lib/Archive/ArchiveWriter.cpp

   1 //===-- ArchiveWriter.cpp - Write LLVM archive files ----------------------===//
   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 // Builds up an LLVM archive file (.a) containing LLVM bitcode.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "ArchiveInternals.h"
  15 #include "llvm/Module.h"
  16 #include "llvm/ADT/OwningPtr.h"
  17 #include "llvm/Bitcode/ReaderWriter.h"
  18 #include "llvm/Support/FileSystem.h"
  19 #include "llvm/Support/MemoryBuffer.h"
  20 #include "llvm/Support/Process.h"
  21 #include "llvm/Support/Signals.h"
  22 #include "llvm/Support/system_error.h"
  23 #include <fstream>
  24 #include <ostream>
  25 #include <iomanip>
  26 using namespace llvm;
  27
  28 // Write an integer using variable bit rate encoding. This saves a few bytes
  29 // per entry in the symbol table.
  30 static inline void writeInteger(unsigned num, std::ofstream& ARFile) {
  31   while (1) {
  32     if (num < 0x80) { // done?
  33       ARFile << (unsigned char)num;
  34       return;
  35     }
  36
  37     // Nope, we are bigger than a character, output the next 7 bits and set the
  38     // high bit to say that there is more coming...
  39     ARFile << (unsigned char)(0x80 | ((unsigned char)num & 0x7F));
  40     num >>= 7;  // Shift out 7 bits now...
  41   }
  42 }
  43
  44 // Compute how many bytes are taken by a given VBR encoded value. This is needed
  45 // to pre-compute the size of the symbol table.
  46 static inline unsigned numVbrBytes(unsigned num) {
  47
  48   // Note that the following nested ifs are somewhat equivalent to a binary
  49   // search. We split it in half by comparing against 2^14 first. This allows
  50   // most reasonable values to be done in 2 comparisons instead of 1 for
  51   // small ones and four for large ones. We expect this to access file offsets
  52   // in the 2^10 to 2^24 range and symbol lengths in the 2^0 to 2^8 range,
  53   // so this approach is reasonable.
  54   if (num < 1<<14) {
  55     if (num < 1<<7)
  56       return 1;
  57     else
  58       return 2;
  59   }
  60   if (num < 1<<21)
  61     return 3;
  62
  63   if (num < 1<<28)
  64     return 4;
  65   return 5; // anything >= 2^28 takes 5 bytes
  66 }
  67
  68 // Create an empty archive.
  69 Archive* Archive::CreateEmpty(const sys::Path& FilePath, LLVMContext& C) {
  70   Archive* result = new Archive(FilePath, C);
  71   return result;
  72 }
  73
  74 // Fill the ArchiveMemberHeader with the information from a member. If
  75 // TruncateNames is true, names are flattened to 15 chars or less. The sz field
  76 // is provided here instead of coming from the mbr because the member might be
  77 // stored compressed and the compressed size is not the ArchiveMember's size.
  78 // Furthermore compressed files have negative size fields to identify them as
  79 // compressed.
  80 bool
  81 Archive::fillHeader(const ArchiveMember &mbr, ArchiveMemberHeader& hdr,
  82                     int sz, bool TruncateNames) const {
  83
  84   // Set the permissions mode, uid and gid
  85   hdr.init();
  86   char buffer[32];
  87   sprintf(buffer, "%-8o", mbr.getMode());
  88   memcpy(hdr.mode,buffer,8);
  89   sprintf(buffer,  "%-6u", mbr.getUser());
  90   memcpy(hdr.uid,buffer,6);
  91   sprintf(buffer,  "%-6u", mbr.getGroup());
  92   memcpy(hdr.gid,buffer,6);
  93
  94   // Set the last modification date
  95   uint64_t secondsSinceEpoch = mbr.getModTime().toEpochTime();
  96   sprintf(buffer,"%-12u", unsigned(secondsSinceEpoch));
  97   memcpy(hdr.date,buffer,12);
  98
  99   // Get rid of trailing blanks in the name
 100   std::string mbrPath = mbr.getPath().str();
 101   size_t mbrLen = mbrPath.length();
 102   while (mbrLen > 0 && mbrPath[mbrLen-1] == ' ') {
 103     mbrPath.erase(mbrLen-1,1);
 104     mbrLen--;
 105   }
 106
 107   // Set the name field in one of its various flavors.
