Home | History | Annotate | Download | only in MC
      1 //===-- llvm/MC/MCMachObjectWriter.h - Mach Object Writer -------*- 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 #ifndef LLVM_MC_MCMACHOBJECTWRITER_H
     11 #define LLVM_MC_MCMACHOBJECTWRITER_H
     12 
     13 #include "llvm/ADT/DenseMap.h"
     14 #include "llvm/ADT/OwningPtr.h"
     15 #include "llvm/ADT/SmallString.h"
     16 #include "llvm/MC/MCExpr.h"
     17 #include "llvm/MC/MCObjectWriter.h"
     18 #include "llvm/Object/MachOFormat.h"
     19 #include "llvm/Support/DataTypes.h"
     20 #include <vector>
     21 
     22 namespace llvm {
     23 
     24 class MCSectionData;
     25 class MachObjectWriter;
     26 
     27 class MCMachObjectTargetWriter {
     28   const unsigned Is64Bit : 1;
     29   const uint32_t CPUType;
     30   const uint32_t CPUSubtype;
     31   // FIXME: Remove this, we should just always use it once we no longer care
     32   // about Darwin 'as' compatibility.
     33   const unsigned UseAggressiveSymbolFolding : 1;
     34   unsigned LocalDifference_RIT;
     35 
     36 protected:
     37   MCMachObjectTargetWriter(bool Is64Bit_, uint32_t CPUType_,
     38                            uint32_t CPUSubtype_,
     39                            bool UseAggressiveSymbolFolding_ = false);
     40 
     41   void setLocalDifferenceRelocationType(unsigned Type) {
     42     LocalDifference_RIT = Type;
     43   }
     44 
     45 public:
     46   virtual ~MCMachObjectTargetWriter();
     47 
     48   /// @name Lifetime Management
     49   /// @{
     50 
     51   virtual void reset() {};
     52 
     53   /// @}
     54 
     55   /// @name Accessors
     56   /// @{
     57 
     58   bool is64Bit() const { return Is64Bit; }
     59   bool useAggressiveSymbolFolding() const { return UseAggressiveSymbolFolding; }
     60   uint32_t getCPUType() const { return CPUType; }
     61   uint32_t getCPUSubtype() const { return CPUSubtype; }
     62   unsigned getLocalDifferenceRelocationType() const {
     63     return LocalDifference_RIT;
     64   }
     65 
     66   /// @}
     67 
     68   /// @name API
     69   /// @{
     70 
     71   virtual void RecordRelocation(MachObjectWriter *Writer,
     72                                 const MCAssembler &Asm,
     73                                 const MCAsmLayout &Layout,
     74                                 const MCFragment *Fragment,
     75                                 const MCFixup &Fixup,
     76                                 MCValue Target,
     77                                 uint64_t &FixedValue) = 0;
     78 
     79   /// @}
     80 };
     81 
     82 class MachObjectWriter : public MCObjectWriter {
     83   /// MachSymbolData - Helper struct for containing some precomputed information
     84   /// on symbols.
     85   struct MachSymbolData {
     86     MCSymbolData *SymbolData;
     87     uint64_t StringIndex;
     88     uint8_t SectionIndex;
     89 
     90     // Support lexicographic sorting.
     91     bool operator<(const MachSymbolData &RHS) const;
     92   };
     93 
     94   /// The target specific Mach-O writer instance.
     95   llvm::OwningPtr<MCMachObjectTargetWriter> TargetObjectWriter;
     96 
     97   /// @name Relocation Data
     98   /// @{
     99 
    100   llvm::DenseMap<const MCSectionData*,
    101                  std::vector<object::macho::RelocationEntry> > Relocations;
    102   llvm::DenseMap<const MCSectionData*, unsigned> IndirectSymBase;
    103 
    104   /// @}
    105   /// @name Symbol Table Data
    106   /// @{
    107 
    108   SmallString<256> StringTable;
    109   std::vector<MachSymbolData> LocalSymbolData;
    110   std::vector<MachSymbolData> ExternalSymbolData;
    111   std::vector<MachSymbolData> UndefinedSymbolData;
    112 
    113   /// @}
    114 
    115 public:
    116   MachObjectWriter(MCMachObjectTargetWriter *MOTW, raw_ostream &_OS,
    117                    bool _IsLittleEndian)
    118     : MCObjectWriter(_OS, _IsLittleEndian), TargetObjectWriter(MOTW) {
    119   }
    120 
    121   /// @name Lifetime management Methods
    122   /// @{
    123 
    124   virtual void reset();
    125 
    126   /// @}
    127 
    128   /// @name Utility Methods
    129   /// @{
    130 
    131   bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind);
    132 
    133   SectionAddrMap SectionAddress;
    134 
    135   SectionAddrMap &getSectionAddressMap() { return SectionAddress; }
    136 
    137   uint64_t getSectionAddress(const MCSectionData* SD) const {
    138     return SectionAddress.lookup(SD);
    139   }
    140   uint64_t getSymbolAddress(const MCSymbolData* SD,
    141                             const MCAsmLayout &Layout) const;
    142 
    143   uint64_t getFragmentAddress(const MCFragment *Fragment,
    144                               const MCAsmLayout &Layout) const;
    145 
    146   uint64_t getPaddingSize(const MCSectionData *SD,
    147                           const MCAsmLayout &Layout) const;
    148 
    149   bool doesSymbolRequireExternRelocation(const MCSymbolData *SD);
    150 
    151   /// @}
    152 
    153   /// @name Target Writer Proxy Accessors
    154   /// @{
    155 
    156   bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
    157   bool isARM() const {
    158     uint32_t CPUType = TargetObjectWriter->getCPUType() &
    159       ~object::mach::CTFM_ArchMask;
    160     return CPUType == object::mach::CTM_ARM;
    161   }
    162 
    163   /// @}
    164 
    165   void WriteHeader(unsigned NumLoadCommands, unsigned LoadCommandsSize,
    166                    bool SubsectionsViaSymbols);
    167 
    168   /// WriteSegmentLoadCommand - Write a segment load command.
