include/llvm/MC/MCMachObjectWriter.h

   1 //===- llvm/MC/MCMachObjectWriter.h - Mach Object Writer --------*- C++ -*-===//
   2 //
   3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
   4 // See https://llvm.org/LICENSE.txt for license information.
   5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
   6 //
   7 //===----------------------------------------------------------------------===//
   8
   9 #ifndef LLVM_MC_MCMACHOBJECTWRITER_H
  10 #define LLVM_MC_MCMACHOBJECTWRITER_H
  11
  12 #include "llvm/ADT/DenseMap.h"
  13 #include "llvm/ADT/StringRef.h"
  14 #include "llvm/BinaryFormat/MachO.h"
  15 #include "llvm/MC/MCExpr.h"
  16 #include "llvm/MC/MCObjectWriter.h"
  17 #include "llvm/MC/MCSection.h"
  18 #include "llvm/MC/StringTableBuilder.h"
  19 #include <cstdint>
  20 #include <memory>
  21 #include <string>
  22 #include <vector>
  23
  24 namespace llvm {
  25
  26 class MachObjectWriter;
  27
  28 class MCMachObjectTargetWriter : public MCObjectTargetWriter {
  29   const unsigned Is64Bit : 1;
  30   const uint32_t CPUType;
  31   const uint32_t CPUSubtype;
  32   unsigned LocalDifference_RIT;
  33
  34 protected:
  35   MCMachObjectTargetWriter(bool Is64Bit_, uint32_t CPUType_,
  36                            uint32_t CPUSubtype_);
  37
  38   void setLocalDifferenceRelocationType(unsigned Type) {
  39     LocalDifference_RIT = Type;
  40   }
  41
  42 public:
  43   virtual ~MCMachObjectTargetWriter();
  44
  45   virtual Triple::ObjectFormatType getFormat() const { return Triple::MachO; }
  46   static bool classof(const MCObjectTargetWriter *W) {
  47     return W->getFormat() == Triple::MachO;
  48   }
  49
  50   /// \name Lifetime Management
  51   /// @{
  52
  53   virtual void reset() {}
  54
  55   /// @}
  56
  57   /// \name Accessors
  58   /// @{
  59
  60   bool is64Bit() const { return Is64Bit; }
  61   uint32_t getCPUType() const { return CPUType; }
  62   uint32_t getCPUSubtype() const { return CPUSubtype; }
  63   unsigned getLocalDifferenceRelocationType() const {
  64     return LocalDifference_RIT;
  65   }
  66
  67   /// @}
  68
  69   /// \name API
  70   /// @{
  71
  72   virtual void recordRelocation(MachObjectWriter *Writer, MCAssembler &Asm,
  73                                 const MCAsmLayout &Layout,
  74                                 const MCFragment *Fragment,
  75                                 const MCFixup &Fixup, MCValue Target,
  76                                 uint64_t &FixedValue) = 0;
  77
  78   /// @}
  79 };
  80
  81 class MachObjectWriter : public MCObjectWriter {
  82   /// Helper struct for containing some precomputed information on symbols.
  83   struct MachSymbolData {
  84     const MCSymbol *Symbol;
  85     uint64_t StringIndex;
  86     uint8_t SectionIndex;
  87
  88     // Support lexicographic sorting.
  89     bool operator<(const MachSymbolData &RHS) const;
  90   };
  91
  92   /// The target specific Mach-O writer instance.
