasInits          = false;
 159             bool                  markNeverUnload   = false;
 160             bool                  mustBuildClosure = _dyldCacheInvalidFormatVersion;
 161             ImageNum              overrideImageNum  = 0;
 162             ImageNum              foundImageNum     = 0;
 163             const MachOAnalyzer*  mh                = nullptr;
 164             const char*           filePath          = nullptr;
 165             LoadedFileInfo        loadedFileInfo;
 166
 167             // look in dyld cache
 168             filePath = possiblePath;
 169             char realPath[MAXPATHLEN];
 170             if ( _dyldImageArray != nullptr ) {
 171                 uint32_t dyldCacheImageIndex;
 172                 bool foundInCache = _dyldCache->hasImagePath(possiblePath, dyldCacheImageIndex);
 173                 if ( !foundInCache && fileFound ) {
 174                     // see if this is an OS dylib/bundle with a pre-built dlopen closure
 175                     // We can only use the pre-built closure if we are dynamic linkage (a dlopen) and
 176                     // there are no roots
 177                     if ( canUseSharedCacheClosure && (linkageType == LinkageType::kDynamic) ) {
 178                         if (const dyld3::closure::Image* otherImage = _dyldCache->findDlopenOtherImage(possiblePath) ) {
 179                             uint64_t expectedInode;
 180                             uint64_t expectedModTime;
 181                             if ( !otherImage->isInvalid()  ) {
 182                                 bool hasInodeInfo = otherImage->hasFileModTimeAndInode(expectedInode, expectedModTime);
 183                                 // use pre-built Image if it does not have mtime/inode or it does and it has matches current file info
 184                                 if ( !hasInodeInfo || ((expectedInode == fileFoundINode) && (expectedModTime == fileFoundMTime)) ) {
 185                                     loadedFileInfo = MachOAnalyzer::load(_diag, _fileSystem, possiblePath, _archs, _platform, realPath);
 186                                     if ( _diag.noError() ) {
 187                                         mh = (const MachOAnalyzer*)loadedFileInfo.fileContent;
 188                                         foundImageNum = otherImage->imageNum();
 189                                         unmapWhenDone = true;
 190                                         contentRebased = false;
 191                                         hasInits = otherImage->hasInitializers() || otherImage->mayHavePlusLoads();
 192                                         // Use the realpath in the case where we loaded a symlink
 193                                         // The closure must have recordered an alias path
 194                                         if (realPath[0] != '\0')
 195                                             filePath = realPath;
 196                                     }
 197                                 }
 198                             }
 199                         }
 200                     }
 201                     // if not found in cache, may be a symlink to something in cache
 202                     if ( mh == nullptr ) {
 203                         if ( _fileSystem.getRealPath(possiblePath, realPath) ) {
 204                             foundInCache = _dyldCache->hasImagePath(realPath, dyldCacheImageIndex);
 205                             if ( foundInCache ) {
 206                                 filePath = realPath;
 207 #if BUILDING_LIBDYLD
 208                                 // handle case where OS dylib was updated after this process launched
 209                                 if ( foundInCache ) {
 210                                     for (BuilderLoadedImage& li: _loadedImages) {
 211                                         if ( strcmp(li.path(), realPath) == 0 ) {
 212                                             foundImage = &li;
 213                                             result = true;
 214                                             stop = true;
 215                                             return;
 216                                         }
 217                                     }
 218                                 }
 219 #endif
 220                             }
 221                         }
 222                     }
 223                 }
 224
 225                 // if using a cached dylib, look to see if there is an override
 226                 if ( foundInCache ) {
 227                     ImageNum dyldCacheImageNum = dyldCacheImageIndex + 1;
 228                     bool useCache = true;
 229                     markNeverUnload = true; // dylibs in cache, or dylibs that override cache should not be unloaded at runtime
 230                     const Image* image = _dyldImageArray->imageForNum(dyldCacheImageNum);
 231                     if ( image->overridableDylib() ) {
 232                         if ( fileFound ) {
 233                             uint64_t expectedInode;
 234                             uint64_t expectedModTime;
 235                             if ( image->hasFileModTimeAndInode(expectedInode, expectedModTime) ) {
 236                                 // macOS where dylibs remain on disk.  only use cache if mtime and inode have not changed
 237                                 useCache = ( (fileFoundINode == expectedInode) && (fileFoundMTime == expectedModTime) );
 238                             }
 239                             else if ( _makingClosuresInCache ) {
 240                                 // during iOS cache build, don't look at files on disk, use ones in cache
 241                                 useCache = true;
 242                             }
 243                             else {
 244                                 // iOS internal build. Any disk on cache overrides cache
 245                                 useCache = false;
 246                             }
 247                         }
 248                         if ( !useCache ) {
 249                             overrideImageNum = dyldCacheImageNum;
 250                             _foundDyldCacheRoots = true;
 251                         }
 252                     }
 253                     if ( useCache ) {
 254                         foundImageNum = dyldCacheImageNum;
 255                         mh = (MachOAnalyzer*)_dyldCache->getIndexedImageEntry(foundImageNum-1, loadedFileInfo.mtime, loadedFileInfo.inode);
 256                         unmapWhenDone = false;
 257                         // if we are building ImageArray in dyld cache, content is not rebased
 258                         contentRebased = !_makingDyldCacheImages && _dyldCacheIsLive;
 259                         hasInits = image->hasInitializers() || image->mayHavePlusLoads();
 260                         // If the cache format is different from dyld/libdyld then we can't use this closure.
 261                         if ( (_dyldCache->header.formatVersion != dyld3::closure::kFormatVersion) || !canUseSharedCacheClosure ) {
 262                             mustBuildClosure = true;
 263                             _foundDyldCacheRoots = true;
 264                         }
 265                     }
 266                 }
 267             }
 268
 269             // If we are building the cache, and don't find an image, then it might be weak so just return
 270             if (_makingDyldCacheImages) {
 271                 addMustBeMissingPath(possiblePath);
 272                 return;
 273             }
 274
 275              // if not found yet, mmap file
 276             if ( mh == nullptr ) {
 277                 loadedFileInfo = MachOAnalyzer::load(_diag, _fileSystem, filePath, _archs, _platform, realPath);
 278                 mh = (const MachOAnalyzer*)loadedFileInfo.fileContent;
 279                 if ( mh == nullptr ) {
 280                     // Don't add must be missing paths for dlopen as we don't cache dlopen closures
 281                     if (_isLaunchClosure) {
 282                         // If we found the file then we want to skip it as its not a valid macho for this platform/arch
 283                         // We can't record skipped file mtime/inode for caches built on a different machine that it runs on.
 284                         // In that case, we expect the file to be mastered out, as otherwise we couldn't detect if its
 285                         // changed or not on the device side
 286                         if (fileFound && inodesMatchRuntime) {
 287                             addSkippedFile(possiblePath, fileFoundINode, fileFoundMTime);
 288                         } else {
 289                             addMustBeMissingPath(possiblePath);
 290                         }
 291                     }
 292                     return;
 293                 }
 294                 if ( linkageType != LinkageType::kDynamic ) {
 295                     // LC_LOAD_DYLIB can only link with dylibs, and DYLD_INSERT_LIBRARIES can only be dylibs
 296                     if ( !mh->isDylib() ) {
 297                         _diag.error("found '%s' which is not a dylib.  Needed by '%s'", filePath, forImageChain.image.path());
 298                         return;
 299                     }
 300                     // verify this is compatable dylib version
 301                     const char* installName;
 302                     uint32_t    foundCompatVers;
 303                     uint32_t    foundCurrentVers;
 304                     mh->getDylibInstallName(&installName, &foundCompatVers, &foundCurrentVers);
 305                     if ( (foundCompatVers < compatVersion) && mh->enforceCompatVersion() ) {
 306                         char foundStr[32];
 307                         char requiredStr[32];
 308                         MachOFile::packedVersionToString(foundCompatVers, foundStr);
 309                         MachOFile::packedVersionToString(compatVersion, requiredStr);
 310                         _diag.error("found '%s' which has compat version (%s) which is less than required (%s).  Needed by '%s'",
 311                                     filePath, foundStr, requiredStr, forImageChain.image.path());
 312                         return;
 313                     }
 314                 }
 315                 else if ( mh->isMainExecutable() ) {
 316                     // when dlopen()ing a main executable, it must be dynamic Position Independent Executable
 317                     if ( !mh->isPIE() || !mh->isDynamicExecutable() ) {
 318                         _diag.error("not PIE");
 319                         return;
 320                     }
 321                 }
 322                 // Use the realpath in the case where we loaded a symlink
 323                 // The closure must have recordered an alias path
 324                 if (realPath[0] != '\0')
 325                     filePath = realPath;
 326                 foundImageNum = _startImageNum + _nextIndex++;
 327                 _foundNonCachedImage = true;
 328                 mustBuildClosure = true;
 329                 unmapWhenDone = true;
 330             } else {
 331                 loadedFileInfo.fileContent = mh;
 332             }
 333
 334             // if path is not original path, or its an inserted path (as forEachInColonList uses a stack temporary)
 335             if ( (filePath != loadPath) || (linkageType == LinkageType::kInserted) ) {
 336                 // possiblePath may be a temporary (stack) string, since we found file at that path, make it permanent
 337                 filePath = strdup_temp(filePath);
 338                 // check if this overrides what would have been found in cache
 339                 // This is the case where we didn't find the image with the path in the shared cache, perhaps as it used library paths
 340                 // but the path we requested had pointed in to the cache
 341                 // FIXME: What if load path is via an @rpath and we will override the cache?
 342                 if ( overrideImageNum == 0 ) {
 343                     if ( _dyldImageArray != nullptr )  {
 344                         uint32_t dyldCacheImageIndex;
 345                         if ( _dyldCache->hasImagePath(loadPath, dyldCacheImageIndex) ) {
 346                             ImageNum possibleOverrideNum = dyldCacheImageIndex+1;
 347                             if ( possibleOverrideNum != foundImageNum )
 348                                 overrideImageNum = possibleOverrideNum;
 349                         }
 350                     }
 351                 }
 352             }
 353
 354             if ( !markNeverUnload ) {
 355                 switch (linkageType) {
 356                     case LinkageType::kStatic:
 357                         // Static linkages can only be unloaded if the image loading us can be unloaded
 358                         markNeverUnload = forImageChain.image.markNeverUnload;
 359                         break;
 360                     case LinkageType::kDynamic:
 361                         markNeverUnload = false;
 362                         break;
 363                     case LinkageType::kInserted:
 364                         // Inserted libraries must never be unloaded
 365                         markNeverUnload = true;
 366                         break;
 367                 };
 368             }
 369
 370             if ( !markNeverUnload ) {
 371                 // If the parent didn't force us to be never unload, other conditions still may
 372                 if ( mh->hasThreadLocalVariables() ) {
 373                     markNeverUnload = true;
 374                 } else if ( mh->hasObjC() && mh->isDylib() ) {
 375                     markNeverUnload = true;
 376                 } else {
 377                     // record if image has DOF sections
 378                     __block bool hasDOFs = false;
 379                     mh->forEachDOFSection(_diag, ^(uint32_t offset) {
 380                         hasDOFs = true;
 381                     });
 382                     if ( hasDOFs )
 383                         markNeverUnload = true;
 384                 }
 385             }
 386
 387             // Set the path again just in case it was strdup'ed.
 388             loadedFileInfo.path = filePath;
 389
 390             // add new entry
 391             BuilderLoadedImage entry;
 392             entry.loadedFileInfo        = loadedFileInfo;
 393             entry.imageNum              = foundImageNum;
 394             entry.unmapWhenDone         = unmapWhenDone;
 395             entry.contentRebased        = contentRebased;
 396             entry.hasInits              = hasInits;
 397             entry.markNeverUnload       = markNeverUnload;
 398             entry.rtldLocal             = false;
 399             entry.isBadImage            = false;
 400             entry.mustBuildClosure      = mustBuildClosure;
 401             entry.hasMissingWeakImports = false;
 402             entry.overrideImageNum      = overrideImageNum;
 403             _loadedImages.push_back(entry);
 404             foundImage = &_loadedImages.back();
 405             if ( isFallbackPath )
 406                 _fallbackPathUsed = true;
 407             stop = true;
 408             result = true;
 409         });
 410         if (result)
 411             stopPathVariant = true;
 412     }, _platform);
 413
 414     // If we found a file, but also had an error, then we must have logged a diagnostic for a file we couldn't use.
 415     // Clear that for now.
 416     // FIXME: Surface this to the user in case they wanted to see the error
 417     if (result && _diag.hasError())
 418         _diag.clearError();
 419
 420     return result;
 421 }
 422
 423 bool ClosureBuilder::expandAtLoaderPath(const char* loadPath, bool fromLCRPATH, const BuilderLoadedImage& loadedImage, char fixedPath[])
 424 {
 425     switch ( _atPathHandling ) {
 426         case AtPath::none:
 427             return false;
 428         case AtPath::onlyInRPaths:
 429             if ( !fromLCRPATH ) {
 430                 // <rdar://42360708> allow @loader_path in LC_LOAD_DYLIB during dlopen()
 431                 if ( _isLaunchClosure )
 432                     return false;
 433             }
 434             break;
 435         case AtPath::all:
 436             break;
 437     }
 438     if ( strncmp(loadPath, "@loader_path/", 13) != 0 )
 439         return false;
 440
 441     strlcpy(fixedPath, loadedImage.path(), PATH_MAX);
 442     char* lastSlash = strrchr(fixedPath, '/');
 443     if ( lastSlash != nullptr ) {
 444         strcpy(lastSlash+1, &loadPath[13]);
 445         return true;
 446     }
 447     return false;
 448 }
 449
 450 bool ClosureBuilder::expandAtExecutablePath(const char* loadPath, bool fromLCRPATH, char fixedPath[])
 451 {
 452     switch ( _atPathHandling ) {
 453         case AtPath::none:
 454             return false;
 455         case AtPath::onlyInRPaths:
 456             if ( !fromLCRPATH )
 457                 return false;
 458             break;
 459         case AtPath::all:
 460             break;
 461     }
 462     if ( strncmp(loadPath, "@executable_path/", 17) != 0 )
 463         return false;
 464
 465     if ( _atPathHandling != AtPath::all )
 466         return false;
 467
 468     strlcpy(fixedPath, _mainProgLoadPath, PATH_MAX);
 469     char* lastSlash = strrchr(fixedPath, '/');
 470     if ( lastSlash != nullptr ) {
 471         strcpy(lastSlash+1, &loadPath[17]);
 472         return true;
 473     }
 474     return false;
 475 }
 476
 477 void ClosureBuilder::forEachResolvedPathVar(const char* loadPath, const LoadedImageChain& forImageChain,
 478                                             bool implictRPath, LinkageType linkageType,
 479                                             void (^handler)(const char* possiblePath, bool& stop))
 480 {
 481     // don't expand @loader_path or @executable_path if disallowed
 482     if ( (_atPathHandling == AtPath::none) && (loadPath[0] == '@') && (loadPath[1] != 'r') ) {
 483         bool stop = false;
 484         handler(loadPath, stop);
 485         return;
 486     }
 487
 488     // quick out if not @ path or not implicit rpath
 489     if ( !implictRPath && (loadPath[0] != '@') ) {
 490         bool stop = false;
 491         handler(loadPath, stop);
 492         return;
 493     }
 494
 495     // expand @loader_path
 496     // Note this isn't supported for DYLD_INSERT_LIBRARIES
 497     BLOCK_ACCCESSIBLE_ARRAY(char, tempPath, PATH_MAX);  // read as:  char tempPath[PATH_MAX];
 498     if ( (linkageType != LinkageType::kInserted) && expandAtLoaderPath(loadPath, false, forImageChain.image, tempPath) ) {
 499         bool stop = false;
 500         handler(tempPath, stop);
 501         return;
 502     }
 503
 504     // expand @executable_path
 505     // Note this is supported for DYLD_INSERT_LIBRARIES
 506     if ( expandAtExecutablePath(loadPath, false, tempPath) ) {
 507         bool stop = false;
 508         handler(tempPath, stop);
 509         return;
 510     }
 511
 512     // expand @rpath
 513     // Note this isn't supported for DYLD_INSERT_LIBRARIES
 514     const char* rpathTail = nullptr;
 515     char implicitRpathBuffer[PATH_MAX];
 516     if ( linkageType != LinkageType::kInserted ) {
 517         if ( strncmp(loadPath, "@rpath/", 7) == 0 ) {
 518             // note: rpathTail starts with '/'
 519             rpathTail = &loadPath[6];
 520         }
 521         else if ( implictRPath ) {
 522             // make rpathTail starts with '/'
 523             strlcpy(implicitRpathBuffer, "/", PATH_MAX);
 524             strlcat(implicitRpathBuffer, loadPath, PATH_MAX);
 525             rpathTail = implicitRpathBuffer;
 526         }
 527     }
 528     if ( rpathTail != nullptr ) {
 529         // rpath is expansion is technically a stack of rpath dirs built starting with main executable and pushing
 530         // LC_RPATHS from each dylib as they are recursively loaded.  Our imageChain represents that stack.
 531         __block bool done = false;
 532         for (const LoadedImageChain* link = &forImageChain; (link != nullptr) && !done; link = link->previous) {
 533             link->image.loadAddress()->forEachRPath(^(const char* rPath, bool& stop) {
 534                 // fprintf(stderr, "LC_RPATH %s from %s\n", rPath, link->image.loadedFileInfo.path);
 535                 if ( expandAtLoaderPath(rPath, true, link->image, tempPath) || expandAtExecutablePath(rPath, true, tempPath) ) {
 536                     // @loader_path allowed and expended
 537                     strlcat(tempPath, rpathTail, PATH_MAX);
 538                     handler(tempPath, stop);
 539                 }
 540                 else if ( rPath[0] == '/' ) {
 541                     // LC_RPATH is an absolute path, not blocked by AtPath::none
 542                     strlcpy(tempPath, rPath, PATH_MAX);
 543                     strlcat(tempPath, rpathTail, PATH_MAX);
 544                     handler(tempPath, stop);
 545                 }
 546                 if (stop)
 547                     done = true;
 548 #if 0
 549                 if ( _fileSystem.fileExists(tempPath) ) {
 550                     stop = true;
 551                     result = strdup_temp(tempPath);
 552                 }
 553                 else {
 554                     // Don't add must be missing paths for dlopen as we don't cache dlopen closures
 555                     if (_isLaunchClosure) {
 556                         addMustBeMissingPath(tempPath);
 557                     }
 558                 }
 559 #endif
 560             });
 561         }
 562         if (done)
 563             return;
 564     }
 565
 566     bool stop = false;
 567     handler(loadPath, stop);
 568 }
 569
 570 const char* ClosureBuilder::strdup_temp(const char* path)
 571 {
 572     if ( _tempPaths == nullptr )
 573         _tempPaths = PathPool::allocate();
 574     return _tempPaths->add(path);
 575 }
 576
 577 void ClosureBuilder::addMustBeMissingPath(const char* path)
 578 {
 579     //fprintf(stderr, "must be missing: %s\n", path);
 580     if ( _mustBeMissingPaths == nullptr )
 581         _mustBeMissingPaths = PathPool::allocate();
 582     _mustBeMissingPaths->add(path);
 583 }
 584
 585 void ClosureBuilder::addSkippedFile(const char* path, uint64_t inode, uint64_t mtime)
 586 {
 587     _skippedFiles.push_back({ strdup_temp(path), inode, mtime });
 588 }
 589
 590 ClosureBuilder::BuilderLoadedImage& ClosureBuilder::findLoadedImage(ImageNum imageNum)
 591 {
 592     for (BuilderLoadedImage& li : _loadedImages) {
 593         if ( li.imageNum == imageNum ) {
 594             return li;
 595         }
 596     }
 597     for (BuilderLoadedImage& li : _loadedImages) {
 598         if ( li.overrideImageNum == imageNum ) {
 599             return li;
 600         }
 601     }
 602     assert(0 && "LoadedImage not found");
 603 }
 604
 605 ClosureBuilder::BuilderLoadedImage& ClosureBuilder::findLoadedImage(const MachOAnalyzer* mh)
 606 {
 607     for (BuilderLoadedImage& li : _loadedImages) {
 608         if ( li.loadAddress() == mh ) {
 609              return li;
 610         }
 611     }
 612     assert(0 && "LoadedImage not found");
 613 }
 614
 615 const MachOAnalyzer* ClosureBuilder::machOForImageNum(ImageNum imageNum)
 616 {
 617     return findLoadedImage(imageNum).loadAddress();
 618 }
 619
 620 const MachOAnalyzer* ClosureBuilder::findDependent(const MachOLoaded* mh, uint32_t depIndex)
 621 {
 622     for (const BuilderLoadedImage& li : _loadedImages) {
 623         if ( li.loadAddress() == mh ) {
 624             if (li.isBadImage) {
 625                 // Bad image duting building group 1 closures, so the dependents array
 626                 // is potentially incomplete.