 108   bool writeLongName = false;
 109   if (mbr.isStringTable()) {
 110     memcpy(hdr.name,ARFILE_STRTAB_NAME,16);
 111   } else if (mbr.isSVR4SymbolTable()) {
 112     memcpy(hdr.name,ARFILE_SVR4_SYMTAB_NAME,16);
 113   } else if (mbr.isBSD4SymbolTable()) {
 114     memcpy(hdr.name,ARFILE_BSD4_SYMTAB_NAME,16);
 115   } else if (mbr.isLLVMSymbolTable()) {
 116     memcpy(hdr.name,ARFILE_LLVM_SYMTAB_NAME,16);
 117   } else if (TruncateNames) {
 118     const char* nm = mbrPath.c_str();
 119     unsigned len = mbrPath.length();
 120     size_t slashpos = mbrPath.rfind('/');
 121     if (slashpos != std::string::npos) {
 122       nm += slashpos + 1;
 123       len -= slashpos +1;
 124     }
 125     if (len > 15)
 126       len = 15;
 127     memcpy(hdr.name,nm,len);
 128     hdr.name[len] = '/';
 129   } else if (mbrPath.length() < 16 && mbrPath.find('/') == std::string::npos) {
 130     memcpy(hdr.name,mbrPath.c_str(),mbrPath.length());
 131     hdr.name[mbrPath.length()] = '/';
 132   } else {
 133     std::string nm = "#1/";
 134     nm += utostr(mbrPath.length());
 135     memcpy(hdr.name,nm.data(),nm.length());
 136     if (sz < 0)
 137       sz -= mbrPath.length();
 138     else
 139       sz += mbrPath.length();
 140     writeLongName = true;
 141   }
 142
 143   // Set the size field
 144   if (sz < 0) {
 145     buffer[0] = '-';
 146     sprintf(&buffer[1],"%-9u",(unsigned)-sz);
 147   } else {
 148     sprintf(buffer, "%-10u", (unsigned)sz);
 149   }
 150   memcpy(hdr.size,buffer,10);
 151
 152   return writeLongName;
 153 }
 154
 155 // Insert a file into the archive before some other member. This also takes care
 156 // of extracting the necessary flags and information from the file.
 157 bool
 158 Archive::addFileBefore(const sys::Path& filePath, iterator where,
 159                         std::string* ErrMsg) {
 160   bool Exists;
 161   if (sys::fs::exists(filePath.str(), Exists) || !Exists) {
 162     if (ErrMsg)
 163       *ErrMsg = "Can not add a non-existent file to archive";
 164     return true;
 165   }
 166
 167   ArchiveMember* mbr = new ArchiveMember(this);
 168
 169   mbr->data = 0;
 170   mbr->path = filePath;
 171   const sys::FileStatus *FSInfo = mbr->path.getFileStatus(false, ErrMsg);
 172   if (!FSInfo) {
 173     delete mbr;
 174     return true;
 175   }
 176   mbr->info = *FSInfo;
 177
 178   unsigned flags = 0;
 179   bool hasSlash = filePath.str().find('/') != std::string::npos;
 180   if (hasSlash)
 181     flags |= ArchiveMember::HasPathFlag;
 182   if (hasSlash || filePath.str().length() > 15)
 183     flags |= ArchiveMember::HasLongFilenameFlag;
 184
 185   sys::LLVMFileType type;
 186   if (sys::fs::identify_magic(mbr->path.str(), type))
 187     type = sys::Unknown_FileType;
 188   switch (type) {
 189     case sys::Bitcode_FileType:
 190       flags |= ArchiveMember::BitcodeFlag;
 191       break;
 192     default:
 193       break;
 194   }
 195   mbr->flags = flags;
 196   members.insert(where,mbr);
 197   return false;
 198 }
 199
 200 // Write one member out to the file.
