launch-cache/MachOLayout.hpp

   1 /* -*- mode: C++; c-basic-offset: 4; tab-width: 4 -*-
   2  *
   3  * Copyright (c) 2006-2011 Apple Inc. All rights reserved.
   4  *
   5  * @APPLE_LICENSE_HEADER_START@
   6  *
   7  * This file contains Original Code and/or Modifications of Original Code
   8  * as defined in and that are subject to the Apple Public Source License
   9  * Version 2.0 (the 'License'). You may not use this file except in
  10  * compliance with the License. Please obtain a copy of the License at
  11  * http://www.opensource.apple.com/apsl/ and read it before using this
  12  * file.
  13  *
  14  * The Original Code and all software distributed under the License are
  15  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  16  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  17  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  18  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  19  * Please see the License for the specific language governing rights and
  20  * limitations under the License.
  21  *
  22  * @APPLE_LICENSE_HEADER_END@
  23  */
  24
  25 #ifndef __MACHO_LAYOUT__
  26 #define __MACHO_LAYOUT__
  27
  28 #include <sys/types.h>
  29 #include <sys/stat.h>
  30 #include <sys/errno.h>
  31 #include <sys/mman.h>
  32 #include <mach/mach.h>
  33 #include <limits.h>
  34 #include <stdarg.h>
  35 #include <stdio.h>
  36 #include <fcntl.h>
  37 #include <errno.h>
  38 #include <unistd.h>
  39 #include <mach-o/loader.h>
  40 #include <mach-o/fat.h>
  41
  42 #include <vector>
  43 #include <set>
  44 #include <unordered_map>
  45
  46 #include "MachOFileAbstraction.hpp"
  47 #include "Architectures.hpp"
  48
  49
  50 void throwf(const char* format, ...) __attribute__((format(printf, 1, 2)));
  51
  52 __attribute__((noreturn))
  53 void throwf(const char* format, ...)
  54 {
  55         va_list list;
  56         char*   p;
  57         va_start(list, format);
  58         vasprintf(&p, format, list);
  59         va_end(list);
  60
  61         const char*     t = p;
  62         throw t;
  63 }
  64
  65
  66 class MachOLayoutAbstraction
  67 {
  68 public:
  69         struct Segment
  70         {
  71         public:
  72                                         Segment(uint64_t addr, uint64_t vmsize, uint64_t offset, uint64_t file_size, uint64_t sectionsSize,
  73                                                         uint64_t sectionsAlignment, uint64_t align,
  74                                                         uint32_t prot, uint32_t sectionCount, const char* segName) : fOrigAddress(addr), fOrigSize(vmsize),
  75                                                         fOrigFileOffset(offset),  fOrigFileSize(file_size), fOrigPermissions(prot),
  76                                                         fSize(vmsize), fFileOffset(offset), fFileSize(file_size), fAlignment(align),
  77                                                         fPermissions(prot), fSectionCount(sectionCount), fSectionsSize(sectionsSize),
  78                                                         fSectionsAlignment(sectionsAlignment), fNewAddress(0), fMappedAddress(NULL) {
  79                                                                 strlcpy(fOrigName, segName, 16);
  80                                                         }
  81
  82                 uint64_t        address() const         { return fOrigAddress; }
  83                 uint64_t        size() const            { return fSize; }
  84                 uint64_t        fileOffset() const      { return fFileOffset; }
  85                 uint64_t        fileSize() const        { return fFileSize; }
  86                 uint32_t        permissions() const { return fPermissions; }
  87                 bool            readable() const        { return fPermissions & VM_PROT_READ; }
  88                 bool            writable() const        { return fPermissions & VM_PROT_WRITE; }
  89                 bool            executable() const      { return fPermissions & VM_PROT_EXECUTE; }
  90                 uint64_t        alignment() const       { return fAlignment; }
  91                 const char* name() const                { return fOrigName; }
  92                 uint64_t        newAddress() const      { return fNewAddress; }
  93                 uint32_t        sectionCount() const{ return fSectionCount; }
  94                 uint64_t        sectionsSize() const{ return fSectionsSize; }
  95                 uint64_t        sectionsAlignment() const { return fSectionsAlignment; }
  96                 void*           mappedAddress() const                   { return fMappedAddress; }
  97                 void            setNewAddress(uint64_t addr)    { fNewAddress = addr; }
  98                 void            setMappedAddress(void* addr)    { fMappedAddress = addr; }
  99                 void            setSize(uint64_t new_size)              { fSize = new_size; }
 100                 void            setFileOffset(uint64_t new_off) { fFileOffset = new_off; }
 101                 void            setFileSize(uint64_t new_size)  { fFileSize = new_size; }
 102                 void            setWritable(bool w)                             { if (w) fPermissions |= VM_PROT_WRITE; else fPermissions &= ~VM_PROT_WRITE; }
 103                 void            setSectionsAlignment(uint64_t v){ fSectionsAlignment = v; }
 104                 void            reset()                                                 { fSize=fOrigSize; fFileOffset=fOrigFileOffset; fFileSize=fOrigFileSize; fPermissions=fOrigPermissions; }
 105         private:
 106                 uint64_t                fOrigAddress;
 107                 uint64_t                fOrigSize;
 108                 uint64_t                fOrigFileOffset;
 109                 uint64_t                fOrigFileSize;
 110                 uint32_t                fOrigPermissions;
 111                 char                    fOrigName[16];
 112                 uint64_t                fSize;
 113                 uint64_t                fFileOffset;