 627                 return nullptr;
 628             }
 629             ImageNum childNum = li.dependents[depIndex].imageNum();
 630             // This is typically something like a missing weak-dylib we are re-exporting a weak-import symbol from
 631             if (childNum == kMissingWeakLinkedImage)
 632                 return nullptr;
 633             return machOForImageNum(childNum);
 634         }
 635     }
 636     return nullptr;
 637 }
 638
 639 ImageNum ClosureBuilder::imageNumForMachO(const MachOAnalyzer* mh)
 640 {
 641     for (const BuilderLoadedImage& li : _loadedImages) {
 642         if ( li.loadAddress() == mh ) {
 643              return li.imageNum;
 644         }
 645     }
 646     assert(0 && "unknown mach-o");
 647     return 0;
 648 }
 649
 650 void ClosureBuilder::recursiveLoadDependents(LoadedImageChain& forImageChain, bool canUseSharedCacheClosure)
 651 {
 652     // if dependents is set, then we have already loaded this
 653     if ( forImageChain.image.dependents.begin() != nullptr )
 654         return;
 655
 656     uintptr_t startDepIndex = _dependencies.count();
 657     // add dependents
 658     __block uint32_t depIndex = 0;
 659     forImageChain.image.loadAddress()->forEachDependentDylib(^(const char* loadPath, bool isWeak, bool isReExport, bool isUpward, uint32_t compatVersion, uint32_t curVersion, bool &stop) {
 660         Image::LinkKind kind = Image::LinkKind::regular;
 661         if ( isWeak )
 662             kind = Image::LinkKind::weak;
 663         else if ( isReExport )
 664             kind = Image::LinkKind::reExport;
 665         else if ( isUpward )
 666             kind = Image::LinkKind::upward;
 667         BuilderLoadedImage* foundImage;
 668         if ( findImage(loadPath, forImageChain, foundImage, LinkageType::kStatic, compatVersion, canUseSharedCacheClosure) ) {
 669             ImageNum foundImageNum = foundImage->imageNum;
 670             if ( _diag.noError() )
 671                 _dependencies.push_back(Image::LinkedImage(kind, foundImageNum));
 672         }
 673         else if ( isWeak ) {
 674             _dependencies.push_back(Image::LinkedImage(Image::LinkKind::weak, kMissingWeakLinkedImage));
 675         }
 676         else {
 677             BLOCK_ACCCESSIBLE_ARRAY(char, extra, 4096);
 678             extra[0] = '\0';
 679             const char* targetLeaf = strrchr(loadPath, '/');
 680             if ( targetLeaf == nullptr )
 681                 targetLeaf = loadPath;
 682             if ( _mustBeMissingPaths != nullptr ) {
 683                 strcpy(extra, ", tried but didn't find: ");
 684                 _mustBeMissingPaths->forEachPath(^(const char* aPath) {
 685                     const char* aLeaf = strrchr(aPath, '/');
 686                     if ( aLeaf == nullptr )
 687                         aLeaf = aPath;
 688                   if ( strcmp(targetLeaf, aLeaf) == 0 ) {
 689                         strlcat(extra, "'", 4096);
 690                         strlcat(extra, aPath, 4096);
 691                         strlcat(extra, "' ", 4096);
 692                     }
 693                 });
 694             }
 695             if ( !_skippedFiles.empty() ) {
 696                 strcpy(extra, ", tried but invalid: ");
 697                 for (const SkippedFile& skippedFile : _skippedFiles) {
 698                     const char* aPath = skippedFile.path;
 699                     const char* aLeaf = strrchr(aPath, '/');
 700                     if ( aLeaf == nullptr )
 701                         aLeaf = aPath;
 702                     if ( strcmp(targetLeaf, aLeaf) == 0 ) {
 703                         strlcat(extra, "'", 4096);
 704                         strlcat(extra, aPath, 4096);
 705                         strlcat(extra, "' ", 4096);
 706                     }
 707                 }
 708             }
 709             if ( _diag.hasError() ) {
 710         #if BUILDING_CACHE_BUILDER
 711                 std::string errorMessageBuffer = _diag.errorMessage();
 712                 const char* msg = errorMessageBuffer.c_str();
 713         #else
 714                 const char* msg = _diag.errorMessage();
 715         #endif
 716                 char msgCopy[strlen(msg)+4];
 717                 strcpy(msgCopy, msg);
 718                 _diag.error("dependent dylib '%s' not found for '%s'. %s", loadPath, forImageChain.image.path(), msgCopy);
 719             }
 720             else {
 721                 _diag.error("dependent dylib '%s' not found for '%s'%s", loadPath, forImageChain.image.path(), extra);
 722             }
 723             if ( _launchErrorInfo != nullptr ) {
 724                 _launchErrorInfo->kind              = DYLD_EXIT_REASON_DYLIB_MISSING;
 725                 _launchErrorInfo->clientOfDylibPath = strdup_temp(forImageChain.image.path());
 726                 _launchErrorInfo->targetDylibPath   = strdup_temp(loadPath);
 727                 _launchErrorInfo->symbol            = nullptr;
 728            }
 729         }
 730         ++depIndex;
 731         if ( _diag.hasError() )
 732             stop = true;
 733     });
 734     if ( _diag.hasError() )
 735         return;
 736     forImageChain.image.dependents = _dependencies.subArray(startDepIndex, depIndex);
 737
 738     // breadth first recurse
 739     for (Image::LinkedImage dep : forImageChain.image.dependents) {
 740         // don't recurse upwards
 741         if ( dep.kind() == Image::LinkKind::upward )
 742             continue;
 743         // don't recurse down missing weak links
 744         if ( (dep.kind() == Image::LinkKind::weak) && (dep.imageNum() == kMissingWeakLinkedImage) )
 745             continue;
 746         BuilderLoadedImage& depLoadedImage = findLoadedImage(dep.imageNum());
 747         LoadedImageChain chain = { &forImageChain, depLoadedImage };
 748         recursiveLoadDependents(chain, canUseSharedCacheClosure);
 749         if ( _diag.hasError() )
 750             break;
 751     }
 752 }
 753
 754 void ClosureBuilder::loadDanglingUpwardLinks(bool canUseSharedCacheClosure)
 755 {
 756     bool danglingFixed;
 757     do {
 758         danglingFixed = false;
 759         for (BuilderLoadedImage& li : _loadedImages) {
 760             if ( li.dependents.begin() == nullptr ) {
 761                 // this image has not have dependents set (probably a dangling upward link or referenced by upward link)
 762                 LoadedImageChain chain = { nullptr, li };
 763                 recursiveLoadDependents(chain, canUseSharedCacheClosure);
 764                 danglingFixed = true;
 765                 break;
 766             }
 767         }
 768     } while (danglingFixed && _diag.noError());
 769 }
 770
 771 bool ClosureBuilder::overridableDylib(const BuilderLoadedImage& forImage)
 772 {
 773     // only set on dylibs in the dyld shared cache
 774     if ( !_makingDyldCacheImages )
 775         return false;
 776
 777     // on macOS dylibs always override cache
 778     if ( _platform == Platform::macOS )
 779         return true;
 780
 781     // on embedded platforms with Internal cache, allow overrides
 782     if ( !_makingCustomerCache )
 783         return true;
 784
 785     // embedded platform customer caches, no overrides
 786     return false;  // FIXME, allow libdispatch.dylib to be overridden
 787 }
 788
 789 void ClosureBuilder::buildImage(ImageWriter& writer, BuilderLoadedImage& forImage)
 790 {
 791     const MachOAnalyzer* macho = forImage.loadAddress();
 792         // set ImageNum
 793     writer.setImageNum(forImage.imageNum);
 794
 795     // set flags
 796     writer.setHasWeakDefs(macho->hasWeakDefs());
 797     writer.setIsBundle(macho->isBundle());
 798     writer.setIsDylib(macho->isDylib());
 799     writer.setIs64(macho->is64());
 800     writer.setIsExecutable(macho->isMainExecutable());
 801     writer.setUses16KPages(macho->uses16KPages());
 802     writer.setOverridableDylib(overridableDylib(forImage));
 803     writer.setInDyldCache(macho->inDyldCache());
 804     if ( macho->hasObjC() ) {
 805         writer.setHasObjC(true);
 806         bool hasPlusLoads = macho->hasPlusLoadMethod(_diag);
 807         writer.setHasPlusLoads(hasPlusLoads);
 808         if ( hasPlusLoads )
 809             forImage.hasInits = true;
 810     }
 811     else {
 812         writer.setHasObjC(false);
 813         writer.setHasPlusLoads(false);
 814     }
 815
 816     if ( forImage.markNeverUnload ) {
 817         writer.setNeverUnload(true);
 818     }
 819
 820 #if BUILDING_DYLD || BUILDING_LIBDYLD
 821     if ( _foundDyldCacheRoots ) {
 822         // If we had roots, then some images are potentially on-disk while others are
 823         // being rebuilt for a new initializer order, but do not exist on disk
 824         if ( macho->inDyldCache() && !_dyldCache->header.dylibsExpectedOnDisk ) {
 825             // don't add file info for shared cache files mastered out of final file system
 826         }
 827         else {
 828             // file is either not in cache or is in cache but not mastered out
 829             writer.setFileInfo(forImage.loadedFileInfo.inode, forImage.loadedFileInfo.mtime);
 830         }
 831     } else {
 832         // shared cache not built by dyld or libdyld.dylib, so must be real file
 833         writer.setFileInfo(forImage.loadedFileInfo.inode, forImage.loadedFileInfo.mtime);
 834     }
 835 #else
 836     if ( _platform == Platform::macOS || MachOFile::isSimulatorPlatform(_platform) ) {
 837         if ( macho->inDyldCache() && !_dyldCache->header.dylibsExpectedOnDisk ) {
 838             // don't add file info for shared cache files mastered out of final file system
 839         }
 840         else {
 841             // file is either not in cache or is in cache but not mastered out
 842             writer.setFileInfo(forImage.loadedFileInfo.inode, forImage.loadedFileInfo.mtime);
 843         }
 844     }
 845     else {
 846         // all other platforms, cache is built off-device, so inodes are not known
 847     }
 848 #endif
 849
 850     // add info on how to load image
 851     if ( !macho->inDyldCache() ) {
 852         writer.setMappingInfo(forImage.loadedFileInfo.sliceOffset, macho->mappedSize());
 853         // add code signature, if signed
 854         uint32_t codeSigFileOffset;
 855         uint32_t codeSigSize;
 856         if ( macho->hasCodeSignature(codeSigFileOffset, codeSigSize) ) {
 857             writer.setCodeSignatureLocation(codeSigFileOffset, codeSigSize);
 858             macho->forEachCDHash(^(const uint8_t *cdHash) {
 859                 writer.addCDHash(cdHash);
 860             });
 861         }
 862         // add FairPlay encryption range if encrypted
 863         uint32_t fairPlayFileOffset;
 864         uint32_t fairPlaySize;
 865         if ( macho->isFairPlayEncrypted(fairPlayFileOffset, fairPlaySize) ) {
 866             writer.setFairPlayEncryptionRange(fairPlayFileOffset, fairPlaySize);
 867         }
 868     }
 869
 870     // set path
 871     writer.addPath(forImage.path());
 872     if ( _aliases != nullptr ) {
 873         for (const CachedDylibAlias& alias : *_aliases) {
 874             if ( strcmp(alias.realPath, forImage.path()) == 0 )
 875                 writer.addPath(alias.aliasPath);
 876         }
 877     }
 878
 879     // set uuid, if has one
 880     uuid_t uuid;
 881     if ( macho->getUuid(uuid) )
 882         writer.setUUID(uuid);
 883
 884     // set dependents
 885     writer.setDependents(forImage.dependents);
 886
 887     // set segments
 888     addSegments(writer, macho);
 889
 890     // record if this dylib overrides something in the cache
 891     if ( forImage.overrideImageNum != 0 ) {
 892         writer.setAsOverrideOf(forImage.overrideImageNum);
 893         const char* overridePath = _dyldImageArray->imageForNum(forImage.overrideImageNum)->path();
 894         writer.addPath(overridePath);
 895         if ( strcmp(overridePath, "/usr/lib/system/libdyld.dylib") == 0 )
 896             _libDyldImageNum = forImage.imageNum;
 897         else if ( strcmp(overridePath, "/usr/lib/libSystem.B.dylib") == 0 )
 898             _libSystemImageNum = forImage.imageNum;
 899     }
 900
 901     // do fix up info for non-cached, and cached if building cache
 902     if ( !macho->inDyldCache() || _makingDyldCacheImages ) {
 903         if ( macho->hasChainedFixups() ) {
 904             addChainedFixupInfo(writer, forImage);
 905         }
 906         else {
 907             if ( _handlers != nullptr ) {
 908                 reportRebasesAndBinds(writer, forImage);
 909             }
 910             else {
 911                 // Note we have to do binds before rebases so that we know if we have missing lazy binds
 912                 addBindInfo(writer, forImage);
 913                 if ( _diag.noError() )
 914                     addRebaseInfo(writer, macho);
 915             }
 916         }
 917     }
 918     if ( _diag.hasError() ) {
 919         writer.setInvalid();
 920         return;
 921     }
 922
 923     // Don't build iOSMac for now.  Just add an invalid placeholder
 924     if ( _makingDyldCacheImages && strncmp(forImage.path(), "/System/iOSSupport/", 19) == 0 ) {
 925         writer.setInvalid();
 926         return;
 927     }
 928
 929     // add initializers
 930     bool contentRebased = forImage.contentRebased;
 931     __block unsigned initCount = 0;
 932     Diagnostics initializerDiag;
 933     macho->forEachInitializer(initializerDiag, contentRebased, ^(uint32_t offset) {
 934         ++initCount;
 935     }, _dyldCache);
 936     if ( initializerDiag.noError() ) {
 937         if ( initCount != 0 ) {
 938             BLOCK_ACCCESSIBLE_ARRAY(uint32_t, initOffsets, initCount);
 939             __block unsigned index = 0;
 940            macho->forEachInitializer(_diag, contentRebased, ^(uint32_t offset) {
 941                 initOffsets[index++] = offset;
 942             }, _dyldCache);
 943             writer.setInitOffsets(initOffsets, initCount);
 944             forImage.hasInits = true;
 945         }
 946     }
 947     else {
 948         // mod_init_func section is malformed, might be self modifying pointers
 949         macho->forEachInitializerPointerSection(_diag, ^(uint32_t sectionOffset, uint32_t sectionSize, const uint8_t* content, bool& stop) {
 950             writer.setInitSectRange(sectionOffset, sectionSize);
 951             forImage.hasInits = true;
 952         });
 953     }
 954
 955
 956     // add terminators (except for dylibs in the cache because they are never unloaded)
 957     if ( !macho->inDyldCache() ) {
 958         __block unsigned termCount = 0;
 959         macho->forEachTerminator(_diag, contentRebased, ^(uint32_t offset) {
 960             ++termCount;
 961         });
 962         if ( termCount != 0 ) {
 963             BLOCK_ACCCESSIBLE_ARRAY(uint32_t, termOffsets, termCount);
 964             __block unsigned index = 0;
 965             macho->forEachTerminator(_diag, contentRebased, ^(uint32_t offset) {
 966                 termOffsets[index++] = offset;
 967             });
 968             writer.setTermOffsets(termOffsets, termCount);
 969         }
 970     }
 971
 972     // record if image has DOF sections
 973     STACK_ALLOC_ARRAY(uint32_t, dofSectionOffsets, 256);
 974     macho->forEachDOFSection(_diag, ^(uint32_t offset) {
 975         dofSectionOffsets.push_back(offset);
 976     });
 977     if ( !dofSectionOffsets.empty() ) {
 978         writer.setDofOffsets(dofSectionOffsets);
 979     }
 980
 981 }
 982
 983 void ClosureBuilder::addSegments(ImageWriter& writer, const MachOAnalyzer* mh)
 984 {
 985     const uint32_t segCount = mh->segmentCount();
 986     if ( mh->inDyldCache() ) {
 987         uint64_t cacheUnslideBaseAddress = _dyldCache->unslidLoadAddress();
 988         BLOCK_ACCCESSIBLE_ARRAY(Image::DyldCacheSegment, segs, segCount);
 989         mh->forEachSegment(^(const MachOAnalyzer::SegmentInfo& info, bool& stop) {
 990             segs[info.segIndex] = { (uint32_t)(info.vmAddr-cacheUnslideBaseAddress), (uint32_t)info.vmSize, info.protections };
 991         });
 992         writer.setCachedSegments(segs, segCount);
 993     }
 994     else {
 995         const uint32_t   pageSize          = (mh->uses16KPages() ? 0x4000 : 0x1000);
 996         __block uint32_t diskSegIndex      = 0;
 997         __block uint32_t totalPageCount    = 0;
 998         __block uint32_t lastFileOffsetEnd = 0;
 999         __block uint64_t lastVmAddrEnd     = 0;
1000         BLOCK_ACCCESSIBLE_ARRAY(Image::DiskSegment, dsegs, segCount*3); // room for padding
1001         mh->forEachSegment(^(const MachOAnalyzer::SegmentInfo& info, bool& stop) {
1002             if ( (info.fileOffset != 0) && (info.fileOffset != lastFileOffsetEnd) ) {
1003                 Image::DiskSegment filePadding;
1004                 filePadding.filePageCount   = (info.fileOffset - lastFileOffsetEnd)/pageSize;
1005                 filePadding.vmPageCount     = 0;
1006                 filePadding.permissions     = 0;
1007                 filePadding.paddingNotSeg   = 1;
1008                 dsegs[diskSegIndex++] = filePadding;
1009             }
1010             if ( (lastVmAddrEnd != 0) && (info.vmAddr != lastVmAddrEnd) ) {
1011                 Image::DiskSegment vmPadding;
1012                 vmPadding.filePageCount   = 0;
1013                 vmPadding.vmPageCount     = (info.vmAddr - lastVmAddrEnd)/pageSize;
1014                 vmPadding.permissions     = 0;
1015                 vmPadding.paddingNotSeg   = 1;
1016                 dsegs[diskSegIndex++] = vmPadding;
1017                 totalPageCount += vmPadding.vmPageCount;
1018             }
1019             {
1020                 Image::DiskSegment segInfo;
1021                 segInfo.filePageCount   = (info.fileSize+pageSize-1)/pageSize;
1022                 segInfo.vmPageCount     = (info.vmSize+pageSize-1)/pageSize;
1023                 segInfo.permissions     = info.protections & 7;
1024                 segInfo.paddingNotSeg   = 0;
1025                 if ( info.readOnlyData )
1026                     segInfo.permissions  = Image::DiskSegment::kReadOnlyDataPermissions;
1027                 dsegs[diskSegIndex++] = segInfo;
1028                 totalPageCount   += segInfo.vmPageCount;
1029                 if ( info.fileSize != 0 )
1030                     lastFileOffsetEnd = (uint32_t)(info.fileOffset + info.fileSize);
1031                 if ( info.vmSize != 0 )
1032                     lastVmAddrEnd     = info.vmAddr + info.vmSize;
1033             }
1034         });
1035         writer.setDiskSegments(dsegs, diskSegIndex);
1036     }
1037 }
1038
1039 static bool isTupleFixup(uint64_t tupleSectVmStartOffset, uint64_t tupleSectVmEndOffset, uint64_t imageOffsetOfFixup, uint32_t entrySize, uint32_t& tupleIndex)
1040 {
1041     if ( imageOffsetOfFixup < tupleSectVmStartOffset )
1042         return false;
1043     if ( imageOffsetOfFixup > tupleSectVmEndOffset )
1044         return false;
1045     uint64_t offsetIntoSection = imageOffsetOfFixup - tupleSectVmStartOffset;
1046     tupleIndex = (uint32_t)(offsetIntoSection/entrySize);
1047     return (tupleIndex*entrySize == offsetIntoSection) || ((tupleIndex*entrySize+entrySize/2) == offsetIntoSection);
1048 }
1049
1050 void ClosureBuilder::addInterposingTuples(LaunchClosureWriter& writer, const Image* image, const MachOAnalyzer* mh)
1051 {
1052     const unsigned pointerSize  = mh->pointerSize();
1053     const uint64_t baseAddress  = mh->preferredLoadAddress();
1054     mh->forEachInterposingSection(_diag, ^(uint64_t sectVmOffset, uint64_t sectVmSize, bool &stop) {
1055         const uint32_t entrySize = 2*pointerSize;
1056         const uint32_t tupleCount = (uint32_t)(sectVmSize/entrySize);
1057         const uint64_t sectVmEndOffset = sectVmOffset + sectVmSize;
1058         BLOCK_ACCCESSIBLE_ARRAY(InterposingTuple, resolvedTuples, tupleCount);
1059         for (uint32_t i=0; i < tupleCount; ++i) {
1060             resolvedTuples[i].stockImplementation.absolute.kind  = Image::ResolvedSymbolTarget::kindAbsolute;
1061             resolvedTuples[i].stockImplementation.absolute.value = 0;
1062             resolvedTuples[i].newImplementation.absolute.kind    = Image::ResolvedSymbolTarget::kindAbsolute;
1063             resolvedTuples[i].newImplementation.absolute.value   = 0;
1064         }
1065         // figure out what the replacement (rebase) and replacement (bind) of the tuple point to
1066         image->forEachFixup(^(uint64_t imageOffsetToRebase, bool& rebaseStop) {
1067             uint32_t tupleIndex;
1068             if ( isTupleFixup(sectVmOffset, sectVmEndOffset, imageOffsetToRebase, entrySize, tupleIndex) ) {
1069                 const void* content = (uint8_t*)mh + imageOffsetToRebase;
1070                 uint64_t unslidTargetAddress = mh->is64() ?  *(uint64_t*)content : *(uint32_t*)content;
1071                 resolvedTuples[tupleIndex].newImplementation.image.kind     = Image::ResolvedSymbolTarget::kindImage;
1072                 resolvedTuples[tupleIndex].newImplementation.image.imageNum = image->imageNum();
1073                 resolvedTuples[tupleIndex].newImplementation.image.offset   = unslidTargetAddress - mh->preferredLoadAddress();
1074             }
1075         },
1076         ^(uint64_t imageOffsetToBind, Image::ResolvedSymbolTarget bindTarget, bool &bindStop) {
1077             uint32_t tupleIndex;
1078             if ( isTupleFixup(sectVmOffset, sectVmEndOffset, imageOffsetToBind, entrySize, tupleIndex) ) {
1079                 resolvedTuples[tupleIndex].stockImplementation = bindTarget;
1080             }
1081         },
1082         ^(uint64_t imageOffsetToStartsInfo, const Array<Image::ResolvedSymbolTarget>& targets, bool& chainStop) {
1083             mh->withChainStarts(_diag, imageOffsetToStartsInfo, ^(const dyld_chained_starts_in_image* startsInfo) {
1084                 mh->forEachFixupInAllChains(_diag, startsInfo, false, ^(MachOLoaded::ChainedFixupPointerOnDisk* fixupLoc, const dyld_chained_starts_in_segment* segInfo, bool& fixupsStop) {
1085                     uint64_t fixupOffset = (uint8_t*)fixupLoc - (uint8_t*)mh;
1086                     uint32_t tupleIndex;
1087                     if ( !isTupleFixup(sectVmOffset, sectVmEndOffset, fixupOffset, entrySize, tupleIndex) )
1088                         return;
1089                     uint32_t bindOrdinal;
1090                     uint64_t rebaseTargetOffset;
1091                     if ( fixupLoc->isBind(segInfo->pointer_format, bindOrdinal) ) {
1092                         if ( bindOrdinal < targets.count() ) {
1093                             resolvedTuples[tupleIndex].stockImplementation = targets[bindOrdinal];
1094                         }
1095                         else {
1096                             _diag.error("out of range bind ordinal %d (max %lu)", bindOrdinal, targets.count());
1097                             fixupsStop = true;
1098                         }
1099                     }
1100                     else if ( fixupLoc->isRebase(segInfo->pointer_format, baseAddress, rebaseTargetOffset) ) {
1101                         resolvedTuples[tupleIndex].newImplementation.image.kind     = Image::ResolvedSymbolTarget::kindImage;
1102                         resolvedTuples[tupleIndex].newImplementation.image.imageNum = image->imageNum();
1103                         resolvedTuples[tupleIndex].newImplementation.image.offset   = rebaseTargetOffset;
1104                     }
1105                 });
1106             });
1107         },
1108         ^(uint64_t imageOffsetToFixup) {
1109             // objc optimisation can't be interposed so nothing to do here.
1110         },
1111         ^(uint64_t imageOffsetToBind, Image::ResolvedSymbolTarget bindTarget, bool &bindStop) {
1112             // objc protocol optimisation fixups can't be interposed so nothing to do here.
1113         },
1114         ^(uint64_t imageOffsetToFixup, uint32_t selectorIndex, bool inSharedCache, bool &fixupStop) {
1115             // objc selector optimisation fixups can't be interposed so nothing to do here.
1116         },
1117         ^(uint64_t imageOffsetToFixup, bool &fixupStop) {
1118             // objc stable Swift optimisation fixups can't be interposed so nothing to do here.
1119         },
1120         ^(uint64_t imageOffsetToFixup, bool &fixupStop) {
1121             // objc method list optimisation fixups can't be interposed so nothing to do here.
