generator.c

   1 /* -*- c-file-style: "linux" -*-
   2
   3    rsync -- fast file replication program
   4
   5    Copyright (C) 1996-2000 by Andrew Tridgell
   6    Copyright (C) Paul Mackerras 1996
   7    Copyright (C) 2002 by Martin Pool <mbp@samba.org>
   8
   9    This program is free software; you can redistribute it and/or modify
  10    it under the terms of the GNU General Public License as published by
  11    the Free Software Foundation; either version 2 of the License, or
  12    (at your option) any later version.
  13
  14    This program is distributed in the hope that it will be useful,
  15    but WITHOUT ANY WARRANTY; without even the implied warranty of
  16    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17    GNU General Public License for more details.
  18
  19    You should have received a copy of the GNU General Public License
  20    along with this program; if not, write to the Free Software
  21    Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
  22 */
  23
  24 #include "rsync.h"
  25
  26 extern int verbose;
  27 extern int dry_run;
  28 extern int relative_paths;
  29 extern int keep_dirlinks;
  30 extern int preserve_links;
  31 extern int am_root;
  32 extern int preserve_devices;
  33 extern int preserve_hard_links;
  34 extern int preserve_perms;
  35 extern int preserve_uid;
  36 extern int preserve_gid;
  37 extern int update_only;
  38 extern int opt_ignore_existing;
  39 extern int inplace;
  40 extern int make_backups;
  41 extern int csum_length;
  42 extern int ignore_times;
  43 extern int size_only;
  44 extern int io_timeout;
  45 extern int protocol_version;
  46 extern int always_checksum;
  47 extern char *partial_dir;
  48 extern char *compare_dest;
  49 extern int link_dest;
  50 extern int whole_file;
  51 extern int local_server;
  52 extern int list_only;
  53 extern int read_batch;
  54 extern int only_existing;
  55 extern int orig_umask;
  56 extern int safe_symlinks;
  57 extern unsigned int block_size;
  58
  59 extern struct exclude_list_struct server_exclude_list;
  60
  61
  62 /* choose whether to skip a particular file */
  63 static int skip_file(char *fname, struct file_struct *file, STRUCT_STAT *st)
  64 {
  65         if (st->st_size != file->length)
  66                 return 0;
  67         if (link_dest) {
  68                 if (preserve_perms
  69                     && (st->st_mode & CHMOD_BITS) != (file->mode & CHMOD_BITS))
  70                         return 0;
  71
  72                 if (am_root && preserve_uid && st->st_uid != file->uid)
  73                         return 0;
  74
  75                 if (preserve_gid && file->gid != GID_NONE
  76                     && st->st_gid != file->gid)
  77                         return 0;
  78         }
  79
  80         /* if always checksum is set then we use the checksum instead
  81            of the file time to determine whether to sync */
  82         if (always_checksum && S_ISREG(st->st_mode)) {
  83                 char sum[MD4_SUM_LENGTH];
  84                 file_checksum(fname,sum,st->st_size);
  85                 return memcmp(sum, file->u.sum, protocol_version < 21 ? 2
  86                                                         : MD4_SUM_LENGTH) == 0;
  87         }
  88
  89         if (size_only)
  90                 return 1;
  91
  92         if (ignore_times)
  93                 return 0;
  94
  95         return cmp_modtime(st->st_mtime, file->modtime) == 0;
  96 }
  97
  98
  99 /*
 100  * NULL sum_struct means we have no checksums
 101  */
 102 void write_sum_head(int f, struct sum_struct *sum)
 103 {
 104         static struct sum_struct null_sum;
 105
 106         if (sum == NULL)
 107                 sum = &null_sum;
 108
 109         write_int(f, sum->count);
 110         write_int(f, sum->blength);
 111         if (protocol_version >= 27)
 112                 write_int(f, sum->s2length);
 113         write_int(f, sum->remainder);
 114 }
 115
 116 /*
 117  * set (initialize) the size entries in the per-file sum_struct
 118  * calculating dynamic block and checksum sizes.
