lib/Archive/ArchiveWriter.cpp

   1 //===-- ArchiveWriter.cpp - Write LLVM archive files ----------------------===//
   2 //
   3 //                     The LLVM Compiler Infrastructure
   4 //
   5 // This file was developed by Reid Spencer and is distributed under the
   6 // University of Illinois Open Source License. See LICENSE.TXT for details.
   7 //
   8 //===----------------------------------------------------------------------===//
   9 //
  10 // Builds up an LLVM archive file (.a) containing LLVM bytecode.
  11 //
  12 //===----------------------------------------------------------------------===//
  13
  14 #include "ArchiveInternals.h"
  15 #include "llvm/Bytecode/Reader.h"
  16 #include "llvm/Support/Compressor.h"
  17 #include "llvm/System/Signals.h"
  18 #include "llvm/System/Process.h"
  19 #include <fstream>
  20 #include <iostream>
  21 #include <iomanip>
  22
  23 using namespace llvm;
  24
  25 // Write an integer using variable bit rate encoding. This saves a few bytes
  26 // per entry in the symbol table.
  27 inline void writeInteger(unsigned num, std::ofstream& ARFile) {
  28   while (1) {
  29     if (num < 0x80) { // done?
  30       ARFile << (unsigned char)num;
  31       return;
  32     }
  33
  34     // Nope, we are bigger than a character, output the next 7 bits and set the
  35     // high bit to say that there is more coming...
  36     ARFile << (unsigned char)(0x80 | ((unsigned char)num & 0x7F));
  37     num >>= 7;  // Shift out 7 bits now...
  38   }
  39 }
  40
  41 // Compute how many bytes are taken by a given VBR encoded value. This is needed
  42 // to pre-compute the size of the symbol table.
  43 inline unsigned numVbrBytes(unsigned num) {
  44
  45   // Note that the following nested ifs are somewhat equivalent to a binary
  46   // search. We split it in half by comparing against 2^14 first. This allows
  47   // most reasonable values to be done in 2 comparisons instead of 1 for
  48   // small ones and four for large ones. We expect this to access file offsets
  49   // in the 2^10 to 2^24 range and symbol lengths in the 2^0 to 2^8 range,
  50   // so this approach is reasonable.
  51   if (num < 1<<14)
  52     if (num < 1<<7)
  53       return 1;
  54     else
  55       return 2;
  56   if (num < 1<<21)
  57     return 3;
  58
  59   if (num < 1<<28)
  60     return 4;
  61   return 5; // anything >= 2^28 takes 5 bytes
  62 }
  63
  64 // Create an empty archive.
  65 Archive*
  66 Archive::CreateEmpty(const sys::Path& FilePath ) {
  67   Archive* result = new Archive(FilePath,false);
  68   return result;
  69 }
  70
  71 // Fill the ArchiveMemberHeader with the information from a member. If
  72 // TruncateNames is true, names are flattened to 15 chars or less. The sz field
  73 // is provided here instead of coming from the mbr because the member might be
  74 // stored compressed and the compressed size is not the ArchiveMember's size.
  75 // Furthermore compressed files have negative size fields to identify them as
  76 // compressed.
  77 bool
  78 Archive::fillHeader(const ArchiveMember &mbr, ArchiveMemberHeader& hdr,
  79                     int sz, bool TruncateNames) const {
  80
  81   // Set the permissions mode, uid and gid
  82   hdr.init();
  83   char buffer[32];
  84   sprintf(buffer, "%-8o", mbr.getMode());
  85   memcpy(hdr.mode,buffer,8);
  86   sprintf(buffer,  "%-6u", mbr.getUser());
  87   memcpy(hdr.uid,buffer,6);
  88   sprintf(buffer,  "%-6u", mbr.getGroup());
  89   memcpy(hdr.gid,buffer,6);
  90
  91   // Set the last modification date
  92   uint64_t secondsSinceEpoch = mbr.getModTime().toEpochTime();
  93   sprintf(buffer,"%-12u", unsigned(secondsSinceEpoch));
  94   memcpy(hdr.date,buffer,12);
  95
  96   // Get rid of trailing blanks in the name
  97   std::string mbrPath = mbr.getPath().toString();
  98   size_t mbrLen = mbrPath.length();
  99   while (mbrLen > 0 && mbrPath[mbrLen-1] == ' ') {
 100     mbrPath.erase(mbrLen-1,1);
 101     mbrLen--;
 102   }
 103
 104   // Set the name field in one of its various flavors.
