ArchiveWriter.cpp revision 321369
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/BinaryFormat/Magic.h" 18#include "llvm/IR/LLVMContext.h" 19#include "llvm/Object/Archive.h" 20#include "llvm/Object/ObjectFile.h" 21#include "llvm/Object/SymbolicFile.h" 22#include "llvm/Support/EndianStream.h" 23#include "llvm/Support/Errc.h" 24#include "llvm/Support/ErrorHandling.h" 25#include "llvm/Support/Format.h" 26#include "llvm/Support/Path.h" 27#include "llvm/Support/ToolOutputFile.h" 28#include "llvm/Support/raw_ostream.h" 29 30#if !defined(_MSC_VER) && !defined(__MINGW32__) 31#include <unistd.h> 32#else 33#include <io.h> 34#endif 35 36using namespace llvm; 37 38NewArchiveMember::NewArchiveMember(MemoryBufferRef BufRef) 39 : Buf(MemoryBuffer::getMemBuffer(BufRef, false)), 40 MemberName(BufRef.getBufferIdentifier()) {} 41 42Expected<NewArchiveMember> 43NewArchiveMember::getOldMember(const object::Archive::Child &OldMember, 44 bool Deterministic) { 45 Expected<llvm::MemoryBufferRef> BufOrErr = OldMember.getMemoryBufferRef(); 46 if (!BufOrErr) 47 return BufOrErr.takeError(); 48 49 NewArchiveMember M; 50 assert(M.IsNew == false); 51 M.Buf = MemoryBuffer::getMemBuffer(*BufOrErr, false); 52 M.MemberName = M.Buf->getBufferIdentifier(); 53 if (!Deterministic) { 54 auto ModTimeOrErr = OldMember.getLastModified(); 55 if (!ModTimeOrErr) 56 return ModTimeOrErr.takeError(); 57 M.ModTime = ModTimeOrErr.get(); 58 Expected<unsigned> UIDOrErr = OldMember.getUID(); 59 if (!UIDOrErr) 60 return UIDOrErr.takeError(); 61 M.UID = UIDOrErr.get(); 62 Expected<unsigned> GIDOrErr = OldMember.getGID(); 63 if (!GIDOrErr) 64 return GIDOrErr.takeError(); 65 M.GID = GIDOrErr.get(); 66 Expected<sys::fs::perms> AccessModeOrErr = OldMember.getAccessMode(); 67 if (!AccessModeOrErr) 68 return AccessModeOrErr.takeError(); 69 M.Perms = AccessModeOrErr.get(); 70 } 71 return std::move(M); 72} 73 74Expected<NewArchiveMember> NewArchiveMember::getFile(StringRef FileName, 75 bool Deterministic) { 76 sys::fs::file_status Status; 77 int FD; 78 if (auto EC = sys::fs::openFileForRead(FileName, FD)) 79 return errorCodeToError(EC); 80 assert(FD != -1); 81 82 if (auto EC = sys::fs::status(FD, Status)) 83 return errorCodeToError(EC); 84 85 // Opening a directory doesn't make sense. Let it fail. 86 // Linux cannot open directories with open(2), although 87 // cygwin and *bsd can. 88 if (Status.type() == sys::fs::file_type::directory_file) 89 return errorCodeToError(make_error_code(errc::is_a_directory)); 90 91 ErrorOr<std::unique_ptr<MemoryBuffer>> MemberBufferOrErr = 92 MemoryBuffer::getOpenFile(FD, FileName, Status.getSize(), false); 93 if (!MemberBufferOrErr) 94 return errorCodeToError(MemberBufferOrErr.getError()); 95 96 if (close(FD) != 0) 97 return errorCodeToError(std::error_code(errno, std::generic_category())); 98 99 NewArchiveMember M; 100 M.IsNew = true; 101 M.Buf = std::move(*MemberBufferOrErr); 102 M.MemberName = M.Buf->getBufferIdentifier(); 103 if (!Deterministic) { 104 M.ModTime = std::chrono::time_point_cast<std::chrono::seconds>( 105 Status.getLastModificationTime()); 106 M.UID = Status.getUser(); 107 M.GID = Status.getGroup(); 