1122         });
1123
1124         // remove any tuples in which both sides are not set (or target is weak-import NULL)
1125         STACK_ALLOC_ARRAY(InterposingTuple, goodTuples, tupleCount);
1126         for (uint32_t i=0; i < tupleCount; ++i) {
1127             if ( (resolvedTuples[i].stockImplementation.image.kind != Image::ResolvedSymbolTarget::kindAbsolute)
1128               && (resolvedTuples[i].newImplementation.image.kind != Image::ResolvedSymbolTarget::kindAbsolute) )
1129                 goodTuples.push_back(resolvedTuples[i]);
1130         }
1131         writer.addInterposingTuples(goodTuples);
1132
1133         // if the target of the interposing is in the dyld shared cache, add a PatchEntry so the cache is fixed up at launch
1134         STACK_ALLOC_ARRAY(Closure::PatchEntry, patches, goodTuples.count());
1135         for (const InterposingTuple& aTuple : goodTuples) {
1136             if ( aTuple.stockImplementation.sharedCache.kind == Image::ResolvedSymbolTarget::kindSharedCache ) {
1137                 uint32_t imageIndex;
1138                 assert(_dyldCache->addressInText((uint32_t)aTuple.stockImplementation.sharedCache.offset, &imageIndex));
1139                 ImageNum imageInCache = imageIndex+1;
1140                 Closure::PatchEntry patch;
1141                 patch.exportCacheOffset      = (uint32_t)aTuple.stockImplementation.sharedCache.offset;
1142                 patch.overriddenDylibInCache = imageInCache;
1143                 patch.replacement            = aTuple.newImplementation;
1144                 patches.push_back(patch);
1145             }
1146         }
1147         writer.addCachePatches(patches);
1148     });
1149 }
1150
1151 void ClosureBuilder::addRebaseInfo(ImageWriter& writer, const MachOAnalyzer* mh)
1152 {
1153         const uint64_t ptrSize = mh->pointerSize();
1154     Image::RebasePattern maxLeapPattern = { 0xFFFFF, 0, 0xF };
1155     const uint64_t maxLeapCount = maxLeapPattern.repeatCount * maxLeapPattern.skipCount;
1156     STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::RebasePattern, rebaseEntries, 1024);
1157     __block uint64_t lastLocation = -ptrSize;
1158         mh->forEachRebase(_diag, !_foundMissingLazyBinds, ^(uint64_t runtimeOffset, bool& stop) {
1159         const uint64_t delta   = runtimeOffset - lastLocation;
1160         const bool     aligned = ((delta % ptrSize) == 0);
1161         if ( delta == ptrSize ) {
1162             // this rebase location is contiguous to previous
1163             if ( rebaseEntries.back().contigCount < 255 ) {
1164                 // just bump previous's contigCount
1165                 rebaseEntries.back().contigCount++;
1166             }
1167             else {
1168                 // previous contiguous run already has max 255, so start a new run
1169                 rebaseEntries.push_back({ 1, 1, 0 });
1170             }
1171         }
1172         else if ( aligned && (delta <= (ptrSize*15)) ) {
1173             // this rebase is within skip distance of last rebase
1174             rebaseEntries.back().skipCount = (uint8_t)((delta-ptrSize)/ptrSize);
1175             int lastIndex = (int)(rebaseEntries.count() - 1);
1176             if ( lastIndex > 1 ) {
1177                 if ( (rebaseEntries[lastIndex].contigCount == rebaseEntries[lastIndex-1].contigCount)
1178                   && (rebaseEntries[lastIndex].skipCount   == rebaseEntries[lastIndex-1].skipCount) ) {
1179                     // this entry as same contig and skip as prev, so remove it and bump repeat count of previous
1180                     rebaseEntries.pop_back();
1181                     rebaseEntries.back().repeatCount += 1;
1182                 }
1183             }
1184             rebaseEntries.push_back({ 1, 1, 0 });
1185         }
1186         else {
1187             uint64_t advanceCount = (delta-ptrSize);
1188             if ( (runtimeOffset < lastLocation) && (lastLocation != -ptrSize) ) {
1189                 // out of rebases! handle this be resting rebase offset to zero
1190                 rebaseEntries.push_back({ 0, 0, 0 });
1191                 advanceCount = runtimeOffset;
1192             }
1193             // if next rebase is too far to reach with one pattern, use series
1194             while ( advanceCount > maxLeapCount ) {
1195                 rebaseEntries.push_back(maxLeapPattern);
1196                 advanceCount -= maxLeapCount;
1197             }
1198             // if next rebase is not reachable with skipCount==1 or skipCount==15, add intermediate
1199             while ( advanceCount > maxLeapPattern.repeatCount ) {
1200                 uint64_t count = advanceCount / maxLeapPattern.skipCount;
1201                 rebaseEntries.push_back({ (uint32_t)count, 0, maxLeapPattern.skipCount });
1202                 advanceCount -= (count*maxLeapPattern.skipCount);
1203             }
1204             if ( advanceCount != 0 )
1205                 rebaseEntries.push_back({ (uint32_t)advanceCount, 0, 1 });
1206             rebaseEntries.push_back({ 1, 1, 0 });
1207         }
1208         lastLocation = runtimeOffset;
1209         });
1210     writer.setRebaseInfo(rebaseEntries);
1211
1212     // i386 programs also use text relocs to rebase stubs
1213     if ( mh->cputype == CPU_TYPE_I386 ) {
1214         STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::TextFixupPattern, textRebases, 512);
1215         __block uint64_t lastOffset = -4;
1216         mh->forEachTextRebase(_diag, ^(uint64_t runtimeOffset, bool& stop) {
1217             if ( textRebases.freeCount() < 2 ) {
1218                 _diag.error("too many text rebase locations (%ld) in %s", textRebases.maxCount(), writer.currentImage()->path());
1219                 stop = true;
1220             }
1221             bool mergedIntoPrevious = false;
1222             if ( (runtimeOffset > lastOffset) && !textRebases.empty() ) {
1223                 uint32_t skipAmount = (uint32_t)(runtimeOffset - lastOffset);
1224                 if ( (textRebases.back().repeatCount == 1) && (textRebases.back().skipCount == 0) ) {
1225                     textRebases.back().repeatCount = 2;
1226                     textRebases.back().skipCount   = skipAmount;
1227                     mergedIntoPrevious             = true;
1228                 }
1229                 else if ( textRebases.back().skipCount == skipAmount ) {
1230                     textRebases.back().repeatCount += 1;
1231                     mergedIntoPrevious = true;
1232                 }
1233             }
1234             if ( !mergedIntoPrevious ) {
1235                 Image::TextFixupPattern pattern;
1236                 pattern.target.raw    = 0;
1237                 pattern.startVmOffset = (uint32_t)runtimeOffset;
1238                 pattern.repeatCount   = 1;
1239                 pattern.skipCount     = 0;
1240                 textRebases.push_back(pattern);
1241             }
1242             lastOffset = runtimeOffset;
1243         });
1244         writer.setTextRebaseInfo(textRebases);
1245     }
1246 }
1247
1248
1249 void ClosureBuilder::forEachBind(BuilderLoadedImage& forImage, void (^handler)(uint64_t runtimeOffset, Image::ResolvedSymbolTarget target, const ResolvedTargetInfo& targetInfo, bool& stop),
1250                                                                void (^strongHandler)(const char* strongSymbolName),
1251                                                                void (^missingLazyBindHandler)())
1252 {
1253     __block int                         lastLibOrdinal  = 256;
1254     __block const char*                 lastSymbolName  = nullptr;
1255     __block uint64_t                    lastAddend      = 0;
1256     __block Image::ResolvedSymbolTarget target;
1257     __block ResolvedTargetInfo          targetInfo;
1258     forImage.loadAddress()->forEachBind(_diag, ^(uint64_t runtimeOffset, int libOrdinal, const char* symbolName, bool weakImport, bool lazyBind, uint64_t addend, bool& stop) {
1259         if ( (symbolName == lastSymbolName) && (libOrdinal == lastLibOrdinal) && (addend == lastAddend) ) {
1260             // same symbol lookup as last location
1261             handler(runtimeOffset, target, targetInfo, stop);
1262         }
1263         else if ( findSymbol(forImage, libOrdinal, symbolName, weakImport, lazyBind, addend, target, targetInfo) ) {
1264             if ( !targetInfo.skippableWeakDef ) {
1265                 handler(runtimeOffset, target, targetInfo, stop);
1266                 lastSymbolName = symbolName;
1267                 lastLibOrdinal = libOrdinal;
1268                 lastAddend     = addend;
1269             }
1270         }
1271         else {
1272             stop = true;
1273         }
1274     }, ^(const char* symbolName) {
1275         strongHandler(symbolName);
1276     }, ^() {
1277         missingLazyBindHandler();
1278     });
1279 }
1280
1281 void ClosureBuilder::addBindInfo(ImageWriter& writer, BuilderLoadedImage& forImage)
1282 {
1283     const uint32_t ptrSize = forImage.loadAddress()->pointerSize();
1284         STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::BindPattern, binds, 512);
1285     __block uint64_t                    lastOffset = -ptrSize;
1286         __block Image::ResolvedSymbolTarget lastTarget = { {0, 0} };
1287     forEachBind(forImage, ^(uint64_t runtimeOffset, Image::ResolvedSymbolTarget target, const ResolvedTargetInfo& targetInfo, bool& stop) {
1288         if ( targetInfo.weakBindCoalese )  {
1289             // may be previous bind to this location
1290             // if so, update that rather create new BindPattern
1291             for (Image::BindPattern& aBind : binds) {
1292                 if ( (aBind.startVmOffset == runtimeOffset) && (aBind.repeatCount == 1)  && (aBind.skipCount == 0) ) {
1293                     aBind.target = target;
1294                     return;
1295                 }
1296             }
1297         }
1298         bool mergedIntoPrevious = false;
1299         if ( !mergedIntoPrevious && (target == lastTarget) && (runtimeOffset > lastOffset) && !binds.empty() ) {
1300             uint64_t skipAmount = (runtimeOffset - lastOffset - ptrSize)/ptrSize;
1301             if ( skipAmount*ptrSize != (runtimeOffset - lastOffset - ptrSize) ) {
1302                 // misaligned pointer means we cannot optimize
1303             }
1304             else {
1305                 if ( (binds.back().repeatCount == 1) && (binds.back().skipCount == 0) && (skipAmount <= 255) ) {
1306                     binds.back().repeatCount = 2;
1307                     binds.back().skipCount   = skipAmount;
1308                     assert(binds.back().skipCount == skipAmount); // check overflow
1309                     mergedIntoPrevious       = true;
1310                 }
1311                 else if ( (binds.back().skipCount == skipAmount) && (binds.back().repeatCount < 0xfff) ) {
1312                     uint32_t prevRepeatCount = binds.back().repeatCount;
1313                     binds.back().repeatCount += 1;
1314                     assert(binds.back().repeatCount > prevRepeatCount); // check overflow
1315                     mergedIntoPrevious       = true;
1316                 }
1317             }
1318         }
1319         if ( (target == lastTarget) && (runtimeOffset == lastOffset) && !binds.empty() ) {
1320             // duplicate bind for same location, ignore this one
1321             mergedIntoPrevious = true;
1322         }
1323         if ( !mergedIntoPrevious ) {
1324             Image::BindPattern pattern;
1325             pattern.target        = target;
1326             pattern.startVmOffset = runtimeOffset;
1327             pattern.repeatCount   = 1;
1328             pattern.skipCount     = 0;
1329             assert(pattern.startVmOffset == runtimeOffset);
1330             binds.push_back(pattern);
1331         }
1332         lastTarget = target;
1333         lastOffset = runtimeOffset;
1334         }, ^(const char* strongSymbolName) {
1335         if ( !_makingDyldCacheImages ) {
1336             // something has a strong symbol definition that may override a weak impl in the dyld cache
1337             Image::ResolvedSymbolTarget strongOverride;
1338             ResolvedTargetInfo          strongTargetInfo;
1339             if ( findSymbolInImage(forImage.loadAddress(), strongSymbolName, 0, false, false, strongOverride, strongTargetInfo) ) {
1340                 for (const BuilderLoadedImage& li : _loadedImages) {
1341                     if ( li.loadAddress()->inDyldCache() && li.loadAddress()->hasWeakDefs() ) {
1342                         Image::ResolvedSymbolTarget implInCache;
1343                         ResolvedTargetInfo          implInCacheInfo;
1344                         if ( findSymbolInImage(li.loadAddress(), strongSymbolName, 0, false, false, implInCache, implInCacheInfo) ) {
1345                             // found another instance in some dylib in dyld cache, will need to patch it
1346                             Closure::PatchEntry patch;
1347                             patch.exportCacheOffset      = (uint32_t)implInCache.sharedCache.offset;
1348                             patch.overriddenDylibInCache = li.imageNum;
1349                             patch.replacement            = strongOverride;
1350                             _weakDefCacheOverrides.push_back(patch);
1351                         }
1352                     }
1353                 }
1354             }
1355         }
1356     }, ^() {
1357         _foundMissingLazyBinds = true;
1358     });
1359
1360     // check for __dyld section in main executable to support licenseware
1361     if ( forImage.loadAddress()->filetype == MH_EXECUTE ) {
1362         forImage.loadAddress()->forEachSection(^(const MachOAnalyzer::SectionInfo& sectInfo, bool malformedSectionRange, bool& stop) {
1363             if ( (strcmp(sectInfo.sectName, "__dyld") == 0) && (strcmp(sectInfo.segInfo.segName, "__DATA") == 0) ) {
1364                 // find dyld3::compatFuncLookup in libdyld.dylib
1365                 assert(_libDyldImageNum != 0);
1366                 Image::ResolvedSymbolTarget lookupFuncTarget;
1367                 ResolvedTargetInfo          lookupFuncInfo;
1368                 if ( findSymbolInImage(findLoadedImage(_libDyldImageNum).loadAddress(), "__ZN5dyld316compatFuncLookupEPKcPPv", 0, false, false, lookupFuncTarget, lookupFuncInfo) ) {
1369                     // add bind to set second pointer in __dyld section to be dyld3::compatFuncLookup
1370                     uint64_t runtimeOffset = sectInfo.sectAddr - forImage.loadAddress()->preferredLoadAddress() + forImage.loadAddress()->pointerSize();
1371                     Image::BindPattern compatFuncPattern;
1372                     compatFuncPattern.target        = lookupFuncTarget;
1373                     compatFuncPattern.startVmOffset = runtimeOffset;
1374                     compatFuncPattern.repeatCount   = 1;
1375                     compatFuncPattern.skipCount     = 0;
1376                     assert(compatFuncPattern.startVmOffset == runtimeOffset);
1377                     binds.push_back(compatFuncPattern);
1378                 }
1379                 else {
1380                     _diag.error("libdyld.dylib is dyld3::compatFuncLookup");
1381                 }
1382             }
1383         });
1384     }
1385
1386     writer.setBindInfo(binds);
1387 }
1388
1389 void ClosureBuilder::reportRebasesAndBinds(ImageWriter& writer, BuilderLoadedImage& forImage)
1390 {
1391     // report all rebases
1392     forImage.loadAddress()->forEachRebase(_diag, true, ^(uint64_t runtimeOffset, bool& stop) {
1393         _handlers->rebase(forImage.imageNum, forImage.loadAddress(), (uint32_t)runtimeOffset);
1394     });
1395
1396     // report all binds
1397     forEachBind(forImage, ^(uint64_t runtimeOffset, Image::ResolvedSymbolTarget target, const ResolvedTargetInfo& targetInfo, bool& stop) {
1398         _handlers->bind(forImage.imageNum, forImage.loadAddress(), (uint32_t)runtimeOffset, target, targetInfo);
1399     },
1400     ^(const char* strongSymbolName) {},
1401     ^() { });
1402
1403     // i386 programs also use text relocs to rebase stubs
1404     if ( forImage.loadAddress()->cputype == CPU_TYPE_I386 ) {
1405         // FIX ME
1406     }
1407 }
1408
1409 // These are mangled symbols for all the variants of operator new and delete
1410 // which a main executable can define (non-weak) and override the
1411 // weak-def implementation in the OS.
1412 static const char* const sTreatAsWeak[] = {
1413     "__Znwm", "__ZnwmRKSt9nothrow_t",
1414     "__Znam", "__ZnamRKSt9nothrow_t",
1415     "__ZdlPv", "__ZdlPvRKSt9nothrow_t", "__ZdlPvm",
1416     "__ZdaPv", "__ZdaPvRKSt9nothrow_t", "__ZdaPvm",
1417     "__ZnwmSt11align_val_t", "__ZnwmSt11align_val_tRKSt9nothrow_t",
1418     "__ZnamSt11align_val_t", "__ZnamSt11align_val_tRKSt9nothrow_t",
1419     "__ZdlPvSt11align_val_t", "__ZdlPvSt11align_val_tRKSt9nothrow_t", "__ZdlPvmSt11align_val_t",
1420     "__ZdaPvSt11align_val_t", "__ZdaPvSt11align_val_tRKSt9nothrow_t", "__ZdaPvmSt11align_val_t"
1421 };
1422
1423
1424 void ClosureBuilder::addChainedFixupInfo(ImageWriter& writer, BuilderLoadedImage& forImage)
1425 {
1426     // build array of targets
1427     STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::ResolvedSymbolTarget, targets,     1024);
1428     STACK_ALLOC_OVERFLOW_SAFE_ARRAY(ResolvedTargetInfo,          targetInfos, 1024);
1429     forImage.loadAddress()->forEachChainedFixupTarget(_diag, ^(int libOrdinal, const char* symbolName, uint64_t addend, bool weakImport, bool& stop) {
1430         Image::ResolvedSymbolTarget target;
1431         ResolvedTargetInfo          targetInfo;
1432         if ( !findSymbol(forImage, libOrdinal, symbolName, weakImport, false, addend, target, targetInfo) ) {
1433             const char* expectedInPath = forImage.loadAddress()->dependentDylibLoadPath(libOrdinal-1);
1434             _diag.error("symbol '%s' not found, expected in '%s', needed by '%s'", symbolName, expectedInPath, forImage.path());
1435             stop = true;
1436             return;
1437         }
1438         if ( libOrdinal == BIND_SPECIAL_DYLIB_WEAK_LOOKUP ) {
1439             // add if not already in array
1440             bool alreadyInArray = false;
1441             for (const char* sym : _weakDefsFromChainedBinds) {
1442                 if ( strcmp(sym, symbolName) == 0 ) {
1443                     alreadyInArray = true;
1444                     break;
1445                 }
1446             }
1447             if ( !alreadyInArray )
1448                 _weakDefsFromChainedBinds.push_back(symbolName);
1449         }
1450         targets.push_back(target);
1451         targetInfos.push_back(targetInfo);
1452     });
1453     if ( _diag.hasError() )
1454         return;
1455
1456     uint64_t chainStartsOffset = forImage.loadAddress()->chainStartsOffset();
1457     if ( _handlers != nullptr ) {
1458         forImage.loadAddress()->withChainStarts(_diag, chainStartsOffset, ^(const dyld_chained_starts_in_image* starts) {
1459             _handlers->chainedBind(forImage.imageNum, forImage.loadAddress(), starts, targets, targetInfos);
1460         });
1461     }
1462     else {
1463         writer.setChainedFixups(chainStartsOffset, targets);
1464     }
1465
1466     // with chained fixups, main executable may define symbol that overrides weak-defs but has no fixup
1467     if ( _isLaunchClosure && forImage.loadAddress()->hasWeakDefs() && forImage.loadAddress()->isMainExecutable() ) {
1468         for (const char* weakSymbolName : sTreatAsWeak) {
1469             Diagnostics exportDiag;
1470             dyld3::MachOAnalyzer::FoundSymbol foundInfo;
1471             if ( forImage.loadAddress()->findExportedSymbol(exportDiag, weakSymbolName, false, foundInfo, nullptr) ) {
1472                 _weakDefsFromChainedBinds.push_back(weakSymbolName);
1473             }
1474         }
1475     }
1476 }
1477
1478
1479 bool ClosureBuilder::findSymbolInImage(const MachOAnalyzer* macho, const char* symbolName, uint64_t addend, bool followReExports,
1480                                        bool weakImport, Image::ResolvedSymbolTarget& target, ResolvedTargetInfo& targetInfo)
1481 {
1482     targetInfo.foundInDylib        = nullptr;
1483     targetInfo.requestedSymbolName = symbolName;
1484     targetInfo.addend              = addend;
1485     targetInfo.weakBindCoalese     = false;
1486     targetInfo.weakBindSameImage   = false;
1487     targetInfo.isWeakDef           = false;
1488     targetInfo.skippableWeakDef    = false;
1489     MachOLoaded::DependentToMachOLoaded reexportFinder = ^(const MachOLoaded* mh, uint32_t depIndex) {
1490         return (const MachOLoaded*)findDependent(mh, depIndex);
1491     };
1492     MachOAnalyzer::DependentToMachOLoaded finder = nullptr;
1493     if ( followReExports )
1494         finder = reexportFinder;
1495
1496     dyld3::MachOAnalyzer::FoundSymbol foundInfo;
1497     if ( macho->findExportedSymbol(_diag, symbolName, weakImport, foundInfo, finder) ) {
1498         const MachOAnalyzer* impDylib = (const MachOAnalyzer*)foundInfo.foundInDylib;
1499         targetInfo.foundInDylib    = foundInfo.foundInDylib;
1500         targetInfo.foundSymbolName = foundInfo.foundSymbolName;
1501         if ( foundInfo.isWeakDef )
1502             targetInfo.isWeakDef = true;
1503         if ( foundInfo.kind == MachOAnalyzer::FoundSymbol::Kind::absolute ) {
1504             target.absolute.kind   = Image::ResolvedSymbolTarget::kindAbsolute;
1505             target.absolute.value  = foundInfo.value + addend;
1506         }
1507         else if ( impDylib->inDyldCache() ) {
1508             target.sharedCache.kind   = Image::ResolvedSymbolTarget::kindSharedCache;
1509             target.sharedCache.offset = (uint8_t*)impDylib - (uint8_t*)_dyldCache + foundInfo.value + addend;
1510         }
1511         else {
1512             target.image.kind     = Image::ResolvedSymbolTarget::kindImage;
1513             target.image.imageNum = findLoadedImage(impDylib).imageNum;
1514             target.image.offset   = foundInfo.value + addend;
1515         }
1516         return true;
1517     }
1518     return false;
1519 }
1520
1521 bool ClosureBuilder::findSymbol(BuilderLoadedImage& fromImage, int libOrdinal, const char* symbolName, bool weakImport, bool lazyBind,
1522                                 uint64_t addend, Image::ResolvedSymbolTarget& target, ResolvedTargetInfo& targetInfo)
1523 {
1524     target.raw                      = 0;
1525     targetInfo.weakBindCoalese      = false;
1526     targetInfo.weakBindSameImage    = false;
1527     targetInfo.isWeakDef            = false;
1528     targetInfo.skippableWeakDef     = false;
1529     targetInfo.requestedSymbolName  = symbolName;
1530     targetInfo.libOrdinal           = libOrdinal;
1531     if ( libOrdinal == BIND_SPECIAL_DYLIB_FLAT_LOOKUP ) {
1532         for (const BuilderLoadedImage& li : _loadedImages) {
1533             if ( !li.rtldLocal && findSymbolInImage(li.loadAddress(), symbolName, addend, true, weakImport, target, targetInfo) )
1534                 return true;
1535         }
1536         if ( weakImport ) {
1537             target.absolute.kind  = Image::ResolvedSymbolTarget::kindAbsolute;
1538             target.absolute.value = 0;
1539             // Record that we found a missing weak import so that the objc optimizer doens't have to check
1540             fromImage.hasMissingWeakImports = true;