 114                 uint64_t                fFileSize;
 115                 uint64_t                fAlignment;
 116                 uint32_t                fPermissions;
 117                 uint32_t                fSectionCount;
 118                 uint64_t                fSectionsSize;
 119                 uint64_t                fSectionsAlignment;
 120                 uint64_t                fNewAddress;
 121                 void*                   fMappedAddress;
 122         };
 123
 124         struct Library
 125         {
 126                 const char*     name;
 127                 uint32_t        currentVersion;
 128                 uint32_t        compatibilityVersion;
 129                 bool            weakImport;
 130         };
 131
 132
 133         virtual ArchPair                                                        getArchPair() const = 0;
 134         virtual const char*                                                     getFilePath() const = 0;
 135         virtual uint64_t                                                        getOffsetInUniversalFile() const        = 0;
 136         virtual uint32_t                                                        getFileType() const     = 0;
 137         virtual uint32_t                                                        getFlags() const = 0;
 138         virtual Library                                                         getID() const = 0;
 139         virtual bool                                                            isDylib() const = 0;
 140         virtual bool                                                            isSplitSeg() const = 0;
 141         virtual bool                                                            hasSplitSegInfo() const = 0;
 142         virtual bool                                                            hasSplitSegInfoV2() const = 0;
 143         virtual bool                                                            inSharableLocation() const = 0;
 144         virtual bool                                                            hasDynamicLookupLinkage() const = 0;
 145         virtual bool                                                            hasMainExecutableLookupLinkage() const = 0;
 146         virtual bool                                                            isTwoLevelNamespace() const     = 0;
 147         virtual bool                                                            hasDyldInfo() const     = 0;
 148         virtual bool                                                            hasMultipleReadWriteSegments() const = 0;
 149         virtual uint32_t                                                        getNameFileOffset() const = 0;
 150         virtual time_t                                                          getLastModTime() const = 0;
 151         virtual ino_t                                                           getInode() const = 0;
 152         virtual std::vector<Segment>&                           getSegments() = 0;
 153         virtual const std::vector<Segment>&                     getSegments() const = 0;
 154         virtual const Segment*                                          getSegment(const char* name) const = 0;
 155         virtual const std::vector<Library>&                     getLibraries() const = 0;
 156         virtual uint64_t                                                        getBaseAddress() const = 0;
 157         virtual uint64_t                                                        getVMSize() const = 0;
 158         virtual uint64_t                                                        getBaseExecutableAddress() const = 0;
 159         virtual uint64_t                                                        getBaseWritableAddress() const = 0;
 160         virtual uint64_t                                                        getBaseReadOnlyAddress() const = 0;
 161         virtual uint64_t                                                        getExecutableVMSize() const = 0;
 162         virtual uint64_t                                                        getWritableVMSize() const = 0;
 163         virtual uint64_t                                                        getReadOnlyVMSize() const = 0;
 164         // need getDyldInfoExports because export info uses ULEB encoding and size could grow
 165         virtual const uint8_t*                                          getDyldInfoExports() const = 0;
 166         virtual void                                                            setDyldInfoExports(const uint8_t* newExports) const = 0;
 167         virtual void                                                            uuid(uuid_t u) const = 0;
 168 };
 169
 170
 171
 172
 173 template <typename A>
 174 class MachOLayout : public MachOLayoutAbstraction
 175 {
 176 public:
 177                                                                                                 MachOLayout(const void* machHeader, uint64_t offset, const char* path,
 178                                                                                                                                         ino_t inode, time_t modTime, uid_t uid);
 179         virtual                                                                         ~MachOLayout() {}
 180
 181         virtual ArchPair                                                        getArchPair() const             { return fArchPair; }
 182         virtual const char*                                                     getFilePath() const             { return fPath; }
 183         virtual uint64_t                                                        getOffsetInUniversalFile() const { return fOffset; }
 184         virtual uint32_t                                                        getFileType() const             { return fFileType; }
 185         virtual uint32_t                                                        getFlags() const                { return fFlags; }
 186         virtual Library                                                         getID() const                   { return fDylibID; }
