Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 1 | //===- MachOObjectFile.cpp - Mach-O object file binding ---------*- 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 defines the MachOObjectFile class, which binds the MachOObject |
| 11 | // class to the generic ObjectFile wrapper. |
| 12 | // |
| 13 | //===----------------------------------------------------------------------===// |
| 14 | |
| 15 | #include "llvm/ADT/Triple.h" |
| 16 | #include "llvm/Object/MachOFormat.h" |
| 17 | #include "llvm/Object/MachOObject.h" |
| 18 | #include "llvm/Object/ObjectFile.h" |
| 19 | #include "llvm/Support/MemoryBuffer.h" |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 20 | #include "llvm/Support/MachO.h" |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 21 | |
| 22 | #include <cctype> |
| 23 | #include <cstring> |
| 24 | #include <limits> |
| 25 | |
| 26 | using namespace llvm; |
| 27 | using namespace object; |
| 28 | |
| 29 | namespace llvm { |
| 30 | |
| 31 | typedef MachOObject::LoadCommandInfo LoadCommandInfo; |
| 32 | |
| 33 | class MachOObjectFile : public ObjectFile { |
| 34 | public: |
Michael J. Spencer | 001c920 | 2011-06-25 17:54:50 +0000 | [diff] [blame] | 35 | MachOObjectFile(MemoryBuffer *Object, MachOObject *MOO, error_code &ec) |
| 36 | : ObjectFile(Binary::isMachO, Object, ec), |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 37 | MachOObj(MOO), |
| 38 | RegisteredStringTable(std::numeric_limits<uint32_t>::max()) {} |
| 39 | |
| 40 | virtual symbol_iterator begin_symbols() const; |
| 41 | virtual symbol_iterator end_symbols() const; |
| 42 | virtual section_iterator begin_sections() const; |
| 43 | virtual section_iterator end_sections() const; |
| 44 | |
| 45 | virtual uint8_t getBytesInAddress() const; |
| 46 | virtual StringRef getFileFormatName() const; |
| 47 | virtual unsigned getArch() const; |
| 48 | |
| 49 | protected: |
| 50 | virtual SymbolRef getSymbolNext(DataRefImpl Symb) const; |
| 51 | virtual StringRef getSymbolName(DataRefImpl Symb) const; |
| 52 | virtual uint64_t getSymbolAddress(DataRefImpl Symb) const; |
| 53 | virtual uint64_t getSymbolSize(DataRefImpl Symb) const; |
| 54 | virtual char getSymbolNMTypeChar(DataRefImpl Symb) const; |
| 55 | virtual bool isSymbolInternal(DataRefImpl Symb) const; |
| 56 | |
| 57 | virtual SectionRef getSectionNext(DataRefImpl Sec) const; |
| 58 | virtual StringRef getSectionName(DataRefImpl Sec) const; |
| 59 | virtual uint64_t getSectionAddress(DataRefImpl Sec) const; |
| 60 | virtual uint64_t getSectionSize(DataRefImpl Sec) const; |
| 61 | virtual StringRef getSectionContents(DataRefImpl Sec) const; |
| 62 | virtual bool isSectionText(DataRefImpl Sec) const; |
| 63 | |
| 64 | private: |
| 65 | MachOObject *MachOObj; |
| 66 | mutable uint32_t RegisteredStringTable; |
| 67 | |
| 68 | void moveToNextSection(DataRefImpl &DRI) const; |
| 69 | void getSymbolTableEntry(DataRefImpl DRI, |
| 70 | InMemoryStruct<macho::SymbolTableEntry> &Res) const; |
| 71 | void moveToNextSymbol(DataRefImpl &DRI) const; |
| 72 | void getSection(DataRefImpl DRI, InMemoryStruct<macho::Section> &Res) const; |
| 73 | }; |
| 74 | |
| 75 | ObjectFile *ObjectFile::createMachOObjectFile(MemoryBuffer *Buffer) { |
Michael J. Spencer | 001c920 | 2011-06-25 17:54:50 +0000 | [diff] [blame] | 76 | error_code ec; |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 77 | std::string Err; |
| 78 | MachOObject *MachOObj = MachOObject::LoadFromBuffer(Buffer, &Err); |
| 79 | if (!MachOObj) |
| 80 | return NULL; |