 105   bool writeLongName = false;
 106   if (mbr.isStringTable()) {
 107     memcpy(hdr.name,ARFILE_STRTAB_NAME,16);
 108   } else if (mbr.isSVR4SymbolTable()) {
 109     memcpy(hdr.name,ARFILE_SVR4_SYMTAB_NAME,16);
 110   } else if (mbr.isBSD4SymbolTable()) {
 111     memcpy(hdr.name,ARFILE_BSD4_SYMTAB_NAME,16);
 112   } else if (mbr.isLLVMSymbolTable()) {
 113     memcpy(hdr.name,ARFILE_LLVM_SYMTAB_NAME,16);
 114   } else if (TruncateNames) {
 115     const char* nm = mbrPath.c_str();
 116     unsigned len = mbrPath.length();
 117     size_t slashpos = mbrPath.rfind('/');
 118     if (slashpos != std::string::npos) {
 119       nm += slashpos + 1;
 120       len -= slashpos +1;
 121     }
 122     if (len > 15)
 123       len = 15;
 124     memcpy(hdr.name,nm,len);
 125     hdr.name[len] = '/';
 126   } else if (mbrPath.length() < 16 && mbrPath.find('/') == std::string::npos) {
 127     memcpy(hdr.name,mbrPath.c_str(),mbrPath.length());
 128     hdr.name[mbrPath.length()] = '/';
 129   } else {
 130     std::string nm = "#1/";
 131     nm += utostr(mbrPath.length());
 132     memcpy(hdr.name,nm.data(),nm.length());
 133     if (sz < 0)
 134       sz -= mbrPath.length();
 135     else
 136       sz += mbrPath.length();
 137     writeLongName = true;
 138   }
 139
 140   // Set the size field
 141   if (sz < 0) {
 142     buffer[0] = '-';
 143     sprintf(&buffer[1],"%-9u",(unsigned)-sz);
 144   } else {
 145     sprintf(buffer, "%-10u", (unsigned)sz);
 146   }
 147   memcpy(hdr.size,buffer,10);
 148
 149   return writeLongName;
 150 }
 151
 152 // Insert a file into the archive before some other member. This also takes care
 153 // of extracting the necessary flags and information from the file.
 154 void
 155 Archive::addFileBefore(const sys::Path& filePath, iterator where) {
 156   assert(filePath.exists() && "Can't add a non-existent file");
 157
 158   ArchiveMember* mbr = new ArchiveMember(this);
 159
 160   mbr->data = 0;
 161   mbr->path = filePath;
 162   std::string err;
 163   if (mbr->path.getFileStatus(mbr->info, &err))
 164     throw err;
 165
 166   unsigned flags = 0;
 167   bool hasSlash = filePath.toString().find('/') != std::string::npos;
 168   if (hasSlash)
 169     flags |= ArchiveMember::HasPathFlag;
 170   if (hasSlash || filePath.toString().length() > 15)
 171     flags |= ArchiveMember::HasLongFilenameFlag;
 172   std::string magic;
 173   mbr->path.getMagicNumber(magic,4);
 174   switch (sys::IdentifyFileType(magic.c_str(),4)) {
 175     case sys::BytecodeFileType:
 176       flags |= ArchiveMember::BytecodeFlag;
 177       break;
 178     case sys::CompressedBytecodeFileType:
 179       flags |= ArchiveMember::CompressedBytecodeFlag;
 180       break;
 181     default:
 182       break;
 183   }
 184   mbr->flags = flags;
 185   members.insert(where,mbr);
 186 }
 187
 188 // Write one member out to the file.