 187         virtual bool                                                            isDylib() const                 { return fIsDylib; }
 188         virtual bool                                                            isSplitSeg() const;
 189         virtual bool                                                            hasSplitSegInfo() const { return fSplitSegInfo != NULL; }
 190         virtual bool                                                            hasSplitSegInfoV2() const{ return fHasSplitSegInfoV2; }
 191         virtual bool                                                            inSharableLocation() const { return fShareableLocation; }
 192         virtual bool                                                            hasDynamicLookupLinkage() const { return fDynamicLookupLinkage; }
 193         virtual bool                                                            hasMainExecutableLookupLinkage() const { return fMainExecutableLookupLinkage; }
 194         virtual bool                                                            isTwoLevelNamespace() const     { return (fFlags & MH_TWOLEVEL); }
 195         virtual bool                                                            hasDyldInfo() const             { return fHasDyldInfo; }
 196         virtual bool                                                            hasMultipleReadWriteSegments() const { return fHasTooManyWritableSegments; }
 197         virtual uint32_t                                                        getNameFileOffset() const{ return fNameFileOffset; }
 198         virtual time_t                                                          getLastModTime() const  { return fMTime; }
 199         virtual ino_t                                                           getInode() const                { return fInode; }
 200         virtual std::vector<Segment>&                           getSegments()                   { return fSegments; }
 201         virtual const std::vector<Segment>&                     getSegments() const             { return fSegments; }
 202         virtual const Segment*                                          getSegment(const char* name) const;
 203         virtual const std::vector<Library>&                     getLibraries() const    { return fLibraries; }
 204         virtual uint64_t                                                        getBaseAddress() const  { return fLowSegment->address(); }
 205         virtual uint64_t                                                        getVMSize() const               { return fVMSize; }
 206         virtual uint64_t                                                        getBaseExecutableAddress() const { return fLowExecutableSegment->address(); }
 207         virtual uint64_t                                                        getBaseWritableAddress() const  { return fLowWritableSegment->address(); }
 208         virtual uint64_t                                                        getBaseReadOnlyAddress() const  { return fLowReadOnlySegment->address(); }
 209         virtual uint64_t                                                        getExecutableVMSize() const             { return fVMExecutableSize; }
 210         virtual uint64_t                                                        getWritableVMSize() const               { return fVMWritablSize; }
 211         virtual uint64_t                                                        getReadOnlyVMSize() const               { return fVMReadOnlySize; }
 212         virtual const uint8_t*                                          getDyldInfoExports() const              { return fDyldInfoExports; }
 213         virtual void                                                            setDyldInfoExports(const uint8_t* newExports) const { fDyldInfoExports = newExports; }
 214         virtual void                                                            uuid(uuid_t u) const { memcpy(u, fUUID, 16); }
 215
 216 private:
 217         typedef typename A::P                                   P;
 218         typedef typename A::P::E                                E;
 219         typedef typename A::P::uint_t                   pint_t;
 220
 221         uint64_t                                                                        segmentSize(const macho_segment_command<typename A::P>* segCmd) const;
 222         uint64_t                                                                        segmentFileSize(const macho_segment_command<typename A::P>* segCmd) const;
 223         uint64_t                                                                        segmentAlignment(const macho_segment_command<typename A::P>* segCmd) const;
 224         uint64_t                                                                        sectionsSize(const macho_segment_command<typename A::P>* segCmd) const;
 225         uint64_t                                                                        sectionsAlignment(const macho_segment_command<typename A::P>* segCmd) const;
 226
 227         bool                                                                            validReadWriteSeg(const Segment& seg) const;
 228
 229         static cpu_type_t                                                       arch();
 230
 231         const char*                                                                     fPath;
 232         uint64_t                                                                        fOffset;
 233         uint32_t                                                                        fFileType;
 234         ArchPair                                                                        fArchPair;
 235         uint32_t                                                                        fFlags;
 236         std::vector<Segment>                                            fSegments;
 237         std::vector<Library>                                            fLibraries;
 238         const Segment*                                                          fLowSegment;
