Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1 | //===- lib/ReaderWriter/MachO/MachONormalizedFileBinaryWriter.cpp ---------===// |
| 2 | // |
| 3 | // The LLVM Linker |
| 4 | // |
| 5 | // This file is distributed under the University of Illinois Open Source |
| 6 | // License. See LICENSE.TXT for details. |
| 7 | // |
| 8 | //===----------------------------------------------------------------------===// |
| 9 | |
| 10 | /// |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 11 | /// \file For mach-o object files, this implementation converts normalized |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 12 | /// mach-o in memory to mach-o binary on disk. |
| 13 | /// |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 14 | /// +---------------+ |
| 15 | /// | binary mach-o | |
| 16 | /// +---------------+ |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 17 | /// ^ |
| 18 | /// | |
| 19 | /// | |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 20 | /// +------------+ |
| 21 | /// | normalized | |
| 22 | /// +------------+ |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 23 | |
| 24 | #include "MachONormalizedFile.h" |
| 25 | #include "MachONormalizedFileBinaryUtils.h" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 26 | #include "lld/Core/Error.h" |
| 27 | #include "lld/Core/LLVM.h" |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 28 | #include "llvm/ADT/ilist.h" |
| 29 | #include "llvm/ADT/ilist_node.h" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 30 | #include "llvm/ADT/SmallString.h" |
| 31 | #include "llvm/ADT/SmallVector.h" |
| 32 | #include "llvm/ADT/StringRef.h" |
| 33 | #include "llvm/Support/Casting.h" |
| 34 | #include "llvm/Support/Debug.h" |
Shankar Easwaran | 2b67fca | 2014-10-18 05:33:55 +0000 | [diff] [blame] | 35 | #include "llvm/Support/Errc.h" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 36 | #include "llvm/Support/ErrorHandling.h" |
| 37 | #include "llvm/Support/FileOutputBuffer.h" |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 38 | #include "llvm/Support/Format.h" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 39 | #include "llvm/Support/Host.h" |
Nick Kledzik | 00a15d9 | 2013-11-09 01:00:51 +0000 | [diff] [blame] | 40 | #include "llvm/Support/LEB128.h" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 41 | #include "llvm/Support/MachO.h" |
| 42 | #include "llvm/Support/MemoryBuffer.h" |
| 43 | #include "llvm/Support/raw_ostream.h" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 44 | #include <functional> |
Nick Kledzik | 07ba512 | 2014-12-02 01:50:44 +0000 | [diff] [blame] | 45 | #include <list> |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 46 | #include <map> |
Rafael Espindola | 54427cc | 2014-06-12 17:15:58 +0000 | [diff] [blame] | 47 | #include <system_error> |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 48 | |
| 49 | using namespace llvm::MachO; |
| 50 | |
| 51 | namespace lld { |
| 52 | namespace mach_o { |
| 53 | namespace normalized { |
| 54 | |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 55 | class ByteBuffer { |
| 56 | public: |
| 57 | ByteBuffer() : _ostream(_bytes) { } |
| 58 | |
| 59 | void append_byte(uint8_t b) { |
| 60 | _ostream << b; |
| 61 | } |
| 62 | void append_uleb128(uint64_t value) { |
| 63 | llvm::encodeULEB128(value, _ostream); |
| 64 | } |
| 65 | void append_uleb128Fixed(uint64_t value, unsigned byteCount) { |
| 66 | unsigned min = llvm::getULEB128Size(value); |
| 67 | assert(min <= byteCount); |
| 68 | unsigned pad = byteCount - min; |
| 69 | llvm::encodeULEB128(value, _ostream, pad); |
| 70 | } |
| 71 | void append_sleb128(int64_t value) { |
| 72 | llvm::encodeSLEB128(value, _ostream); |
| 73 | } |
| 74 | void append_string(StringRef str) { |
| 75 | _ostream << str; |
| 76 | append_byte(0); |
| 77 | } |
| 78 | void align(unsigned alignment) { |
| 79 | while ( (_ostream.tell() % alignment) != 0 ) |
| 80 | append_byte(0); |
| 81 | } |
| 82 | size_t size() { |
| 83 | return _ostream.tell(); |
| 84 | } |
| 85 | const uint8_t *bytes() { |
| 86 | return reinterpret_cast<const uint8_t*>(_ostream.str().data()); |
| 87 | } |
| 88 | |
| 89 | private: |
| 90 | SmallVector<char, 128> _bytes; |
| 91 | // Stream ivar must be after SmallVector ivar to construct properly. |
| 92 | llvm::raw_svector_ostream _ostream; |
| 93 | }; |
| 94 | |
| 95 | struct TrieNode; // Forward declaration. |
| 96 | |
| 97 | struct TrieEdge : public llvm::ilist_node<TrieEdge> { |
| 98 | TrieEdge(StringRef s, TrieNode *node) : _subString(s), _child(node) {} |
| 99 | |
| 100 | StringRef _subString; |
| 101 | struct TrieNode *_child; |
| 102 | }; |
| 103 | |
| 104 | } // namespace normalized |
| 105 | } // namespace mach_o |
| 106 | } // namespace lld |
| 107 | |
| 108 | |
| 109 | namespace llvm { |
| 110 | using lld::mach_o::normalized::TrieEdge; |
| 111 | template <> |
| 112 | struct ilist_traits<TrieEdge> |
| 113 | : public ilist_default_traits<TrieEdge> { |
| 114 | private: |
| 115 | mutable ilist_half_node<TrieEdge> Sentinel; |
| 116 | public: |
| 117 | TrieEdge *createSentinel() const { |
| 118 | return static_cast<TrieEdge*>(&Sentinel); |
| 119 | } |
| 120 | void destroySentinel(TrieEdge *) const {} |
| 121 | |
| 122 | TrieEdge *provideInitialHead() const { return createSentinel(); } |
| 123 | TrieEdge *ensureHead(TrieEdge*) const { return createSentinel(); } |
| 124 | static void noteHead(TrieEdge*, TrieEdge*) {} |
| 125 | void deleteNode(TrieEdge *N) {} |
| 126 | |
| 127 | private: |
| 128 | void createNode(const TrieEdge &); |
| 129 | }; |
| 130 | } // namespace llvm |
| 131 | |
| 132 | |
| 133 | namespace lld { |
| 134 | namespace mach_o { |
| 135 | namespace normalized { |
| 136 | |
| 137 | struct TrieNode { |
| 138 | typedef llvm::ilist<TrieEdge> TrieEdgeList; |
| 139 | |
| 140 | TrieNode(StringRef s) |
| 141 | : _cummulativeString(s), _address(0), _flags(0), _other(0), |
| 142 | _trieOffset(0), _hasExportInfo(false) {} |
| 143 | ~TrieNode() = default; |
| 144 | |
| 145 | void addSymbol(const Export &entry, BumpPtrAllocator &allocator, |
| 146 | std::vector<TrieNode *> &allNodes); |
| 147 | bool updateOffset(uint32_t &offset); |
| 148 | void appendToByteBuffer(ByteBuffer &out); |
| 149 | |
| 150 | private: |
| 151 | StringRef _cummulativeString; |
| 152 | TrieEdgeList _children; |
| 153 | uint64_t _address; |
| 154 | uint64_t _flags; |
| 155 | uint64_t _other; |
| 156 | StringRef _importedName; |
| 157 | uint32_t _trieOffset; |
| 158 | bool _hasExportInfo; |
| 159 | }; |
| 160 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 161 | /// Utility class for writing a mach-o binary file given an in-memory |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 162 | /// normalized file. |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 163 | class MachOFileLayout { |
| 164 | public: |
Joey Gouly | b275d7f | 2013-12-23 23:29:50 +0000 | [diff] [blame] | 165 | /// All layout computation is done in the constructor. |
| 166 | MachOFileLayout(const NormalizedFile &file); |
| 167 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 168 | /// Returns the final file size as computed in the constructor. |
| 169 | size_t size() const; |
| 170 | |
Nick Kledzik | 2fcbe82 | 2014-07-30 00:58:06 +0000 | [diff] [blame] | 171 | // Returns size of the mach_header and load commands. |
| 172 | size_t headerAndLoadCommandsSize() const; |
| 173 | |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 174 | /// Writes the normalized file as a binary mach-o file to the specified |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 175 | /// path. This does not have a stream interface because the generated |
| 176 | /// file may need the 'x' bit set. |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 177 | std::error_code writeBinary(StringRef path); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 178 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 179 | private: |
| 180 | uint32_t loadCommandsSize(uint32_t &count); |
| 181 | void buildFileOffsets(); |
| 182 | void writeMachHeader(); |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 183 | std::error_code writeLoadCommands(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 184 | void writeSectionContent(); |
| 185 | void writeRelocations(); |
| 186 | void writeSymbolTable(); |
| 187 | void writeRebaseInfo(); |
| 188 | void writeBindingInfo(); |
| 189 | void writeLazyBindingInfo(); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 190 | void writeExportInfo(); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 191 | void writeDataInCodeInfo(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 192 | void writeLinkEditContent(); |
| 193 | void buildLinkEditInfo(); |
| 194 | void buildRebaseInfo(); |
| 195 | void buildBindInfo(); |
| 196 | void buildLazyBindInfo(); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 197 | void buildExportTrie(); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 198 | void computeDataInCodeSize(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 199 | void computeSymbolTableSizes(); |
| 200 | void buildSectionRelocations(); |
| 201 | void appendSymbols(const std::vector<Symbol> &symbols, |
| 202 | uint32_t &symOffset, uint32_t &strOffset); |
| 203 | uint32_t indirectSymbolIndex(const Section §, uint32_t &index); |
| 204 | uint32_t indirectSymbolElementSize(const Section §); |
| 205 | |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 206 | // For use as template parameter to load command methods. |
| 207 | struct MachO64Trait { |
| 208 | typedef llvm::MachO::segment_command_64 command; |
| 209 | typedef llvm::MachO::section_64 section; |
| 210 | enum { LC = llvm::MachO::LC_SEGMENT_64 }; |
| 211 | }; |
| 212 | |