108 M.Perms = Status.permissions(); 109 } 110 return std::move(M); 111} 112 113template <typename T> 114static void printWithSpacePadding(raw_fd_ostream &OS, T Data, unsigned Size, 115 bool MayTruncate = false) { 116 uint64_t OldPos = OS.tell(); 117 OS << Data; 118 unsigned SizeSoFar = OS.tell() - OldPos; 119 if (Size > SizeSoFar) { 120 OS.indent(Size - SizeSoFar); 121 } else if (Size < SizeSoFar) { 122 assert(MayTruncate && "Data doesn't fit in Size"); 123 // Some of the data this is used for (like UID) can be larger than the 124 // space available in the archive format. Truncate in that case. 125 OS.seek(OldPos + Size); 126 } 127} 128 129static bool isBSDLike(object::Archive::Kind Kind) { 130 switch (Kind) { 131 case object::Archive::K_GNU: 132 return false; 133 case object::Archive::K_BSD: 134 case object::Archive::K_DARWIN: 135 return true; 136 case object::Archive::K_MIPS64: 137 case object::Archive::K_DARWIN64: 138 case object::Archive::K_COFF: 139 break; 140 } 141 llvm_unreachable("not supported for writting"); 142} 143 144static void print32(raw_ostream &Out, object::Archive::Kind Kind, 145 uint32_t Val) { 146 if (isBSDLike(Kind)) 147 support::endian::Writer<support::little>(Out).write(Val); 148 else 149 support::endian::Writer<support::big>(Out).write(Val); 150} 151 152static void printRestOfMemberHeader( 153 raw_fd_ostream &Out, const sys::TimePoint<std::chrono::seconds> &ModTime, 154 unsigned UID, unsigned GID, unsigned Perms, unsigned Size) { 155 printWithSpacePadding(Out, sys::toTimeT(ModTime), 12); 156 printWithSpacePadding(Out, UID, 6, true); 157 printWithSpacePadding(Out, GID, 6, true); 158 printWithSpacePadding(Out, format("%o", Perms), 8); 159 printWithSpacePadding(Out, Size, 10); 160 Out << "`\n"; 161} 162 163static void 164printGNUSmallMemberHeader(raw_fd_ostream &Out, StringRef Name, 165 const sys::TimePoint<std::chrono::seconds> &ModTime, 166 unsigned UID, unsigned GID, unsigned Perms, 167 unsigned Size) { 168 printWithSpacePadding(Out, Twine(Name) + "/", 16); 169 printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size); 170} 171 172static void 173printBSDMemberHeader(raw_fd_ostream &Out, StringRef Name, 174 const sys::TimePoint<std::chrono::seconds> &ModTime, 175 unsigned UID, unsigned GID, unsigned Perms, 176 unsigned Size) { 177 uint64_t PosAfterHeader = Out.tell() + 60 + Name.size(); 178 // Pad so that even 64 bit object files are aligned. 179 unsigned Pad = OffsetToAlignment(PosAfterHeader, 8); 180 unsigned NameWithPadding = Name.size() + Pad; 181 printWithSpacePadding(Out, Twine("#1/") + Twine(NameWithPadding), 16); 182 printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, 183 NameWithPadding + Size); 184 Out << Name; 185 assert(PosAfterHeader == Out.tell()); 186 while (Pad--) 187 Out.write(uint8_t(0)); 188} 189 190static bool useStringTable(bool Thin, StringRef Name) { 191 return Thin || Name.size() >= 16 || Name.contains('/'); 192} 193 194static void 195printMemberHeader(raw_fd_ostream &Out, object::Archive::Kind Kind, bool Thin, 196 StringRef Name, 197 std::vector<unsigned>::iterator &StringMapIndexIter, 198 const