1 //===- ArchiveWriter.cpp - ar File Format implementation --------*- 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 writeArchive function. 11 // 12 //===----------------------------------------------------------------------===// 13 14 #include "llvm/Object/ArchiveWriter.h" 15 #include "llvm/ADT/ArrayRef.h" 16 #include "llvm/ADT/StringRef.h" 17 #include "llvm/IR/LLVMContext.h" 18 #include "llvm/Object/Archive.h" 19 #include "llvm/Object/ObjectFile.h" 20 #include "llvm/Object/SymbolicFile.h" 21 #include "llvm/Support/EndianStream.h" 22 #include "llvm/Support/Errc.h" 23 #include "llvm/Support/ErrorHandling.h" 24 #include "llvm/Support/Format.h" 25 #include "llvm/Support/Path.h" 26 #include "llvm/Support/ToolOutputFile.h" 27 #include "llvm/Support/raw_ostream.h" 28 29 #if !defined(_MSC_VER) && !defined(__MINGW32__) 30 #include <unistd.h> 31 #else 32 #include <io.h> 33 #endif 34 35 using namespace llvm; 36 37 NewArchiveMember::NewArchiveMember(MemoryBufferRef BufRef) 38 : Buf(MemoryBuffer::getMemBuffer(BufRef, false)) {} 39 40 Expected<NewArchiveMember> 41 NewArchiveMember::getOldMember(const object::Archive::Child &OldMember, 42 bool Deterministic) { 43 Expected<llvm::MemoryBufferRef> BufOrErr = OldMember.getMemoryBufferRef(); 44 if (!BufOrErr) 45 return BufOrErr.takeError(); 46 47 NewArchiveMember M; 48 assert(M.IsNew == false); 49 M.Buf = MemoryBuffer::getMemBuffer(*BufOrErr, false); 50 if (!Deterministic) { 51 auto ModTimeOrErr = OldMember.getLastModified(); 52 if (!ModTimeOrErr) 53 return ModTimeOrErr.takeError(); 54 M.ModTime = ModTimeOrErr.get(); 55 Expected<unsigned> UIDOrErr = OldMember.getUID(); 56 if (!UIDOrErr) 57 return UIDOrErr.takeError(); 58 M.UID = UIDOrErr.get(); 59 Expected<unsigned> GIDOrErr = OldMember.getGID(); 60 if (!GIDOrErr) 61 return GIDOrErr.takeError(); 62 M.GID = GIDOrErr.get(); 63 Expected<sys::fs::perms> AccessModeOrErr = OldMember.getAccessMode(); 64 if (!AccessModeOrErr) 65 return AccessModeOrErr.takeError(); 66 M.Perms = AccessModeOrErr.get(); 67 } 68 return std::move(M); 69 } 70 71 Expected<NewArchiveMember> NewArchiveMember::getFile(StringRef FileName, 72 bool Deterministic) { 73 sys::fs::file_status Status; 74 int FD; 75 if (auto EC = sys::fs::openFileForRead(FileName, FD)) 76 return errorCodeToError(EC); 77 assert(FD != -1); 78 79 if (auto EC = sys::fs::status(FD, Status)) 80 return errorCodeToError(EC); 81 82 // Opening a directory doesn't make sense. Let it fail. 83 // Linux cannot open directories with open(2), although 84 // cygwin and *bsd can. 85 if (Status.type() == sys::fs::file_type::directory_file) 86 return errorCodeToError(make_error_code(errc::is_a_directory)); 87 88 ErrorOr<std::unique_ptr<MemoryBuffer>> MemberBufferOrErr = 89 MemoryBuffer::getOpenFile(FD, FileName, Status.getSize(), false); 90 if (!MemberBufferOrErr) 91 return errorCodeToError(MemberBufferOrErr.getError()); 92 93 if (close(FD) != 0) 94 return errorCodeToError(std::error_code(errno, std::generic_category())); 95 96 NewArchiveMember M; 97 M.IsNew = true; 98 M.Buf = std::move(*MemberBufferOrErr); 99 if (!Deterministic) { 100 M.ModTime = std::chrono::time_point_cast<std::chrono::seconds>( 101 Status.getLastModificationTime()); 102 M.UID = Status.getUser(); 103 M.GID = Status.getGroup(); 104 M.Perms = Status.permissions(); 105 } 106 return std::move(M); 107 } 108 109 template <typename T> 110 static void printWithSpacePadding(raw_fd_ostream &OS, T Data, unsigned Size, 111 bool MayTruncate = false) { 112 uint64_t OldPos = OS.tell(); 113 OS << Data; 114 unsigned SizeSoFar = OS.tell() - OldPos; 115 if (Size > SizeSoFar) { 116 OS.indent(Size - SizeSoFar); 117 } else if (Size < SizeSoFar) { 118 assert(MayTruncate && "Data doesn't fit in Size"); 119 // Some of the data this is used for (like UID) can be larger than the 120 // space available in the archive format. Truncate in that case. 121 OS.seek(OldPos + Size); 122 } 123 } 124 125 static void print32(raw_ostream &Out, object::Archive::Kind Kind, 126 uint32_t Val) { 127 if (Kind == object::Archive::K_GNU) 128 support::endian::Writer<support::big>(Out).write(Val); 129 else 130 support::endian::Writer<support::little>(Out).write(Val); 131 } 132 133 static void printRestOfMemberHeader( 134 raw_fd_ostream &Out, const sys::TimePoint<std::chrono::seconds> &ModTime, 135 unsigned UID, unsigned GID, unsigned Perms, unsigned Size) { 136 printWithSpacePadding(Out, sys::toTimeT(ModTime), 12); 137 printWithSpacePadding(Out, UID, 6, true); 138 printWithSpacePadding(Out, GID, 6, true); 139 printWithSpacePadding(Out, format("%o", Perms), 8); 140 printWithSpacePadding(Out, Size, 10); 141 Out << "`\n"; 142 } 143 144 static void 145 printGNUSmallMemberHeader(raw_fd_ostream &Out, StringRef Name, 146 const sys::TimePoint<std::chrono::seconds> &ModTime, 147 unsigned UID, unsigned GID, unsigned Perms, 148 unsigned Size) { 149 printWithSpacePadding(Out, Twine(Name) + "/", 16); 150 printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size); 151 } 152 153 static void 154 printBSDMemberHeader(raw_fd_ostream &Out, StringRef Name, 155 const sys::TimePoint<std::chrono::seconds> &ModTime, 156 unsigned UID, unsigned GID, unsigned Perms, 157 unsigned Size) { 158 uint64_t PosAfterHeader = Out.tell() + 60 + Name.size(); 159 // Pad so that even 64 bit object files are aligned. 160 unsigned Pad = OffsetToAlignment(PosAfterHeader, 8); 161 unsigned NameWithPadding = Name.size() + Pad; 162 printWithSpacePadding(Out, Twine("#1/") + Twine(NameWithPadding), 16); 163 printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, 164 NameWithPadding + Size); 165 Out << Name; 166 assert(PosAfterHeader == Out.tell()); 167 while (Pad--) 168 Out.write(uint8_t(0)); 169 } 170 171 static bool useStringTable(bool Thin, StringRef Name) { 172 return Thin || Name.size() >= 16; 173 } 174 175 static void 176 printMemberHeader(raw_fd_ostream &Out, object::Archive::Kind Kind, bool Thin, 177 StringRef Name, 178 std::vector<unsigned>::iterator &StringMapIndexIter, 179 const sys::TimePoint<std::chrono::seconds> &ModTime, 180 unsigned UID, unsigned GID, unsigned Perms, unsigned Size) { 181 if (Kind == object::Archive::K_BSD) 182 return printBSDMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size); 183 if (!useStringTable(Thin, Name)) 184 return printGNUSmallMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size); 185 Out << '/'; 186 printWithSpacePadding(Out, *StringMapIndexIter++, 15); 187 printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size); 188 } 189 190 // Compute the relative path from From to To. 191 static std::string computeRelativePath(StringRef From, StringRef To) { 192 if (sys::path::is_absolute(From) || sys::path::is_absolute(To)) 193 return To; 194 195 StringRef DirFrom = sys::path::parent_path(From); 196 auto FromI = sys::path::begin(DirFrom); 197 auto ToI = sys::path::begin(To); 198 while (*FromI == *ToI) { 199 ++FromI; 200 ++ToI; 201 } 202 203 SmallString<128> Relative; 204 for (auto FromE = sys::path::end(DirFrom); FromI != FromE; ++FromI) 205 sys::path::append(Relative, ".."); 206 207 for (auto ToE = sys::path::end(To); ToI != ToE; ++ToI) 208 sys::path::append(Relative, *ToI); 209 210 #ifdef LLVM_ON_WIN32 211 // Replace backslashes with slashes so that the path is portable between *nix 212 // and Windows. 