| 213 | // For use as template parameter to load command methods. |
| 214 | struct MachO32Trait { |
| 215 | typedef llvm::MachO::segment_command command; |
| 216 | typedef llvm::MachO::section section; |
| 217 | enum { LC = llvm::MachO::LC_SEGMENT }; |
| 218 | }; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 219 | |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 220 | template <typename T> |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 221 | std::error_code writeSingleSegmentLoadCommand(uint8_t *&lc); |
| 222 | template <typename T> std::error_code writeSegmentLoadCommands(uint8_t *&lc); |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 223 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 224 | uint32_t pointerAlign(uint32_t value); |
| 225 | static StringRef dyldPath(); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 226 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 227 | struct SegExtraInfo { |
| 228 | uint32_t fileOffset; |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 229 | uint32_t fileSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 230 | std::vector<const Section*> sections; |
| 231 | }; |
| 232 | typedef std::map<const Segment*, SegExtraInfo> SegMap; |
| 233 | struct SectionExtraInfo { |
| 234 | uint32_t fileOffset; |
| 235 | }; |
| 236 | typedef std::map<const Section*, SectionExtraInfo> SectionMap; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 237 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 238 | const NormalizedFile &_file; |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 239 | std::error_code _ec; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 240 | uint8_t *_buffer; |
| 241 | const bool _is64; |
| 242 | const bool _swap; |
| 243 | const bool _bigEndianArch; |
| 244 | uint64_t _seg1addr; |
| 245 | uint32_t _startOfLoadCommands; |
| 246 | uint32_t _countOfLoadCommands; |
| 247 | uint32_t _endOfLoadCommands; |
| 248 | uint32_t _startOfRelocations; |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 249 | uint32_t _startOfDataInCode; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 250 | uint32_t _startOfSymbols; |
| 251 | uint32_t _startOfIndirectSymbols; |
| 252 | uint32_t _startOfSymbolStrings; |
| 253 | uint32_t _endOfSymbolStrings; |
| 254 | uint32_t _symbolTableLocalsStartIndex; |
| 255 | uint32_t _symbolTableGlobalsStartIndex; |
| 256 | uint32_t _symbolTableUndefinesStartIndex; |
| 257 | uint32_t _symbolStringPoolSize; |
| 258 | uint32_t _symbolTableSize; |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 259 | uint32_t _dataInCodeSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 260 | uint32_t _indirectSymbolTableCount; |
| 261 | // Used in object file creation only |
| 262 | uint32_t _startOfSectionsContent; |
| 263 | uint32_t _endOfSectionsContent; |
| 264 | // Used in final linked image only |
| 265 | uint32_t _startOfLinkEdit; |
| 266 | uint32_t _startOfRebaseInfo; |
| 267 | uint32_t _endOfRebaseInfo; |
| 268 | uint32_t _startOfBindingInfo; |
| 269 | uint32_t _endOfBindingInfo; |
| 270 | uint32_t _startOfLazyBindingInfo; |
| 271 | uint32_t _endOfLazyBindingInfo; |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 272 | uint32_t _startOfExportTrie; |
| 273 | uint32_t _endOfExportTrie; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 274 | uint32_t _endOfLinkEdit; |
| 275 | uint64_t _addressOfLinkEdit; |
| 276 | SegMap _segInfo; |
| 277 | SectionMap _sectInfo; |
| 278 | ByteBuffer _rebaseInfo; |
| 279 | ByteBuffer _bindingInfo; |
| 280 | ByteBuffer _lazyBindingInfo; |
| 281 | ByteBuffer _weakBindingInfo; |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 282 | ByteBuffer _exportTrie; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 283 | }; |
| 284 | |
| 285 | size_t headerAndLoadCommandsSize(const NormalizedFile &file) { |
| 286 | MachOFileLayout layout(file); |
Nick Kledzik | 2fcbe82 | 2014-07-30 00:58:06 +0000 | [diff] [blame] | 287 | return layout.headerAndLoadCommandsSize(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 288 | } |
| 289 | |
| 290 | StringRef MachOFileLayout::dyldPath() { |
| 291 | return "/usr/lib/dyld"; |
| 292 | } |
| 293 | |
| 294 | uint32_t MachOFileLayout::pointerAlign(uint32_t value) { |
Rui Ueyama | 489a806 | 2016-01-14 20:53:50 +0000 | [diff] [blame] | 295 | return llvm::alignTo(value, _is64 ? 8 : 4); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 296 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 297 | |
| 298 | |
Nick Kledzik | 2fcbe82 | 2014-07-30 00:58:06 +0000 | [diff] [blame] | 299 | size_t MachOFileLayout::headerAndLoadCommandsSize() const { |
| 300 | return _endOfLoadCommands; |
| 301 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 302 | |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 303 | MachOFileLayout::MachOFileLayout(const NormalizedFile &file) |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 304 | : _file(file), |
| 305 | _is64(MachOLinkingContext::is64Bit(file.arch)), |
| 306 | _swap(!MachOLinkingContext::isHostEndian(file.arch)), |
| 307 | _bigEndianArch(MachOLinkingContext::isBigEndian(file.arch)), |
| 308 | _seg1addr(INT64_MAX) { |
| 309 | _startOfLoadCommands = _is64 ? sizeof(mach_header_64) : sizeof(mach_header); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 310 | const size_t segCommandBaseSize = |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 311 | (_is64 ? sizeof(segment_command_64) : sizeof(segment_command)); |
| 312 | const size_t sectsSize = (_is64 ? sizeof(section_64) : sizeof(section)); |
| 313 | if (file.fileType == llvm::MachO::MH_OBJECT) { |
| 314 | // object files have just one segment load command containing all sections |
| 315 | _endOfLoadCommands = _startOfLoadCommands |
| 316 | + segCommandBaseSize |
| 317 | + file.sections.size() * sectsSize |
| 318 | + sizeof(symtab_command); |
| 319 | _countOfLoadCommands = 2; |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 320 | if (!_file.dataInCode.empty()) { |
| 321 | _endOfLoadCommands += sizeof(linkedit_data_command); |
| 322 | _countOfLoadCommands++; |
| 323 | } |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 324 | // Assign file offsets to each section. |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 325 | _startOfSectionsContent = _endOfLoadCommands; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 326 | unsigned relocCount = 0; |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 327 | uint64_t offset = _startOfSectionsContent; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 328 | for (const Section § : file.sections) { |
Lang Hames | ac2adce | 2015-12-11 23:25:09 +0000 | [diff] [blame] | 329 | if (isZeroFillSection(sect.type)) |
| 330 | _sectInfo[§].fileOffset = 0; |
| 331 | else { |
Rui Ueyama | 489a806 | 2016-01-14 20:53:50 +0000 | [diff] [blame] | 332 | offset = llvm::alignTo(offset, sect.alignment); |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 333 | _sectInfo[§].fileOffset = offset; |
| 334 | offset += sect.content.size(); |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 335 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 336 | relocCount += sect.relocations.size(); |
| 337 | } |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 338 | _endOfSectionsContent = offset; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 339 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 340 | computeSymbolTableSizes(); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 341 | computeDataInCodeSize(); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 342 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 343 | // Align start of relocations. |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 344 | _startOfRelocations = pointerAlign(_endOfSectionsContent); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 345 | _startOfDataInCode = _startOfRelocations + relocCount * 8; |
| 346 | _startOfSymbols = _startOfDataInCode + _dataInCodeSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 347 | // Add Indirect symbol table. |
| 348 | _startOfIndirectSymbols = _startOfSymbols + _symbolTableSize; |
| 349 | // Align start of symbol table and symbol strings. |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 350 | _startOfSymbolStrings = _startOfIndirectSymbols |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 351 | + pointerAlign(_indirectSymbolTableCount * sizeof(uint32_t)); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 352 | _endOfSymbolStrings = _startOfSymbolStrings |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 353 | + pointerAlign(_symbolStringPoolSize); |
| 354 | _endOfLinkEdit = _endOfSymbolStrings; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 355 | DEBUG_WITH_TYPE("MachOFileLayout", |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 356 | llvm::dbgs() << "MachOFileLayout()\n" |
| 357 | << " startOfLoadCommands=" << _startOfLoadCommands << "\n" |
| 358 | << " countOfLoadCommands=" << _countOfLoadCommands << "\n" |
| 359 | << " endOfLoadCommands=" << _endOfLoadCommands << "\n" |
| 360 | << " startOfRelocations=" << _startOfRelocations << "\n" |
| 361 | << " startOfSymbols=" << _startOfSymbols << "\n" |
| 362 | << " startOfSymbolStrings=" << _startOfSymbolStrings << "\n" |
| 363 | << " endOfSymbolStrings=" << _endOfSymbolStrings << "\n" |
| 364 | << " startOfSectionsContent=" << _startOfSectionsContent << "\n" |
| 365 | << " endOfSectionsContent=" << _endOfSectionsContent << "\n"); |
| 366 | } else { |
| 367 | // Final linked images have one load command per segment. |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 368 | _endOfLoadCommands = _startOfLoadCommands |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 369 | + loadCommandsSize(_countOfLoadCommands); |
| 370 | |