1541             return true;
1542         }
1543         // <rdar://problem/44315944> closures should bind missing lazy-bind symbols to a missing symbol handler in libdyld in flat namespace
1544         if ( lazyBind && _allowMissingLazies ) {
1545             if ( findMissingSymbolHandler(target, targetInfo) )
1546                 return true;
1547         }
1548         _diag.error("symbol '%s' not found, expected in flat namespace by '%s'", symbolName, fromImage.path());
1549     }
1550     else if ( libOrdinal == BIND_SPECIAL_DYLIB_WEAK_LOOKUP ) {
1551         // to resolve weakDef coalesing, we need to search all images in order and use first definition
1552         // but, if first found is a weakDef, a later non-weak def overrides that
1553         bool foundWeakDefImpl   = false;
1554         bool foundStrongDefImpl = false;
1555         bool foundImpl          = false;
1556
1557         if ( _makingDyldCacheImages ) {
1558             // _loadedImages is all dylibs in the dyld cache, it is not load-order, so need alterate weak-def binding algorithm
1559             // look first in /usr/lib/libc++, most will be here
1560             for (const BuilderLoadedImage& li : _loadedImages) {
1561                 if ( li.loadAddress()->hasWeakDefs() && (strncmp(li.path(), "/usr/lib/libc++", 15) == 0) ) {
1562                    if ( findSymbolInImage(li.loadAddress(), symbolName, addend, false, weakImport, target, targetInfo) ) {
1563                         foundImpl = true;
1564                         break;
1565                     }
1566                 }
1567             }
1568             // if not found, try looking in the images itself, most custom weak-def symbols have a copy in the image itself
1569             if ( !foundImpl ) {
1570                 if ( findSymbolInImage(fromImage.loadAddress(), symbolName, addend, false, weakImport, target, targetInfo) ) {
1571                     foundImpl = true;
1572                 }
1573             }
1574             // if still not found, then this is the rare case of a simple use of a weak-def symbol
1575             if ( !foundImpl ) {
1576                 // look in all direct dependents
1577                 for (Image::LinkedImage child : fromImage.dependents) {
1578                     if (child.imageNum() == kMissingWeakLinkedImage)
1579                         continue;
1580                     BuilderLoadedImage& childLi = findLoadedImage(child.imageNum());
1581                     if ( childLi.loadAddress()->hasWeakDefs() && findSymbolInImage(childLi.loadAddress(), symbolName, addend, false, weakImport, target, targetInfo) ) {
1582                         foundImpl = true;
1583                         break;
1584                     }
1585                 }
1586             }
1587             targetInfo.weakBindCoalese = true;
1588         }
1589         else {
1590             // walk images in load-order to find first that implements this symbol
1591             Image::ResolvedSymbolTarget  aTarget;
1592             ResolvedTargetInfo           aTargetInfo;
1593             STACK_ALLOC_ARRAY(const BuilderLoadedImage*, cachedDylibsUsingSymbol, 1024);
1594             for (const BuilderLoadedImage& li : _loadedImages) {
1595                 // only search images with weak-defs that were not loaded with RTLD_LOCAL
1596                 if ( li.loadAddress()->hasWeakDefs() && !li.rtldLocal ) {
1597                     if ( findSymbolInImage(li.loadAddress(), symbolName, addend, false, weakImport, aTarget, aTargetInfo) ) {
1598                         foundImpl = true;
1599                         // with non-chained images, weak-defs first have a rebase to their local impl, and a weak-bind which allows earlier impls to override
1600                         if ( !li.loadAddress()->hasChainedFixups() && (aTargetInfo.foundInDylib == fromImage.loadAddress()) )
1601                             targetInfo.weakBindSameImage = true;
1602                         if ( aTargetInfo.isWeakDef ) {
1603                             // found a weakDef impl, if this is first found, set target to this
1604                             if ( !foundWeakDefImpl && !foundStrongDefImpl ) {
1605                                 target      = aTarget;
1606                                 targetInfo  = aTargetInfo;
1607                             }
1608                             foundWeakDefImpl = true;
1609                         }
1610                         else {
1611                             // found a non-weak impl, use this (unless early strong found)
1612                             if ( !foundStrongDefImpl ) {
1613                                 target      = aTarget;
1614                                 targetInfo  = aTargetInfo;
1615                             }
1616                             foundStrongDefImpl = true;
1617                         }
1618                     }
1619                     if ( foundImpl && li.loadAddress()->inDyldCache() )
1620                         cachedDylibsUsingSymbol.push_back(&li);
1621                 }
1622             }
1623
1624             // now that final target found, if any dylib in dyld cache uses that symbol name, redirect it to new target
1625             if ( !cachedDylibsUsingSymbol.empty() ) {
1626                 for (const BuilderLoadedImage* li : cachedDylibsUsingSymbol) {
1627                     Image::ResolvedSymbolTarget implInCache;
1628                     ResolvedTargetInfo          implInCacheInfo;
1629                     if ( findSymbolInImage(li->loadAddress(), symbolName, addend, false, weakImport, implInCache, implInCacheInfo) ) {
1630                         if ( implInCache != target ) {
1631                             // found another instance in some dylib in dyld cache, will need to patch it
1632                             Closure::PatchEntry patch;
1633                             patch.exportCacheOffset      = (uint32_t)implInCache.sharedCache.offset;
1634                             patch.overriddenDylibInCache = li->imageNum;
1635                             patch.replacement            = target;
1636                             _weakDefCacheOverrides.push_back(patch);
1637                         }
1638                     }
1639                 }
1640             }
1641             targetInfo.weakBindCoalese = true;
1642         }
1643
1644         if ( foundImpl )
1645             return true;
1646         if ( weakImport ) {
1647             target.absolute.kind  = Image::ResolvedSymbolTarget::kindAbsolute;
1648             target.absolute.value = 0;
1649             return true;
1650         }
1651         if ( ! fromImage.loadAddress()->hasChainedFixups() ) {
1652             // support old binaries where symbols have been stripped and have weak_bind to itself
1653             targetInfo.skippableWeakDef = true;
1654             return true;
1655         }
1656
1657         _diag.error("symbol '%s' not found, expected to be weak-def coalesced by '%s'", symbolName, fromImage.path());
1658     }
1659     else {
1660         const BuilderLoadedImage* targetLoadedImage = nullptr;
1661         if ( (libOrdinal > 0) && (libOrdinal <= (int)fromImage.dependents.count()) ) {
1662             ImageNum childNum = fromImage.dependents[libOrdinal - 1].imageNum();
1663             if ( childNum != kMissingWeakLinkedImage ) {
1664                 targetLoadedImage = &findLoadedImage(childNum);
1665             }
1666         }
1667         else if ( libOrdinal == BIND_SPECIAL_DYLIB_SELF ) {
1668             targetLoadedImage = &fromImage;
1669         }
1670         else if ( libOrdinal == BIND_SPECIAL_DYLIB_MAIN_EXECUTABLE ) {
1671             targetLoadedImage = &_loadedImages[_mainProgLoadIndex];
1672         }
1673         else {
1674             _diag.error("unknown special ordinal %d in %s", libOrdinal, fromImage.path());
1675             return false;
1676         }
1677
1678         if ( targetLoadedImage != nullptr ) {
1679             if ( findSymbolInImage(targetLoadedImage->loadAddress(), symbolName, addend, true, weakImport, target, targetInfo) )
1680                 return true;
1681         }
1682
1683         if ( weakImport ) {
1684             target.absolute.kind  = Image::ResolvedSymbolTarget::kindAbsolute;
1685             target.absolute.value = 0;
1686             // Record that we found a missing weak import so that the objc optimizer doens't have to check
1687             fromImage.hasMissingWeakImports = true;
1688             return true;
1689         }
1690
1691         // <rdar://problem/43315403> closures should bind missing lazy-bind symbols to a missing symbol handler in libdyld
1692         if ( lazyBind && _allowMissingLazies ) {
1693             if ( findMissingSymbolHandler(target, targetInfo) )
1694                 return true;
1695         }
1696
1697         // symbol not found and not weak or lazy so error out
1698         const char* expectedInPath = targetLoadedImage ? targetLoadedImage->path() : "unknown";
1699         _diag.error("symbol '%s' not found, expected in '%s', needed by '%s'", symbolName, expectedInPath, fromImage.path());
1700         if ( _launchErrorInfo != nullptr ) {
1701             _launchErrorInfo->kind              = DYLD_EXIT_REASON_SYMBOL_MISSING;
1702             _launchErrorInfo->clientOfDylibPath = strdup_temp(fromImage.path());
1703             _launchErrorInfo->targetDylibPath   = strdup_temp(expectedInPath);
1704             _launchErrorInfo->symbol            = symbolName;
1705         }
1706     }
1707     return false;
1708 }
1709
1710
1711 bool ClosureBuilder::findMissingSymbolHandler(Image::ResolvedSymbolTarget& target, ResolvedTargetInfo& targetInfo)
1712 {
1713    for (BuilderLoadedImage& li : _loadedImages) {
1714          if ( li.loadAddress()->isDylib() && (strcmp(li.loadAddress()->installName(), "/usr/lib/system/libdyld.dylib") == 0) ) {
1715             if ( findSymbolInImage(li.loadAddress(), "__dyld_missing_symbol_abort", 0, false, false, target, targetInfo) ) {
1716                 return true;
1717             }
1718             break;
1719         }
1720     }
1721     return false;
1722 }
1723
1724 void ClosureBuilder::depthFirstRecurseSetInitInfo(uint32_t loadIndex, InitInfo initInfos[], uint32_t& initOrder, bool& hasError)
1725 {
1726     if ( initInfos[loadIndex].visited )
1727         return;
1728     initInfos[loadIndex].visited        = true;
1729     initInfos[loadIndex].danglingUpward = false;
1730
1731     if (_loadedImages[loadIndex].isBadImage) {
1732         hasError = true;
1733         return;
1734     }
1735
1736     for (const Image::LinkedImage& dep : _loadedImages[loadIndex].dependents) {
1737         if ( dep.imageNum() == kMissingWeakLinkedImage )
1738             continue;
1739         ClosureBuilder::BuilderLoadedImage& depLi = findLoadedImage(dep.imageNum());
1740         uint32_t depLoadIndex = (uint32_t)_loadedImages.index(depLi);
1741         if ( dep.kind() == Image::LinkKind::upward ) {
1742             if ( !initInfos[depLoadIndex].visited )
1743                 initInfos[depLoadIndex].danglingUpward = true;
1744         }
1745         else {
1746             depthFirstRecurseSetInitInfo(depLoadIndex, initInfos, initOrder, hasError);
1747             if (hasError)
1748                 return;
1749         }
1750     }
1751     initInfos[loadIndex].initOrder = initOrder++;
1752 }
1753
1754 void ClosureBuilder::computeInitOrder(ImageWriter& imageWriter, uint32_t loadIndex)
1755 {
1756     // allocate array to track initializers
1757     InitInfo initInfos[_loadedImages.count()];
1758     bzero(initInfos, sizeof(initInfos));
1759
1760     // recurse all images and build initializer list from bottom up
1761     uint32_t initOrder = 1;
1762     bool hasMissingDependent = false;
1763     depthFirstRecurseSetInitInfo(loadIndex, initInfos, initOrder, hasMissingDependent);
1764     if (hasMissingDependent) {
1765         imageWriter.setInvalid();
1766         return;
1767     }
1768
1769     // any images not visited yet are are danging, force add them to end of init list
1770     for (uint32_t i=0; i < (uint32_t)_loadedImages.count(); ++i) {
1771         if ( !initInfos[i].visited && initInfos[i].danglingUpward ) {
1772             depthFirstRecurseSetInitInfo(i, initInfos, initOrder, hasMissingDependent);
1773         }
1774     }
1775
1776     if (hasMissingDependent) {
1777         imageWriter.setInvalid();
1778         return;
1779     }
1780
1781     // build array of just images with initializer
1782     STACK_ALLOC_ARRAY(uint32_t, indexOfImagesWithInits, _loadedImages.count());
1783     uint32_t index = 0;
1784     for (const BuilderLoadedImage& li : _loadedImages) {
1785         if ( initInfos[index].visited && li.hasInits ) {
1786             indexOfImagesWithInits.push_back(index);
1787         }
1788         ++index;
1789     }
1790
1791     // bubble sort (FIXME)
1792     if ( indexOfImagesWithInits.count() > 1 ) {
1793         for (uint32_t i=0; i < indexOfImagesWithInits.count()-1; ++i) {
1794             for (uint32_t j=0; j < indexOfImagesWithInits.count()-i-1; ++j) {
1795                 if ( initInfos[indexOfImagesWithInits[j]].initOrder > initInfos[indexOfImagesWithInits[j+1]].initOrder ) {
1796                     uint32_t temp               = indexOfImagesWithInits[j];
1797                     indexOfImagesWithInits[j]   = indexOfImagesWithInits[j+1];
1798                     indexOfImagesWithInits[j+1] = temp;
1799                 }
1800             }
1801         }
1802     }
1803
1804     // copy ImageNum of each image with initializers into array
1805     ImageNum initNums[indexOfImagesWithInits.count()];
1806     for (uint32_t i=0; i < indexOfImagesWithInits.count(); ++i) {
1807         initNums[i] = _loadedImages[indexOfImagesWithInits[i]].imageNum;
1808     }
1809
1810     // add to closure info
1811     imageWriter.setInitsOrder(initNums, (uint32_t)indexOfImagesWithInits.count());
1812 }
1813
1814 void ClosureBuilder::addClosureInfo(LaunchClosureWriter& closureWriter)
1815 {
1816     // record which is libSystem
1817     assert(_libSystemImageNum != 0);
1818         closureWriter.setLibSystemImageNum(_libSystemImageNum);
1819
1820     // record which is libdyld
1821     assert(_libDyldImageNum != 0);
1822     Image::ResolvedSymbolTarget entryLocation;
1823     ResolvedTargetInfo          entryInfo;
1824     if ( findSymbolInImage(findLoadedImage(_libDyldImageNum).loadAddress(), "__ZN5dyld318entryVectorForDyldE", 0, false, false, entryLocation, entryInfo) ) {
1825         const dyld3::LibDyldEntryVector* libDyldEntry = nullptr;
1826         switch ( entryLocation.image.kind ) {
1827             case Image::ResolvedSymbolTarget::kindSharedCache:
1828                 libDyldEntry = (dyld3::LibDyldEntryVector*)((uint8_t*)_dyldCache + entryLocation.sharedCache.offset);
1829                 break;
1830             case Image::ResolvedSymbolTarget::kindImage:
1831                 libDyldEntry = (dyld3::LibDyldEntryVector*)((uint8_t*)findLoadedImage(entryLocation.image.imageNum).loadAddress() + entryLocation.image.offset);
1832                 break;
1833         }
1834         if ( (libDyldEntry != nullptr) && ((libDyldEntry->binaryFormatVersion & LibDyldEntryVector::kBinaryFormatVersionMask) == dyld3::closure::kFormatVersion) )
1835             closureWriter.setLibDyldEntry(entryLocation);
1836         else
1837             _diag.error("libdyld.dylib entry vector is incompatible");
1838     }
1839     else {
1840         _diag.error("libdyld.dylib is missing entry vector");
1841     }
1842
1843     // record which is main executable
1844     ImageNum mainProgImageNum = _loadedImages[_mainProgLoadIndex].imageNum;
1845     closureWriter.setTopImageNum(mainProgImageNum);
1846
1847     // add entry
1848     uint32_t    entryOffset;
1849     bool        usesCRT;
1850     if ( _loadedImages[_mainProgLoadIndex].loadAddress()->getEntry(entryOffset, usesCRT) ) {
1851         Image::ResolvedSymbolTarget location;
1852         location.image.kind     = Image::ResolvedSymbolTarget::kindImage;
1853         location.image.imageNum = mainProgImageNum;
1854         location.image.offset   = entryOffset;
1855         if ( usesCRT )
1856             closureWriter.setStartEntry(location);
1857         else
1858             closureWriter.setMainEntry(location);
1859     }
1860
1861     // add env vars that must match at launch time
1862     _pathOverrides.forEachEnvVar(^(const char* envVar) {
1863         closureWriter.addEnvVar(envVar);
1864     });
1865
1866     // add list of files which must be missing
1867     STACK_ALLOC_OVERFLOW_SAFE_ARRAY(const char*, paths, 8192);
1868     if ( _mustBeMissingPaths != nullptr ) {
1869         _mustBeMissingPaths->forEachPath(^(const char* aPath) {
1870             paths.push_back(aPath);
1871         });
1872     }
1873         closureWriter.setMustBeMissingFiles(paths);
1874
1875     // add list of files which must be be present with a specific inode/mtime
1876     if (!_skippedFiles.empty())
1877         closureWriter.setMustExistFiles(_skippedFiles);
1878 }
1879 void ClosureBuilder::invalidateInitializerRoots()
1880 {
1881     while (true) {
1882         bool madeChange = false;
1883         for (uintptr_t loadedImageIndex = _alreadyInitedIndex; loadedImageIndex != _loadedImages.count(); ++loadedImageIndex) {
1884             BuilderLoadedImage& li = _loadedImages[loadedImageIndex];
1885             if ( li.mustBuildClosure ) {
1886                 // Already invalidated
1887                 continue;
1888             }
1889             for (Image::LinkedImage depIndex : li.dependents) {
1890                 if ( depIndex.imageNum() == kMissingWeakLinkedImage )
1891                     continue;
1892                 BuilderLoadedImage& depImage = findLoadedImage(depIndex.imageNum());
1893                 // If a dependent is bad, or a new image num, or an override, then we need this image to get a new closure
1894                 if ( depImage.mustBuildClosure ) {
1895                     li.mustBuildClosure = true;    // mark bad
1896                     madeChange = true;
1897                 }
1898             }
1899         }
1900         if (!madeChange)
1901             break;
1902         // If we made a change, then we detected an existing image with a dependent which needed to be rebuilt.
1903         // This corresponds to a root of the shared cache where the existing image is a shared cache one and the root is the depImage
1904         _foundDyldCacheRoots = true;
1905     }
1906 }
1907
1908 size_t ClosureBuilder::HashCString::hash(const char* v) {
1909     // FIXME: Use hash<string_view> when it has the correct visibility markup
1910     return __gnu_cxx::hash<const char*>{}(v);
1911 }
1912
1913 bool ClosureBuilder::EqualCString::equal(const char* s1, const char* s2) {
1914     return strcmp(s1, s2) == 0;
1915 }
1916
1917
1918 struct HashUInt64 {
1919     static size_t hash(const uint64_t& v) {
1920         return std::hash<uint64_t>{}(v);
1921     }
1922 };
1923
1924 struct EqualUInt64 {
1925     static bool equal(uint64_t s1, uint64_t s2) {
1926         return s1 == s2;
1927     }
1928 };
1929
1930 void ClosureBuilder::writeClassOrProtocolHashTable(bool classes, Array<ObjCOptimizerImage>& objcImages) {
1931     __block MultiMap<const char*, dyld3::closure::Image::ObjCClassImageOffset, HashCString, EqualCString> seenClassesMap;
1932     __block Map<const char*, dyld3::closure::Image::ObjCClassNameImageOffset, HashCString, EqualCString> classNameMap;
1933     __block OverflowSafeArray<const char*> classNames;
1934
1935     // Note we walk the images backwards as we want them in load order to match the order they are registered with objc
1936     for (size_t imageIndex = 0, reverseIndex = (objcImages.count() - 1); imageIndex != objcImages.count(); ++imageIndex, --reverseIndex) {
1937         if (objcImages[reverseIndex].diag.hasError())
1938             continue;
1939         ObjCOptimizerImage& image = objcImages[reverseIndex];
1940         const OverflowSafeArray<ObjCOptimizerImage::SeenClass>& seenClasses = classes ? image.seenClasses : image.seenProtocols;
1941
1942         for (const ObjCOptimizerImage::SeenClass& seenClass : seenClasses) {
1943             closure::Image::ObjCClassNameImageOffset    classNameTarget = seenClass.first;
1944             dyld3::closure::Image::ObjCClassImageOffset classDataTarget = seenClass.second;
1945             Image::ObjCClassImage                       classImage      = _objcClassesHashTableImages[classNameTarget.classNameImageIndex];
1946
1947             const BuilderLoadedImage& li    = findLoadedImage(classImage.imageNum);
1948             const dyld3::MachOAnalyzer* ma  = li.loadAddress();
1949
1950             const char* className   = ((const char*)ma) + classImage.offsetOfClassNames + classNameTarget.classNameImageOffset;
1951             //uint64_t nameVMAddr     = ma->preferredLoadAddress() + classImage.offsetOfClassNames + classNameTarget.classNameImageOffset;
1952             //printf("%s: 0x%08llx = '%s'\n", li.path(), nameVMAddr, className);
1953             seenClassesMap.insert({ className, classDataTarget });
1954
1955             // Also track the name
1956             auto itAndInserted = classNameMap.insert({ className, dyld3::closure::Image::ObjCClassNameImageOffset() });
1957             if (itAndInserted.second) {
1958                 // We inserted the class name so we need to add it to the strings for the closure hash table
1959                 classNames.push_back(className);
1960
1961                 // We already computed a class name target in a previous loop so use that one
1962                 itAndInserted.first->second = seenClass.first;
1963
1964                 // If we are processing protocols, and this is the first one we've seen, then track its ISA to be fixed up
1965                 if ( !classes ) {
1966                     uint64_t protocolVMOffset = classImage.offsetOfClasses + classDataTarget.classData.imageOffset;
1967                     image.protocolISAFixups.push_back(protocolVMOffset);
1968                 }
1969             }
1970         }
1971     }
1972
1973     __block uint32_t duplicateCount = 0;
1974     seenClassesMap.forEachEntry(^(const char *const &key, const Image::ObjCClassImageOffset **values,
1975                                   uint64_t valuesCount) {
1976         if (valuesCount != 1)
1977             duplicateCount += valuesCount;
1978     });
1979
1980     // If we have closure class names, we need to make a hash table for them.
1981     OverflowSafeArray<uint8_t>& hashTable = classes ? _objcClassesHashTable : _objcProtocolsHashTable;
1982     if (!classNames.empty()) {
1983         objc_opt::perfect_hash phash;
1984         objc_opt::make_perfect(classNames, phash);
1985         size_t size = ObjCClassOpt::size(phash, duplicateCount);
1986         hashTable.resize(size);
1987         //printf("Class table size: %lld\n", size);
1988         ObjCClassOpt* classesHashTable = (ObjCClassOpt*)hashTable.begin();
1989         classesHashTable->write(phash, classNameMap.array(), seenClassesMap, duplicateCount);
1990     }
1991 }
1992
1993 bool ClosureBuilder::optimizeObjC(Array<ImageWriter>& writers) {
1994     if ( _dyldCache == nullptr )
1995         return false;
1996
1997     // If we have the read only data, make sure it has a valid selector table inside.