213 std::replace(Relative.begin(), Relative.end(), '\\', '/'); 214 #endif 215 216 return Relative.str(); 217 } 218 219 static void writeStringTable(raw_fd_ostream &Out, StringRef ArcName, 220 ArrayRef<NewArchiveMember> Members, 221 std::vector<unsigned> &StringMapIndexes, 222 bool Thin) { 223 unsigned StartOffset = 0; 224 for (const NewArchiveMember &M : Members) { 225 StringRef Path = M.Buf->getBufferIdentifier(); 226 StringRef Name = sys::path::filename(Path); 227 if (!useStringTable(Thin, Name)) 228 continue; 229 if (StartOffset == 0) { 230 printWithSpacePadding(Out, "//", 58); 231 Out << "`\n"; 232 StartOffset = Out.tell(); 233 } 234 StringMapIndexes.push_back(Out.tell() - StartOffset); 235 236 if (Thin) { 237 if (M.IsNew) 238 Out << computeRelativePath(ArcName, Path); 239 else 240 Out << M.Buf->getBufferIdentifier(); 241 } else 242 Out << Name; 243 244 Out << "/\n"; 245 } 246 if (StartOffset == 0) 247 return; 248 if (Out.tell() % 2) 249 Out << '\n'; 250 int Pos = Out.tell(); 251 Out.seek(StartOffset - 12); 252 printWithSpacePadding(Out, Pos - StartOffset, 10); 253 Out.seek(Pos); 254 } 255 256 static sys::TimePoint<std::chrono::seconds> now(bool Deterministic) { 257 using namespace std::chrono; 258 259 if (!Deterministic) 260 return time_point_cast<seconds>(system_clock::now()); 261 return sys::TimePoint<seconds>(); 262 } 263 264 // Returns the offset of the first reference to a member offset. 265 static ErrorOr<unsigned> 266 writeSymbolTable(raw_fd_ostream &Out, object::Archive::Kind Kind, 267 ArrayRef<NewArchiveMember> Members, 268 std::vector<unsigned> &MemberOffsetRefs, bool Deterministic) { 269 unsigned HeaderStartOffset = 0; 270 unsigned BodyStartOffset = 0; 271 SmallString<128> NameBuf; 272 raw_svector_ostream NameOS(NameBuf); 273 LLVMContext Context; 274 for (unsigned MemberNum = 0, N = Members.size(); MemberNum < N; ++MemberNum) { 275 MemoryBufferRef MemberBuffer = Members[MemberNum].Buf->getMemBufferRef(); 276 Expected<std::unique_ptr<object::SymbolicFile>> ObjOrErr = 277 object::SymbolicFile::createSymbolicFile( 278 MemberBuffer, sys::fs::file_magic::unknown, &Context); 279 if (!ObjOrErr) { 280 // FIXME: check only for "not an object file" errors. 281 consumeError(ObjOrErr.takeError()); 282 continue; 283 } 284 object::SymbolicFile &Obj = *ObjOrErr.get(); 285 286 if (!HeaderStartOffset) { 287 HeaderStartOffset = Out.tell(); 288 if (Kind == object::Archive::K_GNU) 289 printGNUSmallMemberHeader(Out, "", now(Deterministic), 0, 0, 0, 0); 290 else 291 printBSDMemberHeader(Out, "__.SYMDEF", now(Deterministic), 0, 0, 0, 0); 292 BodyStartOffset = Out.tell(); 293 print32(Out, Kind, 0); // number of entries or bytes 294 } 295 296 for (const object::BasicSymbolRef &S : Obj.symbols()) { 297 uint32_t Symflags = S.getFlags(); 298 if (Symflags & object::SymbolRef::SF_FormatSpecific) 299 continue; 300 if (!(Symflags & object::SymbolRef::SF_Global)) 301 continue; 302 if (Symflags & object::SymbolRef::SF_Undefined) 303 continue; 304 305 unsigned NameOffset = NameOS.tell(); 306 if (auto EC = S.printName(NameOS)) 307 return EC; 308 NameOS << '\0'; 309 MemberOffsetRefs.push_back(MemberNum); 310 if (Kind == object::Archive::K_BSD) 311 print32(Out, Kind, NameOffset); 312 print32(Out, Kind, 0); // member offset 313 } 314 } 315 316 if (HeaderStartOffset == 0) 317 return 0; 318 319 StringRef StringTable = NameOS.str(); 320 if (Kind == object::Archive::K_BSD) 321 print32(Out, Kind, StringTable.size()); // byte count of the string table 322 Out << StringTable; 323 324 // ld64 requires the next member header to start at an offset that is 325 // 4 bytes aligned. 326 unsigned Pad = OffsetToAlignment(Out.tell(), 4); 327 while (Pad--) 328 Out.write(uint8_t(0)); 329 330 // Patch up the size of the symbol table now that we know how big it is. 331 unsigned Pos = Out.tell(); 332 const unsigned MemberHeaderSize = 60; 333 Out.seek(HeaderStartOffset + 48); // offset of the size field. 