| 371 | // Assign section file offsets. |
| 372 | buildFileOffsets(); |
| 373 | buildLinkEditInfo(); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 374 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 375 | // LINKEDIT of final linked images has in order: |
| 376 | // rebase info, binding info, lazy binding info, weak binding info, |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 377 | // data-in-code, symbol table, indirect symbol table, symbol table strings. |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 378 | _startOfRebaseInfo = _startOfLinkEdit; |
| 379 | _endOfRebaseInfo = _startOfRebaseInfo + _rebaseInfo.size(); |
| 380 | _startOfBindingInfo = _endOfRebaseInfo; |
| 381 | _endOfBindingInfo = _startOfBindingInfo + _bindingInfo.size(); |
| 382 | _startOfLazyBindingInfo = _endOfBindingInfo; |
| 383 | _endOfLazyBindingInfo = _startOfLazyBindingInfo + _lazyBindingInfo.size(); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 384 | _startOfExportTrie = _endOfLazyBindingInfo; |
| 385 | _endOfExportTrie = _startOfExportTrie + _exportTrie.size(); |
| 386 | _startOfDataInCode = _endOfExportTrie; |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 387 | _startOfSymbols = _startOfDataInCode + _dataInCodeSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 388 | _startOfIndirectSymbols = _startOfSymbols + _symbolTableSize; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 389 | _startOfSymbolStrings = _startOfIndirectSymbols |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 390 | + pointerAlign(_indirectSymbolTableCount * sizeof(uint32_t)); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 391 | _endOfSymbolStrings = _startOfSymbolStrings |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 392 | + pointerAlign(_symbolStringPoolSize); |
| 393 | _endOfLinkEdit = _endOfSymbolStrings; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 394 | DEBUG_WITH_TYPE("MachOFileLayout", |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 395 | llvm::dbgs() << "MachOFileLayout()\n" |
| 396 | << " startOfLoadCommands=" << _startOfLoadCommands << "\n" |
| 397 | << " countOfLoadCommands=" << _countOfLoadCommands << "\n" |
| 398 | << " endOfLoadCommands=" << _endOfLoadCommands << "\n" |
| 399 | << " startOfLinkEdit=" << _startOfLinkEdit << "\n" |
| 400 | << " startOfRebaseInfo=" << _startOfRebaseInfo << "\n" |
| 401 | << " endOfRebaseInfo=" << _endOfRebaseInfo << "\n" |
| 402 | << " startOfBindingInfo=" << _startOfBindingInfo << "\n" |
| 403 | << " endOfBindingInfo=" << _endOfBindingInfo << "\n" |
| 404 | << " startOfLazyBindingInfo=" << _startOfLazyBindingInfo << "\n" |
| 405 | << " endOfLazyBindingInfo=" << _endOfLazyBindingInfo << "\n" |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 406 | << " startOfExportTrie=" << _startOfExportTrie << "\n" |
| 407 | << " endOfExportTrie=" << _endOfExportTrie << "\n" |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 408 | << " startOfDataInCode=" << _startOfDataInCode << "\n" |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 409 | << " startOfSymbols=" << _startOfSymbols << "\n" |
| 410 | << " startOfSymbolStrings=" << _startOfSymbolStrings << "\n" |
| 411 | << " endOfSymbolStrings=" << _endOfSymbolStrings << "\n" |
| 412 | << " addressOfLinkEdit=" << _addressOfLinkEdit << "\n"); |
| 413 | } |
| 414 | } |
| 415 | |
| 416 | uint32_t MachOFileLayout::loadCommandsSize(uint32_t &count) { |
| 417 | uint32_t size = 0; |
| 418 | count = 0; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 419 | |
| 420 | const size_t segCommandSize = |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 421 | (_is64 ? sizeof(segment_command_64) : sizeof(segment_command)); |
| 422 | const size_t sectionSize = (_is64 ? sizeof(section_64) : sizeof(section)); |
| 423 | |
| 424 | // Add LC_SEGMENT for each segment. |
| 425 | size += _file.segments.size() * segCommandSize; |
| 426 | count += _file.segments.size(); |
| 427 | // Add section record for each section. |
| 428 | size += _file.sections.size() * sectionSize; |
| 429 | // Add one LC_SEGMENT for implicit __LINKEDIT segment |
| 430 | size += segCommandSize; |
| 431 | ++count; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 432 | |
Tim Northover | 301c4e6 | 2014-07-01 08:15:41 +0000 | [diff] [blame] | 433 | // If creating a dylib, add LC_ID_DYLIB. |
| 434 | if (_file.fileType == llvm::MachO::MH_DYLIB) { |
| 435 | size += sizeof(dylib_command) + pointerAlign(_file.installName.size() + 1); |
| 436 | ++count; |
| 437 | } |
| 438 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 439 | // Add LC_DYLD_INFO |
| 440 | size += sizeof(dyld_info_command); |
| 441 | ++count; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 442 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 443 | // Add LC_SYMTAB |
| 444 | size += sizeof(symtab_command); |
| 445 | ++count; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 446 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 447 | // Add LC_DYSYMTAB |
| 448 | if (_file.fileType != llvm::MachO::MH_PRELOAD) { |
| 449 | size += sizeof(dysymtab_command); |
| 450 | ++count; |
| 451 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 452 | |
Pete Cooper | 354809e | 2016-02-03 22:28:29 +0000 | [diff] [blame] | 453 | // If main executable add LC_LOAD_DYLINKER |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 454 | if (_file.fileType == llvm::MachO::MH_EXECUTE) { |
| 455 | size += pointerAlign(sizeof(dylinker_command) + dyldPath().size()+1); |
| 456 | ++count; |
Pete Cooper | 354809e | 2016-02-03 22:28:29 +0000 | [diff] [blame] | 457 | } |
| 458 | |
| 459 | // Add LC_VERSION_MIN_MACOSX, LC_VERSION_MIN_IPHONEOS, LC_VERSION_MIN_WATCHOS, |
| 460 | // LC_VERSION_MIN_TVOS |
| 461 | if (_file.hasMinVersionLoadCommand) { |
| 462 | size += sizeof(version_min_command); |
| 463 | ++count; |
| 464 | } |
| 465 | |
| 466 | // If main executable add LC_MAIN |
| 467 | if (_file.fileType == llvm::MachO::MH_EXECUTE) { |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 468 | size += sizeof(entry_point_command); |
| 469 | ++count; |
| 470 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 471 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 472 | // Add LC_LOAD_DYLIB for each dependent dylib. |
| 473 | for (const DependentDylib &dep : _file.dependentDylibs) { |
| 474 | size += sizeof(dylib_command) + pointerAlign(dep.path.size()+1); |
| 475 | ++count; |
| 476 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 477 | |
Jean-Daniel Dupas | 23dd15e | 2014-12-18 21:33:38 +0000 | [diff] [blame] | 478 | // Add LC_RPATH |
| 479 | for (const StringRef &path : _file.rpaths) { |
Lang Hames | 2ed3bf9 | 2015-10-29 16:50:26 +0000 | [diff] [blame] | 480 | size += pointerAlign(sizeof(rpath_command) + path.size() + 1); |
Jean-Daniel Dupas | 23dd15e | 2014-12-18 21:33:38 +0000 | [diff] [blame] | 481 | ++count; |
| 482 | } |
| 483 | |
Nick Kledzik | 54ce2958 | 2014-10-28 22:21:10 +0000 | [diff] [blame] | 484 | // Add LC_DATA_IN_CODE if needed |
| 485 | if (!_file.dataInCode.empty()) { |
| 486 | size += sizeof(linkedit_data_command); |
| 487 | ++count; |
| 488 | } |
| 489 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 490 | return size; |
| 491 | } |
| 492 | |
| 493 | static bool overlaps(const Segment &s1, const Segment &s2) { |
| 494 | if (s2.address >= s1.address+s1.size) |
| 495 | return false; |
| 496 | if (s1.address >= s2.address+s2.size) |
| 497 | return false; |
| 498 | return true; |
| 499 | } |
| 500 | |
| 501 | static bool overlaps(const Section &s1, const Section &s2) { |
| 502 | if (s2.address >= s1.address+s1.content.size()) |
| 503 | return false; |
| 504 | if (s1.address >= s2.address+s2.content.size()) |
| 505 | return false; |
| 506 | return true; |
| 507 | } |
| 508 | |
| 509 | void MachOFileLayout::buildFileOffsets() { |
| 510 | // Verify no segments overlap |
| 511 | for (const Segment &sg1 : _file.segments) { |
| 512 | for (const Segment &sg2 : _file.segments) { |
| 513 | if (&sg1 == &sg2) |
| 514 | continue; |
| 515 | if (overlaps(sg1,sg2)) { |
Rafael Espindola | 372bc70 | 2014-06-13 17:20:48 +0000 | [diff] [blame] | 516 | _ec = make_error_code(llvm::errc::executable_format_error); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 517 | return; |
| 518 | } |
| 519 | } |
| 520 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 521 | |
| 522 | // Verify no sections overlap |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 523 | for (const Section &s1 : _file.sections) { |
| 524 | for (const Section &s2 : _file.sections) { |
| 525 | if (&s1 == &s2) |
| 526 | continue; |
| 527 | if (overlaps(s1,s2)) { |
Rafael Espindola | 372bc70 | 2014-06-13 17:20:48 +0000 | [diff] [blame] | 528 | _ec = make_error_code(llvm::errc::executable_format_error); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 529 | return; |
| 530 | } |
| 531 | } |
| 532 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 533 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 534 | // Build side table of extra info about segments and sections. |
| 535 | SegExtraInfo t; |
| 536 | t.fileOffset = 0; |
| 537 | for (const Segment &sg : _file.segments) { |
| 538 | _segInfo[&sg] = t; |
| 539 | } |
| 540 | SectionExtraInfo t2; |
| 541 | t2.fileOffset = 0; |
| 542 | // Assign sections to segments. |
| 543 | for (const Section &s : _file.sections) { |
| 544 | _sectInfo[&s] = t2; |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 545 | bool foundSegment = false; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 546 | for (const Segment &sg : _file.segments) { |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 547 | if (sg.name.equals(s.segmentName)) { |