Michael J. Spencer | 001c920 | 2011-06-25 17:54:50 +0000 | [diff] [blame] | 81 | return new MachOObjectFile(Buffer, MachOObj, ec); |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 82 | } |
| 83 | |
| 84 | /*===-- Symbols -----------------------------------------------------------===*/ |
| 85 | |
| 86 | void MachOObjectFile::moveToNextSymbol(DataRefImpl &DRI) const { |
| 87 | uint32_t LoadCommandCount = MachOObj->getHeader().NumLoadCommands; |
| 88 | while (DRI.d.a < LoadCommandCount) { |
| 89 | LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a); |
| 90 | if (LCI.Command.Type == macho::LCT_Symtab) { |
| 91 | InMemoryStruct<macho::SymtabLoadCommand> SymtabLoadCmd; |
| 92 | MachOObj->ReadSymtabLoadCommand(LCI, SymtabLoadCmd); |
| 93 | if (DRI.d.b < SymtabLoadCmd->NumSymbolTableEntries) |
| 94 | return; |
| 95 | } |
| 96 | |
| 97 | DRI.d.a++; |
| 98 | DRI.d.b = 0; |
| 99 | } |
| 100 | } |
| 101 | |
| 102 | void MachOObjectFile::getSymbolTableEntry(DataRefImpl DRI, |
| 103 | InMemoryStruct<macho::SymbolTableEntry> &Res) const { |
| 104 | InMemoryStruct<macho::SymtabLoadCommand> SymtabLoadCmd; |
| 105 | LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a); |
| 106 | MachOObj->ReadSymtabLoadCommand(LCI, SymtabLoadCmd); |
| 107 | |
| 108 | if (RegisteredStringTable != DRI.d.a) { |
| 109 | MachOObj->RegisterStringTable(*SymtabLoadCmd); |
| 110 | RegisteredStringTable = DRI.d.a; |
| 111 | } |
| 112 | |
| 113 | MachOObj->ReadSymbolTableEntry(SymtabLoadCmd->SymbolTableOffset, DRI.d.b, |
| 114 | Res); |
| 115 | } |
| 116 | |
| 117 | |
| 118 | SymbolRef MachOObjectFile::getSymbolNext(DataRefImpl DRI) const { |
| 119 | DRI.d.b++; |
| 120 | moveToNextSymbol(DRI); |
| 121 | return SymbolRef(DRI, this); |
| 122 | } |
| 123 | |
| 124 | StringRef MachOObjectFile::getSymbolName(DataRefImpl DRI) const { |
| 125 | InMemoryStruct<macho::SymbolTableEntry> Entry; |
| 126 | getSymbolTableEntry(DRI, Entry); |
| 127 | return MachOObj->getStringAtIndex(Entry->StringIndex); |
| 128 | } |
| 129 | |
| 130 | uint64_t MachOObjectFile::getSymbolAddress(DataRefImpl DRI) const { |
| 131 | InMemoryStruct<macho::SymbolTableEntry> Entry; |
| 132 | getSymbolTableEntry(DRI, Entry); |
| 133 | return Entry->Value; |
| 134 | } |
| 135 | |
| 136 | uint64_t MachOObjectFile::getSymbolSize(DataRefImpl DRI) const { |
| 137 | return UnknownAddressOrSize; |
| 138 | } |
| 139 | |
| 140 | char MachOObjectFile::getSymbolNMTypeChar(DataRefImpl DRI) const { |
| 141 | InMemoryStruct<macho::SymbolTableEntry> Entry; |
| 142 | getSymbolTableEntry(DRI, Entry); |
| 143 | |
| 144 | char Char; |
| 145 | switch (Entry->Type & macho::STF_TypeMask) { |
| 146 | case macho::STT_Undefined: |
| 147 | Char = 'u'; |
| 148 | break; |
| 149 | case macho::STT_Absolute: |
| 150 | case macho::STT_Section: |
| 151 | Char = 's'; |
| 152 | break; |
| 153 | default: |
| 154 | Char = '?'; |
| 155 | break; |
| 156 | } |
| 157 | |
| 158 | if (Entry->Flags & (macho::STF_External | macho::STF_PrivateExtern)) |
| 159 | Char = toupper(Char); |
| 160 | return Char; |
| 161 | } |
| 162 | |
| 163 | bool MachOObjectFile::isSymbolInternal(DataRefImpl DRI) const { |
| 164 | InMemoryStruct<macho::SymbolTableEntry> Entry; |
| 165 | getSymbolTableEntry(DRI, Entry); |
| 166 | return Entry->Flags & macho::STF_StabsEntryMask; |
| 167 | } |
| 168 | |
| 169 | ObjectFile::symbol_iterator MachOObjectFile::begin_symbols() const { |
| 170 | // DRI.d.a = segment number; DRI.d.b = symbol index. |
| 171 | DataRefImpl DRI; |
| 172 | DRI.d.a = DRI.d.b = 0; |
| 173 | moveToNextSymbol(DRI); |
| 174 | return symbol_iterator(SymbolRef(DRI, this)); |
| 175 | } |
| 176 | |
| 177 | ObjectFile::symbol_iterator MachOObjectFile::end_symbols() const { |