1998     const objc_opt::objc_clsopt_t* objcClassOpt = nullptr;
1999     const objc_opt::objc_selopt_t* objcSelOpt = nullptr;
2000     const objc_opt::objc_protocolopt2_t* objcProtocolOpt = nullptr;
2001     if (const objc_opt::objc_opt_t* optObjCHeader = _dyldCache->objcOpt()) {
2002         objcClassOpt = optObjCHeader->clsopt();
2003         objcSelOpt = optObjCHeader->selopt();
2004         objcProtocolOpt = optObjCHeader->protocolopt2();
2005     }
2006
2007     if ( !objcClassOpt || !objcSelOpt || !objcProtocolOpt )
2008         return false;
2009
2010     // We have 24 bits of index in SelectorReferenceFixup so we can't handle a
2011     // shared cache selector table larger than that
2012     if ( objcSelOpt->usedCount() >= (1 << 24) )
2013         return false;
2014
2015     // Make sure we have the pointers section with the pointer to the protocol class
2016     const void* objcOptPtrs = _dyldCache->objcOptPtrs();
2017     if ( objcOptPtrs == nullptr )
2018         return false;
2019
2020     uint32_t pointerSize = _loadedImages.begin()->loadAddress()->pointerSize();
2021     uint64_t classProtocolVMAddr = (pointerSize == 8) ? *(uint64_t*)objcOptPtrs : *(uint32_t*)objcOptPtrs;
2022
2023     Image::ResolvedSymbolTarget objcProtocolClassTarget;
2024     objcProtocolClassTarget.sharedCache.kind = Image::ResolvedSymbolTarget::kindSharedCache;
2025     if ( _dyldCacheIsLive ) {
2026         objcProtocolClassTarget.sharedCache.offset = classProtocolVMAddr - (uint64_t)_dyldCache;
2027     } else {
2028         objcProtocolClassTarget.sharedCache.offset = classProtocolVMAddr - _dyldCache->unslidLoadAddress();
2029     }
2030
2031     STACK_ALLOC_OVERFLOW_SAFE_ARRAY(ObjCOptimizerImage, objcImages, 32);
2032     ArrayFinalizer<ObjCOptimizerImage> scopedCleanup(objcImages,
2033                                                      ^(ObjCOptimizerImage& objcImage) {
2034                                                          objcImage.~ObjCOptimizerImage();
2035                                                      });
2036
2037     // Find all the images with valid objc info
2038     // Also add shared cache images to a map so that we can see them later for looking up classes
2039     Map<const dyld3::MachOAnalyzer*, bool, HashPointer, EqualPointer> sharedCacheImagesMap;
2040     for (size_t imageIndex = 0, writerIndex = 0; imageIndex != _loadedImages.count(); ++imageIndex) {
2041         BuilderLoadedImage& li = _loadedImages[imageIndex];
2042
2043         // Skip shared cache images as even if they need a new closure, the objc runtime can still use
2044         // the optimized shared cache tables.
2045         if ( li.loadAddress()->inDyldCache() ) {
2046             sharedCacheImagesMap.insert({ li.loadAddress(), true });
2047             // Bump the writer index if we have a writer for this image
2048             if ( li.mustBuildClosure )
2049                 ++writerIndex;
2050             continue;
2051         }
2052         // Images which don't need a closure can be skipped.  They are from the shared cache
2053         if ( !li.mustBuildClosure )
2054             continue;
2055
2056         // If we have a root of libobjc, just give up for now
2057         if ( !strcmp(li.path(), "/usr/lib/libobjc.A.dylib"))
2058             return false;
2059
2060         ImageWriter& writer = writers[writerIndex];
2061         ++writerIndex;
2062
2063         const dyld3::MachOAnalyzer* ma = li.loadAddress();
2064
2065         // Skip images with chained fixups other than arm64e legacy fixups until we can test them
2066         // FIXME: Handle chained fixups
2067         if ( ma->hasChainedFixups() ) {
2068             switch ( ma->chainedPointerFormat() ) {
2069                 case DYLD_CHAINED_PTR_ARM64E:
2070                 case DYLD_CHAINED_PTR_64:
2071                     // We've tested the 64-bit chained fixups.
2072                     break;
2073                 case DYLD_CHAINED_PTR_32:
2074                 case DYLD_CHAINED_PTR_32_CACHE:
2075                 case DYLD_CHAINED_PTR_32_FIRMWARE:
2076                     // FIXME: Test 32-bit chained fixups then enable this.
2077                     continue;
2078             }
2079         }
2080
2081         const MachOAnalyzer::ObjCImageInfo* objcImageInfo = ma->objcImageInfo();
2082         if ( objcImageInfo == nullptr )
2083             continue;
2084
2085         // This image is good so record it for use later.
2086         objcImages.default_constuct_back();
2087         ObjCOptimizerImage& image = objcImages.back();
2088         image.loadedImage   = &li;
2089         image.writer        = &writer;
2090
2091         // Find FairPlay encryption range if encrypted
2092         uint32_t fairPlayFileOffset;
2093         uint32_t fairPlaySize;
2094         if ( ma->isFairPlayEncrypted(fairPlayFileOffset, fairPlaySize) ) {
2095             image.fairplayFileOffsetStart   = fairPlayFileOffset;
2096             image.fairplayFileOffsetEnd     = fairPlayFileOffset;
2097         }
2098
2099         // Set the offset to the objc image info
2100         image.objcImageInfoVMOffset = (uint64_t)objcImageInfo - (uint64_t)ma;
2101     }
2102
2103     OverflowSafeArray<const char*>                                                          closureSelectorStrings;
2104     Map<const char*, dyld3::closure::Image::ObjCImageOffset, HashCString, EqualCString>     closureSelectorMap;
2105     OverflowSafeArray<const char*>                                                          closureDuplicateSharedCacheClassNames;
2106     Map<const char*, dyld3::closure::Image::ObjCDuplicateClass, HashCString, EqualCString>  closureDuplicateSharedCacheClassMap;
2107     for (ObjCOptimizerImage& image : objcImages) {
2108         optimizeObjCClasses(objcClassOpt, sharedCacheImagesMap, closureDuplicateSharedCacheClassMap, image);
2109         if (image.diag.hasError())
2110             continue;
2111
2112         optimizeObjCProtocols(objcProtocolOpt, sharedCacheImagesMap, image);
2113         if (image.diag.hasError())
2114             continue;
2115
2116         optimizeObjCSelectors(objcSelOpt, closureSelectorMap, image);
2117         if (image.diag.hasError())
2118             continue;
2119
2120         // If this image is still valid, then add its intermediate results to the main tables
2121
2122         // Class results
2123         for (auto nameAndDataVMOffset : image.classesNameAndDataVMOffsets) {
2124             uint64_t nameVMOffset = nameAndDataVMOffset.first;
2125             uint64_t dataVMOffset = nameAndDataVMOffset.second;
2126             _objcClassesHashTableImages.push_back({ image.loadedImage->imageNum, (uint32_t)nameVMOffset, (uint32_t)dataVMOffset });
2127         }
2128         image.classesNameAndDataVMOffsets.clear();
2129
2130         for (const auto& stringAndDuplicate : image.classSharedCacheDuplicates) {
2131             closureDuplicateSharedCacheClassMap[stringAndDuplicate.first] = stringAndDuplicate.second;
2132             closureDuplicateSharedCacheClassNames.push_back(stringAndDuplicate.first);
2133         }
2134
2135         // Selector results
2136         // Note we don't need to add the selector binds here.  Its easier just to process them later from each image
2137         for (const auto& stringAndTarget : image.selectorMap) {
2138             closureSelectorMap[stringAndTarget.first] = stringAndTarget.second;
2139             closureSelectorStrings.push_back(stringAndTarget.first);
2140         }
2141         if (image.methodNameVMOffset)
2142             _objcSelectorsHashTableImages.push_back({ image.loadedImage->imageNum, (uint32_t)*image.methodNameVMOffset });
2143     }
2144
2145     // If we successfully analyzed the classes and selectors, we can now emit their data
2146     // Set all the writers to have optimized objc
2147     for (ObjCOptimizerImage& image : objcImages) {
2148         if (image.diag.hasError())
2149             continue;
2150         image.writer->setHasPrecomputedObjC(true);
2151     }
2152
2153     // Write out the class table
2154     writeClassOrProtocolHashTable(true, objcImages);
2155
2156     // Write out the protocol table
2157     writeClassOrProtocolHashTable(false, objcImages);
2158
2159     // If we have closure duplicate classes, we need to make a hash table for them.
2160     closure::ObjCStringTable* duplicateClassesTable = nullptr;
2161     if (!closureDuplicateSharedCacheClassNames.empty()) {
2162         objc_opt::perfect_hash phash;
2163         objc_opt::make_perfect(closureDuplicateSharedCacheClassNames, phash);
2164         size_t size = ObjCStringTable::size(phash);
2165         _objcClassesDuplicatesHashTable.resize(size);
2166         //printf("Duplicate classes table size: %lld\n", size);
2167         duplicateClassesTable = (closure::ObjCClassDuplicatesOpt*)_objcClassesDuplicatesHashTable.begin();
2168         duplicateClassesTable->write(phash, closureDuplicateSharedCacheClassMap.array());
2169     }
2170
2171     // If we have closure selectors, we need to make a hash table for them.
2172     closure::ObjCStringTable* selectorStringTable = nullptr;
2173     if (!closureSelectorStrings.empty()) {
2174         objc_opt::perfect_hash phash;
2175         objc_opt::make_perfect(closureSelectorStrings, phash);
2176         size_t size = ObjCStringTable::size(phash);
2177         _objcSelectorsHashTable.resize(size);
2178         //printf("Selector table size: %lld\n", size);
2179         selectorStringTable = (closure::ObjCStringTable*)_objcSelectorsHashTable.begin();
2180         selectorStringTable->write(phash, closureSelectorMap.array());
2181     }
2182
2183     // Add fixups for the image info, protocol ISAs, and selector refs
2184     for (ObjCOptimizerImage& image : objcImages) {
2185         if (image.diag.hasError())
2186             continue;
2187
2188         // Protocol ISA references
2189         STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::ProtocolISAFixup, protocolFixups, 512);
2190         if ( !image.protocolISAFixups.empty() ) {
2191
2192             __block uint64_t lastOffset = -pointerSize;
2193             for (uint64_t runtimeOffset : image.protocolISAFixups) {
2194                 bool mergedIntoPrevious = false;
2195                 if ( (runtimeOffset > lastOffset) && !protocolFixups.empty() ) {
2196                     uint64_t skipAmount = (runtimeOffset - lastOffset - pointerSize)/pointerSize;
2197                     if ( skipAmount*pointerSize != (runtimeOffset - lastOffset - pointerSize) ) {
2198                         // misaligned pointer means we cannot optimize
2199                     }
2200                     else {
2201                         if ( (protocolFixups.back().repeatCount == 1) && (protocolFixups.back().skipCount == 0) && (skipAmount <= 255) ) {
2202                             protocolFixups.back().repeatCount = 2;
2203                             protocolFixups.back().skipCount   = skipAmount;
2204                             assert(protocolFixups.back().skipCount == skipAmount); // check overflow
2205                             mergedIntoPrevious       = true;
2206                         }
2207                         else if ( (protocolFixups.back().skipCount == skipAmount) && (protocolFixups.back().repeatCount < 0xfff) ) {
2208                             uint32_t prevRepeatCount = protocolFixups.back().repeatCount;
2209                             protocolFixups.back().repeatCount += 1;
2210                             assert(protocolFixups.back().repeatCount > prevRepeatCount); // check overflow
2211                             mergedIntoPrevious       = true;
2212                         }
2213                     }
2214                 }
2215                 if ( !mergedIntoPrevious ) {
2216                     Image::ProtocolISAFixup pattern;
2217                     pattern.startVmOffset = runtimeOffset;
2218                     pattern.repeatCount   = 1;
2219                     pattern.skipCount     = 0;
2220                     assert(pattern.startVmOffset == runtimeOffset);
2221                     protocolFixups.push_back(pattern);
2222                 }
2223                 lastOffset = runtimeOffset;
2224             }
2225         }
2226
2227         // Selector references
2228         STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::SelectorReferenceFixup, selRefFixups, 512);
2229         if ( !image.selectorFixups.empty() ) {
2230             uint64_t prevVMOffset = 0;
2231             const uint64_t maxChainOffset = (4 * ((1 << 7) - 1));
2232             for (const ObjCOptimizerImage::SelectorFixup& selectorFixup : image.selectorFixups) {
2233                 assert( (selectorFixup.fixupVMOffset & 3) == 0 );
2234                 if ( (selectorFixup.fixupVMOffset - prevVMOffset) <= maxChainOffset ) {
2235                     // Add this to the previous chain
2236                     selRefFixups.back().chainEntry.next = (uint32_t)(selectorFixup.fixupVMOffset - prevVMOffset) / 4;
2237                 } else {
2238                     // Need to start a new chain as the previous offset can't reach
2239                     Image::SelectorReferenceFixup fixup;
2240                     fixup.chainStartVMOffset = selectorFixup.fixupVMOffset;
2241                     selRefFixups.push_back(fixup);
2242                 }
2243
2244                 if ( selectorFixup.isSharedCache ) {
2245                     // If the entry is in the shared cache then we already have the index for it
2246                     Image::SelectorReferenceFixup fixup;
2247                     fixup.chainEntry.index          = selectorFixup.sharedCache.selectorTableIndex;
2248                     fixup.chainEntry.next           = 0;
2249                     fixup.chainEntry.inSharedCache  = 1;
2250                     selRefFixups.push_back(fixup);
2251                 } else {
2252                     // We had to record the string for the closure table entries as we don't know the
2253                     // index until now
2254                     uint32_t selectorTableIndex = selectorStringTable->getIndex(selectorFixup.image.selectorString);
2255                     assert(selectorTableIndex != ObjCSelectorOpt::indexNotFound);
2256                     Image::SelectorReferenceFixup fixup;
2257                     fixup.chainEntry.index          = selectorTableIndex;
2258                     fixup.chainEntry.next           = 0;
2259                     fixup.chainEntry.inSharedCache  = 0;
2260                     selRefFixups.push_back(fixup);
2261                 }
2262
2263                 prevVMOffset = selectorFixup.fixupVMOffset;
2264             }
2265         }
2266
2267         // Stable Swift fixups
2268         STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::ClassStableSwiftFixup, stableSwiftFixups, 512);
2269         if ( !image.classStableSwiftFixups.empty() ) {
2270
2271             __block uint64_t lastOffset = -pointerSize;
2272             for (uint64_t runtimeOffset : image.classStableSwiftFixups) {
2273                 bool mergedIntoPrevious = false;
2274                 if ( (runtimeOffset > lastOffset) && !stableSwiftFixups.empty() ) {
2275                     uint64_t skipAmount = (runtimeOffset - lastOffset - pointerSize)/pointerSize;
2276                     if ( skipAmount*pointerSize != (runtimeOffset - lastOffset - pointerSize) ) {
2277                         // misaligned pointer means we cannot optimize
2278                     }
2279                     else {
2280                         if ( (stableSwiftFixups.back().repeatCount == 1) && (stableSwiftFixups.back().skipCount == 0) && (skipAmount <= 255) ) {
2281                             stableSwiftFixups.back().repeatCount = 2;
2282                             stableSwiftFixups.back().skipCount   = skipAmount;
2283                             assert(stableSwiftFixups.back().skipCount == skipAmount); // check overflow
2284                             mergedIntoPrevious       = true;
2285                         }
2286                         else if ( (stableSwiftFixups.back().skipCount == skipAmount) && (stableSwiftFixups.back().repeatCount < 0xfff) ) {
2287                             uint32_t prevRepeatCount = stableSwiftFixups.back().repeatCount;
2288                             stableSwiftFixups.back().repeatCount += 1;
2289                             assert(stableSwiftFixups.back().repeatCount > prevRepeatCount); // check overflow
2290                             mergedIntoPrevious       = true;
2291                         }
2292                     }
2293                 }
2294                 if ( !mergedIntoPrevious ) {
2295                     Image::ClassStableSwiftFixup pattern;
2296                     pattern.startVmOffset = runtimeOffset;
2297                     pattern.repeatCount   = 1;
2298                     pattern.skipCount     = 0;
2299                     assert(pattern.startVmOffset == runtimeOffset);
2300                     stableSwiftFixups.push_back(pattern);
2301                 }
2302                 lastOffset = runtimeOffset;
2303             }
2304         }
2305
2306         // Method list fixups
2307         // TODO: Implement this
2308         STACK_ALLOC_OVERFLOW_SAFE_ARRAY(Image::MethodListFixup, methodListFixups, 512);
2309
2310         image.writer->setObjCFixupInfo(objcProtocolClassTarget, image.objcImageInfoVMOffset, protocolFixups,
2311                                        selRefFixups, stableSwiftFixups, methodListFixups);
2312     }
2313
2314     return true;
2315 }
2316
2317 void ClosureBuilder::optimizeObjCSelectors(const objc_opt::objc_selopt_t* objcSelOpt,
2318                                            const Map<const char*, dyld3::closure::Image::ObjCImageOffset, HashCString, EqualCString>& closureSelectorMap,
2319                                            ObjCOptimizerImage& image) {
2320
2321     BuilderLoadedImage& li = *image.loadedImage;
2322
2323     const dyld3::MachOAnalyzer* ma = li.loadAddress();
2324     uint32_t pointerSize = ma->pointerSize();
2325     const uint64_t loadAddress = ma->preferredLoadAddress();
2326
2327     // The legacy (objc1) codebase uses a bunch of sections we don't want to reason about.  If we see them just give up.
2328     __block bool foundBadSection = false;
2329     ma->forEachSection(^(const MachOAnalyzer::SectionInfo &sectInfo, bool malformedSectionRange, bool &stop) {
2330         if ( strcmp(sectInfo.segInfo.segName, "__OBJC") != 0 )
2331             return;
2332         if (strcmp(sectInfo.sectName, "__module_info") == 0) {
2333             foundBadSection = true;
2334             stop = true;
2335             return;
2336         }
2337         if (strcmp(sectInfo.sectName, "__protocol") == 0) {
2338             foundBadSection = true;
2339             stop = true;
2340             return;
2341         }
2342         if (strcmp(sectInfo.sectName, "__message_refs") == 0) {
2343             foundBadSection = true;
2344             stop = true;
2345             return;
2346         }
2347     });
2348     if (foundBadSection) {
2349         image.diag.error("Old objc section");
2350         return;
2351     }
2352
2353     __block MachOAnalyzer::SectionCache selectorStringSectionCache(ma);
2354
2355     uint32_t sharedCacheSentinelIndex = objcSelOpt->getSentinelIndex();
2356
2357     auto visitReferenceToObjCSelector = ^void(uint64_t selectorStringVMAddr, uint64_t selectorReferenceVMAddr) {
2358
2359         uint64_t selectorUseImageOffset = selectorReferenceVMAddr - loadAddress;
2360         if ( (selectorUseImageOffset & 3) != 0 ) {
2361             image.diag.error("Unaligned selector reference fixup");
2362             return;
2363         }
2364
2365         // Image::SelectorReferenceFixup only has a 32-bit reach
2366         if ( selectorUseImageOffset >= (1ULL << 32) ) {
2367             image.diag.error("Selector reference fixup exceeds supported vm offset");
2368             return;
2369         }
2370
2371         // Get the section for the name
2372         const char* selectorString = nullptr;
2373         MachOAnalyzer::PrintableStringResult selectorStringResult = MachOAnalyzer::PrintableStringResult::UnknownSection;
2374         __block uint64_t selectorStringSectionStartVMAddr = 0;
2375         auto selectorStringSectionHandler = ^bool(const MachOAnalyzer::SectionInfo& sectInfo) {
2376
2377             // We only have 24-bits in ObjCClassNameImageOffset to index in to the strings
2378             if (sectInfo.sectSize >= Image::ObjCImageOffset::maximumOffset) {
2379                 return false;
2380             }
2381
2382             // We use 32-bit offsets so make sure the section is no larger than that.
2383             uint64_t classNameVMOffset = sectInfo.sectAddr - loadAddress;
2384             if (classNameVMOffset >= (1ULL << 32)) {
2385                 return false;
2386             }
2387
2388             selectorStringSectionStartVMAddr = sectInfo.sectAddr;
2389             return true;
2390         };
2391         selectorString = ma->getPrintableString(selectorStringVMAddr, selectorStringResult,
2392                                                 &selectorStringSectionCache, selectorStringSectionHandler);
2393
2394         if ( selectorStringResult != MachOAnalyzer::PrintableStringResult::CanPrint ) {
2395             image.diag.error("Invalid selector string for objc optimisation");
2396             return;
2397         }
2398
2399         uint32_t cacheSelectorIndex = objcSelOpt->getIndexForKey(selectorString);
2400         //printf("selector: %p -> %p %s\n", methodName, cacheSelector, selectorString);
2401
2402         if ( cacheSelectorIndex != sharedCacheSentinelIndex ) {
2403             // We got the selector from the cache so add a fixup to point there.
2404             ObjCOptimizerImage::SelectorFixup fixup;
2405             fixup.isSharedCache                     = true;
2406             fixup.fixupVMOffset                     = (uint32_t)selectorUseImageOffset;
2407             fixup.sharedCache.selectorTableIndex    = cacheSelectorIndex;
2408
2409             //printf("Overriding fixup at 0x%08llX to cache offset 0x%08llX\n", selectorUseImageOffset, (uint64_t)cacheSelector - (uint64_t)_dyldCache);
2410             image.selectorFixups.push_back(fixup);
2411             return;
2412         }
2413
2414         // See if this selector is already in the closure map from a previous image
2415         auto closureSelectorIt = closureSelectorMap.find(selectorString);
2416         if (closureSelectorIt != closureSelectorMap.end()) {
2417             // This selector was found in a previous image, so use it here.
2418             ObjCOptimizerImage::SelectorFixup fixup;
2419             fixup.isSharedCache         = false;
2420             fixup.fixupVMOffset         = (uint32_t)selectorUseImageOffset;
2421             fixup.image.selectorString  = selectorString;
2422
2423             //printf("Overriding fixup at 0x%08llX to '%s' offset 0x%08llX\n", selectorUseImageOffset, findLoadedImage(target.image.imageNum).path(), target.image.offset);
2424             image.selectorFixups.push_back(fixup);
2425             return;
2426         }
2427
2428         // See if this selector is already in the map for this image
2429         auto itAndInserted = image.selectorMap.insert({ selectorString, dyld3::closure::Image::ObjCImageOffset() });
2430         if (itAndInserted.second) {
2431             // We added the selector so its pointing in to our own image.
2432             // We don't need to add a fixup to our image, but we do need to
2433             // populate the data for other images later to point here.
2434             // First put our image in the list if its not already there.
2435             uint64_t methodNameVMOffset = selectorStringSectionStartVMAddr - loadAddress;
2436             if (!image.methodNameVMOffset) {
2437                 if ( _objcSelectorsHashTableImages.count() == Image::ObjCImageOffset::maximumImageIndex ) {
2438                     image.diag.error("Out of space for selector hash images");
2439                     return;
2440                 }
2441                 image.methodNameVMOffset = methodNameVMOffset;
2442             } else {
2443                 // If we already set the offset to the start of the method names section, double check that
2444                 // the section we are in right now is the same as that one.  Otherwise we don't have the code
2445                 // to handle both right now.
