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/SmallString.h"
     15 #include "llvm/MC/MCExpr.h"
     16 #include "llvm/MC/MCObjectWriter.h"
     17 #include "llvm/MC/StringTableBuilder.h"
     18 #include "llvm/Support/DataTypes.h"
     19 #include "llvm/Support/MachO.h"
     20 #include <vector>
     21 
     22 namespace llvm {
     23 
     24 class MachObjectWriter;
     25 
     26 class MCMachObjectTargetWriter {
     27   const unsigned Is64Bit : 1;
     28   const uint32_t CPUType;
     29   const uint32_t CPUSubtype;
     30   unsigned LocalDifference_RIT;
     31 
     32 protected:
     33   MCMachObjectTargetWriter(bool Is64Bit_, uint32_t CPUType_,
     34                            uint32_t CPUSubtype_);
     35 
     36   void setLocalDifferenceRelocationType(unsigned Type) {
     37     LocalDifference_RIT = Type;
     38   }
     39 
     40 public:
     41   virtual ~MCMachObjectTargetWriter();
     42 
     43   /// \name Lifetime Management
     44   /// @{
     45 
     46   virtual void reset() {}
     47 
     48   /// @}
     49 
     50   /// \name Accessors
     51   /// @{
     52 
     53   bool is64Bit() const { return Is64Bit; }
     54   uint32_t getCPUType() const { return CPUType; }
     55   uint32_t getCPUSubtype() const { return CPUSubtype; }
     56   unsigned getLocalDifferenceRelocationType() const {
     57     return LocalDifference_RIT;
     58   }
     59 
     60   /// @}
     61 
     62   /// \name API
     63   /// @{
     64 
     65   virtual void recordRelocation(MachObjectWriter *Writer, MCAssembler &Asm,
     66                                 const MCAsmLayout &Layout,
     67                                 const MCFragment *Fragment,
     68                                 const MCFixup &Fixup, MCValue Target,
     69                                 uint64_t &FixedValue) = 0;
     70 
     71   /// @}
     72 };
     73 
     74 class MachObjectWriter : public MCObjectWriter {
     75   /// Helper struct for containing some precomputed information on symbols.
     76   struct MachSymbolData {
     77     const MCSymbol *Symbol;
     78     uint64_t StringIndex;
     79     uint8_t SectionIndex;
     80 
     81     // Support lexicographic sorting.
     82     bool operator<(const MachSymbolData &RHS) const;
     83   };
     84 
     85   /// The target specific Mach-O writer instance.
     86   std::unique_ptr<MCMachObjectTargetWriter> TargetObjectWriter;
     87 
     88   /// \name Relocation Data
     89   /// @{
     90 
     91   struct RelAndSymbol {
     92     const MCSymbol *Sym;
     93     MachO::any_relocation_info MRE;
     94     RelAndSymbol(const MCSymbol *Sym, const MachO::any_relocation_info &MRE)
     95         : Sym(Sym), MRE(MRE) {}
     96   };
     97 
     98   llvm::DenseMap<const MCSection *, std::vector<RelAndSymbol>> Relocations;
     99   llvm::DenseMap<const MCSection *, unsigned> IndirectSymBase;
    100 
    101   SectionAddrMap SectionAddress;
    102 
    103   /// @}
    104   /// \name Symbol Table Data
    105   /// @{
    106 
    107   StringTableBuilder StringTable{StringTableBuilder::MachO};
    108   std::vector<MachSymbolData> LocalSymbolData;
    109   std::vector<MachSymbolData> ExternalSymbolData;
    110   std::vector<MachSymbolData> UndefinedSymbolData;
    111 
    112   /// @}
    113 
    114   MachSymbolData *findSymbolData(const MCSymbol &Sym);
    115 
    116 public:
    117   MachObjectWriter(MCMachObjectTargetWriter *MOTW, raw_pwrite_stream &OS,
    118                    bool IsLittleEndian)
    119       : MCObjectWriter(OS, IsLittleEndian), TargetObjectWriter(MOTW) {}
    120 
    121   const MCSymbol &findAliasedSymbol(const MCSymbol &Sym) const;
    122 
    123   /// \name Lifetime management Methods
    124   /// @{
    125 
    126   void reset() override;
    127 
    128   /// @}
    129 
    130   /// \name Utility Methods
    131   /// @{
    132 
    133   bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind);
    134 
    135   SectionAddrMap &getSectionAddressMap() { return SectionAddress; }
    136 
    137   uint64_t getSectionAddress(const MCSection *Sec) const {
    138     return SectionAddress.lookup(Sec);
    139   }
    140   uint64_t getSymbolAddress(const MCSymbol &S, const MCAsmLayout &Layout) const;
    141 
    142   uint64_t getFragmentAddress(const MCFragment *Fragment,
    143                               const MCAsmLayout &Layout) const;
    144 
    145   uint64_t getPaddingSize(const MCSection *SD, const MCAsmLayout &Layout) const;
    146 
    147   bool doesSymbolRequireExternRelocation(const MCSymbol &S);
    148 
    149   /// @}
    150 
    151   /// \name Target Writer Proxy Accessors
    152   /// @{
    153 
    154   bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
    155   bool isX86_64() const {
    156     uint32_t CPUType = TargetObjectWriter->getCPUType();
    157     return CPUType == MachO::CPU_TYPE_X86_64;
    158   }
    159 
    160   /// @}
    161 
    162   void writeHeader(MachO::HeaderFileType Type, unsigned NumLoadCommands,
    163                    unsigned LoadCommandsSize, bool SubsectionsViaSymbols);
    164 
    165   /// Write a segment load command.