 189 bool
 190 Archive::writeMember(
 191   const ArchiveMember& member,
 192   std::ofstream& ARFile,
 193   bool CreateSymbolTable,
 194   bool TruncateNames,
 195   bool ShouldCompress,
 196   std::string* error
 197 ) {
 198
 199   unsigned filepos = ARFile.tellp();
 200   filepos -= 8;
 201
 202   // Get the data and its size either from the
 203   // member's in-memory data or directly from the file.
 204   size_t fSize = member.getSize();
 205   const char* data = (const char*)member.getData();
 206   sys::MappedFile* mFile = 0;
 207   if (!data) {
 208     mFile = new sys::MappedFile(member.getPath());
 209     data = (const char*) mFile->map();
 210     fSize = mFile->size();
 211   }
 212
 213   // Now that we have the data in memory, update the
 214   // symbol table if its a bytecode file.
 215   if (CreateSymbolTable &&
 216       (member.isBytecode() || member.isCompressedBytecode())) {
 217     std::vector<std::string> symbols;
 218     std::string FullMemberName = archPath.toString() + "(" +
 219       member.getPath().toString()
 220       + ")";
 221     ModuleProvider* MP = GetBytecodeSymbols(
 222       (const unsigned char*)data,fSize,FullMemberName, symbols);
 223
 224     // If the bytecode parsed successfully
 225     if ( MP ) {
 226       for (std::vector<std::string>::iterator SI = symbols.begin(),
 227            SE = symbols.end(); SI != SE; ++SI) {
 228
 229         std::pair<SymTabType::iterator,bool> Res =
 230           symTab.insert(std::make_pair(*SI,filepos));
 231
 232         if (Res.second) {
 233           symTabSize += SI->length() +
 234                         numVbrBytes(SI->length()) +
 235                         numVbrBytes(filepos);
 236         }
 237       }
 238       // We don't need this module any more.
 239       delete MP;
 240     } else {
 241       if (mFile != 0) {
 242         mFile->close();
 243         delete mFile;
 244       }
 245       if (error)
 246         *error = "Can't parse bytecode member: " + member.getPath().toString();
 247     }
 248   }
 249
 250   // Determine if we actually should compress this member
 251   bool willCompress =
 252       (ShouldCompress &&
 253       !member.isCompressed() &&
 254       !member.isCompressedBytecode() &&
 255       !member.isLLVMSymbolTable() &&
 256       !member.isSVR4SymbolTable() &&
 257       !member.isBSD4SymbolTable());
 258
 259   // Perform the compression. Note that if the file is uncompressed bytecode
 260   // then we turn the file into compressed bytecode rather than treating it as
 261   // compressed data. This is necessary since it allows us to determine that the
 262   // file contains bytecode instead of looking like a regular compressed data
 263   // member. A compressed bytecode file has its content compressed but has a
 264   // magic number of "llvc". This acounts for the +/-4 arithmetic in the code
 265   // below.
 266   int hdrSize;
 267   if (willCompress) {
 268     char* output = 0;
 269     if (member.isBytecode()) {
 270       data +=4;
 271       fSize -= 4;
 272     }
 273     fSize = Compressor::compressToNewBuffer(data,fSize,output,error);
 274     if (fSize == 0)
 275       return false;
 276     data = output;
 277     if (member.isBytecode())
 278       hdrSize = -fSize-4;
 279     else
 280       hdrSize = -fSize;
 281   } else {
 282     hdrSize = fSize;
 283   }
 284
 285   // Compute the fields of the header
 286   ArchiveMemberHeader Hdr;
 287   bool writeLongName = fillHeader(member,Hdr,hdrSize,TruncateNames);
 288
 289   // Write header to archive file
 290   ARFile.write((char*)&Hdr, sizeof(Hdr));
 291
 292   // Write the long filename if its long
 293   if (writeLongName) {
 294     ARFile.write(member.getPath().toString().data(),
 295                  member.getPath().toString().length());
 296   }
 297
 298   // Make sure we write the compressed bytecode magic number if we should.