 201 bool
 202 Archive::writeMember(
 203   const ArchiveMember& member,
 204   std::ofstream& ARFile,
 205   bool CreateSymbolTable,
 206   bool TruncateNames,
 207   bool ShouldCompress,
 208   std::string* ErrMsg
 209 ) {
 210
 211   unsigned filepos = ARFile.tellp();
 212   filepos -= 8;
 213
 214   // Get the data and its size either from the
 215   // member's in-memory data or directly from the file.
 216   size_t fSize = member.getSize();
 217   const char *data = (const char*)member.getData();
 218   MemoryBuffer *mFile = 0;
 219   if (!data) {
 220     OwningPtr<MemoryBuffer> File;
 221     if (error_code ec = MemoryBuffer::getFile(member.getPath().c_str(), File)) {
 222       if (ErrMsg)
 223         *ErrMsg = ec.message();
 224       return true;
 225     }
 226     mFile = File.take();
 227     data = mFile->getBufferStart();
 228     fSize = mFile->getBufferSize();
 229   }
 230
 231   // Now that we have the data in memory, update the
 232   // symbol table if it's a bitcode file.
 233   if (CreateSymbolTable && member.isBitcode()) {
 234     std::vector<std::string> symbols;
 235     std::string FullMemberName = archPath.str() + "(" + member.getPath().str()
 236       + ")";
 237     Module* M =
 238       GetBitcodeSymbols(data, fSize, FullMemberName, Context, symbols, ErrMsg);
 239
 240     // If the bitcode parsed successfully
 241     if ( M ) {
 242       for (std::vector<std::string>::iterator SI = symbols.begin(),
 243            SE = symbols.end(); SI != SE; ++SI) {
 244
 245         std::pair<SymTabType::iterator,bool> Res =
 246           symTab.insert(std::make_pair(*SI,filepos));
 247
 248         if (Res.second) {
 249           symTabSize += SI->length() +
 250                         numVbrBytes(SI->length()) +
 251                         numVbrBytes(filepos);
 252         }
 253       }
 254       // We don't need this module any more.
 255       delete M;
 256     } else {
 257       delete mFile;
 258       if (ErrMsg)
 259         *ErrMsg = "Can't parse bitcode member: " + member.getPath().str()
 260           + ": " + *ErrMsg;
 261       return true;
 262     }
 263   }
 264
 265   int hdrSize = fSize;
 266
 267   // Compute the fields of the header
 268   ArchiveMemberHeader Hdr;
 269   bool writeLongName = fillHeader(member,Hdr,hdrSize,TruncateNames);
 270
 271   // Write header to archive file
 272   ARFile.write((char*)&Hdr, sizeof(Hdr));
 273
 274   // Write the long filename if its long
 275   if (writeLongName) {
 276     ARFile.write(member.getPath().str().data(),
 277                  member.getPath().str().length());
 278   }
 279
 280   // Write the (possibly compressed) member's content to the file.
 281   ARFile.write(data,fSize);
 282
 283   // Make sure the member is an even length
 284   if ((ARFile.tellp() & 1) == 1)
 285     ARFile << ARFILE_PAD;
 286
 287   // Close the mapped file if it was opened
 288   delete mFile;
 289   return false;
 290 }
 291
 292 // Write out the LLVM symbol table as an archive member to the file.