 239         const Segment*                                                          fLowExecutableSegment;
 240         const Segment*                                                          fLowWritableSegment;
 241         const Segment*                                                          fLowReadOnlySegment;
 242         Library                                                                         fDylibID;
 243         uint32_t                                                                        fNameFileOffset;
 244         time_t                                                                          fMTime;
 245         ino_t                                                                           fInode;
 246         uint64_t                                                                        fVMSize;
 247         uint64_t                                                                        fVMExecutableSize;
 248         uint64_t                                                                        fVMWritablSize;
 249         uint64_t                                                                        fVMReadOnlySize;
 250         const macho_linkedit_data_command<P>*           fSplitSegInfo;
 251         bool                                                                            fHasSplitSegInfoV2;
 252         bool                                                                            fShareableLocation;
 253         bool                                                                            fDynamicLookupLinkage;
 254         bool                                                                            fMainExecutableLookupLinkage;
 255         bool                                                                            fIsDylib;
 256         bool                                                                            fHasDyldInfo;
 257         bool                                                                            fHasTooManyWritableSegments;
 258         mutable const uint8_t*                                          fDyldInfoExports;
 259         uuid_t                                                                          fUUID;
 260 };
 261
 262
 263
 264 class UniversalMachOLayout
 265 {
 266 public:
 267                                                                                                 UniversalMachOLayout(const char* path, const std::set<ArchPair>* onlyArchs=NULL);
 268                                                                                                 ~UniversalMachOLayout() {}
 269
 270         static const UniversalMachOLayout&                      find(const char* path, const std::set<ArchPair>* onlyArchs=NULL);
 271         const MachOLayoutAbstraction*                           getSlice(ArchPair ap) const;
 272         const std::vector<MachOLayoutAbstraction*>&     allLayouts() const { return fLayouts; }
 273
 274 private:
 275         class CStringHash {
 276         public:
 277                 size_t operator()(const char* __s) const {
 278                         size_t __h = 0;
 279                         for ( ; *__s; ++__s)
 280                                 __h = 5 * __h + *__s;
 281                         return __h;
 282                 };
 283         };
 284         struct CStringEquals {
 285                 bool operator()(const char* left, const char* right) const { return (strcmp(left, right) == 0); }
 286         };
 287         typedef std::unordered_map<const char*, const UniversalMachOLayout*, CStringHash, CStringEquals> PathToNode;
 288
 289         static bool                                     requestedSlice(const std::set<ArchPair>* onlyArchs, cpu_type_t cpuType, cpu_subtype_t cpuSubType);
 290
 291         static PathToNode                                                       fgLayoutCache;
 292         const char*                                                                     fPath;
 293         std::vector<MachOLayoutAbstraction*>            fLayouts;
 294 };
 295
 296 UniversalMachOLayout::PathToNode UniversalMachOLayout::fgLayoutCache;
 297
 298
 299
 300
 301 const MachOLayoutAbstraction* UniversalMachOLayout::getSlice(ArchPair ap) const
 302 {
 303         // use matching cputype and cpusubtype
 304         for(std::vector<MachOLayoutAbstraction*>::const_iterator it=fLayouts.begin(); it != fLayouts.end(); ++it) {
 305                 const MachOLayoutAbstraction* layout = *it;
 306                 if ( layout->getArchPair().arch == ap.arch ) {
 307             switch ( ap.arch ) {
 308                 case CPU_TYPE_ARM:
 309                                 case CPU_TYPE_X86_64:
 310                    if ( (layout->getArchPair().subtype & ~CPU_SUBTYPE_MASK) == (ap.subtype & ~CPU_SUBTYPE_MASK) )
 311                         return layout;
 312                     break;
 313                  default:
 314                     return layout;
 315             }
 316         }
 317         }
 318         // if requesting x86_64h and it did not exist, try x86_64 as a fallback
 319         if ((ap.arch == CPU_TYPE_X86_64) && (ap.subtype == CPU_SUBTYPE_X86_64_H)) {
 320                 ap.subtype = CPU_SUBTYPE_X86_64_ALL;
 321                 return this->getSlice(ap);
 322         }
 323         return NULL;
 324 }
 325
 326
 327 const UniversalMachOLayout& UniversalMachOLayout::find(const char* path, const std::set<ArchPair>* onlyArchs)
 328 {
 329         // look in cache
 330         PathToNode::iterator pos = fgLayoutCache.find(path);
 331         if ( pos != fgLayoutCache.end() )
 332                 return *pos->second;
 333
 334         // create UniversalMachOLayout
 335         const UniversalMachOLayout* result = new UniversalMachOLayout(path, onlyArchs);
 336
 337         // add it to cache
 338         fgLayoutCache[result->fPath] = result;
 339
 340         return *result;
 341 }
 342
 343
 344 bool UniversalMachOLayout::requestedSlice(const std::set<ArchPair>* onlyArchs, cpu_type_t cpuType, cpu_subtype_t cpuSubType)
 345 {
 346         if ( onlyArchs == NULL )
 347                 return true;