 299   if (willCompress && member.isBytecode())
 300     ARFile.write("llvc",4);
 301
 302   // Write the (possibly compressed) member's content to the file.
 303   ARFile.write(data,fSize);
 304
 305   // Make sure the member is an even length
 306   if ((ARFile.tellp() & 1) == 1)
 307     ARFile << ARFILE_PAD;
 308
 309   // Free the compressed data, if necessary
 310   if (willCompress) {
 311     free((void*)data);
 312   }
 313
 314   // Close the mapped file if it was opened
 315   if (mFile != 0) {
 316     mFile->close();
 317     delete mFile;
 318   }
 319   return true;
 320 }
 321
 322 // Write out the LLVM symbol table as an archive member to the file.
 323 void
 324 Archive::writeSymbolTable(std::ofstream& ARFile) {
 325
 326   // Construct the symbol table's header
 327   ArchiveMemberHeader Hdr;
 328   Hdr.init();
 329   memcpy(Hdr.name,ARFILE_LLVM_SYMTAB_NAME,16);
 330   uint64_t secondsSinceEpoch = sys::TimeValue::now().toEpochTime();
 331   char buffer[32];
 332   sprintf(buffer, "%-8o", 0644);
 333   memcpy(Hdr.mode,buffer,8);
 334   sprintf(buffer, "%-6u", sys::Process::GetCurrentUserId());
 335   memcpy(Hdr.uid,buffer,6);
 336   sprintf(buffer, "%-6u", sys::Process::GetCurrentGroupId());
 337   memcpy(Hdr.gid,buffer,6);
 338   sprintf(buffer,"%-12u", unsigned(secondsSinceEpoch));
 339   memcpy(Hdr.date,buffer,12);
 340   sprintf(buffer,"%-10u",symTabSize);
 341   memcpy(Hdr.size,buffer,10);
 342
 343   // Write the header
 344   ARFile.write((char*)&Hdr, sizeof(Hdr));
 345
 346   // Save the starting position of the symbol tables data content.
 347   unsigned startpos = ARFile.tellp();
 348
 349   // Write out the symbols sequentially
 350   for ( Archive::SymTabType::iterator I = symTab.begin(), E = symTab.end();
 351         I != E; ++I)
 352   {
 353     // Write out the file index
 354     writeInteger(I->second, ARFile);
 355     // Write out the length of the symbol
 356     writeInteger(I->first.length(), ARFile);
 357     // Write out the symbol
 358     ARFile.write(I->first.data(), I->first.length());
 359   }
 360
 361   // Now that we're done with the symbol table, get the ending file position
 362   unsigned endpos = ARFile.tellp();
 363
 364   // Make sure that the amount we wrote is what we pre-computed. This is
 365   // critical for file integrity purposes.
 366   assert(endpos - startpos == symTabSize && "Invalid symTabSize computation");
 367
 368   // Make sure the symbol table is even sized
 369   if (symTabSize % 2 != 0 )
 370     ARFile << ARFILE_PAD;
 371 }
 372
 373 // Write the entire archive to the file specified when the archive was created.
 374 // This writes to a temporary file first. Options are for creating a symbol
 375 // table, flattening the file names (no directories, 15 chars max) and
 376 // compressing each archive member.
 377 bool
 378 Archive::writeToDisk(bool CreateSymbolTable, bool TruncateNames, bool Compress,
 379                      std::string* error)
 380 {
 381   // Make sure they haven't opened up the file, not loaded it,
 382   // but are now trying to write it which would wipe out the file.
 383   assert(!(members.empty() && mapfile->size() > 8) &&
 384          "Can't write an archive not opened for writing");
 385
 386   // Create a temporary file to store the archive in
 387   sys::Path TmpArchive = archPath;
 388   TmpArchive.createTemporaryFileOnDisk();
 389
 390   // Make sure the temporary gets removed if we crash
 391   sys::RemoveFileOnSignal(TmpArchive);
 392
 393   // Create archive file for output.
 394   std::ios::openmode io_mode = std::ios::out | std::ios::trunc |
 395                                std::ios::binary;
 396   std::ofstream ArchiveFile(TmpArchive.c_str(), io_mode);
 397
 398   // Check for errors opening or creating archive file.
