Home | History | Annotate | Download | only in Object
      1 //===- MachO.h - MachO object file implementation ---------------*- C++ -*-===//
      2 //
      3 //                     The LLVM Compiler Infrastructure
      4 //
      5 // This file is distributed under the University of Illinois Open Source
      6 // License. See LICENSE.TXT for details.
      7 //
      8 //===----------------------------------------------------------------------===//
      9 //
     10 // This file declares the MachOObjectFile class, which implement the ObjectFile
     11 // interface for MachO files.
     12 //
     13 //===----------------------------------------------------------------------===//
     14 
     15 #ifndef LLVM_OBJECT_MACHO_H
     16 #define LLVM_OBJECT_MACHO_H
     17 
     18 #include "llvm/ADT/ArrayRef.h"
     19 #include "llvm/ADT/SmallVector.h"
     20 #include "llvm/ADT/Triple.h"
     21 #include "llvm/Object/ObjectFile.h"
     22 #include "llvm/Support/MachO.h"
     23 
     24 namespace llvm {
     25 namespace object {
     26 
     27 /// DiceRef - This is a value type class that represents a single
     28 /// data in code entry in the table in a Mach-O object file.
     29 class DiceRef {
     30   DataRefImpl DicePimpl;
     31   const ObjectFile *OwningObject;
     32 
     33 public:
     34   DiceRef() : OwningObject(nullptr) { }
     35 
     36   DiceRef(DataRefImpl DiceP, const ObjectFile *Owner);
     37 
     38   bool operator==(const DiceRef &Other) const;
     39   bool operator<(const DiceRef &Other) const;
     40 
     41   void moveNext();
     42 
     43   std::error_code getOffset(uint32_t &Result) const;
     44   std::error_code getLength(uint16_t &Result) const;
     45   std::error_code getKind(uint16_t &Result) const;
     46 
     47   DataRefImpl getRawDataRefImpl() const;
     48   const ObjectFile *getObjectFile() const;
     49 };
     50 typedef content_iterator<DiceRef> dice_iterator;
     51 
     52 /// ExportEntry encapsulates the current-state-of-the-walk used when doing a
     53 /// non-recursive walk of the trie data structure.  This allows you to iterate
     54 /// across all exported symbols using:
     55 ///      for (const llvm::object::ExportEntry &AnExport : Obj->exports()) {
     56 ///      }
     57 class ExportEntry {
     58 public:
     59   ExportEntry(ArrayRef<uint8_t> Trie);
     60 
     61   StringRef name() const;
     62   uint64_t flags() const;
     63   uint64_t address() const;
     64   uint64_t other() const;
     65   StringRef otherName() const;
     66   uint32_t nodeOffset() const;
     67 
     68   bool operator==(const ExportEntry &) const;
     69 
     70   void moveNext();
     71 
     72 private:
     73   friend class MachOObjectFile;
     74   void moveToFirst();
     75   void moveToEnd();
     76   uint64_t readULEB128(const uint8_t *&p);
     77   void pushDownUntilBottom();
     78   void pushNode(uint64_t Offset);
     79 
     80   // Represents a node in the mach-o exports trie.