sys::TimePoint<std::chrono::seconds> &ModTime, 199 unsigned UID, unsigned GID, unsigned Perms, unsigned Size) { 200 if (isBSDLike(Kind)) 201 return printBSDMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size); 202 if (!useStringTable(Thin, Name)) 203 return printGNUSmallMemberHeader(Out, Name, ModTime, UID, GID, Perms, Size); 204 Out << '/'; 205 printWithSpacePadding(Out, *StringMapIndexIter++, 15); 206 printRestOfMemberHeader(Out, ModTime, UID, GID, Perms, Size); 207} 208 209// Compute the relative path from From to To. 210static std::string computeRelativePath(StringRef From, StringRef To) { 211 if (sys::path::is_absolute(From) || sys::path::is_absolute(To)) 212 return To; 213 214 StringRef DirFrom = sys::path::parent_path(From); 215 auto FromI = sys::path::begin(DirFrom); 216 auto ToI = sys::path::begin(To); 217 while (*FromI == *ToI) { 218 ++FromI; 219 ++ToI; 220 } 221 222 SmallString<128> Relative; 223 for (auto FromE = sys::path::end(DirFrom); FromI != FromE; ++FromI) 224 sys::path::append(Relative, ".."); 225 226 for (auto ToE = sys::path::end(To); ToI != ToE; ++ToI) 227 sys::path::append(Relative, *ToI); 228 229#ifdef LLVM_ON_WIN32 230 // Replace backslashes with slashes so that the path is portable between *nix 231 // and Windows. 232 std::replace(Relative.begin(), Relative.end(), '\\', '/'); 233#endif 234 235 return Relative.str(); 236} 237 238static void writeStringTable(raw_fd_ostream &Out, StringRef ArcName, 239 ArrayRef<NewArchiveMember> Members, 240 std::vector<unsigned> &StringMapIndexes, 241 bool Thin) { 242 unsigned StartOffset = 0; 243 for (const NewArchiveMember &M : Members) { 244 StringRef Path = M.Buf->getBufferIdentifier(); 245 StringRef Name = M.MemberName; 246 if (!useStringTable(Thin, Name)) 247 continue; 248 if (StartOffset == 0) { 249 printWithSpacePadding(Out, "//", 58); 250 Out << "`\n"; 251 StartOffset = Out.tell(); 252 } 253 StringMapIndexes.push_back(Out.tell() - StartOffset); 254 255 if (Thin) { 256 if (M.IsNew) 257 Out << computeRelativePath(ArcName, Path); 258 else 259 Out << M.Buf->getBufferIdentifier(); 260 } else 261 Out << Name; 262 263 Out << "/\n"; 264 } 265 if (StartOffset == 0) 266 return; 267 if (Out.tell() % 2) 268 Out << '\n'; 269 int Pos = Out.tell(); 270 Out.seek(StartOffset - 12); 271 printWithSpacePadding(Out, Pos - StartOffset, 10); 272 Out.seek(Pos); 273} 274 275static sys::TimePoint<std::chrono::seconds> now(bool Deterministic) { 276 using namespace std::chrono; 277 278 if (!Deterministic) 279 return time_point_cast<seconds>(system_clock::now()); 280 return sys::TimePoint<seconds>(); 281} 282 283// Returns the offset of the first reference to a member offset. 284static ErrorOr<unsigned> 285writeSymbolTable(raw_fd_ostream &Out, object::Archive::Kind Kind, 286 ArrayRef<NewArchiveMember> Members, 287 std::vector<unsigned> &MemberOffsetRefs, bool Deterministic) { 288 unsigned HeaderStartOffset = 0; 289 unsigned BodyStartOffset = 0; 290 SmallString<128> NameBuf; 291 raw_svector_ostream NameOS(NameBuf); 292 LLVMContext Context; 293 for (unsigned MemberNum = 0, N = Members.size(); MemberNum < N; ++MemberNum) { 294 MemoryBufferRef MemberBuffer = Members[MemberNum].Buf->getMemBufferRef(); 295 Expected<std::unique_ptr<object::SymbolicFile>> ObjOrErr = 296 object::SymbolicFile::createSymbolicFile( 297 MemberBuffer, llvm::file_magic::unknown, &Context); 298 if (!ObjOrErr) { 299 // FIXME: check only for "not an object file" errors. 