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