  93   std::unique_ptr<MCMachObjectTargetWriter> TargetObjectWriter;
  94
  95   /// \name Relocation Data
  96   /// @{
  97
  98   struct RelAndSymbol {
  99     const MCSymbol *Sym;
 100     MachO::any_relocation_info MRE;
 101     RelAndSymbol(const MCSymbol *Sym, const MachO::any_relocation_info &MRE)
 102         : Sym(Sym), MRE(MRE) {}
 103   };
 104
 105   DenseMap<const MCSection *, std::vector<RelAndSymbol>> Relocations;
 106   DenseMap<const MCSection *, unsigned> IndirectSymBase;
 107
 108   SectionAddrMap SectionAddress;
 109
 110   /// @}
 111   /// \name Symbol Table Data
 112   /// @{
 113
 114   StringTableBuilder StringTable{StringTableBuilder::MachO};
 115   std::vector<MachSymbolData> LocalSymbolData;
 116   std::vector<MachSymbolData> ExternalSymbolData;
 117   std::vector<MachSymbolData> UndefinedSymbolData;
 118
 119   /// @}
 120
 121   MachSymbolData *findSymbolData(const MCSymbol &Sym);
 122
 123   void writeWithPadding(StringRef Str, uint64_t Size);
 124
 125 public:
 126   MachObjectWriter(std::unique_ptr<MCMachObjectTargetWriter> MOTW,
 127                    raw_pwrite_stream &OS, bool IsLittleEndian)
 128       : TargetObjectWriter(std::move(MOTW)),
 129         W(OS, IsLittleEndian ? support::little : support::big) {}
 130
 131   support::endian::Writer W;
 132
 133   const MCSymbol &findAliasedSymbol(const MCSymbol &Sym) const;
 134
 135   /// \name Lifetime management Methods
 136   /// @{
 137
 138   void reset() override;
 139
 140   /// @}
 141
 142   /// \name Utility Methods
 143   /// @{
 144
 145   bool isFixupKindPCRel(const MCAssembler &Asm, unsigned Kind);
 146
 147   SectionAddrMap &getSectionAddressMap() { return SectionAddress; }
 148
 149   uint64_t getSectionAddress(const MCSection *Sec) const {
 150     return SectionAddress.lookup(Sec);
 151   }
 152   uint64_t getSymbolAddress(const MCSymbol &S, const MCAsmLayout &Layout) const;
 153
 154   uint64_t getFragmentAddress(const MCFragment *Fragment,
 155                               const MCAsmLayout &Layout) const;
 156
 157   uint64_t getPaddingSize(const MCSection *SD, const MCAsmLayout &Layout) const;
 158
 159   bool doesSymbolRequireExternRelocation(const MCSymbol &S);
 160
 161   /// @}
 162
 163   /// \name Target Writer Proxy Accessors
 164   /// @{
 165
 166   bool is64Bit() const { return TargetObjectWriter->is64Bit(); }
 167   bool isX86_64() const {
 168     uint32_t CPUType = TargetObjectWriter->getCPUType();
 169     return CPUType == MachO::CPU_TYPE_X86_64;
 170   }
 171
 172   /// @}
 173
 174   void writeHeader(MachO::HeaderFileType Type, unsigned NumLoadCommands,
 175                    unsigned LoadCommandsSize, bool SubsectionsViaSymbols);
 176
 177   /// Write a segment load command.
 178   ///
 179   /// \param NumSections The number of sections in this segment.
 180   /// \param SectionDataSize The total size of the sections.
 181   void writeSegmentLoadCommand(StringRef Name, unsigned NumSections,
 182                                uint64_t VMAddr, uint64_t VMSize,
 183                                uint64_t SectionDataStartOffset,
 184                                uint64_t SectionDataSize, uint32_t MaxProt,
 185                                uint32_t InitProt);
 186
 187   void writeSection(const MCAsmLayout &Layout, const MCSection &Sec,
 188                     uint64_t VMAddr, uint64_t FileOffset, unsigned Flags,
 189                     uint64_t RelocationsStart, unsigned NumRelocations);
 190
 191   void writeSymtabLoadCommand(uint32_t SymbolOffset, uint32_t NumSymbols,
 192                               uint32_t StringTableOffset,
 193                               uint32_t StringTableSize);
 194
 195   void writeDysymtabLoadCommand(
 196       uint32_t FirstLocalSymbol, uint32_t NumLocalSymbols,
 197       uint32_t FirstExternalSymbol, uint32_t NumExternalSymbols,
 198       uint32_t FirstUndefinedSymbol, uint32_t NumUndefinedSymbols,
 199       uint32_t IndirectSymbolOffset, uint32_t NumIndirectSymbols);
 200
 201   void writeNlist(MachSymbolData &MSD, const MCAsmLayout &Layout);
 202
 203   void writeLinkeditLoadCommand(uint32_t Type, uint32_t DataOffset,
 204                                 uint32_t DataSize);
 205
 206   void writeLinkerOptionsLoadCommand(const std::vector<std::string> &Options);
 207
 208   // FIXME: We really need to improve the relocation validation. Basically, we
 209   // want to implement a separate computation which evaluates the relocation
 210   // entry as the linker would, and verifies that the resultant fixup value is
 211   // exactly what the encoder wanted. This will catch several classes of
 212   // problems:
 213   //
 214   //  - Relocation entry bugs, the two algorithms are unlikely to have the same
 215   //    exact bug.