| 548 | if ((s.address >= sg.address) |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 549 | && (s.address+s.content.size() <= sg.address+sg.size)) { |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 550 | _segInfo[&sg].sections.push_back(&s); |
| 551 | foundSegment = true; |
| 552 | break; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 553 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 554 | } |
| 555 | } |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 556 | if (!foundSegment) { |
| 557 | _ec = make_error_code(llvm::errc::executable_format_error); |
| 558 | return; |
| 559 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 560 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 561 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 562 | // Assign file offsets. |
| 563 | uint32_t fileOffset = 0; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 564 | DEBUG_WITH_TYPE("MachOFileLayout", |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 565 | llvm::dbgs() << "buildFileOffsets()\n"); |
| 566 | for (const Segment &sg : _file.segments) { |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 567 | _segInfo[&sg].fileOffset = fileOffset; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 568 | if ((_seg1addr == INT64_MAX) && sg.access) |
| 569 | _seg1addr = sg.address; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 570 | DEBUG_WITH_TYPE("MachOFileLayout", |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 571 | llvm::dbgs() << " segment=" << sg.name |
| 572 | << ", fileOffset=" << _segInfo[&sg].fileOffset << "\n"); |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 573 | |
| 574 | uint32_t segFileSize = 0; |
Nick Kledzik | 761d654 | 2014-10-24 22:19:22 +0000 | [diff] [blame] | 575 | // A segment that is not zero-fill must use a least one page of disk space. |
| 576 | if (sg.access) |
| 577 | segFileSize = _file.pageSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 578 | for (const Section *s : _segInfo[&sg].sections) { |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 579 | uint32_t sectOffset = s->address - sg.address; |
| 580 | uint32_t sectFileSize = |
Lang Hames | ac2adce | 2015-12-11 23:25:09 +0000 | [diff] [blame] | 581 | isZeroFillSection(s->type) ? 0 : s->content.size(); |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 582 | segFileSize = std::max(segFileSize, sectOffset + sectFileSize); |
| 583 | |
| 584 | _sectInfo[s].fileOffset = _segInfo[&sg].fileOffset + sectOffset; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 585 | DEBUG_WITH_TYPE("MachOFileLayout", |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 586 | llvm::dbgs() << " section=" << s->sectionName |
| 587 | << ", fileOffset=" << fileOffset << "\n"); |
| 588 | } |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 589 | |
Rui Ueyama | 489a806 | 2016-01-14 20:53:50 +0000 | [diff] [blame] | 590 | _segInfo[&sg].fileSize = llvm::alignTo(segFileSize, _file.pageSize); |
| 591 | fileOffset = llvm::alignTo(fileOffset + segFileSize, _file.pageSize); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 592 | _addressOfLinkEdit = sg.address + sg.size; |
| 593 | } |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 594 | _startOfLinkEdit = fileOffset; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 595 | } |
| 596 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 597 | size_t MachOFileLayout::size() const { |
| 598 | return _endOfSymbolStrings; |
| 599 | } |
| 600 | |
| 601 | void MachOFileLayout::writeMachHeader() { |
| 602 | mach_header *mh = reinterpret_cast<mach_header*>(_buffer); |
| 603 | mh->magic = _is64 ? llvm::MachO::MH_MAGIC_64 : llvm::MachO::MH_MAGIC; |
| 604 | mh->cputype = MachOLinkingContext::cpuTypeFromArch(_file.arch); |
| 605 | mh->cpusubtype = MachOLinkingContext::cpuSubtypeFromArch(_file.arch); |
| 606 | mh->filetype = _file.fileType; |
| 607 | mh->ncmds = _countOfLoadCommands; |
| 608 | mh->sizeofcmds = _endOfLoadCommands - _startOfLoadCommands; |
| 609 | mh->flags = _file.flags; |
| 610 | if (_swap) |
| 611 | swapStruct(*mh); |
| 612 | } |
| 613 | |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 614 | uint32_t MachOFileLayout::indirectSymbolIndex(const Section §, |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 615 | uint32_t &index) { |
| 616 | if (sect.indirectSymbols.empty()) |
| 617 | return 0; |
| 618 | uint32_t result = index; |
| 619 | index += sect.indirectSymbols.size(); |
| 620 | return result; |
| 621 | } |
| 622 | |
| 623 | uint32_t MachOFileLayout::indirectSymbolElementSize(const Section §) { |
| 624 | if (sect.indirectSymbols.empty()) |
| 625 | return 0; |
| 626 | if (sect.type != S_SYMBOL_STUBS) |
| 627 | return 0; |
| 628 | return sect.content.size() / sect.indirectSymbols.size(); |
| 629 | } |
| 630 | |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 631 | template <typename T> |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 632 | std::error_code MachOFileLayout::writeSingleSegmentLoadCommand(uint8_t *&lc) { |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 633 | typename T::command* seg = reinterpret_cast<typename T::command*>(lc); |
| 634 | seg->cmd = T::LC; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 635 | seg->cmdsize = sizeof(typename T::command) |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 636 | + _file.sections.size() * sizeof(typename T::section); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 637 | uint8_t *next = lc + seg->cmdsize; |
| 638 | memset(seg->segname, 0, 16); |
| 639 | seg->vmaddr = 0; |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 640 | seg->vmsize = _file.sections.back().address |
| 641 | + _file.sections.back().content.size(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 642 | seg->fileoff = _endOfLoadCommands; |
| 643 | seg->filesize = seg->vmsize; |
| 644 | seg->maxprot = VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE; |
| 645 | seg->initprot = VM_PROT_READ|VM_PROT_WRITE|VM_PROT_EXECUTE; |
| 646 | seg->nsects = _file.sections.size(); |
| 647 | seg->flags = 0; |
| 648 | if (_swap) |
| 649 | swapStruct(*seg); |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 650 | typename T::section *sout = reinterpret_cast<typename T::section*> |
| 651 | (lc+sizeof(typename T::command)); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 652 | uint32_t relOffset = _startOfRelocations; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 653 | uint32_t indirectSymRunningIndex = 0; |
| 654 | for (const Section &sin : _file.sections) { |
| 655 | setString16(sin.sectionName, sout->sectname); |
| 656 | setString16(sin.segmentName, sout->segname); |
| 657 | sout->addr = sin.address; |
| 658 | sout->size = sin.content.size(); |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 659 | sout->offset = _sectInfo[&sin].fileOffset; |
Rui Ueyama | f217ef0 | 2015-03-26 02:03:44 +0000 | [diff] [blame] | 660 | sout->align = llvm::Log2_32(sin.alignment); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 661 | sout->reloff = sin.relocations.empty() ? 0 : relOffset; |
| 662 | sout->nreloc = sin.relocations.size(); |
| 663 | sout->flags = sin.type | sin.attributes; |
| 664 | sout->reserved1 = indirectSymbolIndex(sin, indirectSymRunningIndex); |
| 665 | sout->reserved2 = indirectSymbolElementSize(sin); |
| 666 | relOffset += sin.relocations.size() * sizeof(any_relocation_info); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 667 | if (_swap) |
| 668 | swapStruct(*sout); |
| 669 | ++sout; |
| 670 | } |
| 671 | lc = next; |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 672 | return std::error_code(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 673 | } |
| 674 | |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 675 | template <typename T> |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 676 | std::error_code MachOFileLayout::writeSegmentLoadCommands(uint8_t *&lc) { |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 677 | uint32_t indirectSymRunningIndex = 0; |
| 678 | for (const Segment &seg : _file.segments) { |
| 679 | // Write segment command with trailing sections. |
| 680 | SegExtraInfo &segInfo = _segInfo[&seg]; |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 681 | typename T::command* cmd = reinterpret_cast<typename T::command*>(lc); |
| 682 | cmd->cmd = T::LC; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 683 | cmd->cmdsize = sizeof(typename T::command) |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 684 | + segInfo.sections.size() * sizeof(typename T::section); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 685 | uint8_t *next = lc + cmd->cmdsize; |
| 686 | setString16(seg.name, cmd->segname); |
| 687 | cmd->vmaddr = seg.address; |
| 688 | cmd->vmsize = seg.size; |
| 689 | cmd->fileoff = segInfo.fileOffset; |
Tim Northover | 08d6a7b | 2014-06-30 09:49:30 +0000 | [diff] [blame] | 690 | cmd->filesize = segInfo.fileSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 691 | cmd->maxprot = seg.access; |
| 692 | cmd->initprot = seg.access; |
| 693 | cmd->nsects = segInfo.sections.size(); |
| 694 | cmd->flags = 0; |
| 695 | if (_swap) |
| 696 | swapStruct(*cmd); |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 697 | typename T::section *sect = reinterpret_cast<typename T::section*> |
| 698 | (lc+sizeof(typename T::command)); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 699 | for (const Section *section : segInfo.sections) { |
| 700 | setString16(section->sectionName, sect->sectname); |
| 701 | setString16(section->segmentName, sect->segname); |
| 702 | sect->addr = section->address; |
| 703 | sect->size = section->content.size(); |
Lang Hames | ac2adce | 2015-12-11 23:25:09 +0000 | [diff] [blame] | 704 | if (isZeroFillSection(section->type)) |
Nick Kledzik | b072c36 | 2014-11-18 00:30:29 +0000 | [diff] [blame] | 705 | sect->offset = 0; |