     81   struct NodeState {
     82     NodeState(const uint8_t *Ptr);
     83     const uint8_t *Start;
     84     const uint8_t *Current;
     85     uint64_t Flags;
     86     uint64_t Address;
     87     uint64_t Other;
     88     const char *ImportName;
     89     unsigned ChildCount;
     90     unsigned NextChildIndex;
     91     unsigned ParentStringLength;
     92     bool IsExportNode;
     93   };
     94 
     95   ArrayRef<uint8_t> Trie;
     96   SmallString<256> CumulativeString;
     97   SmallVector<NodeState, 16> Stack;
     98   bool Malformed;
     99   bool Done;
    100 };
    101 typedef content_iterator<ExportEntry> export_iterator;
    102 
    103 /// MachORebaseEntry encapsulates the current state in the decompression of
    104 /// rebasing opcodes. This allows you to iterate through the compressed table of
    105 /// rebasing using:
    106 ///    for (const llvm::object::MachORebaseEntry &Entry : Obj->rebaseTable()) {
    107 ///    }
    108 class MachORebaseEntry {
    109 public:
    110   MachORebaseEntry(ArrayRef<uint8_t> opcodes, bool is64Bit);
    111 
    112   uint32_t segmentIndex() const;
    113   uint64_t segmentOffset() const;
    114   StringRef typeName() const;
    115 
    116   bool operator==(const MachORebaseEntry &) const;
    117 
    118   void moveNext();
    119 
    120 private:
    121   friend class MachOObjectFile;
    122   void moveToFirst();
    123   void moveToEnd();
    124   uint64_t readULEB128();
    125 
    126   ArrayRef<uint8_t> Opcodes;
    127   const uint8_t *Ptr;
    128   uint64_t SegmentOffset;
    129   uint32_t SegmentIndex;
    130   uint64_t RemainingLoopCount;
    131   uint64_t AdvanceAmount;
    132   uint8_t  RebaseType;
    133   uint8_t  PointerSize;
    134   bool     Malformed;
    135   bool     Done;
    136 };
    137 typedef content_iterator<MachORebaseEntry> rebase_iterator;
    138 
    139 /// MachOBindEntry encapsulates the current state in the decompression of
    140 /// binding opcodes. This allows you to iterate through the compressed table of
    141 /// bindings using:
    142 ///    for (const llvm::object::MachOBindEntry &Entry : Obj->bindTable()) {
    143 ///    }
    144 class MachOBindEntry {
    145 public:
    146   enum class Kind { Regular, Lazy, Weak };
    147 
    148   MachOBindEntry(ArrayRef<uint8_t> Opcodes, bool is64Bit, MachOBindEntry::Kind);
    149 
    150   uint32_t segmentIndex() const;
    151   uint64_t segmentOffset() const;
    152   StringRef typeName() const;
    153   StringRef symbolName() const;
    154   uint32_t flags() const;
    155   int64_t addend() const;
    156   int ordinal() const;
    157 
    158   bool operator==(const MachOBindEntry &) const;
    159 
    160   void moveNext();
    161 
    162 private:
    163   friend class MachOObjectFile;
    164   void moveToFirst();
    165   void moveToEnd();
    166   uint64_t readULEB128();
    167   int64_t readSLEB128();
    168 
    169   ArrayRef<uint8_t> Opcodes;
    170   const uint8_t *Ptr;
    171   uint64_t SegmentOffset;
    172   uint32_t SegmentIndex;
    173   StringRef SymbolName;
    174   int      Ordinal;
    175   uint32_t Flags;
    176   int64_t  Addend;
    177   uint64_t RemainingLoopCount;
    178   uint64_t AdvanceAmount;
    179   uint8_t  BindType;
    180   uint8_t  PointerSize;
    181   Kind     TableKind;
    182   bool     Malformed;
    183   bool     Done;
    184 };
    185 typedef content_iterator<MachOBindEntry> bind_iterator;
    186 
    187 class MachOObjectFile : public ObjectFile {
    188 public:
    189   struct LoadCommandInfo {
    190     const char *Ptr;      // Where in memory the load command is.
    191     MachO::load_command C; // The command itself.
    192   };
    193   typedef SmallVector<LoadCommandInfo, 4> LoadCommandList;
    194   typedef LoadCommandList::const_iterator load_command_iterator;
    195 
    196   static Expected<std::unique_ptr<MachOObjectFile>>
    197   create(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits);
    198 
    199   void moveSymbolNext(DataRefImpl &Symb) const override;
    200 
    201   uint64_t getNValue(DataRefImpl Sym) const;
    202   Expected<StringRef> getSymbolName(DataRefImpl Symb) const override;
    203 
    204   // MachO specific.