2446                 if (*image.methodNameVMOffset != methodNameVMOffset) {
2447                     image.diag.error("Cannot handle more than one selector strings section");
2448                     return;
2449                 }
2450             }
2451
2452             dyld3::closure::Image::ObjCImageOffset target;
2453             target.imageIndex   = (uint32_t)_objcSelectorsHashTableImages.count();
2454             target.imageOffset  = (uint32_t)(selectorStringVMAddr - selectorStringSectionStartVMAddr);
2455             itAndInserted.first->second = target;
2456             return;
2457         }
2458
2459         // This selector was found elsewhere in our image.  If this reference already points to the same
2460         // selector string as we found before (and it should!) then we have nothing to do.  Otherwise we
2461         // need to add a fixup here to make sure we point to our chosen definition.
2462         uint32_t imageOffset = (uint32_t)(selectorStringVMAddr - loadAddress);
2463         if ( imageOffset == (*image.methodNameVMOffset + itAndInserted.first->second.imageOffset) )
2464             return;
2465
2466         ObjCOptimizerImage::SelectorFixup fixup;
2467         fixup.isSharedCache         = false;
2468         fixup.fixupVMOffset         = (uint32_t)selectorUseImageOffset;
2469         fixup.image.selectorString  = selectorString;
2470
2471         //printf("Overriding fixup at 0x%08llX to '%s' offset 0x%08llX\n", selectorUseImageOffset, findLoadedImage(target.image.imageNum).path(), target.image.offset);
2472         image.selectorFixups.push_back(fixup);
2473     };
2474
2475     auto visitMethod = ^(uint64_t methodVMAddr, const dyld3::MachOAnalyzer::ObjCMethod& method) {
2476         visitReferenceToObjCSelector(method.nameVMAddr, method.nameLocationVMAddr);
2477     };
2478
2479     auto visitClass = ^(Diagnostics& diag, uint64_t classVMAddr,
2480                         uint64_t classSuperclassVMAddr, uint64_t classDataVMAddr,
2481                         const dyld3::MachOAnalyzer::ObjCClassInfo& objcClass, bool isMetaClass) {
2482         ma->forEachObjCMethod(objcClass.baseMethodsVMAddr(pointerSize), li.contentRebased,
2483                               visitMethod);
2484     };
2485
2486     auto visitCategory = ^(Diagnostics& diag, uint64_t categoryVMAddr,
2487                            const dyld3::MachOAnalyzer::ObjCCategory& objcCategory) {
2488         ma->forEachObjCMethod(objcCategory.instanceMethodsVMAddr, li.contentRebased,
2489                               visitMethod);
2490         ma->forEachObjCMethod(objcCategory.classMethodsVMAddr, li.contentRebased,
2491                               visitMethod);
2492     };
2493     auto visitProtocol = ^(Diagnostics& diag, uint64_t protocolVMAddr,
2494                            const dyld3::MachOAnalyzer::ObjCProtocol& objCProtocol) {
2495         ma->forEachObjCMethod(objCProtocol.instanceMethodsVMAddr, li.contentRebased,
2496                               visitMethod);
2497         ma->forEachObjCMethod(objCProtocol.classMethodsVMAddr, li.contentRebased,
2498                               visitMethod);
2499         ma->forEachObjCMethod(objCProtocol.optionalInstanceMethodsVMAddr, li.contentRebased,
2500                               visitMethod);
2501         ma->forEachObjCMethod(objCProtocol.optionalClassMethodsVMAddr, li.contentRebased,
2502                               visitMethod);
2503     };
2504
2505     // Walk the class list
2506     ma->forEachObjCClass(image.diag, li.contentRebased, visitClass);
2507     if (image.diag.hasError())
2508         return;
2509
2510     // Walk the category list
2511     ma->forEachObjCCategory(image.diag, li.contentRebased, visitCategory);
2512     if (image.diag.hasError())
2513         return;
2514
2515     // Walk the protocol list
2516     ma->forEachObjCProtocol(image.diag, li.contentRebased, visitProtocol);
2517     if (image.diag.hasError())
2518         return;
2519
2520     // Visit the selector refs
2521     ma->forEachObjCSelectorReference(image.diag, li.contentRebased, ^(uint64_t selRefVMAddr, uint64_t selRefTargetVMAddr) {
2522         visitReferenceToObjCSelector(selRefTargetVMAddr, selRefVMAddr);
2523     });
2524     if (image.diag.hasError())
2525         return;
2526
2527     // Visit the message refs
2528     // Note this isn't actually supported in libobjc any more.  Its logic for deciding whether to support it is if this is true:
2529     // #if (defined(__x86_64__) && (TARGET_OS_OSX || TARGET_OS_SIMULATOR))
2530     // So to keep it simple, lets only do this walk if we are x86_64
2531     if ( ma->isArch("x86_64") || ma->isArch("x86_64h") ) {
2532         if (ma->hasObjCMessageReferences()) {
2533             image.diag.error("Cannot handle message refs");
2534             return;
2535         }
2536     }
2537 }
2538
2539 static const dyld3::MachOAnalyzer* getMachHeaderFromObjCHeaderInfo(const void* opaqueHeaderInfo, uint32_t pointerSize) {
2540     if (pointerSize == 8) {
2541         typedef int64_t PtrTy;
2542         struct HeaderInfo {
2543             PtrTy mhdr_offset;     // offset to mach_header_64
2544             PtrTy info_offset;     // offset to objc_image_info *
2545         };
2546         const HeaderInfo* headerInfo = (const HeaderInfo*)opaqueHeaderInfo;
2547         return (const dyld3::MachOAnalyzer*)(((const uint8_t*)&headerInfo->mhdr_offset) + headerInfo->mhdr_offset);
2548     } else {
2549         typedef int32_t PtrTy;
2550         struct HeaderInfo {
2551             PtrTy mhdr_offset;     // offset to mach_header
2552             PtrTy info_offset;     // offset to objc_image_info *
2553         };
2554         const HeaderInfo* headerInfo = (const HeaderInfo*)opaqueHeaderInfo;
2555         return (const dyld3::MachOAnalyzer*)(((const uint8_t*)&headerInfo->mhdr_offset) + headerInfo->mhdr_offset);
2556     }
2557 }
2558
2559 void ClosureBuilder::addDuplicateObjCClassWarning(const char* className,
2560                                                   const char* duplicateDefinitionPath,
2561                                                   const char* canonicalDefinitionPath)
2562 {
2563     if ( _objcDuplicateClassWarnings == nullptr )
2564         _objcDuplicateClassWarnings = PathPool::allocate();
2565     // Use a diagnostic to give us a buffer we can safely print to
2566     Diagnostics diag;
2567     diag.error("Class %s is implemented in both %s and %s. One of the two will be used. Which one is undefined.",
2568                className, canonicalDefinitionPath, duplicateDefinitionPath);
2569 #if BUILDING_CACHE_BUILDER
2570     _objcDuplicateClassWarnings->add(diag.errorMessage().c_str());
2571 #else
2572     _objcDuplicateClassWarnings->add(diag.errorMessage());
2573 #endif
2574 }
2575
2576 void ClosureBuilder::optimizeObjCClasses(const objc_opt::objc_clsopt_t* objcClassOpt,
2577                                          const Map<const dyld3::MachOAnalyzer*, bool, HashPointer, EqualPointer>& sharedCacheImagesMap,
2578                                          const Map<const char*, dyld3::closure::Image::ObjCDuplicateClass, HashCString, EqualCString>& duplicateSharedCacheClasses,
2579                                          ObjCOptimizerImage& image) {
2580
2581     BuilderLoadedImage& li = *image.loadedImage;
2582     OverflowSafeArray<ObjCOptimizerImage::SeenClass>& seenClasses = image.seenClasses;
2583
2584     const dyld3::MachOAnalyzer* ma = li.loadAddress();
2585     const uint32_t pointerSize = ma->pointerSize();
2586     const uint64_t loadAddress = ma->preferredLoadAddress();
2587
2588     // Keep track of any missing weak imports so that we can tell if the superclasses are nil
2589     // This is necessary as the shared cache will be marked with 'no missing weak superclasses'
2590     // and so we need to continue to satisfy that constraint
2591     __block Map<uint64_t, bool, HashUInt64, EqualUInt64> missingWeakImportOffets;
2592     if (li.hasMissingWeakImports) {
2593         if (ma->hasChainedFixups()) {
2594             const Image* closureImage = image.writer->currentImage();
2595
2596             const Array<Image::ResolvedSymbolTarget> targets = closureImage->chainedTargets();
2597             if ( !targets.empty() ) {
2598                 ma->withChainStarts(_diag, closureImage->chainedStartsOffset(), ^(const dyld_chained_starts_in_image* startsInfo) {
2599                     ma->forEachFixupInAllChains(_diag, startsInfo, false, ^(MachOLoaded::ChainedFixupPointerOnDisk* fixupLoc,
2600                                                                             const dyld_chained_starts_in_segment* segInfo, bool& fixupsStop) {
2601                         uint64_t fixupOffset = (uint8_t*)fixupLoc - (uint8_t*)ma;
2602                         uint32_t bindOrdinal;
2603                         if ( fixupLoc->isBind(segInfo->pointer_format, bindOrdinal) ) {
2604                             if ( bindOrdinal < targets.count() ) {
2605                                 const Image::ResolvedSymbolTarget& target = targets[bindOrdinal];
2606                                 if ( (target.absolute.kind == Image::ResolvedSymbolTarget::kindAbsolute) && (target.absolute.value == 0) )
2607                                     missingWeakImportOffets[fixupOffset] = true;
2608                             }
2609                             else {
2610                                 image.diag.error("out of range bind ordinal %d (max %lu)", bindOrdinal, targets.count());
2611                                 fixupsStop = true;
2612                             }
2613                         }
2614                     });
2615                 });
2616                 if (image.diag.hasError())
2617                     return;
2618             }
2619         } else {
2620             forEachBind(li, ^(uint64_t runtimeOffset, Image::ResolvedSymbolTarget target, const ResolvedTargetInfo& targetInfo, bool& stop) {
2621                 if ( (target.absolute.kind == Image::ResolvedSymbolTarget::kindAbsolute) && (target.absolute.value == 0) )
2622                     missingWeakImportOffets[runtimeOffset] = true;
2623             }, ^(const char *strongSymbolName) {
2624             }, ^() { });
2625         }
2626     }
2627
2628     // Class names and data may be in different sections depending on swift vs objc so handle multiple sections
2629     __block MachOAnalyzer::SectionCache classNameSectionCache(ma);
2630     __block MachOAnalyzer::SectionCache classSectionCache(ma);
2631
2632     ma->forEachObjCClass(image.diag, li.contentRebased, ^(Diagnostics &diag, uint64_t classVMAddr,
2633                                                           uint64_t classSuperclassVMAddr, uint64_t classDataVMAddr,
2634                                                           const MachOAnalyzer::ObjCClassInfo &objcClass, bool isMetaClass) {
2635         if (isMetaClass) return;
2636
2637         // Make sure the superclass pointer is not nil
2638         uint64_t superclassRuntimeOffset = classSuperclassVMAddr - loadAddress;
2639         if (missingWeakImportOffets.find(superclassRuntimeOffset) != missingWeakImportOffets.end()) {
2640             diag.error("Missing weak superclass");
2641             return;
2642         }
2643
2644         // Does this class need to be fixed up for stable Swift ABI.
2645         // Note the order matches the objc runtime in that we always do this fix before checking for dupes,
2646         // but after excluding classes with missing weak superclasses.
2647         if (objcClass.isUnfixedBackwardDeployingStableSwift()) {
2648             // Class really is stable Swift, pretending to be pre-stable.
2649             // Fix its lie.  This involves fixing the FAST bits on the class data value, so record that vmaddr
2650             image.classStableSwiftFixups.push_back(classDataVMAddr - loadAddress);
2651         }
2652
2653         // Get the section for the name
2654         const char* className = nullptr;
2655         MachOAnalyzer::PrintableStringResult classNameResult = MachOAnalyzer::PrintableStringResult::UnknownSection;
2656         __block uint64_t classNameSectionStartVMAddr = 0;
2657         auto classNameSectionHandler = ^bool(const MachOAnalyzer::SectionInfo& sectInfo) {
2658             // We only have 24-bits in ObjCClassNameImageOffset to index in to the strings
2659             if (sectInfo.sectSize >= Image::ObjCClassNameImageOffset::maximumOffset) {
2660                 return false;
2661             }
2662
2663             // We use 32-bit offsets so make sure the section is no larger than that.
2664             uint64_t classNameVMOffset = sectInfo.sectAddr - loadAddress;
2665             if (classNameVMOffset >= (1ULL << 32)) {
2666                 return false;
2667             }
2668
2669             classNameSectionStartVMAddr = sectInfo.sectAddr;
2670             return true;
2671         };
2672         uint64_t classNameVMAddr = objcClass.nameVMAddr(pointerSize);
2673         className = ma->getPrintableString(classNameVMAddr, classNameResult,
2674                                            &classNameSectionCache, classNameSectionHandler);
2675
2676         if ( classNameResult != MachOAnalyzer::PrintableStringResult::CanPrint ) {
2677             diag.error("Invalid class name for objc optimisation");
2678             return;
2679         }
2680
2681         // If the class also exists in a shared cache image which is loaded, then objc
2682         // would have found that one, regardless of load order.  So we can just skip this one.
2683         {
2684             void *cls;
2685             void *hi;
2686             uint32_t index;
2687             uint32_t count = objcClassOpt->getClassHeaderAndIndex(className, cls, hi, index);
2688             if (count == 1) {
2689                 // exactly one matching class.  Check if its loaded
2690                 const dyld3::MachOAnalyzer* sharedCacheMA = getMachHeaderFromObjCHeaderInfo(hi, pointerSize);
2691                 if (sharedCacheImagesMap.find(sharedCacheMA) != sharedCacheImagesMap.end()) {
2692                     addDuplicateObjCClassWarning(className, li.path(), sharedCacheMA->installName());
2693
2694                     // We have a duplicate class, so check if we've already got it in our map.
2695                     if ( duplicateSharedCacheClasses.find(className) == duplicateSharedCacheClasses.end() ) {
2696                         // We haven't seen this one yet
2697                         Image::ObjCDuplicateClass duplicateClass;
2698                         duplicateClass.sharedCacheClassOptIndex         = index;
2699                         duplicateClass.sharedCacheClassDuplicateIndex   = 0;
2700                         image.classSharedCacheDuplicates.insert({ className, duplicateClass });
2701                     }
2702                 }
2703             }
2704             else if (count > 1) {
2705                 // more than one matching class - find one that is loaded
2706                 void *clslist[count];
2707                 void *hilist[count];
2708                 objcClassOpt->getClassesAndHeaders(className, clslist, hilist);
2709                 for (uint32_t i = 0; i < count; i++) {
2710                     const dyld3::MachOAnalyzer* sharedCacheMA = getMachHeaderFromObjCHeaderInfo(hilist[i], pointerSize);
2711                     if (sharedCacheImagesMap.find(sharedCacheMA) != sharedCacheImagesMap.end()) {
2712                         addDuplicateObjCClassWarning(className, li.path(), sharedCacheMA->installName());
2713
2714                         // We have a duplicate class, so check if we've already got it in our map.
2715                         if ( duplicateSharedCacheClasses.find(className) == duplicateSharedCacheClasses.end() ) {
2716                             // We haven't seen this one yet
2717                             Image::ObjCDuplicateClass duplicateClass;
2718                             duplicateClass.sharedCacheClassOptIndex         = index;
2719                             duplicateClass.sharedCacheClassDuplicateIndex   = i;
2720                             image.classSharedCacheDuplicates.insert({ className, duplicateClass });
2721                         }
2722
2723                         break;
2724                     }
2725                 }
2726             }
2727         }
2728
2729         // Get the section for the class itself
2730         __block uint64_t classSectionStartVMAddr = 0;
2731         auto classSectionHandler = ^bool(const MachOAnalyzer::SectionInfo& sectInfo) {
2732             // We only have 23-bits in ObjCClassImageOffset to index in to the classes
2733             if (sectInfo.sectSize > Image::ObjCClassImageOffset::maximumOffset) {
2734                 return false;
2735             }
2736
2737             // We use 32-bit offsets so make sure the section is no larger than that.
2738             uint64_t classDatasVMOffset = sectInfo.sectAddr - loadAddress;
2739             if (classDatasVMOffset >= (1ULL << 32)) {
2740                 return false;
2741             }
2742
2743             classSectionStartVMAddr = sectInfo.sectAddr;
2744             return true;
2745         };
2746         if (!classSectionCache.findSectionForVMAddr(classVMAddr, classSectionHandler)) {
2747             diag.error("Invalid class for objc optimisation");
2748             return;
2749         }
2750
2751         // Make sure we have an entry for our images offsets for later
2752         uint64_t classNameSectionVMOffset = classNameSectionStartVMAddr - loadAddress;
2753         uint64_t classSectionVMOffset = classSectionStartVMAddr - loadAddress;
2754         uint64_t hashTableVMOffsetsIndex = 0;
2755         for (auto nameAndDataVMOffset : image.classesNameAndDataVMOffsets) {
2756             if ( (nameAndDataVMOffset.first == classNameSectionVMOffset) && (nameAndDataVMOffset.second == classSectionVMOffset) )
2757                 break;
2758             ++hashTableVMOffsetsIndex;
2759         }
2760
2761         if (hashTableVMOffsetsIndex == image.classesNameAndDataVMOffsets.count()) {
2762             // Didn't find an image entry with this offset.  Add one if we have space
2763             uint64_t totalHashTableImages = image.classesNameAndDataVMOffsets.count() + _objcClassesHashTableImages.count();
2764             if ( totalHashTableImages == Image::ObjCClassNameImageOffset::maximumImageIndex ) {
2765                 // No more space.  We need to give up
2766                 diag.error("No more space for class hash table image");
2767                 return;
2768             }
2769             image.classesNameAndDataVMOffsets.push_back({ classNameSectionVMOffset, classSectionVMOffset });
2770         }
2771
2772         hashTableVMOffsetsIndex += _objcClassesHashTableImages.count();
2773
2774         uint64_t classNameOffset = classNameVMAddr - classNameSectionStartVMAddr;
2775         uint64_t classDataOffset = classVMAddr - classSectionStartVMAddr;
2776
2777         closure::Image::ObjCClassNameImageOffset classNameTarget;
2778         classNameTarget.classNameImageIndex   = (uint32_t)hashTableVMOffsetsIndex;
2779         classNameTarget.classNameImageOffset  = (uint32_t)classNameOffset;
2780
2781         dyld3::closure::Image::ObjCClassImageOffset classDataTarget;
2782         classDataTarget.classData.imageIndex   = (uint32_t)hashTableVMOffsetsIndex;
2783         classDataTarget.classData.imageOffset  = (uint32_t)classDataOffset;
2784         classDataTarget.classData.isDuplicate  = 0;
2785
2786         seenClasses.push_back({ classNameTarget, classDataTarget });
2787     });
2788 }
2789
2790 void ClosureBuilder::optimizeObjCProtocols(const objc_opt::objc_protocolopt2_t* objcProtocolOpt,
2791                                            const Map<const dyld3::MachOAnalyzer*, bool, HashPointer, EqualPointer>& sharedCacheImagesMap,
2792                                            ObjCOptimizerImage& image) {
2793
2794     BuilderLoadedImage& li = *image.loadedImage;
2795     OverflowSafeArray<ObjCOptimizerImage::SeenClass>& seenProtocols = image.seenProtocols;
2796
2797     const dyld3::MachOAnalyzer* ma = li.loadAddress();
2798     const uint32_t pointerSize = ma->pointerSize();
2799     const uint64_t loadAddress = ma->preferredLoadAddress();
2800
2801     // Protocol names and data may be in different sections depending on swift vs objc so handle multiple sections
2802     __block MachOAnalyzer::SectionCache protocolNameSectionCache(ma);
2803     __block MachOAnalyzer::SectionCache protocolSectionCache(ma);
2804
2805     ma->forEachObjCProtocol(image.diag, li.contentRebased, ^(Diagnostics &diag, uint64_t protocolVMAddr,
2806                                                              const dyld3::MachOAnalyzer::ObjCProtocol &objCProtocol) {
2807         if ( objCProtocol.requiresObjCReallocation ) {
2808             // We can't optimize this protocol as the runtime needs all fields to be present
2809             diag.error("Protocol is too small to be optimized");
2810             return;
2811         }
2812         if ( objCProtocol.isaVMAddr != 0 ) {
2813             // We can't optimize this protocol if it has an ISA as we want to override it
2814             diag.error("Protocol ISA cannot be non-zero");
2815             return;
2816         }
2817
2818         // Get the section for the name
2819         const char* protocolName = nullptr;
2820         MachOAnalyzer::PrintableStringResult protocolNameResult = MachOAnalyzer::PrintableStringResult::UnknownSection;
2821         __block uint64_t protocolNameSectionStartVMAddr = 0;
2822         auto protocolNameSectionHandler = ^bool(const MachOAnalyzer::SectionInfo& sectInfo) {
2823             // We only have 24-bits in ObjCClassNameImageOffset to index in to the strings
2824             if (sectInfo.sectSize >= Image::ObjCClassNameImageOffset::maximumOffset) {
2825                 return false;
2826             }
2827
2828             // We use 32-bit offsets so make sure the section is no larger than that.
2829             uint64_t protocolNameVMOffset = sectInfo.sectAddr - loadAddress;
2830             if (protocolNameVMOffset >= (1ULL << 32)) {
2831                 return false;
2832             }
2833
2834             protocolNameSectionStartVMAddr = sectInfo.sectAddr;
2835             return true;
2836         };
2837         uint64_t protocolNameVMAddr = objCProtocol.nameVMAddr;
2838         protocolName = ma->getPrintableString(protocolNameVMAddr, protocolNameResult,
2839                                               &protocolNameSectionCache, protocolNameSectionHandler);
2840
2841         if ( protocolNameResult != MachOAnalyzer::PrintableStringResult::CanPrint ) {
2842             diag.error("Invalid protocol name for objc optimisation");
2843             return;
2844         }
2845
2846         // If the protocol also exists in a shared cache image which is loaded, then objc
2847         // would have found that one, regardless of load order.  So we can just skip this one.
2848         {
2849             void *cls;
2850             void *hi;
2851             uint32_t count = objcProtocolOpt->getClassAndHeader(protocolName, cls, hi);
2852             if (count == 1) {
2853                 // exactly one matching protocol.  Check if its loaded
2854                 if (sharedCacheImagesMap.find(getMachHeaderFromObjCHeaderInfo(hi, pointerSize)) != sharedCacheImagesMap.end())
2855                     return;
2856             }
2857             else if (count > 1) {
2858                 // more than one matching protocol - find one that is loaded
2859                 void *clslist[count];
2860                 void *hilist[count];
2861                 objcProtocolOpt->getClassesAndHeaders(protocolName, clslist, hilist);
2862                 for (uint32_t i = 0; i < count; i++) {
2863                     if (sharedCacheImagesMap.find(getMachHeaderFromObjCHeaderInfo(hilist[i], pointerSize)) != sharedCacheImagesMap.end())
2864                         return;
2865                 }
2866             }
2867         }
2868
2869         // Get the section for the protocol itself
2870         __block uint64_t protocolSectionStartVMAddr = 0;
2871         auto protocolSectionHandler = ^bool(const MachOAnalyzer::SectionInfo& sectInfo) {
2872             // We only have 23-bits in ObjCClassImageOffset to index in to the protocols
2873             if (sectInfo.sectSize > Image::ObjCClassImageOffset::maximumOffset) {
2874                 return false;
2875             }
2876
2877             // We use 32-bit offsets so make sure the section is no larger than that.