| 706 | else |
| 707 | sect->offset = section->address - seg.address + segInfo.fileOffset; |
Rui Ueyama | f217ef0 | 2015-03-26 02:03:44 +0000 | [diff] [blame] | 708 | sect->align = llvm::Log2_32(section->alignment); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 709 | sect->reloff = 0; |
| 710 | sect->nreloc = 0; |
| 711 | sect->flags = section->type | section->attributes; |
| 712 | sect->reserved1 = indirectSymbolIndex(*section, indirectSymRunningIndex); |
| 713 | sect->reserved2 = indirectSymbolElementSize(*section); |
| 714 | if (_swap) |
| 715 | swapStruct(*sect); |
| 716 | ++sect; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 717 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 718 | lc = reinterpret_cast<uint8_t*>(next); |
| 719 | } |
| 720 | // Add implicit __LINKEDIT segment |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 721 | size_t linkeditSize = _endOfLinkEdit - _startOfLinkEdit; |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 722 | typename T::command* cmd = reinterpret_cast<typename T::command*>(lc); |
| 723 | cmd->cmd = T::LC; |
| 724 | cmd->cmdsize = sizeof(typename T::command); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 725 | uint8_t *next = lc + cmd->cmdsize; |
| 726 | setString16("__LINKEDIT", cmd->segname); |
| 727 | cmd->vmaddr = _addressOfLinkEdit; |
Rui Ueyama | 489a806 | 2016-01-14 20:53:50 +0000 | [diff] [blame] | 728 | cmd->vmsize = llvm::alignTo(linkeditSize, _file.pageSize); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 729 | cmd->fileoff = _startOfLinkEdit; |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 730 | cmd->filesize = linkeditSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 731 | cmd->maxprot = VM_PROT_READ; |
| 732 | cmd->initprot = VM_PROT_READ; |
| 733 | cmd->nsects = 0; |
| 734 | cmd->flags = 0; |
| 735 | if (_swap) |
| 736 | swapStruct(*cmd); |
| 737 | lc = next; |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 738 | return std::error_code(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 739 | } |
| 740 | |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 741 | std::error_code MachOFileLayout::writeLoadCommands() { |
| 742 | std::error_code ec; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 743 | uint8_t *lc = &_buffer[_startOfLoadCommands]; |
| 744 | if (_file.fileType == llvm::MachO::MH_OBJECT) { |
| 745 | // Object files have one unnamed segment which holds all sections. |
| 746 | if (_is64) |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 747 | ec = writeSingleSegmentLoadCommand<MachO64Trait>(lc); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 748 | else |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 749 | ec = writeSingleSegmentLoadCommand<MachO32Trait>(lc); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 750 | // Add LC_SYMTAB with symbol table info |
| 751 | symtab_command* st = reinterpret_cast<symtab_command*>(lc); |
| 752 | st->cmd = LC_SYMTAB; |
| 753 | st->cmdsize = sizeof(symtab_command); |
| 754 | st->symoff = _startOfSymbols; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 755 | st->nsyms = _file.localSymbols.size() + _file.globalSymbols.size() |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 756 | + _file.undefinedSymbols.size(); |
| 757 | st->stroff = _startOfSymbolStrings; |
| 758 | st->strsize = _endOfSymbolStrings - _startOfSymbolStrings; |
| 759 | if (_swap) |
| 760 | swapStruct(*st); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 761 | lc += sizeof(symtab_command); |
| 762 | // Add LC_DATA_IN_CODE if needed. |
| 763 | if (_dataInCodeSize != 0) { |
| 764 | linkedit_data_command* dl = reinterpret_cast<linkedit_data_command*>(lc); |
| 765 | dl->cmd = LC_DATA_IN_CODE; |
| 766 | dl->cmdsize = sizeof(linkedit_data_command); |
| 767 | dl->dataoff = _startOfDataInCode; |
| 768 | dl->datasize = _dataInCodeSize; |
| 769 | if (_swap) |
| 770 | swapStruct(*dl); |
| 771 | lc += sizeof(linkedit_data_command); |
| 772 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 773 | } else { |
| 774 | // Final linked images have sections under segments. |
| 775 | if (_is64) |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 776 | ec = writeSegmentLoadCommands<MachO64Trait>(lc); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 777 | else |
Nick Kledzik | 29f749e | 2013-11-09 00:07:28 +0000 | [diff] [blame] | 778 | ec = writeSegmentLoadCommands<MachO32Trait>(lc); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 779 | |
Tim Northover | 301c4e6 | 2014-07-01 08:15:41 +0000 | [diff] [blame] | 780 | // Add LC_ID_DYLIB command for dynamic libraries. |
| 781 | if (_file.fileType == llvm::MachO::MH_DYLIB) { |
| 782 | dylib_command *dc = reinterpret_cast<dylib_command*>(lc); |
| 783 | StringRef path = _file.installName; |
| 784 | uint32_t size = sizeof(dylib_command) + pointerAlign(path.size() + 1); |
| 785 | dc->cmd = LC_ID_DYLIB; |
| 786 | dc->cmdsize = size; |
| 787 | dc->dylib.name = sizeof(dylib_command); // offset |
Jean-Daniel Dupas | edefccc | 2014-12-20 09:22:56 +0000 | [diff] [blame] | 788 | // needs to be some constant value different than the one in LC_LOAD_DYLIB |
| 789 | dc->dylib.timestamp = 1; |
Nick Kledzik | 5b9e48b | 2014-11-19 02:21:53 +0000 | [diff] [blame] | 790 | dc->dylib.current_version = _file.currentVersion; |
| 791 | dc->dylib.compatibility_version = _file.compatVersion; |
Tim Northover | 301c4e6 | 2014-07-01 08:15:41 +0000 | [diff] [blame] | 792 | if (_swap) |
| 793 | swapStruct(*dc); |
| 794 | memcpy(lc + sizeof(dylib_command), path.begin(), path.size()); |
| 795 | lc[sizeof(dylib_command) + path.size()] = '\0'; |
| 796 | lc += size; |
| 797 | } |
| 798 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 799 | // Add LC_DYLD_INFO_ONLY. |
| 800 | dyld_info_command* di = reinterpret_cast<dyld_info_command*>(lc); |
| 801 | di->cmd = LC_DYLD_INFO_ONLY; |
| 802 | di->cmdsize = sizeof(dyld_info_command); |
| 803 | di->rebase_off = _rebaseInfo.size() ? _startOfRebaseInfo : 0; |
| 804 | di->rebase_size = _rebaseInfo.size(); |
| 805 | di->bind_off = _bindingInfo.size() ? _startOfBindingInfo : 0; |
| 806 | di->bind_size = _bindingInfo.size(); |
| 807 | di->weak_bind_off = 0; |
| 808 | di->weak_bind_size = 0; |
| 809 | di->lazy_bind_off = _lazyBindingInfo.size() ? _startOfLazyBindingInfo : 0; |
| 810 | di->lazy_bind_size = _lazyBindingInfo.size(); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 811 | di->export_off = _exportTrie.size() ? _startOfExportTrie : 0; |
| 812 | di->export_size = _exportTrie.size(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 813 | if (_swap) |
| 814 | swapStruct(*di); |
| 815 | lc += sizeof(dyld_info_command); |
| 816 | |
| 817 | // Add LC_SYMTAB with symbol table info. |
| 818 | symtab_command* st = reinterpret_cast<symtab_command*>(lc); |
| 819 | st->cmd = LC_SYMTAB; |
| 820 | st->cmdsize = sizeof(symtab_command); |
| 821 | st->symoff = _startOfSymbols; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 822 | st->nsyms = _file.localSymbols.size() + _file.globalSymbols.size() |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 823 | + _file.undefinedSymbols.size(); |
| 824 | st->stroff = _startOfSymbolStrings; |
| 825 | st->strsize = _endOfSymbolStrings - _startOfSymbolStrings; |
| 826 | if (_swap) |
| 827 | swapStruct(*st); |
| 828 | lc += sizeof(symtab_command); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 829 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 830 | // Add LC_DYSYMTAB |
| 831 | if (_file.fileType != llvm::MachO::MH_PRELOAD) { |
| 832 | dysymtab_command* dst = reinterpret_cast<dysymtab_command*>(lc); |
| 833 | dst->cmd = LC_DYSYMTAB; |
| 834 | dst->cmdsize = sizeof(dysymtab_command); |
| 835 | dst->ilocalsym = _symbolTableLocalsStartIndex; |
| 836 | dst->nlocalsym = _file.localSymbols.size(); |
| 837 | dst->iextdefsym = _symbolTableGlobalsStartIndex; |
| 838 | dst->nextdefsym = _file.globalSymbols.size(); |
| 839 | dst->iundefsym = _symbolTableUndefinesStartIndex; |
| 840 | dst->nundefsym = _file.undefinedSymbols.size(); |
| 841 | dst->tocoff = 0; |
| 842 | dst->ntoc = 0; |
| 843 | dst->modtaboff = 0; |
| 844 | dst->nmodtab = 0; |
| 845 | dst->extrefsymoff = 0; |
| 846 | dst->nextrefsyms = 0; |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 847 | dst->indirectsymoff = _startOfIndirectSymbols; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 848 | dst->nindirectsyms = _indirectSymbolTableCount; |
| 849 | dst->extreloff = 0; |
| 850 | dst->nextrel = 0; |
| 851 | dst->locreloff = 0; |
| 852 | dst->nlocrel = 0; |
| 853 | if (_swap) |
| 854 | swapStruct(*dst); |
| 855 | lc += sizeof(dysymtab_command); |
| 856 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 857 | |
Pete Cooper | 354809e | 2016-02-03 22:28:29 +0000 | [diff] [blame] | 858 | // If main executable, add LC_LOAD_DYLINKER |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 859 | if (_file.fileType == llvm::MachO::MH_EXECUTE) { |
| 860 | // Build LC_LOAD_DYLINKER load command. |
| 861 | uint32_t size=pointerAlign(sizeof(dylinker_command)+dyldPath().size()+1); |
| 862 | dylinker_command* dl = reinterpret_cast<dylinker_command*>(lc); |
| 863 | dl->cmd = LC_LOAD_DYLINKER; |
| 864 | dl->cmdsize = size; |
| 865 | dl->name = sizeof(dylinker_command); // offset |
| 866 | if (_swap) |
| 867 | swapStruct(*dl); |
| 868 | memcpy(lc+sizeof(dylinker_command), dyldPath().data(), dyldPath().size()); |
| 869 | lc[sizeof(dylinker_command)+dyldPath().size()] = '\0'; |
| 870 | lc += size; |
Pete Cooper | 354809e | 2016-02-03 22:28:29 +0000 | [diff] [blame] | 871 | } |
| 872 | |
| 873 | // Add LC_VERSION_MIN_MACOSX, LC_VERSION_MIN_IPHONEOS, LC_VERSION_MIN_WATCHOS, |
| 874 | // LC_VERSION_MIN_TVOS |
