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