| 178 | DataRefImpl DRI; |
| 179 | DRI.d.a = MachOObj->getHeader().NumLoadCommands; |
| 180 | DRI.d.b = 0; |
| 181 | return symbol_iterator(SymbolRef(DRI, this)); |
| 182 | } |
| 183 | |
| 184 | |
| 185 | /*===-- Sections ----------------------------------------------------------===*/ |
| 186 | |
| 187 | void MachOObjectFile::moveToNextSection(DataRefImpl &DRI) const { |
| 188 | uint32_t LoadCommandCount = MachOObj->getHeader().NumLoadCommands; |
| 189 | while (DRI.d.a < LoadCommandCount) { |
| 190 | LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a); |
| 191 | if (LCI.Command.Type == macho::LCT_Segment) { |
| 192 | InMemoryStruct<macho::SegmentLoadCommand> SegmentLoadCmd; |
| 193 | MachOObj->ReadSegmentLoadCommand(LCI, SegmentLoadCmd); |
| 194 | if (DRI.d.b < SegmentLoadCmd->NumSections) |
| 195 | return; |
| 196 | } else if (LCI.Command.Type == macho::LCT_Segment64) { |
| 197 | InMemoryStruct<macho::Segment64LoadCommand> Segment64LoadCmd; |
| 198 | MachOObj->ReadSegment64LoadCommand(LCI, Segment64LoadCmd); |
| 199 | if (DRI.d.b < Segment64LoadCmd->NumSections) |
| 200 | return; |
| 201 | } |
| 202 | |
| 203 | DRI.d.a++; |
| 204 | DRI.d.b = 0; |
| 205 | } |
| 206 | } |
| 207 | |
| 208 | SectionRef MachOObjectFile::getSectionNext(DataRefImpl DRI) const { |
| 209 | DRI.d.b++; |
| 210 | moveToNextSection(DRI); |
| 211 | return SectionRef(DRI, this); |
| 212 | } |
| 213 | |
| 214 | void |
| 215 | MachOObjectFile::getSection(DataRefImpl DRI, |
| 216 | InMemoryStruct<macho::Section> &Res) const { |
| 217 | InMemoryStruct<macho::SegmentLoadCommand> SLC; |
| 218 | LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a); |
| 219 | MachOObj->ReadSegmentLoadCommand(LCI, SLC); |
| 220 | MachOObj->ReadSection(LCI, DRI.d.b, Res); |
| 221 | } |
| 222 | |
| 223 | StringRef MachOObjectFile::getSectionName(DataRefImpl DRI) const { |
| 224 | InMemoryStruct<macho::SegmentLoadCommand> SLC; |
| 225 | LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a); |
| 226 | MachOObj->ReadSegmentLoadCommand(LCI, SLC); |
| 227 | InMemoryStruct<macho::Section> Sect; |
| 228 | MachOObj->ReadSection(LCI, DRI.d.b, Sect); |
| 229 | |
| 230 | static char Result[34]; |
| 231 | strcpy(Result, SLC->Name); |
| 232 | strcat(Result, ","); |
| 233 | strcat(Result, Sect->Name); |
| 234 | return StringRef(Result); |
| 235 | } |
| 236 | |
| 237 | uint64_t MachOObjectFile::getSectionAddress(DataRefImpl DRI) const { |
| 238 | InMemoryStruct<macho::Section> Sect; |
| 239 | getSection(DRI, Sect); |
| 240 | return Sect->Address; |
| 241 | } |
| 242 | |
| 243 | uint64_t MachOObjectFile::getSectionSize(DataRefImpl DRI) const { |
| 244 | InMemoryStruct<macho::Section> Sect; |
| 245 | getSection(DRI, Sect); |
| 246 | return Sect->Size; |
| 247 | } |
| 248 | |
| 249 | StringRef MachOObjectFile::getSectionContents(DataRefImpl DRI) const { |
| 250 | InMemoryStruct<macho::Section> Sect; |
| 251 | getSection(DRI, Sect); |
| 252 | return MachOObj->getData(Sect->Offset, Sect->Size); |
| 253 | } |
| 254 | |
| 255 | bool MachOObjectFile::isSectionText(DataRefImpl DRI) const { |
| 256 | InMemoryStruct<macho::SegmentLoadCommand> SLC; |
| 257 | LoadCommandInfo LCI = MachOObj->getLoadCommandInfo(DRI.d.a); |
| 258 | MachOObj->ReadSegmentLoadCommand(LCI, SLC); |
| 259 | return !strcmp(SLC->Name, "__TEXT"); |
| 260 | } |
| 261 | |
| 262 | ObjectFile::section_iterator MachOObjectFile::begin_sections() const { |
| 263 | DataRefImpl DRI; |
| 264 | DRI.d.a = DRI.d.b = 0; |
| 265 | moveToNextSection(DRI); |
| 266 | return section_iterator(SectionRef(DRI, this)); |
| 267 | } |
| 268 | |