| 875 | if (_file.hasMinVersionLoadCommand) { |
| 876 | version_min_command *vm = reinterpret_cast<version_min_command*>(lc); |
| 877 | switch (_file.os) { |
| 878 | case MachOLinkingContext::OS::unknown: |
| 879 | // TODO: We need to emit the load command if we managed to derive |
| 880 | // a platform from one of the files we are linking. |
| 881 | llvm_unreachable("Version commands for unknown OS aren't supported"); |
| 882 | break; |
| 883 | case MachOLinkingContext::OS::macOSX: |
| 884 | vm->cmd = LC_VERSION_MIN_MACOSX; |
| 885 | vm->cmdsize = sizeof(version_min_command); |
| 886 | vm->version = _file.minOSverson; |
| 887 | vm->sdk = _file.sdkVersion; |
| 888 | break; |
| 889 | case MachOLinkingContext::OS::iOS: |
| 890 | case MachOLinkingContext::OS::iOS_simulator: |
| 891 | vm->cmd = LC_VERSION_MIN_MACOSX; |
| 892 | vm->cmdsize = sizeof(version_min_command); |
| 893 | vm->version = _file.minOSverson; |
| 894 | vm->sdk = _file.sdkVersion; |
| 895 | break; |
| 896 | } |
| 897 | if (_swap) |
| 898 | swapStruct(*vm); |
| 899 | lc += sizeof(version_min_command); |
| 900 | } |
| 901 | |
| 902 | // If main executable, add LC_MAIN. |
| 903 | if (_file.fileType == llvm::MachO::MH_EXECUTE) { |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 904 | // Build LC_MAIN load command. |
| 905 | entry_point_command* ep = reinterpret_cast<entry_point_command*>(lc); |
| 906 | ep->cmd = LC_MAIN; |
| 907 | ep->cmdsize = sizeof(entry_point_command); |
| 908 | ep->entryoff = _file.entryAddress - _seg1addr; |
Lang Hames | 65a64c9 | 2015-05-20 22:10:50 +0000 | [diff] [blame] | 909 | ep->stacksize = _file.stackSize; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 910 | if (_swap) |
| 911 | swapStruct(*ep); |
| 912 | lc += sizeof(entry_point_command); |
| 913 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 914 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 915 | // Add LC_LOAD_DYLIB commands |
| 916 | for (const DependentDylib &dep : _file.dependentDylibs) { |
| 917 | dylib_command* dc = reinterpret_cast<dylib_command*>(lc); |
| 918 | uint32_t size = sizeof(dylib_command) + pointerAlign(dep.path.size()+1); |
Nick Kledzik | 5172067 | 2014-10-16 19:31:28 +0000 | [diff] [blame] | 919 | dc->cmd = dep.kind; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 920 | dc->cmdsize = size; |
| 921 | dc->dylib.name = sizeof(dylib_command); // offset |
Jean-Daniel Dupas | edefccc | 2014-12-20 09:22:56 +0000 | [diff] [blame] | 922 | // needs to be some constant value different than the one in LC_ID_DYLIB |
Nick Kledzik | 5b9e48b | 2014-11-19 02:21:53 +0000 | [diff] [blame] | 923 | dc->dylib.timestamp = 2; |
| 924 | dc->dylib.current_version = dep.currentVersion; |
| 925 | dc->dylib.compatibility_version = dep.compatVersion; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 926 | if (_swap) |
| 927 | swapStruct(*dc); |
| 928 | memcpy(lc+sizeof(dylib_command), dep.path.begin(), dep.path.size()); |
| 929 | lc[sizeof(dylib_command)+dep.path.size()] = '\0'; |
| 930 | lc += size; |
| 931 | } |
Jean-Daniel Dupas | 23dd15e | 2014-12-18 21:33:38 +0000 | [diff] [blame] | 932 | |
| 933 | // Add LC_RPATH |
| 934 | for (const StringRef &path : _file.rpaths) { |
| 935 | rpath_command *rpc = reinterpret_cast<rpath_command *>(lc); |
Lang Hames | 2ed3bf9 | 2015-10-29 16:50:26 +0000 | [diff] [blame] | 936 | uint32_t size = pointerAlign(sizeof(rpath_command) + path.size() + 1); |
Jean-Daniel Dupas | 23dd15e | 2014-12-18 21:33:38 +0000 | [diff] [blame] | 937 | rpc->cmd = LC_RPATH; |
| 938 | rpc->cmdsize = size; |
| 939 | rpc->path = sizeof(rpath_command); // offset |
| 940 | if (_swap) |
| 941 | swapStruct(*rpc); |
| 942 | memcpy(lc+sizeof(rpath_command), path.begin(), path.size()); |
| 943 | lc[sizeof(rpath_command)+path.size()] = '\0'; |
| 944 | lc += size; |
| 945 | } |
| 946 | |
Nick Kledzik | 54ce2958 | 2014-10-28 22:21:10 +0000 | [diff] [blame] | 947 | // Add LC_DATA_IN_CODE if needed. |
| 948 | if (_dataInCodeSize != 0) { |
| 949 | linkedit_data_command* dl = reinterpret_cast<linkedit_data_command*>(lc); |
| 950 | dl->cmd = LC_DATA_IN_CODE; |
| 951 | dl->cmdsize = sizeof(linkedit_data_command); |
| 952 | dl->dataoff = _startOfDataInCode; |
| 953 | dl->datasize = _dataInCodeSize; |
| 954 | if (_swap) |
| 955 | swapStruct(*dl); |
| 956 | lc += sizeof(linkedit_data_command); |
| 957 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 958 | } |
| 959 | return ec; |
| 960 | } |
| 961 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 962 | void MachOFileLayout::writeSectionContent() { |
| 963 | for (const Section &s : _file.sections) { |
| 964 | // Copy all section content to output buffer. |
Lang Hames | ac2adce | 2015-12-11 23:25:09 +0000 | [diff] [blame] | 965 | if (isZeroFillSection(s.type)) |
Nick Kledzik | 61fdef6 | 2014-05-15 20:59:23 +0000 | [diff] [blame] | 966 | continue; |
Nick Kledzik | 1bebb28 | 2014-09-09 23:52:59 +0000 | [diff] [blame] | 967 | if (s.content.empty()) |
| 968 | continue; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 969 | uint32_t offset = _sectInfo[&s].fileOffset; |
| 970 | uint8_t *p = &_buffer[offset]; |
| 971 | memcpy(p, &s.content[0], s.content.size()); |
| 972 | p += s.content.size(); |
| 973 | } |
| 974 | } |
| 975 | |
| 976 | void MachOFileLayout::writeRelocations() { |
| 977 | uint32_t relOffset = _startOfRelocations; |
| 978 | for (Section sect : _file.sections) { |
| 979 | for (Relocation r : sect.relocations) { |
| 980 | any_relocation_info* rb = reinterpret_cast<any_relocation_info*>( |
| 981 | &_buffer[relOffset]); |
| 982 | *rb = packRelocation(r, _swap, _bigEndianArch); |
| 983 | relOffset += sizeof(any_relocation_info); |
| 984 | } |
| 985 | } |
| 986 | } |
| 987 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 988 | void MachOFileLayout::appendSymbols(const std::vector<Symbol> &symbols, |
| 989 | uint32_t &symOffset, uint32_t &strOffset) { |
| 990 | for (const Symbol &sym : symbols) { |
| 991 | if (_is64) { |
| 992 | nlist_64* nb = reinterpret_cast<nlist_64*>(&_buffer[symOffset]); |
| 993 | nb->n_strx = strOffset - _startOfSymbolStrings; |
| 994 | nb->n_type = sym.type | sym.scope; |
| 995 | nb->n_sect = sym.sect; |
| 996 | nb->n_desc = sym.desc; |
| 997 | nb->n_value = sym.value; |
| 998 | if (_swap) |
| 999 | swapStruct(*nb); |
| 1000 | symOffset += sizeof(nlist_64); |
| 1001 | } else { |
| 1002 | nlist* nb = reinterpret_cast<nlist*>(&_buffer[symOffset]); |
| 1003 | nb->n_strx = strOffset - _startOfSymbolStrings; |
| 1004 | nb->n_type = sym.type | sym.scope; |
| 1005 | nb->n_sect = sym.sect; |
| 1006 | nb->n_desc = sym.desc; |
| 1007 | nb->n_value = sym.value; |
| 1008 | if (_swap) |
| 1009 | swapStruct(*nb); |
| 1010 | symOffset += sizeof(nlist); |
| 1011 | } |
| 1012 | memcpy(&_buffer[strOffset], sym.name.begin(), sym.name.size()); |
| 1013 | strOffset += sym.name.size(); |
| 1014 | _buffer[strOffset++] ='\0'; // Strings in table have nul terminator. |
| 1015 | } |
| 1016 | } |
| 1017 | |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 1018 | void MachOFileLayout::writeDataInCodeInfo() { |
| 1019 | uint32_t offset = _startOfDataInCode; |
| 1020 | for (const DataInCode &entry : _file.dataInCode) { |
| 1021 | data_in_code_entry *dst = reinterpret_cast<data_in_code_entry*>( |
| 1022 | &_buffer[offset]); |
| 1023 | dst->offset = entry.offset; |
| 1024 | dst->length = entry.length; |
| 1025 | dst->kind = entry.kind; |
| 1026 | if (_swap) |
| 1027 | swapStruct(*dst); |
| 1028 | offset += sizeof(data_in_code_entry); |
| 1029 | } |
| 1030 | } |
| 1031 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1032 | void MachOFileLayout::writeSymbolTable() { |
| 1033 | // Write symbol table and symbol strings in parallel. |
| 1034 | uint32_t symOffset = _startOfSymbols; |
| 1035 | uint32_t strOffset = _startOfSymbolStrings; |
| 1036 | _buffer[strOffset++] = '\0'; // Reserve n_strx offset of zero to mean no name. |
| 1037 | appendSymbols(_file.localSymbols, symOffset, strOffset); |
| 1038 | appendSymbols(_file.globalSymbols, symOffset, strOffset); |
| 1039 | appendSymbols(_file.undefinedSymbols, symOffset, strOffset); |
| 1040 | // Write indirect symbol table array. |
| 1041 | uint32_t *indirects = reinterpret_cast<uint32_t*> |
| 1042 | (&_buffer[_startOfIndirectSymbols]); |
| 1043 | if (_file.fileType == llvm::MachO::MH_OBJECT) { |
| 1044 | // Object files have sections in same order as input normalized file. |
| 1045 | for (const Section §ion : _file.sections) { |
| 1046 | for (uint32_t index : section.indirectSymbols) { |
| 1047 | if (_swap) |
Artyom Skrobov | 17587fb | 2014-06-14 12:40:04 +0000 | [diff] [blame] | 1048 | *indirects++ = llvm::sys::getSwappedBytes(index); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1049 | else |
| 1050 | *indirects++ = index; |
| 1051 | } |
| 1052 | } |
| 1053 | } else { |
| 1054 | // Final linked images must sort sections from normalized file. |
| 1055 | for (const Segment &seg : _file.segments) { |
| 1056 | SegExtraInfo &segInfo = _segInfo[&seg]; |
| 1057 | for (const Section *section : segInfo.sections) { |
| 1058 | for (uint32_t index : section->indirectSymbols) { |
| 1059 | if (_swap) |
Artyom Skrobov | 17587fb | 2014-06-14 12:40:04 +0000 | [diff] [blame] | 1060 | *indirects++ = llvm::sys::getSwappedBytes(index); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1061 | else |
| 1062 | *indirects++ = index; |
| 1063 | } |
| 1064 | } |
| 1065 | } |
| 1066 | } |
| 1067 | } |
| 1068 | |
| 1069 | void MachOFileLayout::writeRebaseInfo() { |
| 1070 | memcpy(&_buffer[_startOfRebaseInfo], _rebaseInfo.bytes(), _rebaseInfo.size()); |
| 1071 | } |
| 1072 | |
| 1073 | void MachOFileLayout::writeBindingInfo() { |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1074 | memcpy(&_buffer[_startOfBindingInfo], |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1075 | _bindingInfo.bytes(), _bindingInfo.size()); |