334 printWithSpacePadding(Out, Pos - MemberHeaderSize - HeaderStartOffset, 10); 335 336 // Patch up the number of symbols. 337 Out.seek(BodyStartOffset); 338 unsigned NumSyms = MemberOffsetRefs.size(); 339 if (Kind == object::Archive::K_GNU) 340 print32(Out, Kind, NumSyms); 341 else 342 print32(Out, Kind, NumSyms * 8); 343 344 Out.seek(Pos); 345 return BodyStartOffset + 4; 346 } 347 348 std::pair<StringRef, std::error_code> 349 llvm::writeArchive(StringRef ArcName, 350 std::vector<NewArchiveMember> &NewMembers, 351 bool WriteSymtab, object::Archive::Kind Kind, 352 bool Deterministic, bool Thin, 353 std::unique_ptr<MemoryBuffer> OldArchiveBuf) { 354 assert((!Thin || Kind == object::Archive::K_GNU) && 355 "Only the gnu format has a thin mode"); 356 SmallString<128> TmpArchive; 357 int TmpArchiveFD; 358 if (auto EC = sys::fs::createUniqueFile(ArcName + ".temp-archive-%%%%%%%.a", 359 TmpArchiveFD, TmpArchive)) 360 return std::make_pair(ArcName, EC); 361 362 tool_output_file Output(TmpArchive, TmpArchiveFD); 363 raw_fd_ostream &Out = Output.os(); 364 if (Thin) 365 Out << "!<thin>\n"; 366 else 367 Out << "!<arch>\n"; 368 369 std::vector<unsigned> MemberOffsetRefs; 370 371 std::vector<std::unique_ptr<MemoryBuffer>> Buffers; 372 std::vector<MemoryBufferRef> Members; 373 std::vector<sys::fs::file_status> NewMemberStatus; 374 375 unsigned MemberReferenceOffset = 0; 376 if (WriteSymtab) { 377 ErrorOr<unsigned> MemberReferenceOffsetOrErr = writeSymbolTable( 378 Out, Kind, NewMembers, MemberOffsetRefs, Deterministic); 379 if (auto EC = MemberReferenceOffsetOrErr.getError()) 380 return std::make_pair(ArcName, EC); 381 MemberReferenceOffset = MemberReferenceOffsetOrErr.get(); 382 } 383 384 std::vector<unsigned> StringMapIndexes; 385 if (Kind != object::Archive::K_BSD) 386 writeStringTable(Out, ArcName, NewMembers, StringMapIndexes, Thin); 387 388 std::vector<unsigned>::iterator StringMapIndexIter = StringMapIndexes.begin(); 389 std::vector<unsigned> MemberOffset; 390 for (const NewArchiveMember &M : NewMembers) { 391 MemoryBufferRef File = M.Buf->getMemBufferRef(); 392 393 unsigned Pos = Out.tell(); 394 MemberOffset.push_back(Pos); 395 396 printMemberHeader(Out, Kind, Thin, 397 sys::path::filename(M.Buf->getBufferIdentifier()), 398 StringMapIndexIter, M.ModTime, M.UID, M.GID, M.Perms, 399 M.Buf->getBufferSize()); 400 401 if (!Thin) 402 Out << File.getBuffer(); 403 404 if (Out.tell() % 2) 405 Out << '\n'; 406 } 407 408 if (MemberReferenceOffset) { 409 Out.seek(MemberReferenceOffset); 410 for (unsigned MemberNum : MemberOffsetRefs) { 411 if (Kind == object::Archive::K_BSD) 412 Out.seek(Out.tell() + 4); // skip over the string offset 413 print32(Out, Kind, MemberOffset[MemberNum]); 414 } 415 } 416 417 Output.keep(); 418 Out.close(); 419 420 // At this point, we no longer need whatever backing memory 421 // was used to generate the NewMembers. On Windows, this buffer 422 // could be a mapped view of the file we want to replace (if 423 // we're updating an existing archive, say). In that case, the 424 // rename would still succeed, but it would leave behind a 425 // temporary file (actually the original file renamed) because 426 // a file cannot be deleted while there's a handle open on it, 427 // only renamed. So by freeing this buffer, this ensures that 428 // the last open handle on the destination file, if any, is 429 // closed before we attempt to rename. 430 OldArchiveBuf.reset(); 431 432 sys::fs::rename(TmpArchive, ArcName); 433 return std::make_pair("", std::error_code()); 434 } 435