| 269 | ObjectFile::section_iterator MachOObjectFile::end_sections() const { |
| 270 | DataRefImpl DRI; |
| 271 | DRI.d.a = MachOObj->getHeader().NumLoadCommands; |
| 272 | DRI.d.b = 0; |
| 273 | return section_iterator(SectionRef(DRI, this)); |
| 274 | } |
| 275 | |
| 276 | /*===-- Miscellaneous -----------------------------------------------------===*/ |
| 277 | |
| 278 | uint8_t MachOObjectFile::getBytesInAddress() const { |
| 279 | return MachOObj->is64Bit() ? 8 : 4; |
| 280 | } |
| 281 | |
| 282 | StringRef MachOObjectFile::getFileFormatName() const { |
| 283 | if (!MachOObj->is64Bit()) { |
| 284 | switch (MachOObj->getHeader().CPUType) { |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 285 | case llvm::MachO::CPUTypeI386: |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 286 | return "Mach-O 32-bit i386"; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 287 | case llvm::MachO::CPUTypeARM: |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 288 | return "Mach-O arm"; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 289 | case llvm::MachO::CPUTypePowerPC: |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 290 | return "Mach-O 32-bit ppc"; |
| 291 | default: |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 292 | assert((MachOObj->getHeader().CPUType & llvm::MachO::CPUArchABI64) == 0 && |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 293 | "64-bit object file when we're not 64-bit?"); |
| 294 | return "Mach-O 32-bit unknown"; |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 295 | } |
| 296 | } |
| 297 | |
| 298 | switch (MachOObj->getHeader().CPUType) { |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 299 | case llvm::MachO::CPUTypeX86_64: |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 300 | return "Mach-O 64-bit x86-64"; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 301 | case llvm::MachO::CPUTypePowerPC64: |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 302 | return "Mach-O 64-bit ppc64"; |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 303 | default: |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 304 | assert((MachOObj->getHeader().CPUType & llvm::MachO::CPUArchABI64) == 1 && |
Eric Christopher | 9ab1d7f | 2011-04-22 04:08:58 +0000 | [diff] [blame] | 305 | "32-bit object file when we're 64-bit?"); |
| 306 | return "Mach-O 64-bit unknown"; |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 307 | } |
| 308 | } |
| 309 | |
| 310 | unsigned MachOObjectFile::getArch() const { |
| 311 | switch (MachOObj->getHeader().CPUType) { |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 312 | case llvm::MachO::CPUTypeI386: |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 313 | return Triple::x86; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 314 | case llvm::MachO::CPUTypeX86_64: |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 315 | return Triple::x86_64; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 316 | case llvm::MachO::CPUTypeARM: |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 317 | return Triple::arm; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 318 | case llvm::MachO::CPUTypePowerPC: |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 319 | return Triple::ppc; |
Eric Christopher | f4b2f93 | 2011-04-22 06:34:01 +0000 | [diff] [blame] | 320 | case llvm::MachO::CPUTypePowerPC64: |
Eric Christopher | 6256b03 | 2011-04-22 03:19:48 +0000 | [diff] [blame] | 321 | return Triple::ppc64; |
| 322 | default: |
| 323 | return Triple::UnknownArch; |
| 324 | } |
| 325 | } |
| 326 | |
| 327 | } // end namespace llvm |
| 328 | |