300 consumeError(ObjOrErr.takeError()); 301 continue; 302 } 303 object::SymbolicFile &Obj = *ObjOrErr.get(); 304 305 if (!HeaderStartOffset) { 306 HeaderStartOffset = Out.tell(); 307 if (isBSDLike(Kind)) 308 printBSDMemberHeader(Out, "__.SYMDEF", now(Deterministic), 0, 0, 0, 0); 309 else 310 printGNUSmallMemberHeader(Out, "", now(Deterministic), 0, 0, 0, 0); 311 BodyStartOffset = Out.tell(); 312 print32(Out, Kind, 0); // number of entries or bytes 313 } 314 315 for (const object::BasicSymbolRef &S : Obj.symbols()) { 316 uint32_t Symflags = S.getFlags(); 317 if (Symflags & object::SymbolRef::SF_FormatSpecific) 318 continue; 319 if (!(Symflags & object::SymbolRef::SF_Global)) 320 continue; 321 if (Symflags & object::SymbolRef::SF_Undefined && 322 !(Symflags & object::SymbolRef::SF_Indirect)) 323 continue; 324 325 unsigned NameOffset = NameOS.tell(); 326 if (auto EC = S.printName(NameOS)) 327 return EC; 328 NameOS << '\0'; 329 MemberOffsetRefs.push_back(MemberNum); 330 if (isBSDLike(Kind)) 331 print32(Out, Kind, NameOffset); 332 print32(Out, Kind, 0); // member offset 333 } 334 } 335 336 if (HeaderStartOffset == 0) 337 return 0; 338 339 // ld64 prefers the cctools type archive which pads its string table to a 340 // boundary of sizeof(int32_t). 341 if (isBSDLike(Kind)) 342 for (unsigned P = OffsetToAlignment(NameOS.tell(), sizeof(int32_t)); P--;) 343 NameOS << '\0'; 344 345 StringRef StringTable = NameOS.str(); 346 if (isBSDLike(Kind)) 347 print32(Out, Kind, StringTable.size()); // byte count of the string table 348 Out << StringTable; 349 // If there are no symbols, emit an empty symbol table, to satisfy Solaris 350 // tools, older versions of which expect a symbol table in a non-empty 351 // archive, regardless of whether there are any symbols in it. 352 if (StringTable.size() == 0) 353 print32(Out, Kind, 0); 354 355 // ld64 requires the next member header to start at an offset that is 356 // 4 bytes aligned. 357 unsigned Pad = OffsetToAlignment(Out.tell(), 4); 358 while (Pad--) 359 Out.write(uint8_t(0)); 360 361 // Patch up the size of the symbol table now that we know how big it is. 362 unsigned Pos = Out.tell(); 363 const unsigned MemberHeaderSize = 60; 364 Out.seek(HeaderStartOffset + 48); // offset of the size field. 365 printWithSpacePadding(Out, Pos - MemberHeaderSize - HeaderStartOffset, 10); 366 367 // Patch up the number of symbols. 368 Out.seek(BodyStartOffset); 369 unsigned NumSyms = MemberOffsetRefs.size(); 370 if (isBSDLike(Kind)) 371 print32(Out, Kind, NumSyms * 8); 372 else 373 print32(Out, Kind, NumSyms); 374 375 Out.seek(Pos); 376 return BodyStartOffset + 4; 377} 378 379std::pair<StringRef, std::error_code> 380llvm::writeArchive(StringRef ArcName, 381 std::vector<NewArchiveMember> &NewMembers, 382 bool WriteSymtab, object::Archive::Kind Kind, 383 bool Deterministic, bool Thin, 384 std::unique_ptr<MemoryBuffer> OldArchiveBuf) { 385 assert((!Thin || !isBSDLike(Kind)) && "Only the gnu format has a thin mode"); 386 SmallString<128> TmpArchive; 387 int TmpArchiveFD; 388 if (auto EC = sys::fs::createUniqueFile(ArcName + ".temp-archive-%%%%%%%.a", 389 TmpArchiveFD, TmpArchive)) 390 return std::make_pair(ArcName, EC); 391 392 tool_output_file Output(TmpArchive, TmpArchiveFD); 393 raw_fd_ostream &Out = Output.os(); 394 if (Thin) 395 Out << "!