 348         // must match cputype and cpusubtype
 349         for (std::set<ArchPair>::const_iterator it = onlyArchs->begin(); it != onlyArchs->end(); ++it) {
 350                 ArchPair anArch = *it;
 351                 if ( cpuType == anArch.arch ) {
 352             switch ( cpuType ) {
 353                 case CPU_TYPE_ARM:
 354                     if ( cpuSubType == anArch.subtype )
 355                         return true;
 356                     break;
 357                 default:
 358                     return true;
 359             }
 360         }
 361         }
 362         return false;
 363 }
 364
 365
 366 UniversalMachOLayout::UniversalMachOLayout(const char* path, const std::set<ArchPair>* onlyArchs)
 367  : fPath(strdup(path))
 368 {
 369         // map in whole file
 370         int fd = ::open(path, O_RDONLY, 0);
 371         if ( fd == -1 ) {
 372                 int err = errno;
 373                 if  ( err == ENOENT )
 374                         throwf("file not found");
 375                 else
 376                         throwf("can't open file, errno=%d", err);
 377         }
 378         struct stat stat_buf;
 379         if ( fstat(fd, &stat_buf) == -1)
 380                 throwf("can't stat open file %s, errno=%d", path, errno);
 381         if ( stat_buf.st_size < 20 )
 382                 throwf("file too small %s", path);
 383         uint8_t* p = (uint8_t*)::mmap(NULL, stat_buf.st_size, PROT_READ, MAP_FILE | MAP_PRIVATE, fd, 0);
 384         if ( p == (uint8_t*)(-1) )
 385                 throwf("can't map file %s, errno=%d", path, errno);
 386         ::close(fd);
 387
 388         try {
 389                 // if fat file, process each architecture
 390                 const fat_header* fh = (fat_header*)p;
 391                 const mach_header* mh = (mach_header*)p;
 392                 if ( fh->magic == OSSwapBigToHostInt32(FAT_MAGIC) ) {
 393                         // Fat header is always big-endian
 394                         const struct fat_arch* slices = (struct fat_arch*)(p + sizeof(struct fat_header));
 395                         const uint32_t sliceCount = OSSwapBigToHostInt32(fh->nfat_arch);
 396                         for (uint32_t i=0; i < sliceCount; ++i) {
 397                                 if ( requestedSlice(onlyArchs, OSSwapBigToHostInt32(slices[i].cputype), OSSwapBigToHostInt32(slices[i].cpusubtype)) ) {
 398                                         uint32_t fileOffset = OSSwapBigToHostInt32(slices[i].offset);
 399                                         if ( fileOffset > stat_buf.st_size ) {
 400                                                 throwf("malformed universal file, slice %u for architecture 0x%08X is beyond end of file: %s",
 401                                                                 i, OSSwapBigToHostInt32(slices[i].cputype), path);
 402                                         }
 403                                         if ( (fileOffset+OSSwapBigToHostInt32(slices[i].size)) > stat_buf.st_size ) {
 404                                                 throwf("malformed universal file, slice %u for architecture 0x%08X is beyond end of file: %s",
 405                                                                 i, OSSwapBigToHostInt32(slices[i].cputype), path);
 406                                         }
 407                                         try {
 408                                                 switch ( OSSwapBigToHostInt32(slices[i].cputype) ) {
 409                                                         case CPU_TYPE_I386:
 410                                                                 fLayouts.push_back(new MachOLayout<x86>(&p[fileOffset], fileOffset, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 411                                                                 break;
 412                                                         case CPU_TYPE_X86_64:
 413                                                                 fLayouts.push_back(new MachOLayout<x86_64>(&p[fileOffset], fileOffset, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 414                                                                 break;
 415                                                         case CPU_TYPE_ARM:
 416                                                                 fLayouts.push_back(new MachOLayout<arm>(&p[fileOffset], fileOffset, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 417                                                                 break;
 418                                                         case CPU_TYPE_ARM64:
 419                                                                 fLayouts.push_back(new MachOLayout<arm64>(&p[fileOffset], fileOffset, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 420                                                                 break;
 421                                                         default:
 422                                                                 throw "unknown slice in fat file";
 423                                                 }
 424                                         }
 425                                         catch (const char* msg) {
 426                                                 fprintf(stderr, "warning: %s for %s\n", msg, path);
 427                                         }
 428                                 }
 429                         }
 430                 }
 431                 else {
 432                         try {
 433                                 if ( (OSSwapLittleToHostInt32(mh->magic) == MH_MAGIC) && (OSSwapLittleToHostInt32(mh->cputype) == CPU_TYPE_I386)) {
 434                                         if ( requestedSlice(onlyArchs, OSSwapLittleToHostInt32(mh->cputype), OSSwapLittleToHostInt32(mh->cpusubtype)) )
 435                                                 fLayouts.push_back(new MachOLayout<x86>(mh, 0, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 436                                 }