    169   ///
    170   /// \param NumSections The number of sections in this segment.
    171   /// \param SectionDataSize The total size of the sections.
    172   void WriteSegmentLoadCommand(unsigned NumSections,
    173                                uint64_t VMSize,
    174                                uint64_t SectionDataStartOffset,
    175                                uint64_t SectionDataSize);
    176 
    177   void WriteSection(const MCAssembler &Asm, const MCAsmLayout &Layout,
    178                     const MCSectionData &SD, uint64_t FileOffset,
    179                     uint64_t RelocationsStart, unsigned NumRelocations);
    180 
    181   void WriteSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
    182                               uint32_t StringTableOffset,
    183                               uint32_t StringTableSize);
    184 
    185   void WriteDysymtabLoadCommand(uint32_t FirstLocalSymbol,
    186                                 uint32_t NumLocalSymbols,
    187                                 uint32_t FirstExternalSymbol,
    188                                 uint32_t NumExternalSymbols,
    189                                 uint32_t FirstUndefinedSymbol,
    190                                 uint32_t NumUndefinedSymbols,
    191                                 uint32_t IndirectSymbolOffset,
    192                                 uint32_t NumIndirectSymbols);
    193 
    194   void WriteNlist(MachSymbolData &MSD, const MCAsmLayout &Layout);
    195 
    196   void WriteLinkeditLoadCommand(uint32_t Type, uint32_t DataOffset,
    197                                 uint32_t DataSize);
    198 
    199   void WriteLinkerOptionsLoadCommand(const std::vector<std::string> &Options);
    200 
    201   // FIXME: We really need to improve the relocation validation. Basically, we
    202   // want to implement a separate computation which evaluates the relocation
    203   // entry as the linker would, and verifies that the resultant fixup value is
    204   // exactly what the encoder wanted. This will catch several classes of
    205   // problems:
    206   //
    207   //  - Relocation entry bugs, the two algorithms are unlikely to have the same
    208   //    exact bug.
    209   //
    210   //  - Relaxation issues, where we forget to relax something.
    211   //
    212   //  - Input errors, where something cannot be correctly encoded. 'as' allows
    213   //    these through in many cases.
    214 
    215   void addRelocation(const MCSectionData *SD,
    216                      object::macho::RelocationEntry &MRE) {
    217     Relocations[SD].push_back(MRE);
    218   }
    219 
    220   void RecordScatteredRelocation(const MCAssembler &Asm,
    221                                  const MCAsmLayout &Layout,
    222                                  const MCFragment *Fragment,
    223                                  const MCFixup &Fixup, MCValue Target,
    224                                  unsigned Log2Size,
    225                                  uint64_t &FixedValue);
    226 
    227   void RecordTLVPRelocation(const MCAssembler &Asm,
    228                             const MCAsmLayout &Layout,
    229                             const MCFragment *Fragment,
    230                             const MCFixup &Fixup, MCValue Target,
    231                             uint64_t &FixedValue);
    232 
    233   void RecordRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
    234                         const MCFragment *Fragment, const MCFixup &Fixup,
    235                         MCValue Target, uint64_t &FixedValue);
    236 
    237   void BindIndirectSymbols(MCAssembler &Asm);
    238 
    239   /// ComputeSymbolTable - Compute the symbol table data
    240   ///
    241   /// \param StringTable [out] - The string table data.
    242   void ComputeSymbolTable(MCAssembler &Asm, SmallString<256> &StringTable,
    243                           std::vector<MachSymbolData> &LocalSymbolData,
    244                           std::vector<MachSymbolData> &ExternalSymbolData,
    245                           std::vector<MachSymbolData> &UndefinedSymbolData);
    246 
    247   void computeSectionAddresses(const MCAssembler &Asm,
    248                                const MCAsmLayout &Layout);
    249 
    250   void markAbsoluteVariableSymbols(MCAssembler &Asm,
    251                                    const MCAsmLayout &Layout);
    252   void ExecutePostLayoutBinding(MCAssembler &Asm, const MCAsmLayout &Layout);
    253 
    254   virtual bool IsSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
    255                                                       const MCSymbolData &DataA,
    256                                                       const MCFragment &FB,
    257                                                       bool InSet,
    258                                                       bool IsPCRel) const;
    259 
    260   void WriteObject(MCAssembler &Asm, const MCAsmLayout &Layout);
    261 };
    262 
    263 
    264 /// \brief Construct a new Mach-O writer instance.
    265 ///
    266 /// This routine takes ownership of the target writer subclass.
    267 ///
    268 /// \param MOTW - The target specific Mach-O writer subclass.
    269 /// \param OS - The stream to write to.
    270 /// \returns The constructed object writer.
    271 MCObjectWriter *createMachObjectWriter(MCMachObjectTargetWriter *MOTW,
    272                                        raw_ostream &OS, bool IsLittleEndian);
    273 
    274 } // End llvm namespace
    275 
    276 #endif
    277