<thin>\n"; 396 else 397 Out << "!<arch>\n"; 398 399 std::vector<unsigned> MemberOffsetRefs; 400 401 unsigned MemberReferenceOffset = 0; 402 if (WriteSymtab) { 403 ErrorOr<unsigned> MemberReferenceOffsetOrErr = writeSymbolTable( 404 Out, Kind, NewMembers, MemberOffsetRefs, Deterministic); 405 if (auto EC = MemberReferenceOffsetOrErr.getError()) 406 return std::make_pair(ArcName, EC); 407 MemberReferenceOffset = MemberReferenceOffsetOrErr.get(); 408 } 409 410 std::vector<unsigned> StringMapIndexes; 411 if (!isBSDLike(Kind)) 412 writeStringTable(Out, ArcName, NewMembers, StringMapIndexes, Thin); 413 414 std::vector<unsigned>::iterator StringMapIndexIter = StringMapIndexes.begin(); 415 std::vector<unsigned> MemberOffset; 416 for (const NewArchiveMember &M : NewMembers) { 417 MemoryBufferRef File = M.Buf->getMemBufferRef(); 418 unsigned Padding = 0; 419 420 unsigned Pos = Out.tell(); 421 MemberOffset.push_back(Pos); 422 423 // ld64 expects the members to be 8-byte aligned for 64-bit content and at 424 // least 4-byte aligned for 32-bit content. Opt for the larger encoding 425 // uniformly. This matches the behaviour with cctools and ensures that ld64 426 // is happy with archives that we generate. 427 if (Kind == object::Archive::K_DARWIN) 428 Padding = OffsetToAlignment(M.Buf->getBufferSize(), 8); 429 430 printMemberHeader(Out, Kind, Thin, M.MemberName, StringMapIndexIter, 431 M.ModTime, M.UID, M.GID, M.Perms, 432 M.Buf->getBufferSize() + Padding); 433 434 if (!Thin) 435 Out << File.getBuffer(); 436 437 while (Padding--) 438 Out << '\n'; 439 if (Out.tell() % 2) 440 Out << '\n'; 441 } 442 443 if (MemberReferenceOffset) { 444 Out.seek(MemberReferenceOffset); 445 for (unsigned MemberNum : MemberOffsetRefs) { 446 if (isBSDLike(Kind)) 447 Out.seek(Out.tell() + 4); // skip over the string offset 448 print32(Out, Kind, MemberOffset[MemberNum]); 449 } 450 } 451 452 Output.keep(); 453 Out.close(); 454 455 // At this point, we no longer need whatever backing memory 456 // was used to generate the NewMembers. On Windows, this buffer 457 // could be a mapped view of the file we want to replace (if 458 // we're updating an existing archive, say). In that case, the 459 // rename would still succeed, but it would leave behind a 460 // temporary file (actually the original file renamed) because 461 // a file cannot be deleted while there's a handle open on it, 462 // only renamed. So by freeing this buffer, this ensures that 463 // the last open handle on the destination file, if any, is 464 // closed before we attempt to rename. 465 OldArchiveBuf.reset(); 466 467 sys::fs::rename(TmpArchive, ArcName); 468 return std::make_pair("", std::error_code()); 469} 470