 437                                 else if ( (OSSwapLittleToHostInt32(mh->magic) == MH_MAGIC_64) && (OSSwapLittleToHostInt32(mh->cputype) == CPU_TYPE_X86_64)) {
 438                                         if ( requestedSlice(onlyArchs, OSSwapLittleToHostInt32(mh->cputype), OSSwapLittleToHostInt32(mh->cpusubtype)) )
 439                                                 fLayouts.push_back(new MachOLayout<x86_64>(mh, 0, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 440                                 }
 441                                 else if ( (OSSwapLittleToHostInt32(mh->magic) == MH_MAGIC) && (OSSwapLittleToHostInt32(mh->cputype) == CPU_TYPE_ARM)) {
 442                                         if ( requestedSlice(onlyArchs, OSSwapLittleToHostInt32(mh->cputype), OSSwapLittleToHostInt32(mh->cpusubtype)) )
 443                                                 fLayouts.push_back(new MachOLayout<arm>(mh, 0, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 444                                 }
 445                                 else if ( (OSSwapLittleToHostInt32(mh->magic) == MH_MAGIC_64) && (OSSwapLittleToHostInt32(mh->cputype) == CPU_TYPE_ARM64)) {
 446                                         if ( requestedSlice(onlyArchs, OSSwapLittleToHostInt32(mh->cputype), OSSwapLittleToHostInt32(mh->cpusubtype)) )
 447                                                 fLayouts.push_back(new MachOLayout<arm64>(mh, 0, fPath, stat_buf.st_ino, stat_buf.st_mtime, stat_buf.st_uid));
 448                                 }
 449                                 else {
 450                                         throw "unknown file format";
 451                                 }
 452                         }
 453                         catch (const char* msg) {
 454                                 fprintf(stderr, "warning: %s for %s\n", msg, path);
 455                         }
 456                 }
 457         }
 458         catch (...) {
 459                 ::munmap(p, stat_buf.st_size);
 460                 throw;
 461         }
 462 }
 463
 464
 465 template <typename A>
 466 uint64_t MachOLayout<A>::segmentSize(const macho_segment_command<typename A::P>* segCmd) const
 467 {
 468         // <rdar://problem/13089366> segments may have 16KB alignment padding at end, if so we can remove that in cache
 469         if ( segCmd->nsects() > 0 ) {
 470                 const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)segCmd + sizeof(macho_segment_command<P>));
 471                 const macho_section<P>* const lastSection = &sectionsStart[segCmd->nsects()-1];
 472                 uint64_t endSectAddr = lastSection->addr() + lastSection->size();
 473                 uint64_t endSectAddrPage = (endSectAddr + 4095) & (-4096);
 474                 if ( endSectAddrPage < (segCmd->vmaddr() + segCmd->vmsize()) ) {
 475                         uint64_t size =  endSectAddrPage - segCmd->vmaddr();
 476                         //if ( size != segCmd->vmsize() )
 477                         //      fprintf(stderr, "trim %s size=0x%08llX instead of 0x%08llX for %s\n",
 478                         //              segCmd->segname(), size, segCmd->vmsize(), getFilePath());
 479                         return size;
 480                 }
 481         }
 482         return segCmd->vmsize();
 483 }
 484
 485 template <typename A>
 486 uint64_t MachOLayout<A>::segmentAlignment(const macho_segment_command<typename A::P>* segCmd) const
 487 {
 488         int p2align = 12;
 489         if ( segCmd->nsects() > 0 ) {
 490                 const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)segCmd + sizeof(macho_segment_command<P>));
 491                 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()-1];
 492                 for (const macho_section<P>*  sect=sectionsStart; sect < sectionsEnd; ++sect) {
 493                         if ( sect->align() > p2align )
 494                                 p2align = sect->align();
 495                 }
 496         }
 497         return (1 << p2align);
 498 }
 499
 500 template <typename A>
 501 uint64_t MachOLayout<A>::segmentFileSize(const macho_segment_command<typename A::P>* segCmd) const
 502 {
 503         // <rdar://problem/13089366> segments may have 16KB alignment padding at end, if so we can remove that in cache
 504         if ( segCmd->nsects() > 0 ) {
 505                 uint64_t endOffset = segCmd->fileoff();
 506                 const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)segCmd + sizeof(macho_segment_command<P>));
 507                 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()];
 508                 for (const macho_section<P>* sect=sectionsStart; sect < sectionsEnd; ++sect) {
 509                         if ( sect->offset() != 0 )
 510                                 endOffset = sect->offset() + sect->size();
 511                 }
 512                 uint64_t size = (endOffset - segCmd->fileoff() + 4095) & (-4096);
 513                 //if ( size != segCmd->filesize() )
 514                 //      fprintf(stderr, "trim %s filesize=0x%08llX instead of 0x%08llX for %s\n",
 515                 //              segCmd->segname(), size, segCmd->filesize(), getFilePath());
 516                 return size;
 517         }
 518         return segCmd->filesize();
 519 }
 520
 521 template <typename A>
 522 uint64_t MachOLayout<A>::sectionsSize(const macho_segment_command<typename A::P>* segCmd) const
 523 {
 524         if ( segCmd->nsects() > 0 ) {
 525                 const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)segCmd + sizeof(macho_segment_command<P>));
 526                 const macho_section<P>* const lastSection = &sectionsStart[segCmd->nsects()-1];
 527                 uint64_t endSectAddr = lastSection->addr() + lastSection->size();
 528                 if ( endSectAddr < (segCmd->vmaddr() + segCmd->vmsize()) ) {
 529                         uint64_t size =  endSectAddr - segCmd->vmaddr();
 530                         return size;
 531                 }
 532         }
 533         return segCmd->vmsize();
 534 }
 535
 536 template <typename A>