2878             uint64_t protocolDatasVMOffset = sectInfo.sectAddr - loadAddress;
2879             if (protocolDatasVMOffset >= (1ULL << 32)) {
2880                 return false;
2881             }
2882
2883             protocolSectionStartVMAddr = sectInfo.sectAddr;
2884             return true;
2885         };
2886         if (!protocolSectionCache.findSectionForVMAddr(protocolVMAddr, protocolSectionHandler)) {
2887             diag.error("Invalid protocol for objc optimisation");
2888             return;
2889         }
2890
2891         // Make sure we have an entry for our images offsets for later
2892         uint64_t protocolNameSectionVMOffset = protocolNameSectionStartVMAddr - loadAddress;
2893         uint64_t protocolSectionVMOffset = protocolSectionStartVMAddr - loadAddress;
2894         uint64_t hashTableVMOffsetsIndex = 0;
2895         for (auto nameAndDataVMOffset : image.classesNameAndDataVMOffsets) {
2896             if ( (nameAndDataVMOffset.first == protocolNameSectionVMOffset) && (nameAndDataVMOffset.second == protocolSectionVMOffset) )
2897                 break;
2898             ++hashTableVMOffsetsIndex;
2899         }
2900
2901         if (hashTableVMOffsetsIndex == image.classesNameAndDataVMOffsets.count()) {
2902             // Didn't find an image entry with this offset.  Add one if we have space
2903             uint64_t totalHashTableImages = image.classesNameAndDataVMOffsets.count() + _objcClassesHashTableImages.count();
2904             if ( totalHashTableImages == Image::ObjCClassNameImageOffset::maximumImageIndex ) {
2905                 // No more space.  We need to give up
2906                 diag.error("No more space for protocol hash table image");
2907                 return;
2908             }
2909             image.classesNameAndDataVMOffsets.push_back({ protocolNameSectionVMOffset, protocolSectionVMOffset });
2910         }
2911
2912         hashTableVMOffsetsIndex += _objcClassesHashTableImages.count();
2913
2914         uint64_t protocolNameOffset = protocolNameVMAddr - protocolNameSectionStartVMAddr;
2915         uint64_t protocolDataOffset = protocolVMAddr - protocolSectionStartVMAddr;
2916
2917         closure::Image::ObjCClassNameImageOffset protocolNameTarget;
2918         protocolNameTarget.classNameImageIndex   = (uint32_t)hashTableVMOffsetsIndex;
2919         protocolNameTarget.classNameImageOffset  = (uint32_t)protocolNameOffset;
2920
2921         dyld3::closure::Image::ObjCClassImageOffset protocolDataTarget;
2922         protocolDataTarget.classData.imageIndex   = (uint32_t)hashTableVMOffsetsIndex;
2923         protocolDataTarget.classData.imageOffset  = (uint32_t)protocolDataOffset;
2924         protocolDataTarget.classData.isDuplicate  = 0;
2925
2926         seenProtocols.push_back({ protocolNameTarget, protocolDataTarget });
2927     });
2928 }
2929
2930 // used at launch by dyld when kernel has already mapped main executable
2931 const LaunchClosure* ClosureBuilder::makeLaunchClosure(const LoadedFileInfo& fileInfo, bool allowInsertFailures)
2932 {
2933     dyld3::ScopedTimer timer(DBG_DYLD_TIMING_BUILD_CLOSURE, 0, 0, 0);
2934     const mach_header* mainMH = (const mach_header*)fileInfo.fileContent;
2935     // set up stack based storage for all arrays
2936     BuilderLoadedImage  loadImagesStorage[512];
2937     Image::LinkedImage  dependenciesStorage[512*8];
2938     InterposingTuple    tuplesStorage[64];
2939     Closure::PatchEntry cachePatchStorage[64];
2940     const char*         weakDefNameStorage[64];
2941     _loadedImages.setInitialStorage(loadImagesStorage, 512);
2942     _dependencies.setInitialStorage(dependenciesStorage, 512*8);
2943     _interposingTuples.setInitialStorage(tuplesStorage, 64);
2944     _weakDefCacheOverrides.setInitialStorage(cachePatchStorage, 64);
2945     _weakDefsFromChainedBinds.setInitialStorage(weakDefNameStorage, 64);
2946     ArrayFinalizer<BuilderLoadedImage> scopedCleanup(_loadedImages, ^(BuilderLoadedImage& li) { if (li.unmapWhenDone) {_fileSystem.unloadFile(li.loadedFileInfo); li.unmapWhenDone=false;} });
2947
2948     const MachOAnalyzer* mainExecutable = MachOAnalyzer::validMainExecutable(_diag, mainMH, fileInfo.path, fileInfo.sliceLen, _archs, _platform);
2949     if ( mainExecutable == nullptr )
2950         return nullptr;
2951     if ( !mainExecutable->isDynamicExecutable() ) {
2952         _diag.error("not a main executable");
2953         return nullptr;
2954     }
2955     _isLaunchClosure    = true;
2956     _allowMissingLazies = true;
2957
2958     _nextIndex = 0;
2959
2960     // add main executable
2961     __block BuilderLoadedImage mainEntry;
2962     mainEntry.loadedFileInfo            = fileInfo;
2963     mainEntry.imageNum                  = 0; // We can't fill this in until we've done inserted dylibs
2964     mainEntry.unmapWhenDone             = false;
2965     mainEntry.contentRebased            = false;
2966     mainEntry.hasInits                  = false;
2967     mainEntry.markNeverUnload           = true;
2968     mainEntry.rtldLocal                 = false;
2969     mainEntry.isBadImage                = false;
2970     mainEntry.mustBuildClosure          = true;
2971     mainEntry.hasMissingWeakImports     = false;
2972     mainEntry.overrideImageNum          = 0;
2973
2974     // Set the executable load path so that @executable_path can use it later
2975     _mainProgLoadPath = fileInfo.path;
2976
2977     // add any DYLD_INSERT_LIBRARIES
2978     _pathOverrides.forEachInsertedDylib(^(const char* dylibPath, bool &stop) {
2979         LoadedImageChain chainMain = { nullptr, mainEntry };
2980         BuilderLoadedImage* foundTopImage;
2981         if ( !findImage(dylibPath, chainMain, foundTopImage, LinkageType::kInserted, 0, true) ) {
2982             if ( !allowInsertFailures ) {
2983                 if ( _diag.noError() )
2984                     _diag.error("could not load inserted dylib %s", dylibPath);
2985                 stop = true;
2986                 return;
2987             }
2988             _diag.clearError(); // FIXME add way to plumb back warning
2989         }
2990     });
2991
2992     if ( _diag.hasError() )
2993         return nullptr;
2994
2995     _mainProgLoadIndex = (uint32_t)_loadedImages.count();
2996     mainEntry.imageNum = _startImageNum + _nextIndex++;
2997     _loadedImages.push_back(mainEntry);
2998
2999         // get mach_headers for all images needed to launch this main executable
3000     LoadedImageChain chainStart = { nullptr, _loadedImages[_mainProgLoadIndex] };
3001     recursiveLoadDependents(chainStart);
3002     if ( _diag.hasError() )
3003         return nullptr;
3004     for (uint32_t i=0; i < _mainProgLoadIndex; ++i) {
3005         LoadedImageChain insertChainStart = { nullptr, _loadedImages[i] };
3006         recursiveLoadDependents(insertChainStart);
3007         if ( _diag.hasError() )
3008             return nullptr;
3009     }
3010     loadDanglingUpwardLinks();
3011
3012     // If we have an on-disk image then we need all images which are dependent on the disk image to get a new
3013     // initializer order.  Its not enough to just do the top level image as we may dlopen while in dlopen
3014     invalidateInitializerRoots();
3015
3016     // now that everything loaded, set _libDyldImageNum and _libSystemImageNum
3017     for (BuilderLoadedImage& li : _loadedImages) {
3018         if ( li.loadAddress()->isDylib() && (strcmp(li.loadAddress()->installName(), "/usr/lib/system/libdyld.dylib") == 0) )
3019             _libDyldImageNum = li.imageNum;
3020         else if ( strcmp(li.path(), "/usr/lib/libSystem.B.dylib") == 0 )
3021             _libSystemImageNum = li.imageNum;
3022     }
3023
3024     // only some images need to go into closure (non-rooted ones from dyld cache do not)
3025     STACK_ALLOC_ARRAY(ImageWriter, writers, _loadedImages.count());
3026     for (BuilderLoadedImage& li : _loadedImages) {
3027         if ( li.mustBuildClosure ) {
3028             writers.push_back(ImageWriter());
3029             buildImage(writers.back(), li);
3030             if ( _diag.hasError() )
3031                 return nullptr;
3032         }
3033    }
3034
3035     bool optimizedObjC = optimizeObjC(writers);
3036
3037     // Note we have to compute the init order after buildImage as buildImage may set hasInits to true
3038     for (uintptr_t imageIndex = 0, writerIndex = 0; imageIndex != _loadedImages.count(); ++imageIndex) {
3039         BuilderLoadedImage& li = _loadedImages[imageIndex];
3040         if ( li.mustBuildClosure ) {
3041             computeInitOrder(writers[writerIndex], (uint32_t)imageIndex);
3042             writerIndex++;
3043         }
3044     }
3045
3046     // combine all Image objects into one ImageArray
3047     ImageArrayWriter imageArrayWriter(_startImageNum, (uint32_t)writers.count(), _foundDyldCacheRoots);
3048     for (ImageWriter& writer : writers) {
3049         imageArrayWriter.appendImage(writer.finalize());
3050         writer.deallocate();
3051     }
3052     const ImageArray* imageArray = imageArrayWriter.finalize();
3053
3054     // merge ImageArray object into LaunchClosure object
3055     __block LaunchClosureWriter closureWriter(imageArray);
3056
3057     if (optimizedObjC) {
3058         if (!_objcSelectorsHashTable.empty())
3059             closureWriter.setObjCSelectorInfo(_objcSelectorsHashTable, _objcSelectorsHashTableImages);
3060
3061         if (!_objcClassesHashTableImages.empty()) {
3062             closureWriter.setObjCClassAndProtocolInfo(_objcClassesHashTable, _objcProtocolsHashTable,
3063                                                       _objcClassesHashTableImages);
3064         }
3065
3066         if ( _objcDuplicateClassWarnings != nullptr ) {
3067             _objcDuplicateClassWarnings->forEachPath(^(const char* warning) {
3068                 closureWriter.addWarning(Closure::Warning::duplicateObjCClass, warning);
3069             });
3070         }
3071
3072         if (!_objcClassesDuplicatesHashTable.empty())
3073             closureWriter.setObjCDuplicateClassesInfo(_objcClassesDuplicatesHashTable);
3074     }
3075
3076     // record shared cache info
3077     if ( _dyldCache != nullptr ) {
3078         // record cache UUID
3079         uuid_t cacheUUID;
3080         _dyldCache->getUUID(cacheUUID);
3081         closureWriter.setDyldCacheUUID(cacheUUID);
3082
3083         // record any cache patching needed because of dylib overriding cache
3084         for (const BuilderLoadedImage& li : _loadedImages) {
3085             if ( li.overrideImageNum != 0 ) {
3086                 uint32_t imageIndex = li.overrideImageNum - (uint32_t)_dyldImageArray->startImageNum();
3087                 STACK_ALLOC_ARRAY(Closure::PatchEntry, patches, _dyldCache->patchableExportCount(imageIndex));
3088                 MachOLoaded::DependentToMachOLoaded reexportFinder = ^(const MachOLoaded* mh, uint32_t depIndex) {
3089                     return (const MachOLoaded*)findDependent(mh, depIndex);
3090                 };
3091                 //fprintf(stderr, "'%s' overrides '%s'\n", li.loadedFileInfo.path, cacheImage->path());
3092                 _dyldCache->forEachPatchableExport(imageIndex, ^(uint32_t cacheOffsetOfImpl, const char* symbolName) {
3093                     dyld3::MachOAnalyzer::FoundSymbol foundInfo;
3094                     Diagnostics                       patchDiag;
3095                     Closure::PatchEntry               patch;
3096                     patch.overriddenDylibInCache  = li.overrideImageNum;
3097                     patch.exportCacheOffset       = cacheOffsetOfImpl;
3098                     if ( li.loadAddress()->findExportedSymbol(patchDiag, symbolName, false, foundInfo, reexportFinder) ) {
3099                         const MachOAnalyzer* impDylib = (const MachOAnalyzer*)foundInfo.foundInDylib;
3100                         patch.replacement.image.kind     = Image::ResolvedSymbolTarget::kindImage;
3101                         patch.replacement.image.imageNum = findLoadedImage(impDylib).imageNum;
3102                         patch.replacement.image.offset   = foundInfo.value;
3103                     }
3104                     else {
3105                         // this means the symbol is missing in the cache override dylib, so set any uses to NULL
3106                         patch.replacement.absolute.kind    = Image::ResolvedSymbolTarget::kindAbsolute;
3107                         patch.replacement.absolute.value   = 0;
3108                     }
3109                     patches.push_back(patch);
3110                 });
3111                 closureWriter.addCachePatches(patches);
3112             }
3113         }
3114
3115         // handle any extra weak-def coalescing needed by chained fixups
3116         if ( !_weakDefsFromChainedBinds.empty() ) {
3117             for (const char* symbolName : _weakDefsFromChainedBinds) {
3118                 Image::ResolvedSymbolTarget cacheOverrideTarget;
3119                 bool haveCacheOverride = false;
3120                 bool foundCachOverrideIsWeakDef = false;
3121                 for (const BuilderLoadedImage& li : _loadedImages) {
3122                     if ( !li.loadAddress()->hasWeakDefs() )
3123                         continue;
3124                     Image::ResolvedSymbolTarget target;
3125                     ResolvedTargetInfo          targetInfo;
3126                     if ( findSymbolInImage(li.loadAddress(), symbolName, 0, false, false, target, targetInfo) ) {
3127                         if ( li.loadAddress()->inDyldCache() ) {
3128                             if ( haveCacheOverride ) {
3129                                 Closure::PatchEntry patch;
3130                                 patch.exportCacheOffset      = (uint32_t)target.sharedCache.offset;
3131                                 patch.overriddenDylibInCache = li.imageNum;
3132                                 patch.replacement            = cacheOverrideTarget;
3133                                 _weakDefCacheOverrides.push_back(patch);
3134                             }
3135                             else {
3136                                 // found first in cached dylib, so no need to patch cache for this symbol
3137                                 break;
3138                             }
3139                         }
3140                         else {
3141                             // found image that exports this symbol and is not in cache
3142                             if ( !haveCacheOverride || (foundCachOverrideIsWeakDef && !targetInfo.isWeakDef) ) {
3143                                 // update cache to use this symbol if it if first found or it is first non-weak found
3144                                 cacheOverrideTarget         = target;
3145                                 foundCachOverrideIsWeakDef  = targetInfo.isWeakDef;
3146                                 haveCacheOverride           = true;
3147                             }
3148                         }
3149                     }
3150                 }
3151             }
3152         }
3153
3154         // record any cache patching needed because weak-def C++ symbols override dyld cache
3155         if ( !_weakDefCacheOverrides.empty() )
3156             closureWriter.addCachePatches(_weakDefCacheOverrides);
3157
3158    }
3159
3160 #if __IPHONE_OS_VERSION_MIN_REQUIRED
3161     // if closure is built on-device for iOS, then record boot UUID
3162     char bootSessionUUID[256] = { 0 };
3163     size_t bootSize = sizeof(bootSessionUUID);
3164     if ( sysctlbyname("kern.bootsessionuuid", bootSessionUUID, &bootSize, NULL, 0) == 0 )
3165         closureWriter.setBootUUID(bootSessionUUID);
3166 #endif
3167
3168      // record any interposing info
3169     imageArray->forEachImage(^(const Image* image, bool &stop) {
3170         if ( !image->inDyldCache() )
3171             addInterposingTuples(closureWriter, image, findLoadedImage(image->imageNum()).loadAddress());
3172     });
3173
3174     // modify fixups in contained Images by applying interposing tuples
3175     closureWriter.applyInterposing((const LaunchClosure*)closureWriter.currentTypedBytes());
3176
3177     // set flags
3178     closureWriter.setUsedAtPaths(_atPathUsed);
3179     closureWriter.setUsedFallbackPaths(_fallbackPathUsed);
3180     closureWriter.setHasInsertedLibraries(_mainProgLoadIndex > 0);
3181     closureWriter.setInitImageCount((uint32_t)_loadedImages.count());
3182
3183     // add other closure attributes
3184     addClosureInfo(closureWriter);
3185
3186     // make result
3187     const LaunchClosure* result = closureWriter.finalize();
3188     imageArrayWriter.deallocate();
3189
3190     timer.setData4(dyld3::DyldTimingBuildClosure::LaunchClosure_Built);
3191
3192     return result;
3193 }
3194
3195 // used by libdyld for dlopen()
3196 const DlopenClosure* ClosureBuilder::makeDlopenClosure(const char* path, const LaunchClosure* mainClosure, const Array<LoadedImage>& alreadyLoadedList,
3197                                                        closure::ImageNum callerImageNum, bool noLoad, bool forceBindLazies, bool canUseSharedCacheClosure, closure::ImageNum* topImageNum)
3198 {
3199     dyld3::ScopedTimer timer(DBG_DYLD_TIMING_BUILD_CLOSURE, 0, 0, 0);
3200     // set up stack based storage for all arrays
3201     BuilderLoadedImage  loadImagesStorage[300];
3202     Image::LinkedImage  dependenciesStorage[128];
3203     Closure::PatchEntry cachePatchStorage[64];
3204     _loadedImages.setInitialStorage(loadImagesStorage, 300);
3205     _dependencies.setInitialStorage(dependenciesStorage, 128);
3206     _weakDefCacheOverrides.setInitialStorage(cachePatchStorage, 64);
3207     ArrayFinalizer<BuilderLoadedImage> scopedCleanup(_loadedImages, ^(BuilderLoadedImage& li) { if (li.unmapWhenDone) {_fileSystem.unloadFile(li.loadedFileInfo); li.unmapWhenDone=false;} });
3208
3209     // fill in builder array from already loaded images
3210     bool cachedDylibsExpectedOnDisk = _dyldCache ? _dyldCache->header.dylibsExpectedOnDisk : true;
3211     uintptr_t callerImageIndex = UINTPTR_MAX;
3212     for (const LoadedImage& ali : alreadyLoadedList) {
3213         const Image*          image       = ali.image();
3214         const MachOAnalyzer*  ma          = (MachOAnalyzer*)(ali.loadedAddress());
3215         bool                  inDyldCache = ma->inDyldCache();
3216         BuilderLoadedImage entry;
3217         ImageNum overrideImageNum;
3218         entry.loadedFileInfo.path        = image->path();
3219         entry.loadedFileInfo.fileContent = ma;
3220         entry.loadedFileInfo.sliceOffset = 0;
3221         entry.loadedFileInfo.inode       = 0;
3222         entry.loadedFileInfo.mtime       = 0;
3223         entry.imageNum                   = image->imageNum();
3224         entry.dependents                 = image->dependentsArray();
3225         entry.unmapWhenDone              = false;
3226         entry.contentRebased             = inDyldCache;
3227         entry.hasInits                   = false;
3228         entry.markNeverUnload            = image->neverUnload();
3229         entry.rtldLocal                  = ali.hideFromFlatSearch();
3230         entry.isBadImage                 = false;
3231         entry.mustBuildClosure           = false;
3232         entry.hasMissingWeakImports      = false;
3233         entry.overrideImageNum           = 0;
3234         if ( !inDyldCache && image->isOverrideOfDyldCacheImage(overrideImageNum) ) {
3235             entry.overrideImageNum  = overrideImageNum;
3236             canUseSharedCacheClosure = false;
3237         }
3238         if ( !inDyldCache || cachedDylibsExpectedOnDisk )
3239             image->hasFileModTimeAndInode(entry.loadedFileInfo.inode, entry.loadedFileInfo.mtime);
3240         if ( entry.imageNum == callerImageNum )
3241             callerImageIndex = _loadedImages.count();
3242         _loadedImages.push_back(entry);
3243    }
3244     _alreadyInitedIndex = (uint32_t)_loadedImages.count();
3245
3246     // find main executable (may be needed for @executable_path)
3247     _isLaunchClosure = false;
3248     for (uint32_t i=0; i < alreadyLoadedList.count(); ++i) {
3249         if ( _loadedImages[i].loadAddress()->isMainExecutable() )  {
3250             _mainProgLoadIndex = i;
3251             _mainProgLoadPath = _loadedImages[i].path();
3252             break;
3253         }
3254     }
3255
3256     // We can't use an existing dlopen closure if the main closure had interposing tuples
3257     if (canUseSharedCacheClosure) {
3258         if (mainClosure->hasInterposings())
3259             canUseSharedCacheClosure = false;
3260     }
3261
3262     // add top level dylib being dlopen()ed
3263     BuilderLoadedImage* foundTopImage;
3264     _nextIndex = 0;
3265     // @rpath has caller's LC_PRATH, then main executable's LC_RPATH
3266     BuilderLoadedImage& callerImage = (callerImageIndex != UINTPTR_MAX) ? _loadedImages[callerImageIndex]  : _loadedImages[_mainProgLoadIndex];
3267     LoadedImageChain chainCaller = { nullptr, callerImage };
3268     LoadedImageChain chainMain = { &chainCaller, _loadedImages[_mainProgLoadIndex] };
3269     if ( !findImage(path, chainMain, foundTopImage, LinkageType::kDynamic, 0, canUseSharedCacheClosure) ) {
3270         // If we didn't find the image, it might be a symlink to something in the dyld cache that is not on disk
3271         if ( (_dyldCache != nullptr) && !_dyldCache->header.dylibsExpectedOnDisk ) {
3272             char resolvedPath[PATH_MAX];
3273             if ( _fileSystem.getRealPath(path, resolvedPath) ) {
3274                 _diag.clearError();
3275                 if ( !findImage(resolvedPath, chainMain, foundTopImage, LinkageType::kDynamic, 0, canUseSharedCacheClosure) ) {
3276                     return nullptr;
3277                 }
3278             } else {
3279                 // We didn't find a new path from realpath
3280                 return nullptr;
3281             }
3282         } else {
3283             // cached dylibs on disk, so don't call realpath() again, it would have been found first call to findImage()
3284             return nullptr;
3285         }
3286     }
3287
3288     // exit early in RTLD_NOLOAD mode
3289     if ( noLoad ) {
3290         timer.setData4(dyld3::DyldTimingBuildClosure::DlopenClosure_NoLoad);
3291         // if no new images added to _loadedImages, then requested path was already loaded
3292         if ( (uint32_t)_loadedImages.count() == _alreadyInitedIndex )
3293             *topImageNum = foundTopImage->imageNum;
3294         else
3295             *topImageNum = 0;
3296         return nullptr;
3297     }
3298
3299     // fast path if roots are not allowed and target is in dyld cache or is other
3300     if ( (_dyldCache != nullptr) && (_dyldCache->header.cacheType == kDyldSharedCacheTypeProduction) ) {
3301         if ( foundTopImage->imageNum < closure::kFirstLaunchClosureImageNum ) {
3302             if (foundTopImage->imageNum < closure::kLastDyldCacheImageNum)
3303                 timer.setData4(dyld3::DyldTimingBuildClosure::DlopenClosure_UsedSharedCacheDylib);
3304             else
3305                 timer.setData4(dyld3::DyldTimingBuildClosure::DlopenClosure_UsedSharedCacheOther);
3306             *topImageNum = foundTopImage->imageNum;
3307             return nullptr;
3308         }
3309     }
3310
3311     // recursive load dependents
3312     // @rpath for stuff top dylib depends on uses LC_RPATH from caller, main exe, and dylib being dlopen()ed
3313     LoadedImageChain chainTopDylib = { &chainMain, *foundTopImage };
3314     recursiveLoadDependents(chainTopDylib, canUseSharedCacheClosure);
3315     if ( _diag.hasError() )
3316         return nullptr;
3317     loadDanglingUpwardLinks(canUseSharedCacheClosure);
3318     if ( _diag.hasError() )
3319         return nullptr;
3320
3321     // RTLD_NOW means fail the dlopen() if a symbol cannot be bound
3322     _allowMissingLazies = !forceBindLazies;
3323
3324     // only some images need to go into closure (ones from dyld cache do not, unless the cache format changed)
3325     STACK_ALLOC_ARRAY(ImageWriter, writers, _loadedImages.count());
3326     if ( _foundNonCachedImage || _foundDyldCacheRoots ) {
3327         // If we have an on-disk image then we need all images which are dependent on the disk image to get a new
3328         // initializer order.  Its not enough to just do the top level image as we may dlopen while in dlopen
3329         invalidateInitializerRoots();
3330
3331         for (uintptr_t loadedImageIndex = 0; loadedImageIndex != _loadedImages.count(); ++loadedImageIndex) {
3332             BuilderLoadedImage& li = _loadedImages[loadedImageIndex];
3333             if ( li.mustBuildClosure ) {
3334                 writers.push_back(ImageWriter());
3335                 buildImage(writers.back(), li);
3336                 if ( _diag.hasError() )
3337                     return nullptr;
3338             }
3339         }
3340
3341         // Note we have to compute the init order after buildImage as buildImage may set hasInits to true
3342         for (uintptr_t imageIndex = 0, writerIndex = 0; imageIndex != _loadedImages.count(); ++imageIndex) {
3343             BuilderLoadedImage& li = _loadedImages[imageIndex];
3344             if ( li.mustBuildClosure ) {
3345                 computeInitOrder(writers[writerIndex], (uint32_t)imageIndex);
3346                 writerIndex++;
3347             }
3348         }
3349     }
3350     if ( _diag.hasError() )
3351         return nullptr;
3352
3353     // check if top image loaded is in shared cache along with everything it depends on
3354     *topImageNum = foundTopImage->imageNum;
3355     if ( _foundNonCachedImage || _foundDyldCacheRoots ) {
3356         if ( canUseSharedCacheClosure && ( foundTopImage->imageNum < closure::kFirstLaunchClosureImageNum ) ) {
3357             // We used a shared cache built closure, but now discovered roots.  We need to try again
3358             topImageNum = 0;
3359             return sRetryDlopenClosure;
3360         }
3361     } else {
3362         if (foundTopImage->imageNum < closure::kLastDyldCacheImageNum)
3363             timer.setData4(dyld3::DyldTimingBuildClosure::DlopenClosure_UsedSharedCacheDylib);
3364         else
3365             timer.setData4(dyld3::DyldTimingBuildClosure::DlopenClosure_UsedSharedCacheOther);
3366         return nullptr;
3367     }
3368
3369     // combine all Image objects into one ImageArray
3370     ImageArrayWriter imageArrayWriter(_startImageNum, (uint32_t)writers.count(), _foundDyldCacheRoots);
3371     for (ImageWriter& writer : writers) {
3372         imageArrayWriter.appendImage(writer.finalize());
3373         writer.deallocate();
3374     }
3375     const ImageArray* imageArray = imageArrayWriter.finalize();
3376
3377     // merge ImageArray object into LaunchClosure object
3378     DlopenClosureWriter closureWriter(imageArray);
3379
3380     // add other closure attributes
3381     closureWriter.setTopImageNum(foundTopImage->imageNum);
3382
3383     // record any cache patching needed because of dylib overriding cache
3384     if ( _dyldCache != nullptr ) {
3385         for (const BuilderLoadedImage& li : _loadedImages) {
3386             if ( (li.overrideImageNum != 0) && (li.imageNum >= _startImageNum) ) {
3387                 const Image* cacheImage = _dyldImageArray->imageForNum(li.overrideImageNum);
3388                 uint32_t imageIndex = cacheImage->imageNum() - (uint32_t)_dyldCache->cachedDylibsImageArray()->startImageNum();
3389                 STACK_ALLOC_ARRAY(Closure::PatchEntry, patches, _dyldCache->patchableExportCount(imageIndex));
3390                 MachOLoaded::DependentToMachOLoaded reexportFinder = ^(const MachOLoaded* mh, uint32_t depIndex) {
3391                     return (const MachOLoaded*)findDependent(mh, depIndex);
3392                 };
3393                 //fprintf(stderr, "'%s' overrides '%s'\n", li.loadedFileInfo.path, cacheImage->path());
3394                 _dyldCache->forEachPatchableExport(imageIndex,
3395                                                    ^(uint32_t cacheOffsetOfImpl, const char* symbolName) {
3396                     dyld3::MachOAnalyzer::FoundSymbol foundInfo;
3397                     Diagnostics                       patchDiag;
3398                     Closure::PatchEntry               patch;
3399                     patch.overriddenDylibInCache  = li.overrideImageNum;
3400                     patch.exportCacheOffset       = cacheOffsetOfImpl;
3401                     if ( li.loadAddress()->findExportedSymbol(patchDiag, symbolName, false, foundInfo, reexportFinder) ) {
3402                         const MachOAnalyzer* impDylib = (const MachOAnalyzer*)foundInfo.foundInDylib;
3403                         patch.replacement.image.kind     = Image::ResolvedSymbolTarget::kindImage;
3404                         patch.replacement.image.imageNum = findLoadedImage(impDylib).imageNum;
3405                         patch.replacement.image.offset   = foundInfo.value;
3406                     }
3407                     else {
3408                         patch.replacement.absolute.kind    = Image::ResolvedSymbolTarget::kindAbsolute;
3409                         patch.replacement.absolute.value   = 0;
3410                     }
3411                     patches.push_back(patch);
3412                 });
3413                 closureWriter.addCachePatches(patches);
3414             }
3415         }
3416     }
3417
3418     // modify fixups in contained Images by applying interposing tuples
3419     closureWriter.applyInterposing(mainClosure);
3420
3421     // Dlopen's should never keep track of missing paths as we don't cache these closures.