| 1076 | } |
| 1077 | |
| 1078 | void MachOFileLayout::writeLazyBindingInfo() { |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1079 | memcpy(&_buffer[_startOfLazyBindingInfo], |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1080 | _lazyBindingInfo.bytes(), _lazyBindingInfo.size()); |
| 1081 | } |
| 1082 | |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1083 | void MachOFileLayout::writeExportInfo() { |
| 1084 | memcpy(&_buffer[_startOfExportTrie], _exportTrie.bytes(), _exportTrie.size()); |
| 1085 | } |
| 1086 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1087 | void MachOFileLayout::buildLinkEditInfo() { |
| 1088 | buildRebaseInfo(); |
| 1089 | buildBindInfo(); |
| 1090 | buildLazyBindInfo(); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1091 | buildExportTrie(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1092 | computeSymbolTableSizes(); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 1093 | computeDataInCodeSize(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1094 | } |
| 1095 | |
| 1096 | void MachOFileLayout::buildSectionRelocations() { |
| 1097 | |
| 1098 | } |
| 1099 | |
| 1100 | void MachOFileLayout::buildRebaseInfo() { |
| 1101 | // TODO: compress rebasing info. |
| 1102 | for (const RebaseLocation& entry : _file.rebasingInfo) { |
| 1103 | _rebaseInfo.append_byte(REBASE_OPCODE_SET_TYPE_IMM | entry.kind); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1104 | _rebaseInfo.append_byte(REBASE_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1105 | | entry.segIndex); |
| 1106 | _rebaseInfo.append_uleb128(entry.segOffset); |
| 1107 | _rebaseInfo.append_uleb128(REBASE_OPCODE_DO_REBASE_IMM_TIMES | 1); |
| 1108 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1109 | _rebaseInfo.append_byte(REBASE_OPCODE_DONE); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1110 | _rebaseInfo.align(_is64 ? 8 : 4); |
| 1111 | } |
| 1112 | |
| 1113 | void MachOFileLayout::buildBindInfo() { |
| 1114 | // TODO: compress bind info. |
Nick Kledzik | f373c77 | 2014-11-11 01:31:18 +0000 | [diff] [blame] | 1115 | uint64_t lastAddend = 0; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1116 | for (const BindLocation& entry : _file.bindingInfo) { |
| 1117 | _bindingInfo.append_byte(BIND_OPCODE_SET_TYPE_IMM | entry.kind); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1118 | _bindingInfo.append_byte(BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1119 | | entry.segIndex); |
| 1120 | _bindingInfo.append_uleb128(entry.segOffset); |
Lang Hames | 5c69200 | 2015-09-28 20:25:14 +0000 | [diff] [blame] | 1121 | if (entry.ordinal > 0) |
| 1122 | _bindingInfo.append_byte(BIND_OPCODE_SET_DYLIB_ORDINAL_IMM | |
| 1123 | (entry.ordinal & 0xF)); |
| 1124 | else |
| 1125 | _bindingInfo.append_byte(BIND_OPCODE_SET_DYLIB_SPECIAL_IMM | |
| 1126 | (entry.ordinal & 0xF)); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1127 | _bindingInfo.append_byte(BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM); |
| 1128 | _bindingInfo.append_string(entry.symbolName); |
Nick Kledzik | f373c77 | 2014-11-11 01:31:18 +0000 | [diff] [blame] | 1129 | if (entry.addend != lastAddend) { |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1130 | _bindingInfo.append_byte(BIND_OPCODE_SET_ADDEND_SLEB); |
| 1131 | _bindingInfo.append_sleb128(entry.addend); |
Nick Kledzik | f373c77 | 2014-11-11 01:31:18 +0000 | [diff] [blame] | 1132 | lastAddend = entry.addend; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1133 | } |
| 1134 | _bindingInfo.append_byte(BIND_OPCODE_DO_BIND); |
| 1135 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1136 | _bindingInfo.append_byte(BIND_OPCODE_DONE); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1137 | _bindingInfo.align(_is64 ? 8 : 4); |
| 1138 | } |
| 1139 | |
| 1140 | void MachOFileLayout::buildLazyBindInfo() { |
| 1141 | for (const BindLocation& entry : _file.lazyBindingInfo) { |
| 1142 | _lazyBindingInfo.append_byte(BIND_OPCODE_SET_TYPE_IMM | entry.kind); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1143 | _lazyBindingInfo.append_byte(BIND_OPCODE_SET_SEGMENT_AND_OFFSET_ULEB |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1144 | | entry.segIndex); |
Nick Kledzik | f373c77 | 2014-11-11 01:31:18 +0000 | [diff] [blame] | 1145 | _lazyBindingInfo.append_uleb128Fixed(entry.segOffset, 5); |
Lang Hames | 5c69200 | 2015-09-28 20:25:14 +0000 | [diff] [blame] | 1146 | if (entry.ordinal > 0) |
| 1147 | _lazyBindingInfo.append_byte(BIND_OPCODE_SET_DYLIB_ORDINAL_IMM | |
| 1148 | (entry.ordinal & 0xF)); |
| 1149 | else |
| 1150 | _lazyBindingInfo.append_byte(BIND_OPCODE_SET_DYLIB_SPECIAL_IMM | |
| 1151 | (entry.ordinal & 0xF)); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1152 | _lazyBindingInfo.append_byte(BIND_OPCODE_SET_SYMBOL_TRAILING_FLAGS_IMM); |
| 1153 | _lazyBindingInfo.append_string(entry.symbolName); |
| 1154 | _lazyBindingInfo.append_byte(BIND_OPCODE_DO_BIND); |
Nick Kledzik | f373c77 | 2014-11-11 01:31:18 +0000 | [diff] [blame] | 1155 | _lazyBindingInfo.append_byte(BIND_OPCODE_DONE); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1156 | } |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1157 | _lazyBindingInfo.append_byte(BIND_OPCODE_DONE); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1158 | _lazyBindingInfo.align(_is64 ? 8 : 4); |
| 1159 | } |
| 1160 | |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 1161 | void TrieNode::addSymbol(const Export& entry, |
| 1162 | BumpPtrAllocator &allocator, |
| 1163 | std::vector<TrieNode*> &allNodes) { |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1164 | StringRef partialStr = entry.name.drop_front(_cummulativeString.size()); |
| 1165 | for (TrieEdge &edge : _children) { |
| 1166 | StringRef edgeStr = edge._subString; |
| 1167 | if (partialStr.startswith(edgeStr)) { |
| 1168 | // Already have matching edge, go down that path. |
| 1169 | edge._child->addSymbol(entry, allocator, allNodes); |
| 1170 | return; |
| 1171 | } |
| 1172 | // See if string has commmon prefix with existing edge. |
| 1173 | for (int n=edgeStr.size()-1; n > 0; --n) { |
| 1174 | if (partialStr.substr(0, n).equals(edgeStr.substr(0, n))) { |
| 1175 | // Splice in new node: was A -> C, now A -> B -> C |
| 1176 | StringRef bNodeStr = edge._child->_cummulativeString; |
| 1177 | bNodeStr = bNodeStr.drop_back(edgeStr.size()-n).copy(allocator); |
Eugene Zelenko | 4154794 | 2015-11-10 22:37:38 +0000 | [diff] [blame] | 1178 | auto *bNode = new (allocator) TrieNode(bNodeStr); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1179 | allNodes.push_back(bNode); |
| 1180 | TrieNode* cNode = edge._child; |
| 1181 | StringRef abEdgeStr = edgeStr.substr(0,n).copy(allocator); |
| 1182 | StringRef bcEdgeStr = edgeStr.substr(n).copy(allocator); |
| 1183 | DEBUG_WITH_TYPE("trie-builder", llvm::dbgs() |
| 1184 | << "splice in TrieNode('" << bNodeStr |
| 1185 | << "') between edge '" |
| 1186 | << abEdgeStr << "' and edge='" |
| 1187 | << bcEdgeStr<< "'\n"); |
| 1188 | TrieEdge& abEdge = edge; |
| 1189 | abEdge._subString = abEdgeStr; |
| 1190 | abEdge._child = bNode; |
Eugene Zelenko | 4154794 | 2015-11-10 22:37:38 +0000 | [diff] [blame] | 1191 | auto *bcEdge = new (allocator) TrieEdge(bcEdgeStr, cNode); |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 1192 | bNode->_children.insert(bNode->_children.end(), bcEdge); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1193 | bNode->addSymbol(entry, allocator, allNodes); |
| 1194 | return; |
| 1195 | } |
| 1196 | } |
| 1197 | } |
| 1198 | if (entry.flags & EXPORT_SYMBOL_FLAGS_REEXPORT) { |
| 1199 | assert(entry.otherOffset != 0); |
| 1200 | } |
| 1201 | if (entry.flags & EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER) { |
| 1202 | assert(entry.otherOffset != 0); |
| 1203 | } |
| 1204 | // No commonality with any existing child, make a new edge. |
Eugene Zelenko | 4154794 | 2015-11-10 22:37:38 +0000 | [diff] [blame] | 1205 | auto *newNode = new (allocator) TrieNode(entry.name.copy(allocator)); |
| 1206 | auto *newEdge = new (allocator) TrieEdge(partialStr, newNode); |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 1207 | _children.insert(_children.end(), newEdge); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1208 | DEBUG_WITH_TYPE("trie-builder", llvm::dbgs() |
| 1209 | << "new TrieNode('" << entry.name << "') with edge '" |
| 1210 | << partialStr << "' from node='" |
| 1211 | << _cummulativeString << "'\n"); |
| 1212 | newNode->_address = entry.offset; |
| 1213 | newNode->_flags = entry.flags | entry.kind; |
| 1214 | newNode->_other = entry.otherOffset; |
| 1215 | if ((entry.flags & EXPORT_SYMBOL_FLAGS_REEXPORT) && !entry.otherName.empty()) |
| 1216 | newNode->_importedName = entry.otherName.copy(allocator); |
| 1217 | newNode->_hasExportInfo = true; |
| 1218 | allNodes.push_back(newNode); |
| 1219 | } |
| 1220 | |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 1221 | bool TrieNode::updateOffset(uint32_t& offset) { |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1222 | uint32_t nodeSize = 1; // Length when no export info |
| 1223 | if (_hasExportInfo) { |
| 1224 | if (_flags & EXPORT_SYMBOL_FLAGS_REEXPORT) { |
| 1225 | nodeSize = llvm::getULEB128Size(_flags); |
| 1226 | nodeSize += llvm::getULEB128Size(_other); // Other contains ordinal. |
| 1227 | nodeSize += _importedName.size(); |
| 1228 | ++nodeSize; // Trailing zero in imported name. |
| 1229 | } else { |
| 1230 | nodeSize = llvm::getULEB128Size(_flags) + llvm::getULEB128Size(_address); |
| 1231 | if (_flags & EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER) |
| 1232 | nodeSize += llvm::getULEB128Size(_other); |