    205   std::error_code getIndirectName(DataRefImpl Symb, StringRef &Res) const;
    206   unsigned getSectionType(SectionRef Sec) const;
    207 
    208   Expected<uint64_t> getSymbolAddress(DataRefImpl Symb) const override;
    209   uint32_t getSymbolAlignment(DataRefImpl Symb) const override;
    210   uint64_t getCommonSymbolSizeImpl(DataRefImpl Symb) const override;
    211   Expected<SymbolRef::Type> getSymbolType(DataRefImpl Symb) const override;
    212   uint32_t getSymbolFlags(DataRefImpl Symb) const override;
    213   Expected<section_iterator> getSymbolSection(DataRefImpl Symb) const override;
    214   unsigned getSymbolSectionID(SymbolRef Symb) const;
    215   unsigned getSectionID(SectionRef Sec) const;
    216 
    217   void moveSectionNext(DataRefImpl &Sec) const override;
    218   std::error_code getSectionName(DataRefImpl Sec,
    219                                  StringRef &Res) const override;
    220   uint64_t getSectionAddress(DataRefImpl Sec) const override;
    221   uint64_t getSectionSize(DataRefImpl Sec) const override;
    222   std::error_code getSectionContents(DataRefImpl Sec,
    223                                      StringRef &Res) const override;
    224   uint64_t getSectionAlignment(DataRefImpl Sec) const override;
    225   bool isSectionCompressed(DataRefImpl Sec) const override;
    226   bool isSectionText(DataRefImpl Sec) const override;
    227   bool isSectionData(DataRefImpl Sec) const override;
    228   bool isSectionBSS(DataRefImpl Sec) const override;
    229   bool isSectionVirtual(DataRefImpl Sec) const override;
    230   bool isSectionBitcode(DataRefImpl Sec) const override;
    231   relocation_iterator section_rel_begin(DataRefImpl Sec) const override;
    232   relocation_iterator section_rel_end(DataRefImpl Sec) const override;
    233 
    234   void moveRelocationNext(DataRefImpl &Rel) const override;
    235   uint64_t getRelocationOffset(DataRefImpl Rel) const override;
    236   symbol_iterator getRelocationSymbol(DataRefImpl Rel) const override;
    237   section_iterator getRelocationSection(DataRefImpl Rel) const;
    238   uint64_t getRelocationType(DataRefImpl Rel) const override;
    239   void getRelocationTypeName(DataRefImpl Rel,
    240                              SmallVectorImpl<char> &Result) const override;
    241   uint8_t getRelocationLength(DataRefImpl Rel) const;
    242 
    243   // MachO specific.
    244   std::error_code getLibraryShortNameByIndex(unsigned Index, StringRef &) const;
    245 
    246   section_iterator getRelocationRelocatedSection(relocation_iterator Rel) const;
    247 
    248   // TODO: Would be useful to have an iterator based version
    249   // of the load command interface too.
    250 
    251   basic_symbol_iterator symbol_begin_impl() const override;
    252   basic_symbol_iterator symbol_end_impl() const override;
    253 
    254   // MachO specific.
    255   basic_symbol_iterator getSymbolByIndex(unsigned Index) const;
    256   uint64_t getSymbolIndex(DataRefImpl Symb) const;
    257 
    258   section_iterator section_begin() const override;
    259   section_iterator section_end() const override;
    260 
    261   uint8_t getBytesInAddress() const override;
    262 
    263   StringRef getFileFormatName() const override;
    264   unsigned getArch() const override;
    265   SubtargetFeatures getFeatures() const override { return SubtargetFeatures(); }
    266   Triple getArchTriple(const char **McpuDefault = nullptr) const;
    267 
    268   relocation_iterator section_rel_begin(unsigned Index) const;
    269   relocation_iterator section_rel_end(unsigned Index) const;
    270 
    271   dice_iterator begin_dices() const;
    272   dice_iterator end_dices() const;
    273 
    274   load_command_iterator begin_load_commands() const;
    275   load_command_iterator end_load_commands() const;
    276   iterator_range<load_command_iterator> load_commands() const;
    277 
    278   /// For use iterating over all exported symbols.
    279   iterator_range<export_iterator> exports() const;
    280 
    281   /// For use examining a trie not in a MachOObjectFile.