3422     assert(_mustBeMissingPaths == nullptr);
3423
3424     // make final DlopenClosure object
3425     const DlopenClosure* result = closureWriter.finalize();
3426     imageArrayWriter.deallocate();
3427     timer.setData4(dyld3::DyldTimingBuildClosure::DlopenClosure_Built);
3428     return result;
3429 }
3430
3431
3432 // used by dyld_closure_util
3433 const LaunchClosure* ClosureBuilder::makeLaunchClosure(const char* mainPath, bool allowInsertFailures)
3434 {
3435     char realerPath[MAXPATHLEN];
3436     closure::LoadedFileInfo loadedFileInfo = MachOAnalyzer::load(_diag, _fileSystem, mainPath, _archs, _platform, realerPath);
3437     const MachOAnalyzer* mh = (const MachOAnalyzer*)loadedFileInfo.fileContent;
3438     loadedFileInfo.path = mainPath;
3439     if (_diag.hasError())
3440         return nullptr;
3441     if (mh == nullptr) {
3442         _diag.error("could not load file");
3443         return nullptr;
3444     }
3445     if (!mh->isDynamicExecutable()) {
3446         _diag.error("file is not an executable");
3447         return nullptr;
3448     }
3449     const_cast<PathOverrides*>(&_pathOverrides)->setMainExecutable(mh, mainPath);
3450     const LaunchClosure* launchClosure = makeLaunchClosure(loadedFileInfo, allowInsertFailures);
3451     loadedFileInfo.unload(loadedFileInfo);
3452     return launchClosure;
3453 }
3454
3455 void ClosureBuilder::setDyldCacheInvalidFormatVersion() {
3456     _dyldCacheInvalidFormatVersion = true;
3457 }
3458
3459
3460 // used by dyld shared cache builder
3461 const ImageArray* ClosureBuilder::makeDyldCacheImageArray(bool customerCache, const Array<CachedDylibInfo>& dylibs, const Array<CachedDylibAlias>& aliases)
3462 {
3463     // because this is run in cache builder using dispatch_apply() there is minimal stack space
3464     // so set up storage for all arrays to be vm_allocated
3465     uintptr_t maxImageCount = dylibs.count() + 16;
3466     _loadedImages.reserve(maxImageCount);
3467     _dependencies.reserve(maxImageCount*16);
3468
3469     _makingDyldCacheImages = true;
3470     _allowMissingLazies    = false;
3471     _makingCustomerCache   = customerCache;
3472     _aliases               = &aliases;
3473
3474     // build _loadedImages[] with every dylib in cache
3475     __block ImageNum imageNum = _startImageNum;
3476     for (const CachedDylibInfo& aDylibInfo : dylibs)  {
3477         BuilderLoadedImage entry;
3478         entry.loadedFileInfo                = aDylibInfo.fileInfo;
3479         entry.imageNum                      = imageNum++;
3480         entry.unmapWhenDone                 = false;
3481         entry.contentRebased                = false;
3482         entry.hasInits                      = false;
3483         entry.markNeverUnload               = true;
3484         entry.rtldLocal                     = false;
3485         entry.isBadImage                    = false;
3486         entry.mustBuildClosure              = false;
3487         entry.hasMissingWeakImports         = false;
3488         entry.overrideImageNum              = 0;
3489         _loadedImages.push_back(entry);
3490     }
3491
3492     // wire up dependencies between cached dylibs
3493     for (BuilderLoadedImage& li : _loadedImages) {
3494         LoadedImageChain chainStart = { nullptr, li };
3495         recursiveLoadDependents(chainStart);
3496         if ( _diag.hasError() )
3497             break;
3498     }
3499     assert(_loadedImages.count() == dylibs.count());
3500
3501     // create an ImageWriter for each cached dylib
3502     STACK_ALLOC_ARRAY(ImageWriter, writers, _loadedImages.count());
3503     for (BuilderLoadedImage& li : _loadedImages) {
3504         writers.push_back(ImageWriter());
3505         buildImage(writers.back(), li);
3506     }
3507
3508     // add initializer order into each dylib
3509     // Note we have to compute the init order after buildImage as buildImage may set hasInits to true
3510     for (const BuilderLoadedImage& li : _loadedImages) {
3511         uint32_t index = li.imageNum - _startImageNum;
3512         computeInitOrder(writers[index], index);
3513     }
3514
3515     // combine all Image objects into one ImageArray
3516     ImageArrayWriter imageArrayWriter(_startImageNum, (uint32_t)writers.count(), _foundDyldCacheRoots);
3517     for (ImageWriter& writer : writers) {
3518         imageArrayWriter.appendImage(writer.finalize());
3519         writer.deallocate();
3520     }
3521     const ImageArray* imageArray = imageArrayWriter.finalize();
3522
3523     return imageArray;
3524 }
3525
3526
3527 #if BUILDING_CACHE_BUILDER
3528 const ImageArray* ClosureBuilder::makeOtherDylibsImageArray(const Array<LoadedFileInfo>& otherDylibs, uint32_t cachedDylibsCount)
3529 {
3530     // because this is run in cache builder using dispatch_apply() there is minimal stack space
3531     // so set up storage for all arrays to be vm_allocated
3532     uintptr_t maxImageCount = otherDylibs.count() + cachedDylibsCount + 128;
3533     _loadedImages.reserve(maxImageCount);
3534     _dependencies.reserve(maxImageCount*16);
3535
3536     // build _loadedImages[] with every dylib in cache, followed by others
3537     _nextIndex = 0;
3538     for (const LoadedFileInfo& aDylibInfo : otherDylibs)  {
3539         BuilderLoadedImage entry;
3540         entry.loadedFileInfo                = aDylibInfo;
3541         entry.imageNum                      = _startImageNum + _nextIndex++;
3542         entry.unmapWhenDone                 = false;
3543         entry.contentRebased                = false;
3544         entry.hasInits                      = false;
3545         entry.markNeverUnload               = false;
3546         entry.rtldLocal                     = false;
3547         entry.isBadImage                    = false;
3548         entry.mustBuildClosure              = false;
3549         entry.hasMissingWeakImports         = false;
3550         entry.overrideImageNum              = 0;
3551         _loadedImages.push_back(entry);
3552     }
3553
3554     // wire up dependencies between cached dylibs
3555     // Note, _loadedImages can grow when we call recursiveLoadDependents so we need
3556     // to check the count on each iteration.
3557     for (uint64_t index = 0; index != _loadedImages.count(); ++index) {
3558         BuilderLoadedImage& li = _loadedImages[index];
3559         LoadedImageChain chainStart = { nullptr, li };
3560         recursiveLoadDependents(chainStart);
3561         if ( _diag.hasError() ) {
3562             _diag.warning("while building dlopen closure for %s: %s", li.loadedFileInfo.path, _diag.errorMessage().c_str());
3563             //fprintf(stderr, "while building dlopen closure for %s: %s\n", li.loadedFileInfo.path, _diag.errorMessage().c_str());
3564             _diag.clearError();
3565             li.isBadImage = true;    // mark bad
3566         }
3567     }
3568
3569     auto invalidateBadImages = [&]() {
3570         // Invalidate images with bad dependencies
3571         while (true) {
3572             bool madeChange = false;
3573             for (BuilderLoadedImage& li : _loadedImages) {
3574                 if (li.isBadImage) {
3575                     // Already invalidated
3576                     continue;
3577                 }
3578                 for (Image::LinkedImage depIndex : li.dependents) {
3579                     if ( depIndex.imageNum() == kMissingWeakLinkedImage )
3580                         continue;
3581                     if ( depIndex.imageNum() >= dyld3::closure::kLastDyldCacheImageNum ) {
3582                         // dlopen closures can only depend on the shared cache.  This is because if foo.dylib links bar.dylib
3583                         // and bar.dylib is loaded in to the launch closure, then the dlopen closure for foo.dylib wouldn't see
3584                         // bar.dylib at the image num in the launch closure
3585                         _diag.warning("while building dlopen closure for %s: dependent dylib is not from shared cache", li.loadedFileInfo.path);
3586                         li.isBadImage = true;    // mark bad
3587                         madeChange = true;
3588                         continue;
3589                     }
3590                     BuilderLoadedImage& depImage = findLoadedImage(depIndex.imageNum());
3591                     if (depImage.isBadImage) {
3592                         _diag.warning("while building dlopen closure for %s: dependent dylib had error", li.loadedFileInfo.path);
3593                         li.isBadImage = true;    // mark bad
3594                         madeChange = true;
3595                     }
3596                 }
3597             }
3598             if (!madeChange)
3599                 break;
3600         }
3601     };
3602
3603     invalidateBadImages();
3604
3605     // create an ImageWriter for each cached dylib
3606     STACK_ALLOC_ARRAY(ImageWriter, writers, _loadedImages.count());
3607     for (BuilderLoadedImage& li : _loadedImages) {
3608         if ( li.isBadImage )  {
3609             writers.push_back(ImageWriter());
3610             writers.back().setInvalid();
3611             continue;
3612         }
3613         if ( li.imageNum < dyld3::closure::kLastDyldCacheImageNum )
3614             continue;
3615         writers.push_back(ImageWriter());
3616         buildImage(writers.back(), li);
3617         if ( _diag.hasError() ) {
3618             _diag.warning("while building dlopen closure for %s: %s", li.loadedFileInfo.path, _diag.errorMessage().c_str());
3619             //fprintf(stderr, "while building dlopen closure for %s: %s\n", li.loadedFileInfo.path, _diag.errorMessage().c_str());
3620             _diag.clearError();
3621             li.isBadImage = true;    // mark bad
3622             writers.back().setInvalid();
3623         }
3624     }
3625
3626     invalidateBadImages();
3627
3628     // add initializer order into each dylib
3629     // Note we have to compute the init order after buildImage as buildImage may set hasInits to true
3630     for (const BuilderLoadedImage& li : _loadedImages) {
3631         if ( li.imageNum < dyld3::closure::kLastDyldCacheImageNum )
3632             continue;
3633         if (li.isBadImage)
3634             continue;
3635         uint32_t index = li.imageNum - _startImageNum;
3636         computeInitOrder(writers[index], index);
3637     }
3638
3639     // combine all Image objects into one ImageArray
3640     ImageArrayWriter imageArrayWriter(_startImageNum, (uint32_t)writers.count(), _foundDyldCacheRoots);
3641     for (ImageWriter& writer : writers) {
3642         imageArrayWriter.appendImage(writer.finalize());
3643         writer.deallocate();
3644     }
3645     const ImageArray* imageArray = imageArrayWriter.finalize();
3646
3647     return imageArray;
3648 }
3649 #endif
3650
3651
3652 bool ClosureBuilder::inLoadedImageArray(const Array<LoadedImage>& loadedList, ImageNum imageNum)
3653 {
3654     for (const LoadedImage& ali : loadedList) {
3655         if ( ali.image()->representsImageNum(imageNum) )
3656             return true;
3657     }
3658     return false;
3659 }
3660
3661 void ClosureBuilder::buildLoadOrderRecurse(Array<LoadedImage>& loadedList, const Array<const ImageArray*>& imagesArrays, const Image* image)
3662 {
3663     // breadth first load
3664     STACK_ALLOC_ARRAY(const Image*, needToRecurse, 256);
3665     image->forEachDependentImage(^(uint32_t dependentIndex, dyld3::closure::Image::LinkKind kind, ImageNum depImageNum, bool &stop) {
3666         if ( !inLoadedImageArray(loadedList, depImageNum) ) {
3667             const Image* depImage = ImageArray::findImage(imagesArrays, depImageNum);
3668             loadedList.push_back(LoadedImage::make(depImage));
3669             needToRecurse.push_back(depImage);
3670         }
3671     });
3672
3673     // recurse load
3674     for (const Image* img : needToRecurse) {
3675         buildLoadOrderRecurse(loadedList, imagesArrays, img);
3676     }
3677 }
3678
3679 void ClosureBuilder::buildLoadOrder(Array<LoadedImage>& loadedList, const Array<const ImageArray*>& imagesArrays, const Closure* toAdd)
3680 {
3681     const dyld3::closure::Image* topImage = ImageArray::findImage(imagesArrays, toAdd->topImage());
3682         loadedList.push_back(LoadedImage::make(topImage));
3683         buildLoadOrderRecurse(loadedList, imagesArrays, topImage);
3684 }
3685
3686
3687
3688 ////////////////////////////  ObjCStringTable ////////////////////////////////////////
3689
3690 template<typename PerfectHashT, typename ImageOffsetT>
3691 void ObjCStringTable::write(const PerfectHashT& phash, const Array<std::pair<const char*, ImageOffsetT>>& strings)
3692 {
3693     ObjCSelectorOpt::StringTarget sentinel = (ObjCSelectorOpt::StringTarget)ImageOffsetT::sentinelValue;
3694     // Set header
3695     capacity = phash.capacity;
3696     occupied = phash.occupied;
3697     shift = phash.shift;
3698     mask = phash.mask;
3699     sentinelTarget = sentinel;
3700     roundedTabSize = std::max(phash.mask+1, 4U);
3701     salt = phash.salt;
3702
3703     // Set hash data
3704     for (uint32_t i = 0; i < 256; i++) {
3705         scramble[i] = phash.scramble[i];
3706     }
3707     for (uint32_t i = 0; i < phash.mask+1; i++) {
3708         tab[i] = phash.tab[i];
3709     }
3710
3711     dyld3::Array<StringTarget> targetsArray = targets();
3712     dyld3::Array<StringHashCheckByte> checkBytesArray = checkBytes();
3713
3714     // Set offsets to the sentinel
3715     for (uint32_t i = 0; i < phash.capacity; i++) {
3716         targetsArray[i] = sentinel;
3717     }
3718     // Set checkbytes to 0
3719     for (uint32_t i = 0; i < phash.capacity; i++) {
3720         checkBytesArray[i] = 0;
3721     }
3722
3723     // Set real string offsets and checkbytes
3724     for (const auto& s : strings) {
3725         assert(s.second.raw != sentinelTarget);
3726         uint32_t h = hash(s.first);
3727         targetsArray[h] = s.second.raw;
3728         checkBytesArray[h] = checkbyte(s.first);
3729     }
3730 }
3731
3732 ////////////////////////////  ObjCClassOpt ////////////////////////////////////////
3733
3734
3735 template<typename PerfectHashT, typename ImageOffsetT, typename ClassesMapT>
3736 void ObjCClassOpt::write(const PerfectHashT& phash, const Array<std::pair<const char*, ImageOffsetT>>& strings,
3737            const ClassesMapT& classes, uint32_t preCalculatedDuplicateCount)
3738 {
3739     ObjCStringTable::write(phash, strings);
3740
3741     __block dyld3::Array<ClassTarget> classOffsetsArray = classOffsets();
3742     __block dyld3::Array<ClassTarget> duplicateOffsetsArray = duplicateOffsets(preCalculatedDuplicateCount);
3743
3744     // Set class offsets to 0
3745     for (uint32_t i = 0; i < capacity; i++) {
3746         classOffsetsArray[i].raw = dyld3::closure::Image::ObjCImageOffset::sentinelValue;
3747     }
3748
3749     classes.forEachEntry(^(const char *const &key, const Image::ObjCClassImageOffset **values, uint64_t valuesCount) {
3750         uint32_t keyIndex = getIndex(key);
3751         assert(keyIndex != indexNotFound);
3752         assert(classOffsetsArray[keyIndex].raw == dyld3::closure::Image::ObjCImageOffset::sentinelValue);
3753
3754         if (valuesCount == 1) {
3755             // Only one entry so write it in to the class offsets directly
3756             Image::ObjCClassImageOffset classImageOffset = *(values[0]);
3757             assert(classImageOffset.classData.isDuplicate == 0);
3758             classOffsetsArray[keyIndex] = classImageOffset;
3759             return;
3760         }
3761
3762         // We have more than one value.  We add a placeholder to the class offsets which tells us the head
3763         // of the linked list of classes in the duplicates array
3764         uint32_t dest = duplicateCount();
3765         duplicateCount() += valuesCount;
3766
3767         Image::ObjCClassImageOffset classImagePlaceholder;
3768         assert(valuesCount < (1 << 8));
3769         classImagePlaceholder.duplicateData.count           = (uint32_t)valuesCount;
3770         classImagePlaceholder.duplicateData.index           = dest;
3771         classImagePlaceholder.duplicateData.isDuplicate     = 1;
3772         classOffsetsArray[keyIndex] = classImagePlaceholder;
3773
3774         for (uint64_t i = 0; i != valuesCount; ++i) {
3775             Image::ObjCClassImageOffset classImageOffset = *(values[i]);
3776             assert(classImageOffset.classData.isDuplicate == 0);
3777             duplicateOffsetsArray.push_back(classImageOffset);
3778         }
3779     });
3780 }
3781
3782 } // namespace closure
3783 } // namespace dyld3