| 1233 | } |
| 1234 | // Overall node size so far is uleb128 of export info + actual export info. |
| 1235 | nodeSize += llvm::getULEB128Size(nodeSize); |
| 1236 | } |
| 1237 | // Compute size of all child edges. |
| 1238 | ++nodeSize; // Byte for number of chidren. |
| 1239 | for (TrieEdge &edge : _children) { |
| 1240 | nodeSize += edge._subString.size() + 1 // String length. |
| 1241 | + llvm::getULEB128Size(edge._child->_trieOffset); // Offset len. |
| 1242 | } |
| 1243 | // On input, 'offset' is new prefered location for this node. |
| 1244 | bool result = (_trieOffset != offset); |
| 1245 | // Store new location in node object for use by parents. |
| 1246 | _trieOffset = offset; |
| 1247 | // Update offset for next iteration. |
| 1248 | offset += nodeSize; |
| 1249 | // Return true if _trieOffset was changed. |
| 1250 | return result; |
| 1251 | } |
| 1252 | |
Pete Cooper | e420dd4 | 2016-01-25 21:50:54 +0000 | [diff] [blame] | 1253 | void TrieNode::appendToByteBuffer(ByteBuffer &out) { |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1254 | if (_hasExportInfo) { |
| 1255 | if (_flags & EXPORT_SYMBOL_FLAGS_REEXPORT) { |
| 1256 | if (!_importedName.empty()) { |
| 1257 | // nodes with re-export info: size, flags, ordinal, import-name |
| 1258 | uint32_t nodeSize = llvm::getULEB128Size(_flags) |
| 1259 | + llvm::getULEB128Size(_other) |
| 1260 | + _importedName.size() + 1; |
| 1261 | assert(nodeSize < 256); |
| 1262 | out.append_byte(nodeSize); |
| 1263 | out.append_uleb128(_flags); |
| 1264 | out.append_uleb128(_other); |
| 1265 | out.append_string(_importedName); |
| 1266 | } else { |
| 1267 | // nodes without re-export info: size, flags, ordinal, empty-string |
| 1268 | uint32_t nodeSize = llvm::getULEB128Size(_flags) |
| 1269 | + llvm::getULEB128Size(_other) + 1; |
| 1270 | assert(nodeSize < 256); |
| 1271 | out.append_byte(nodeSize); |
| 1272 | out.append_uleb128(_flags); |
| 1273 | out.append_uleb128(_other); |
| 1274 | out.append_byte(0); |
| 1275 | } |
| 1276 | } else if ( _flags & EXPORT_SYMBOL_FLAGS_STUB_AND_RESOLVER ) { |
| 1277 | // Nodes with export info: size, flags, address, other |
| 1278 | uint32_t nodeSize = llvm::getULEB128Size(_flags) |
| 1279 | + llvm::getULEB128Size(_address) |
| 1280 | + llvm::getULEB128Size(_other); |
| 1281 | assert(nodeSize < 256); |
| 1282 | out.append_byte(nodeSize); |
| 1283 | out.append_uleb128(_flags); |
| 1284 | out.append_uleb128(_address); |
| 1285 | out.append_uleb128(_other); |
| 1286 | } else { |
| 1287 | // Nodes with export info: size, flags, address |
| 1288 | uint32_t nodeSize = llvm::getULEB128Size(_flags) |
| 1289 | + llvm::getULEB128Size(_address); |
| 1290 | assert(nodeSize < 256); |
| 1291 | out.append_byte(nodeSize); |
| 1292 | out.append_uleb128(_flags); |
| 1293 | out.append_uleb128(_address); |
| 1294 | } |
| 1295 | } else { |
| 1296 | // Node with no export info. |
| 1297 | uint32_t nodeSize = 0; |
| 1298 | out.append_byte(nodeSize); |
| 1299 | } |
| 1300 | // Add number of children. |
| 1301 | assert(_children.size() < 256); |
| 1302 | out.append_byte(_children.size()); |
| 1303 | // Append each child edge substring and node offset. |
| 1304 | for (TrieEdge &edge : _children) { |
| 1305 | out.append_string(edge._subString); |
| 1306 | out.append_uleb128(edge._child->_trieOffset); |
| 1307 | } |
| 1308 | } |
| 1309 | |
| 1310 | void MachOFileLayout::buildExportTrie() { |
| 1311 | if (_file.exportInfo.empty()) |
| 1312 | return; |
| 1313 | |
| 1314 | // For all temporary strings and objects used building trie. |
| 1315 | BumpPtrAllocator allocator; |
| 1316 | |
| 1317 | // Build trie of all exported symbols. |
Eugene Zelenko | 4154794 | 2015-11-10 22:37:38 +0000 | [diff] [blame] | 1318 | auto *rootNode = new (allocator) TrieNode(StringRef()); |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1319 | std::vector<TrieNode*> allNodes; |
| 1320 | allNodes.reserve(_file.exportInfo.size()*2); |
| 1321 | allNodes.push_back(rootNode); |
| 1322 | for (const Export& entry : _file.exportInfo) { |
| 1323 | rootNode->addSymbol(entry, allocator, allNodes); |
| 1324 | } |
| 1325 | |
| 1326 | // Assign each node in the vector an offset in the trie stream, iterating |
| 1327 | // until all uleb128 sizes have stabilized. |
| 1328 | bool more; |
| 1329 | do { |
| 1330 | uint32_t offset = 0; |
| 1331 | more = false; |
| 1332 | for (TrieNode* node : allNodes) { |
| 1333 | if (node->updateOffset(offset)) |
| 1334 | more = true; |
| 1335 | } |
| 1336 | } while (more); |
| 1337 | |
| 1338 | // Serialize trie to ByteBuffer. |
| 1339 | for (TrieNode* node : allNodes) { |
| 1340 | node->appendToByteBuffer(_exportTrie); |
| 1341 | } |
| 1342 | _exportTrie.align(_is64 ? 8 : 4); |
| 1343 | } |
| 1344 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1345 | void MachOFileLayout::computeSymbolTableSizes() { |
| 1346 | // MachO symbol tables have three ranges: locals, globals, and undefines |
| 1347 | const size_t nlistSize = (_is64 ? sizeof(nlist_64) : sizeof(nlist)); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1348 | _symbolTableSize = nlistSize * (_file.localSymbols.size() |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1349 | + _file.globalSymbols.size() |
| 1350 | + _file.undefinedSymbols.size()); |
Lang Hames | 201c08f | 2015-12-10 00:12:24 +0000 | [diff] [blame] | 1351 | _symbolStringPoolSize = 1; // Always reserve 1-byte for the empty string. |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1352 | for (const Symbol &sym : _file.localSymbols) { |
| 1353 | _symbolStringPoolSize += (sym.name.size()+1); |
| 1354 | } |
| 1355 | for (const Symbol &sym : _file.globalSymbols) { |
| 1356 | _symbolStringPoolSize += (sym.name.size()+1); |
| 1357 | } |
| 1358 | for (const Symbol &sym : _file.undefinedSymbols) { |
| 1359 | _symbolStringPoolSize += (sym.name.size()+1); |
| 1360 | } |
| 1361 | _symbolTableLocalsStartIndex = 0; |
| 1362 | _symbolTableGlobalsStartIndex = _file.localSymbols.size(); |
Shankar Easwaran | 3d8de47 | 2014-01-27 03:09:26 +0000 | [diff] [blame] | 1363 | _symbolTableUndefinesStartIndex = _symbolTableGlobalsStartIndex |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1364 | + _file.globalSymbols.size(); |
| 1365 | |
| 1366 | _indirectSymbolTableCount = 0; |
| 1367 | for (const Section § : _file.sections) { |
| 1368 | _indirectSymbolTableCount += sect.indirectSymbols.size(); |
| 1369 | } |
| 1370 | } |
| 1371 | |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 1372 | void MachOFileLayout::computeDataInCodeSize() { |
| 1373 | _dataInCodeSize = _file.dataInCode.size() * sizeof(data_in_code_entry); |
| 1374 | } |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1375 | |
| 1376 | void MachOFileLayout::writeLinkEditContent() { |
| 1377 | if (_file.fileType == llvm::MachO::MH_OBJECT) { |
| 1378 | writeRelocations(); |
Nick Kledzik | 2192137 | 2014-07-24 23:06:56 +0000 | [diff] [blame] | 1379 | writeDataInCodeInfo(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1380 | writeSymbolTable(); |
| 1381 | } else { |
| 1382 | writeRebaseInfo(); |
| 1383 | writeBindingInfo(); |
| 1384 | writeLazyBindingInfo(); |
| 1385 | // TODO: add weak binding info |
Nick Kledzik | 141330a | 2014-09-03 19:52:50 +0000 | [diff] [blame] | 1386 | writeExportInfo(); |
Nick Kledzik | 54ce2958 | 2014-10-28 22:21:10 +0000 | [diff] [blame] | 1387 | writeDataInCodeInfo(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1388 | writeSymbolTable(); |
| 1389 | } |
| 1390 | } |
| 1391 | |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 1392 | std::error_code MachOFileLayout::writeBinary(StringRef path) { |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1393 | // Check for pending error from constructor. |
| 1394 | if (_ec) |
| 1395 | return _ec; |
| 1396 | // Create FileOutputBuffer with calculated size. |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1397 | unsigned flags = 0; |
| 1398 | if (_file.fileType != llvm::MachO::MH_OBJECT) |
| 1399 | flags = llvm::FileOutputBuffer::F_executable; |
Rafael Espindola | bdc8f2f | 2015-08-13 00:31:46 +0000 | [diff] [blame] | 1400 | ErrorOr<std::unique_ptr<llvm::FileOutputBuffer>> fobOrErr = |
| 1401 | llvm::FileOutputBuffer::create(path, size(), flags); |
| 1402 | if (std::error_code ec = fobOrErr.getError()) |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1403 | return ec; |
Rafael Espindola | bdc8f2f | 2015-08-13 00:31:46 +0000 | [diff] [blame] | 1404 | std::unique_ptr<llvm::FileOutputBuffer> &fob = *fobOrErr; |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1405 | // Write content. |
| 1406 | _buffer = fob->getBufferStart(); |
| 1407 | writeMachHeader(); |
Rafael Espindola | bdc8f2f | 2015-08-13 00:31:46 +0000 | [diff] [blame] | 1408 | std::error_code ec = writeLoadCommands(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1409 | if (ec) |
| 1410 | return ec; |
| 1411 | writeSectionContent(); |
| 1412 | writeLinkEditContent(); |
| 1413 | fob->commit(); |
| 1414 | |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 1415 | return std::error_code(); |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1416 | } |
| 1417 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1418 | /// Takes in-memory normalized view and writes a mach-o object file. |
Rafael Espindola | b1a4d3a | 2014-06-12 14:53:47 +0000 | [diff] [blame] | 1419 | std::error_code writeBinary(const NormalizedFile &file, StringRef path) { |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1420 | MachOFileLayout layout(file); |
| 1421 | return layout.writeBinary(path); |
| 1422 | } |
| 1423 | |
Nick Kledzik | e34182f | 2013-11-06 21:36:55 +0000 | [diff] [blame] | 1424 | } // namespace normalized |
| 1425 | } // namespace mach_o |
| 1426 | } // namespace lld |