 293 void
 294 Archive::writeSymbolTable(std::ofstream& ARFile) {
 295
 296   // Construct the symbol table's header
 297   ArchiveMemberHeader Hdr;
 298   Hdr.init();
 299   memcpy(Hdr.name,ARFILE_LLVM_SYMTAB_NAME,16);
 300   uint64_t secondsSinceEpoch = sys::TimeValue::now().toEpochTime();
 301   char buffer[32];
 302   sprintf(buffer, "%-8o", 0644);
 303   memcpy(Hdr.mode,buffer,8);
 304   sprintf(buffer, "%-6u", sys::Process::GetCurrentUserId());
 305   memcpy(Hdr.uid,buffer,6);
 306   sprintf(buffer, "%-6u", sys::Process::GetCurrentGroupId());
 307   memcpy(Hdr.gid,buffer,6);
 308   sprintf(buffer,"%-12u", unsigned(secondsSinceEpoch));
 309   memcpy(Hdr.date,buffer,12);
 310   sprintf(buffer,"%-10u",symTabSize);
 311   memcpy(Hdr.size,buffer,10);
 312
 313   // Write the header
 314   ARFile.write((char*)&Hdr, sizeof(Hdr));
 315
 316 #ifndef NDEBUG
 317   // Save the starting position of the symbol tables data content.
 318   unsigned startpos = ARFile.tellp();
 319 #endif
 320
 321   // Write out the symbols sequentially
 322   for ( Archive::SymTabType::iterator I = symTab.begin(), E = symTab.end();
 323         I != E; ++I)
 324   {
 325     // Write out the file index
 326     writeInteger(I->second, ARFile);
 327     // Write out the length of the symbol
 328     writeInteger(I->first.length(), ARFile);
 329     // Write out the symbol
 330     ARFile.write(I->first.data(), I->first.length());
 331   }
 332
 333 #ifndef NDEBUG
 334   // Now that we're done with the symbol table, get the ending file position
 335   unsigned endpos = ARFile.tellp();
 336 #endif
 337
 338   // Make sure that the amount we wrote is what we pre-computed. This is
 339   // critical for file integrity purposes.
 340   assert(endpos - startpos == symTabSize && "Invalid symTabSize computation");
 341
 342   // Make sure the symbol table is even sized
 343   if (symTabSize % 2 != 0 )
 344     ARFile << ARFILE_PAD;
 345 }
 346
 347 // Write the entire archive to the file specified when the archive was created.
 348 // This writes to a temporary file first. Options are for creating a symbol
 349 // table, flattening the file names (no directories, 15 chars max) and
 350 // compressing each archive member.
 351 bool
 352 Archive::writeToDisk(bool CreateSymbolTable, bool TruncateNames, bool Compress,
 353                      std::string* ErrMsg)
 354 {
 355   // Make sure they haven't opened up the file, not loaded it,
 356   // but are now trying to write it which would wipe out the file.
 357   if (members.empty() && mapfile && mapfile->getBufferSize() > 8) {
 358     if (ErrMsg)
 359       *ErrMsg = "Can't write an archive not opened for writing";
 360     return true;
 361   }
 362
 363   // Create a temporary file to store the archive in
 364   sys::Path TmpArchive = archPath;
 365   if (TmpArchive.createTemporaryFileOnDisk(ErrMsg))
 366     return true;
 367
 368   // Make sure the temporary gets removed if we crash
 369   sys::RemoveFileOnSignal(TmpArchive);
 370
 371   // Create archive file for output.
 372   std::ios::openmode io_mode = std::ios::out | std::ios::trunc |
 373                                std::ios::binary;
 374   std::ofstream ArchiveFile(TmpArchive.c_str(), io_mode);
 375
 376   // Check for errors opening or creating archive file.
 377   if (!ArchiveFile.is_open() || ArchiveFile.bad()) {
 378     TmpArchive.eraseFromDisk();
 379     if (ErrMsg)
 380       *ErrMsg = "Error opening archive file: " + archPath.str();
 381     return true;
 382   }
 383
 384   // If we're creating a symbol table, reset it now
 385   if (CreateSymbolTable) {
 386     symTabSize = 0;
 387     symTab.clear();
 388   }
 389
 390   // Write magic string to archive.