 537 uint64_t MachOLayout<A>::sectionsAlignment(const macho_segment_command<typename A::P>* segCmd) const
 538 {
 539         int p2align = 4;
 540         if ( hasSplitSegInfoV2() && (segCmd->nsects() > 0) ) {
 541                 const macho_section<P>* const sectionsStart = (macho_section<P>*)((uint8_t*)segCmd + sizeof(macho_segment_command<P>));
 542                 const macho_section<P>* const sectionsEnd = &sectionsStart[segCmd->nsects()-1];
 543                 for (const macho_section<P>*  sect=sectionsStart; sect < sectionsEnd; ++sect) {
 544                         if ( sect->align() > p2align )
 545                                 p2align = sect->align();
 546                 }
 547         }
 548         return (1 << p2align);
 549 }
 550
 551
 552
 553 template <typename A>
 554 MachOLayout<A>::MachOLayout(const void* machHeader, uint64_t offset, const char* path, ino_t inode, time_t modTime, uid_t uid)
 555  : fPath(path), fOffset(offset), fArchPair(0,0), fMTime(modTime), fInode(inode), fSplitSegInfo(NULL), fHasSplitSegInfoV2(false),
 556    fShareableLocation(false), fDynamicLookupLinkage(false), fMainExecutableLookupLinkage(false), fIsDylib(false),
 557         fHasDyldInfo(false), fHasTooManyWritableSegments(false), fDyldInfoExports(NULL)
 558 {
 559         fDylibID.name = NULL;
 560         fDylibID.currentVersion = 0;
 561         fDylibID.compatibilityVersion = 0;
 562         bzero(fUUID, sizeof(fUUID));
 563
 564         const macho_header<P>* mh = (const macho_header<P>*)machHeader;
 565         if ( mh->cputype() != arch() )
 566                 throw "Layout object is wrong architecture";
 567         switch ( mh->filetype() ) {
 568                 case MH_DYLIB:
 569                         fIsDylib = true;
 570                         break;
 571                 case MH_BUNDLE:
 572                 case MH_EXECUTE:
 573                 case MH_DYLIB_STUB:
 574                 case MH_DYLINKER:
 575                         break;
 576                 default:
 577                         throw "file is not a mach-o final linked image";
 578         }
 579         fFlags = mh->flags();
 580         fFileType = mh->filetype();
 581         fArchPair.arch = mh->cputype();
 582         fArchPair.subtype = mh->cpusubtype();
 583
 584         const macho_dyld_info_command<P>* dyldInfo = NULL;
 585         const macho_symtab_command<P>* symbolTableCmd = NULL;
 586         const macho_dysymtab_command<P>* dynamicSymbolTableCmd = NULL;
 587         const macho_load_command<P>* const cmds = (macho_load_command<P>*)((uint8_t*)mh + sizeof(macho_header<P>));
 588         const uint32_t cmd_count = mh->ncmds();
 589         const macho_load_command<P>* cmd = cmds;
 590         for (uint32_t i = 0; i < cmd_count; ++i) {
 591                 switch ( cmd->cmd() ) {
 592                         case LC_ID_DYLIB:
 593                                 {
 594                                         macho_dylib_command<P>* dylib  = (macho_dylib_command<P>*)cmd;
 595                                         fDylibID.name = strdup(dylib->name());
 596                                         fDylibID.currentVersion = dylib->current_version();
 597                                         fDylibID.compatibilityVersion = dylib->compatibility_version();
 598                                         fNameFileOffset = dylib->name() - (char*)machHeader;
 599                                         fShareableLocation = ( (strncmp(fDylibID.name, "/usr/lib/", 9) == 0) || (strncmp(fDylibID.name, "/System/Library/", 16) == 0) );
 600                                 }
 601                                 break;
 602                         case LC_LOAD_DYLIB:
 603                         case LC_LOAD_WEAK_DYLIB:
 604                         case LC_REEXPORT_DYLIB:
 605                         case LC_LOAD_UPWARD_DYLIB:
 606                                 {
 607                                         macho_dylib_command<P>* dylib = (macho_dylib_command<P>*)cmd;
 608                                         Library lib;
 609                                         lib.name = strdup(dylib->name());
 610                                         lib.currentVersion = dylib->current_version();
 611                                         lib.compatibilityVersion = dylib->compatibility_version();
 612                                         lib.weakImport = ( cmd->cmd() == LC_LOAD_WEAK_DYLIB );
 613                                         fLibraries.push_back(lib);
 614                                 }
 615                                 break;
 616                         case LC_SEGMENT_SPLIT_INFO:
 617                                 fSplitSegInfo = (macho_linkedit_data_command<P>*)cmd;
 618                                 break;
 619                         case macho_segment_command<P>::CMD:
 620                                 {
 621                                         const macho_segment_command<P>* segCmd = (macho_segment_command<P>*)cmd;
 622                                         fSegments.push_back(Segment(segCmd->vmaddr(), segmentSize(segCmd), segCmd->fileoff(),
 623                                                                 segmentFileSize(segCmd), sectionsSize(segCmd), sectionsAlignment(segCmd),
 624                                                                 segmentAlignment(segCmd), segCmd->initprot(),
 625                                                                 segCmd->nsects(), segCmd->segname()));
 626                                 }
 627                                 break;
 628                         case LC_SYMTAB:
 629                                 symbolTableCmd = (macho_symtab_command<P>*)cmd;
 630                                 break;
 631                         case LC_DYSYMTAB:
 632                                 dynamicSymbolTableCmd = (macho_dysymtab_command<P>*)cmd;
 633                                 break;
 634                         case LC_DYLD_INFO:
 635                         case LC_DYLD_INFO_ONLY:
 636                                 fHasDyldInfo = true;
 637                                 dyldInfo = (struct macho_dyld_info_command<P>*)cmd;
