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