    282   static iterator_range<export_iterator> exports(ArrayRef<uint8_t> Trie);
    283 
    284   /// For use iterating over all rebase table entries.
    285   iterator_range<rebase_iterator> rebaseTable() const;
    286 
    287   /// For use examining rebase opcodes not in a MachOObjectFile.
    288   static iterator_range<rebase_iterator> rebaseTable(ArrayRef<uint8_t> Opcodes,
    289                                                      bool is64);
    290 
    291   /// For use iterating over all bind table entries.
    292   iterator_range<bind_iterator> bindTable() const;
    293 
    294   /// For use iterating over all lazy bind table entries.
    295   iterator_range<bind_iterator> lazyBindTable() const;
    296 
    297   /// For use iterating over all lazy bind table entries.
    298   iterator_range<bind_iterator> weakBindTable() const;
    299 
    300   /// For use examining bind opcodes not in a MachOObjectFile.
    301   static iterator_range<bind_iterator> bindTable(ArrayRef<uint8_t> Opcodes,
    302                                                  bool is64,
    303                                                  MachOBindEntry::Kind);
    304 
    305 
    306   // In a MachO file, sections have a segment name. This is used in the .o
    307   // files. They have a single segment, but this field specifies which segment
    308   // a section should be put in in the final object.
    309   StringRef getSectionFinalSegmentName(DataRefImpl Sec) const;
    310 
    311   // Names are stored as 16 bytes. These returns the raw 16 bytes without
    312   // interpreting them as a C string.
    313   ArrayRef<char> getSectionRawName(DataRefImpl Sec) const;
    314   ArrayRef<char> getSectionRawFinalSegmentName(DataRefImpl Sec) const;
    315 
    316   // MachO specific Info about relocations.
    317   bool isRelocationScattered(const MachO::any_relocation_info &RE) const;
    318   unsigned getPlainRelocationSymbolNum(
    319                                     const MachO::any_relocation_info &RE) const;
    320   bool getPlainRelocationExternal(const MachO::any_relocation_info &RE) const;
    321   bool getScatteredRelocationScattered(
    322                                     const MachO::any_relocation_info &RE) const;
    323   uint32_t getScatteredRelocationValue(
    324                                     const MachO::any_relocation_info &RE) const;
    325   uint32_t getScatteredRelocationType(
    326                                     const MachO::any_relocation_info &RE) const;
    327   unsigned getAnyRelocationAddress(const MachO::any_relocation_info &RE) const;
    328   unsigned getAnyRelocationPCRel(const MachO::any_relocation_info &RE) const;
    329   unsigned getAnyRelocationLength(const MachO::any_relocation_info &RE) const;
    330   unsigned getAnyRelocationType(const MachO::any_relocation_info &RE) const;
    331   SectionRef getAnyRelocationSection(const MachO::any_relocation_info &RE) const;
    332 
    333   // MachO specific structures.
    334   MachO::section getSection(DataRefImpl DRI) const;
    335   MachO::section_64 getSection64(DataRefImpl DRI) const;
    336   MachO::section getSection(const LoadCommandInfo &L, unsigned Index) const;
    337   MachO::section_64 getSection64(const LoadCommandInfo &L,unsigned Index) const;
    338   MachO::nlist getSymbolTableEntry(DataRefImpl DRI) const;
    339   MachO::nlist_64 getSymbol64TableEntry(DataRefImpl DRI) const;
    340 
    341   MachO::linkedit_data_command
    342   getLinkeditDataLoadCommand(const LoadCommandInfo &L) const;
    343   MachO::segment_command
    344   getSegmentLoadCommand(const LoadCommandInfo &L) const;
    345   MachO::segment_command_64
    346   getSegment64LoadCommand(const LoadCommandInfo &L) const;
    347   MachO::linker_option_command
    348   getLinkerOptionLoadCommand(const LoadCommandInfo &L) const;