 638                                 break;
 639                         case LC_UUID:
 640                                 {
 641                                         const macho_uuid_command<P>* uc = (macho_uuid_command<P>*)cmd;
 642                                         memcpy(&fUUID, uc->uuid(), 16);
 643                                 }
 644                                 break;
 645                 }
 646                 cmd = (const macho_load_command<P>*)(((uint8_t*)cmd)+cmd->cmdsize());
 647         }
 648
 649         fLowSegment = NULL;
 650         fLowExecutableSegment = NULL;
 651         fLowWritableSegment = NULL;
 652         fLowReadOnlySegment = NULL;
 653         fVMExecutableSize = 0;
 654         fVMWritablSize = 0;
 655         fVMReadOnlySize = 0;
 656         fVMSize = 0;
 657         const Segment* highSegment = NULL;
 658         for(std::vector<Segment>::const_iterator it = fSegments.begin(); it != fSegments.end(); ++it) {
 659                 const Segment& seg = *it;
 660                 if ( (fLowSegment == NULL) || (seg.address() < fLowSegment->address()) )
 661                         fLowSegment = &seg;
 662                 if ( (highSegment == NULL) || (seg.address() > highSegment->address()) )
 663                         highSegment = &seg;
 664                 if ( seg.executable() ) {
 665                         if ( (fLowExecutableSegment == NULL) || (seg.address() < fLowExecutableSegment->address()) )
 666                                 fLowExecutableSegment = &seg;
 667                         fVMExecutableSize += seg.size();
 668                 }
 669                 else if ( seg.writable()) {
 670                         if ( (fLowWritableSegment == NULL) || (seg.address() < fLowWritableSegment->address()) )
 671                                 fLowWritableSegment = &seg;
 672                         fVMWritablSize += seg.size();
 673                         if ( !validReadWriteSeg(seg) ) {
 674                                 fHasTooManyWritableSegments = true;
 675                         }
 676                 }
 677                 else {
 678                         if ( (fLowReadOnlySegment == NULL) || (seg.address() < fLowReadOnlySegment->address()) )
 679                                 fLowReadOnlySegment = &seg;
 680                         fVMReadOnlySize += seg.size();
 681                 }
 682         }
 683         if ( (highSegment != NULL) && (fLowSegment != NULL) )
 684                 fVMSize = (highSegment->address() + highSegment->size() - fLowSegment->address() + 4095) & (-4096);
 685
 686         // scan undefines looking, for magic ordinals
 687         if ( (symbolTableCmd != NULL) && (dynamicSymbolTableCmd != NULL) ) {
 688                 const macho_nlist<P>* symbolTable = (macho_nlist<P>*)((uint8_t*)machHeader + symbolTableCmd->symoff());
 689                 const uint32_t startUndefs = dynamicSymbolTableCmd->iundefsym();
 690                 const uint32_t endUndefs = startUndefs + dynamicSymbolTableCmd->nundefsym();
 691                 for (uint32_t i=startUndefs; i < endUndefs; ++i) {
 692                         uint8_t ordinal = GET_LIBRARY_ORDINAL(symbolTable[i].n_desc());
 693                         if ( ordinal == DYNAMIC_LOOKUP_ORDINAL )
 694                                 fDynamicLookupLinkage = true;
 695                         else if ( ordinal == EXECUTABLE_ORDINAL )
 696                                 fMainExecutableLookupLinkage = true;
 697                 }
 698         }
 699
 700         if ( dyldInfo != NULL ) {
 701                 if ( dyldInfo->export_off() != 0 ) {
 702                         fDyldInfoExports = (uint8_t*)machHeader + dyldInfo->export_off();
 703                 }
 704         }
 705
 706         if ( fSplitSegInfo != NULL ) {
 707                 const uint8_t* infoStart = (uint8_t*)machHeader + fSplitSegInfo->dataoff();
 708                 fHasSplitSegInfoV2 = ( *infoStart == DYLD_CACHE_ADJ_V2_FORMAT );
 709                 if ( !fHasSplitSegInfoV2 ) {
 710                         // split seg version not known when segments created
 711                         // v1 does not support packing, so simulate that by forcing alignment
 712                         for (Segment& seg : fSegments) {
 713                                 seg.setSectionsAlignment(4096);
 714                         }
 715                 }
 716         }
 717
 718 }
 719
 720 template <> cpu_type_t MachOLayout<x86>::arch()     { return CPU_TYPE_I386; }
 721 template <> cpu_type_t MachOLayout<x86_64>::arch()  { return CPU_TYPE_X86_64; }
 722 template <> cpu_type_t MachOLayout<arm>::arch()         { return CPU_TYPE_ARM; }
 723 template <> cpu_type_t MachOLayout<arm64>::arch()       { return CPU_TYPE_ARM64; }
 724
 725 template <>
 726 bool MachOLayout<x86>::validReadWriteSeg(const Segment& seg) const
 727 {
 728         return (strcmp(seg.name(), "__DATA") == 0) || (strcmp(seg.name(), "__OBJC") == 0);
 729 }
 730
 731 template <typename A>
 732 bool MachOLayout<A>::validReadWriteSeg(const Segment& seg) const
 733 {
 734         return (strcmp(seg.name(), "__DATA") == 0);
 735 }
 736
 737
 738 template <>
 739 bool MachOLayout<x86>::isSplitSeg() const
 740 {
 741         return ( (this->getFlags() & MH_SPLIT_SEGS) != 0 );
 742 }
 743
 744 template <>
 745 bool MachOLayout<arm>::isSplitSeg() const
 746 {
 747         return ( (this->getFlags() & MH_SPLIT_SEGS) != 0 );
 748 }
 749
 750 template <typename A>
 751 bool MachOLayout<A>::isSplitSeg() const
 752 {
 753         return false;
 754 }
 755
 756 template <typename A>
 757 const MachOLayoutAbstraction::Segment* MachOLayout<A>::getSegment(const char* name) const
 758 {
 759         for(std::vector<Segment>::const_iterator it = fSegments.begin(); it != fSegments.end(); ++it) {
 760                 const Segment& seg = *it;
 761                 if ( strcmp(seg.name(), name) == 0 )
 762                         return &seg;
 763         }
 764         return NULL;
 765 }
 766
 767
 768
 769 #endif // __MACHO_LAYOUT__
 770
 771
 772