 391   ArchiveFile << ARFILE_MAGIC;
 392
 393   // Loop over all member files, and write them out. Note that this also
 394   // builds the symbol table, symTab.
 395   for (MembersList::iterator I = begin(), E = end(); I != E; ++I) {
 396     if (writeMember(*I, ArchiveFile, CreateSymbolTable,
 397                      TruncateNames, Compress, ErrMsg)) {
 398       TmpArchive.eraseFromDisk();
 399       ArchiveFile.close();
 400       return true;
 401     }
 402   }
 403
 404   // Close archive file.
 405   ArchiveFile.close();
 406
 407   // Write the symbol table
 408   if (CreateSymbolTable) {
 409     // At this point we have written a file that is a legal archive but it
 410     // doesn't have a symbol table in it. To aid in faster reading and to
 411     // ensure compatibility with other archivers we need to put the symbol
 412     // table first in the file. Unfortunately, this means mapping the file
 413     // we just wrote back in and copying it to the destination file.
 414     sys::Path FinalFilePath = archPath;
 415
 416     // Map in the archive we just wrote.
 417     {
 418     OwningPtr<MemoryBuffer> arch;
 419     if (error_code ec = MemoryBuffer::getFile(TmpArchive.c_str(), arch)) {
 420       if (ErrMsg)
 421         *ErrMsg = ec.message();
 422       return true;
 423     }
 424     const char* base = arch->getBufferStart();
 425
 426     // Open another temporary file in order to avoid invalidating the
 427     // mmapped data
 428     if (FinalFilePath.createTemporaryFileOnDisk(ErrMsg))
 429       return true;
 430     sys::RemoveFileOnSignal(FinalFilePath);
 431
 432     std::ofstream FinalFile(FinalFilePath.c_str(), io_mode);
 433     if (!FinalFile.is_open() || FinalFile.bad()) {
 434       TmpArchive.eraseFromDisk();
 435       if (ErrMsg)
 436         *ErrMsg = "Error opening archive file: " + FinalFilePath.str();
 437       return true;
 438     }
 439
 440     // Write the file magic number
 441     FinalFile << ARFILE_MAGIC;
 442
 443     // If there is a foreign symbol table, put it into the file now. Most
 444     // ar(1) implementations require the symbol table to be first but llvm-ar
 445     // can deal with it being after a foreign symbol table. This ensures
 446     // compatibility with other ar(1) implementations as well as allowing the
 447     // archive to store both native .o and LLVM .bc files, both indexed.
 448     if (foreignST) {
 449       if (writeMember(*foreignST, FinalFile, false, false, false, ErrMsg)) {
 450         FinalFile.close();
 451         TmpArchive.eraseFromDisk();
 452         return true;
 453       }
 454     }
 455
 456     // Put out the LLVM symbol table now.
 457     writeSymbolTable(FinalFile);
 458
 459     // Copy the temporary file contents being sure to skip the file's magic
 460     // number.
 461     FinalFile.write(base + sizeof(ARFILE_MAGIC)-1,
 462       arch->getBufferSize()-sizeof(ARFILE_MAGIC)+1);
 463
 464     // Close up shop
 465     FinalFile.close();
 466     } // free arch.
 467
 468     // Move the final file over top of TmpArchive
 469     if (FinalFilePath.renamePathOnDisk(TmpArchive, ErrMsg))
 470       return true;
 471   }
 472
 473   // Before we replace the actual archive, we need to forget all the
 474   // members, since they point to data in that old archive. We need to do
 475   // this because we cannot replace an open file on Windows.
 476   cleanUpMemory();
 477
 478   if (TmpArchive.renamePathOnDisk(archPath, ErrMsg))
 479     return true;
 480
 481   // Set correct read and write permissions after temporary file is moved
 482   // to final destination path.
 483   if (archPath.makeReadableOnDisk(ErrMsg))
 484     return true;
 485   if (archPath.makeWriteableOnDisk(ErrMsg))
 486     return true;
 487
 488   return false;
 489 }