 216   //
 217   //  - Relaxation issues, where we forget to relax something.
 218   //
 219   //  - Input errors, where something cannot be correctly encoded. 'as' allows
 220   //    these through in many cases.
 221
 222   // Add a relocation to be output in the object file. At the time this is
 223   // called, the symbol indexes are not know, so if the relocation refers
 224   // to a symbol it should be passed as \p RelSymbol so that it can be updated
 225   // afterwards. If the relocation doesn't refer to a symbol, nullptr should be
 226   // used.
 227   void addRelocation(const MCSymbol *RelSymbol, const MCSection *Sec,
 228                      MachO::any_relocation_info &MRE) {
 229     RelAndSymbol P(RelSymbol, MRE);
 230     Relocations[Sec].push_back(P);
 231   }
 232
 233   void recordScatteredRelocation(const MCAssembler &Asm,
 234                                  const MCAsmLayout &Layout,
 235                                  const MCFragment *Fragment,
 236                                  const MCFixup &Fixup, MCValue Target,
 237                                  unsigned Log2Size, uint64_t &FixedValue);
 238
 239   void recordTLVPRelocation(const MCAssembler &Asm, const MCAsmLayout &Layout,
 240                             const MCFragment *Fragment, const MCFixup &Fixup,
 241                             MCValue Target, uint64_t &FixedValue);
 242
 243   void recordRelocation(MCAssembler &Asm, const MCAsmLayout &Layout,
 244                         const MCFragment *Fragment, const MCFixup &Fixup,
 245                         MCValue Target, uint64_t &FixedValue) override;
 246
 247   void bindIndirectSymbols(MCAssembler &Asm);
 248
 249   /// Compute the symbol table data.
 250   void computeSymbolTable(MCAssembler &Asm,
 251                           std::vector<MachSymbolData> &LocalSymbolData,
 252                           std::vector<MachSymbolData> &ExternalSymbolData,
 253                           std::vector<MachSymbolData> &UndefinedSymbolData);
 254
 255   void computeSectionAddresses(const MCAssembler &Asm,
 256                                const MCAsmLayout &Layout);
 257
 258   void executePostLayoutBinding(MCAssembler &Asm,
 259                                 const MCAsmLayout &Layout) override;
 260
 261   bool isSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
 262                                               const MCSymbol &A,
 263                                               const MCSymbol &B,
 264                                               bool InSet) const override;
 265
 266   bool isSymbolRefDifferenceFullyResolvedImpl(const MCAssembler &Asm,
 267                                               const MCSymbol &SymA,
 268                                               const MCFragment &FB, bool InSet,
 269                                               bool IsPCRel) const override;
 270
 271   uint64_t writeObject(MCAssembler &Asm, const MCAsmLayout &Layout) override;
 272 };
 273
 274 /// Construct a new Mach-O writer instance.
 275 ///
 276 /// This routine takes ownership of the target writer subclass.
 277 ///
 278 /// \param MOTW - The target specific Mach-O writer subclass.
 279 /// \param OS - The stream to write to.
 280 /// \returns The constructed object writer.
 281 std::unique_ptr<MCObjectWriter>
 282 createMachObjectWriter(std::unique_ptr<MCMachObjectTargetWriter> MOTW,
 283                        raw_pwrite_stream &OS, bool IsLittleEndian);
 284
 285 } // end namespace llvm
 286
 287 #endif // LLVM_MC_MCMACHOBJECTWRITER_H