    349   MachO::version_min_command
    350   getVersionMinLoadCommand(const LoadCommandInfo &L) const;
    351   MachO::dylib_command
    352   getDylibIDLoadCommand(const LoadCommandInfo &L) const;
    353   MachO::dyld_info_command
    354   getDyldInfoLoadCommand(const LoadCommandInfo &L) const;
    355   MachO::dylinker_command
    356   getDylinkerCommand(const LoadCommandInfo &L) const;
    357   MachO::uuid_command
    358   getUuidCommand(const LoadCommandInfo &L) const;
    359   MachO::rpath_command
    360   getRpathCommand(const LoadCommandInfo &L) const;
    361   MachO::source_version_command
    362   getSourceVersionCommand(const LoadCommandInfo &L) const;
    363   MachO::entry_point_command
    364   getEntryPointCommand(const LoadCommandInfo &L) const;
    365   MachO::encryption_info_command
    366   getEncryptionInfoCommand(const LoadCommandInfo &L) const;
    367   MachO::encryption_info_command_64
    368   getEncryptionInfoCommand64(const LoadCommandInfo &L) const;
    369   MachO::sub_framework_command
    370   getSubFrameworkCommand(const LoadCommandInfo &L) const;
    371   MachO::sub_umbrella_command
    372   getSubUmbrellaCommand(const LoadCommandInfo &L) const;
    373   MachO::sub_library_command
    374   getSubLibraryCommand(const LoadCommandInfo &L) const;
    375   MachO::sub_client_command
    376   getSubClientCommand(const LoadCommandInfo &L) const;
    377   MachO::routines_command
    378   getRoutinesCommand(const LoadCommandInfo &L) const;
    379   MachO::routines_command_64
    380   getRoutinesCommand64(const LoadCommandInfo &L) const;
    381   MachO::thread_command
    382   getThreadCommand(const LoadCommandInfo &L) const;
    383 
    384   MachO::any_relocation_info getRelocation(DataRefImpl Rel) const;
    385   MachO::data_in_code_entry getDice(DataRefImpl Rel) const;
    386   const MachO::mach_header &getHeader() const;
    387   const MachO::mach_header_64 &getHeader64() const;
    388   uint32_t
    389   getIndirectSymbolTableEntry(const MachO::dysymtab_command &DLC,
    390                               unsigned Index) const;
    391   MachO::data_in_code_entry getDataInCodeTableEntry(uint32_t DataOffset,
    392                                                     unsigned Index) const;
    393   MachO::symtab_command getSymtabLoadCommand() const;
    394   MachO::dysymtab_command getDysymtabLoadCommand() const;
    395   MachO::linkedit_data_command getDataInCodeLoadCommand() const;
    396   MachO::linkedit_data_command getLinkOptHintsLoadCommand() const;
    397   ArrayRef<uint8_t> getDyldInfoRebaseOpcodes() const;
    398   ArrayRef<uint8_t> getDyldInfoBindOpcodes() const;
    399   ArrayRef<uint8_t> getDyldInfoWeakBindOpcodes() const;
    400   ArrayRef<uint8_t> getDyldInfoLazyBindOpcodes() const;
    401   ArrayRef<uint8_t> getDyldInfoExportsTrie() const;
    402   ArrayRef<uint8_t> getUuid() const;
    403 
    404   StringRef getStringTableData() const;
    405   bool is64Bit() const;
    406   void ReadULEB128s(uint64_t Index, SmallVectorImpl<uint64_t> &Out) const;
    407 
    408   static StringRef guessLibraryShortName(StringRef Name, bool &isFramework,
    409                                          StringRef &Suffix);
    410 
    411   static Triple::ArchType getArch(uint32_t CPUType);
    412   static Triple getArchTriple(uint32_t CPUType, uint32_t CPUSubType,
    413                               const char **McpuDefault = nullptr);
    414   static bool isValidArch(StringRef ArchFlag);
    415   static Triple getHostArch();
    416 
    417   bool isRelocatableObject() const override;
    418 
    419   bool hasPageZeroSegment() const { return HasPageZeroSegment; }
    420 
    421   static bool classof(const Binary *v) {
    422     return v->isMachO();
    423   }
    424 
    425   static uint32_t