    166   ///
    167   /// \param NumSections The number of sections in this segment.
    168   /// \param SectionDataSize The total size of the sections.
    169   void writeSegmentLoadCommand(StringRef Name, unsigned NumSections,
    170                                uint64_t VMAddr, uint64_t VMSize,
    171                                uint64_t SectionDataStartOffset,
    172                                uint64_t SectionDataSize, uint32_t MaxProt,
    173                                uint32_t InitProt);
    174 
    175   void writeSection(const MCAsmLayout &Layout, const MCSection &Sec,
    176                     uint64_t VMAddr, uint64_t FileOffset, unsigned Flags,
    177                     uint64_t RelocationsStart, unsigned NumRelocations);
    178 
    179   void writeSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
    180                               uint32_t StringTableOffset,
    181                               uint32_t StringTableSize);
    182 
    183   void writeDysymtabLoadCommand(
    184       uint32_t FirstLocalSymbol, uint32_t NumLocalSymbols,
    185       uint32_t FirstExternalSymbol, uint32_t NumExternalSymbols,
    186       uint32_t FirstUndefinedSymbol, uint32_t NumUndefinedSymbols,
    187       uint32_t IndirectSymbolOffset, uint32_t NumIndirectSymbols);
    188 
    189   void writeNlist(MachSymbolData &MSD, const MCAsmLayout &Layout);
    190 
    191   void writeLinkeditLoadCommand(uint32_t Type, uint32_t DataOffset,
    192                                 uint32_t DataSize);
    193 
    194   void writeLinkerOptionsLoadCommand(const std::vector<std::string> &Options);
    195 
    196   // FIXME: We really need to improve the relocation validation. Basically, we
    197   // want to implement a separate computation which evaluates the relocation
    198   // entry as the linker would, and verifies that the resultant fixup value is
    199   // exactly what the encoder wanted. This will catch several classes of
    200   // problems:
    201   //
    202   //  - Relocation entry bugs, the two algorithms are unlikely to have the same
    203   //    exact bug.
    204   //
    205   //  - Relaxation issues, where we forget to relax something.
    206   //
    207   //  - Input errors, where something cannot be correctly encoded. 'as' allows
    208   //    these through in many cases.
    209 
    210   // Add a relocation to be output in the object file. At the time this is
    211   // called, the symbol indexes are not know, so if the relocation refers
    212   // to a symbol it should be passed as \p RelSymbol so that it can be updated
    213   // afterwards. If the relocation doesn't refer to a symbol, nullptr should be
    214   // used.
    215   void addRelocation(const MCSymbol *RelSymbol, const MCSection *Sec,
    216                      MachO::any_relocation_info &MRE) {
    217     RelAndSymbol P(RelSymbol, MRE);
    218     Relocations[Sec].push_back(P);
    219   }
    220 
    221   void recordScatteredRelocation(const MCAssembler &Asm,
    222                                  const MCAsmLayout &Layout,
    223                                  const MCFragment *Fragment,
    224                                  const MCFixup &Fixup, MCValue Target,
    225                                  unsigned Log2Size, uint64_t &FixedValue);
    226 
    227   void recordTLVPRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
    228                             const MCFragment *Fragment, const MCFixup &Fixup,
    229                             MCValue Target, uint64_t &FixedValue);
    230 
    231   void recordRelocation(MCAssembler &Asm, const MCAsmLayout &Layout,
    232                         const MCFragment *Fragment, const MCFixup &Fixup,
    233                         MCValue Target, bool &IsPCRel,
    234                         uint64_t &FixedValue) override;
    235 
    236   void bindIndirectSymbols(MCAssembler &Asm);
    237 
    238   /// Compute the symbol table data.
    239   void computeSymbolTable(MCAssembler &Asm,
    240                           std::vector<MachSymbolData> &LocalSymbolData,
    241                           std::vector<MachSymbolData> &ExternalSymbolData,
    242                           std::vector<MachSymbolData> &UndefinedSymbolData);
    243 
    244   void computeSectionAddresses(const MCAssembler &Asm,
    245                                const MCAsmLayout &Layout);
    246 
    247   void executePostLayoutBinding(MCAssembler &Asm,
    248                                 const MCAsmLayout &Layout) override;
    249 
    250   bool isSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
    251                                               const MCSymbol &A,
    252                                               const MCSymbol &B,
    253                                               bool InSet) const override;
    254 
    255   bool isSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
    256                                               const MCSymbol &SymA,
    257                                               const MCFragment &FB, bool InSet,
    258                                               bool IsPCRel) const override;
    259 
    260   void writeObject(MCAssembler &Asm, const MCAsmLayout &Layout) override;
    261 };
    262 
    263 /// Construct a new Mach-O writer instance.
    264 ///
    265 /// This routine takes ownership of the target writer subclass.
    266 ///
    267 /// \param MOTW - The target specific Mach-O writer subclass.
    268 /// \param OS - The stream to write to.
    269 /// \returns The constructed object writer.
    270 MCObjectWriter *createMachObjectWriter(MCMachObjectTargetWriter *MOTW,
    271                                        raw_pwrite_stream &OS,
    272                                        bool IsLittleEndian);
    273 
    274 } // End llvm namespace
    275 
    276 #endif
    277