 399   if (!ArchiveFile.is_open() || ArchiveFile.bad()) {
 400     if (TmpArchive.exists())
 401       TmpArchive.eraseFromDisk();
 402     if (error)
 403       *error = "Error opening archive file: " + archPath.toString();
 404     return false;
 405   }
 406
 407   // If we're creating a symbol table, reset it now
 408   if (CreateSymbolTable) {
 409     symTabSize = 0;
 410     symTab.clear();
 411   }
 412
 413   // Write magic string to archive.
 414   ArchiveFile << ARFILE_MAGIC;
 415
 416   // Loop over all member files, and write them out. Note that this also
 417   // builds the symbol table, symTab.
 418   for (MembersList::iterator I = begin(), E = end(); I != E; ++I) {
 419     if (!writeMember(*I, ArchiveFile, CreateSymbolTable,
 420                      TruncateNames, Compress, error)) {
 421       if (TmpArchive.exists())
 422         TmpArchive.eraseFromDisk();
 423       ArchiveFile.close();
 424       return false;
 425     }
 426   }
 427
 428   // Close archive file.
 429   ArchiveFile.close();
 430
 431   // Write the symbol table
 432   if (CreateSymbolTable) {
 433     // At this point we have written a file that is a legal archive but it
 434     // doesn't have a symbol table in it. To aid in faster reading and to
 435     // ensure compatibility with other archivers we need to put the symbol
 436     // table first in the file. Unfortunately, this means mapping the file
 437     // we just wrote back in and copying it to the destination file.
 438
 439     // Map in the archive we just wrote.
 440     sys::MappedFile arch(TmpArchive);
 441     const char* base = (const char*) arch.map();
 442
 443     // Open another temporary file in order to avoid invalidating the
 444     // mmapped data
 445     sys::Path FinalFilePath = archPath;
 446     FinalFilePath.createTemporaryFileOnDisk();
 447     sys::RemoveFileOnSignal(FinalFilePath);
 448
 449     std::ofstream FinalFile(FinalFilePath.c_str(), io_mode);
 450     if (!FinalFile.is_open() || FinalFile.bad()) {
 451       if (TmpArchive.exists())
 452         TmpArchive.eraseFromDisk();
 453       if (error)
 454         *error = "Error opening archive file: " + FinalFilePath.toString();
 455       return false;
 456     }
 457
 458     // Write the file magic number
 459     FinalFile << ARFILE_MAGIC;
 460
 461     // If there is a foreign symbol table, put it into the file now. Most
 462     // ar(1) implementations require the symbol table to be first but llvm-ar
 463     // can deal with it being after a foreign symbol table. This ensures
 464     // compatibility with other ar(1) implementations as well as allowing the
 465     // archive to store both native .o and LLVM .bc files, both indexed.
 466     if (foreignST) {
 467       if (!writeMember(*foreignST, FinalFile, false, false, false, error)) {
 468         FinalFile.close();
 469         if (TmpArchive.exists())
 470           TmpArchive.eraseFromDisk();
 471         return false;
 472       }
 473     }
 474
 475     // Put out the LLVM symbol table now.
 476     writeSymbolTable(FinalFile);
 477
 478     // Copy the temporary file contents being sure to skip the file's magic
 479     // number.
 480     FinalFile.write(base + sizeof(ARFILE_MAGIC)-1,
 481       arch.size()-sizeof(ARFILE_MAGIC)+1);
 482
 483     // Close up shop
 484     FinalFile.close();
 485     arch.close();
 486
 487     // Move the final file over top of TmpArchive
 488     FinalFilePath.renamePathOnDisk(TmpArchive);
 489   }
 490
 491   // Before we replace the actual archive, we need to forget all the
 492   // members, since they point to data in that old archive. We need to do
 493   // this because we cannot replace an open file on Windows.
 494   cleanUpMemory();
 495
 496   TmpArchive.renamePathOnDisk(archPath);
 497
 498   return true;
 499 }