    426   getVersionMinMajor(MachO::version_min_command &C, bool SDK) {
    427     uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
    428     return (VersionOrSDK >> 16) & 0xffff;
    429   }
    430 
    431   static uint32_t
    432   getVersionMinMinor(MachO::version_min_command &C, bool SDK) {
    433     uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
    434     return (VersionOrSDK >> 8) & 0xff;
    435   }
    436 
    437   static uint32_t
    438   getVersionMinUpdate(MachO::version_min_command &C, bool SDK) {
    439     uint32_t VersionOrSDK = (SDK) ? C.sdk : C.version;
    440     return VersionOrSDK & 0xff;
    441   }
    442 
    443 private:
    444 
    445   MachOObjectFile(MemoryBufferRef Object, bool IsLittleEndian, bool Is64Bits,
    446                   Error &Err);
    447 
    448   uint64_t getSymbolValueImpl(DataRefImpl Symb) const override;
    449 
    450   union {
    451     MachO::mach_header_64 Header64;
    452     MachO::mach_header Header;
    453   };
    454   typedef SmallVector<const char*, 1> SectionList;
    455   SectionList Sections;
    456   typedef SmallVector<const char*, 1> LibraryList;
    457   LibraryList Libraries;
    458   LoadCommandList LoadCommands;
    459   typedef SmallVector<StringRef, 1> LibraryShortName;
    460   mutable LibraryShortName LibrariesShortNames;
    461   const char *SymtabLoadCmd;
    462   const char *DysymtabLoadCmd;
    463   const char *DataInCodeLoadCmd;
    464   const char *LinkOptHintsLoadCmd;
    465   const char *DyldInfoLoadCmd;
    466   const char *UuidLoadCmd;
    467   bool HasPageZeroSegment;
    468 };
    469 
    470 /// DiceRef
    471 inline DiceRef::DiceRef(DataRefImpl DiceP, const ObjectFile *Owner)
    472   : DicePimpl(DiceP) , OwningObject(Owner) {}
    473 
    474 inline bool DiceRef::operator==(const DiceRef &Other) const {
    475   return DicePimpl == Other.DicePimpl;
    476 }
    477 
    478 inline bool DiceRef::operator<(const DiceRef &Other) const {
    479   return DicePimpl < Other.DicePimpl;
    480 }
    481 
    482 inline void DiceRef::moveNext() {
    483   const MachO::data_in_code_entry *P =
    484     reinterpret_cast<const MachO::data_in_code_entry *>(DicePimpl.p);
    485   DicePimpl.p = reinterpret_cast<uintptr_t>(P + 1);
    486 }
    487 
    488 // Since a Mach-O data in code reference, a DiceRef, can only be created when
    489 // the OwningObject ObjectFile is a MachOObjectFile a static_cast<> is used for
    490 // the methods that get the values of the fields of the reference.
    491 
    492 inline std::error_code DiceRef::getOffset(uint32_t &Result) const {
    493   const MachOObjectFile *MachOOF =
    494     static_cast<const MachOObjectFile *>(OwningObject);
    495   MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
    496   Result = Dice.offset;
    497   return std::error_code();
    498 }
    499 
    500 inline std::error_code DiceRef::getLength(uint16_t &Result) const {
    501   const MachOObjectFile *MachOOF =
    502     static_cast<const MachOObjectFile *>(OwningObject);
    503   MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
    504   Result = Dice.length;
    505   return std::error_code();
    506 }
    507 
    508 inline std::error_code DiceRef::getKind(uint16_t &Result) const {
    509   const MachOObjectFile *MachOOF =
    510     static_cast<const MachOObjectFile *>(OwningObject);
    511   MachO::data_in_code_entry Dice = MachOOF->getDice(DicePimpl);
    512   Result = Dice.kind;
    513   return std::error_code();
    514 }
    515 
    516 inline DataRefImpl DiceRef::getRawDataRefImpl() const {
    517   return DicePimpl;
    518 }
    519 
    520 inline const ObjectFile *DiceRef::getObjectFile() const {
    521   return OwningObject;
    522 }
    523 
    524 }
    525 }
    526 
    527 #endif
    528