 119  *
 120  * This is only called from generate_and_send_sums() but is a separate
 121  * function to encapsulate the logic.
 122  *
 123  * The block size is a rounded square root of file length.
 124  *
 125  * The checksum size is determined according to:
 126  *     blocksum_bits = BLOCKSUM_EXP + 2*log2(file_len) - log2(block_len)
 127  * provided by Donovan Baarda which gives a probability of rsync
 128  * algorithm corrupting data and falling back using the whole md4
 129  * checksums.
 130  *
 131  * This might be made one of several selectable heuristics.
 132  */
 133
 134 static void sum_sizes_sqroot(struct sum_struct *sum, uint64 len)
 135 {
 136         unsigned int blength;
 137         int s2length;
 138         uint32 c;
 139         uint64 l;
 140
 141         if (block_size) {
 142                 blength = block_size;
 143         } else if (len <= BLOCK_SIZE * BLOCK_SIZE) {
 144                 blength = BLOCK_SIZE;
 145         } else {
 146                 l = len;
 147                 c = 1;
 148                 while (l >>= 2) {
 149                         c <<= 1;
 150                 }
 151                 blength = 0;
 152                 do {
 153                         blength |= c;
 154                         if (len < (uint64)blength * blength)
 155                                 blength &= ~c;
 156                         c >>= 1;
 157                 } while (c >= 8);       /* round to multiple of 8 */
 158                 blength = MAX(blength, BLOCK_SIZE);
 159         }
 160
 161         if (protocol_version < 27) {
 162                 s2length = csum_length;
 163         } else if (csum_length == SUM_LENGTH) {
 164                 s2length = SUM_LENGTH;
 165         } else {
 166                 int b = BLOCKSUM_BIAS;
 167                 l = len;
 168                 while (l >>= 1) {
 169                         b += 2;
 170                 }
 171                 c = blength;
 172                 while (c >>= 1 && b) {
 173                         b--;
 174                 }
 175                 s2length = (b + 1 - 32 + 7) / 8; /* add a bit,
 176                                                   * subtract rollsum,
 177                                                   * round up
 178                                                   *    --optimize in compiler--
 179                                                   */
 180                 s2length = MAX(s2length, csum_length);
 181                 s2length = MIN(s2length, SUM_LENGTH);
 182         }
 183
 184         sum->flength    = len;
 185         sum->blength    = blength;
 186         sum->s2length   = s2length;
 187         sum->count      = (len + (blength - 1)) / blength;
 188         sum->remainder  = (len % blength);
 189
 190         if (sum->count && verbose > 2) {
 191                 rprintf(FINFO, "count=%.0f rem=%u blength=%u s2length=%d flength=%.0f\n",
 192                         (double)sum->count, sum->remainder, sum->blength,
 193                         sum->s2length, (double)sum->flength);
 194         }
 195 }
 196
 197
 198 /*
 199  * Generate and send a stream of signatures/checksums that describe a buffer
 200  *
 201  * Generate approximately one checksum every block_len bytes.
 202  */
 203 static void generate_and_send_sums(int fd, OFF_T len, int f_out, int f_copy)
 204 {
 205         size_t i;
 206         struct map_struct *mapbuf;
 207         struct sum_struct sum;
 208         OFF_T offset = 0;
 209
 210         sum_sizes_sqroot(&sum, len);
 211
 212         if (len > 0)
 213                 mapbuf = map_file(fd, len, MAX_MAP_SIZE, sum.blength);
 214         else
 215                 mapbuf = NULL;
 216
 217         write_sum_head(f_out, &sum);
 218
 219         for (i = 0; i < sum.count; i++) {
 220                 unsigned int n1 = MIN(len, sum.blength);
 221                 char *map = map_ptr(mapbuf, offset, n1);
 222                 uint32 sum1 = get_checksum1(map, n1);
 223                 char sum2[SUM_LENGTH];
 224
 225                 if (f_copy >= 0)
 226                         full_write(f_copy, map, n1);
 227
 228                 get_checksum2(map, n1, sum2);
 229
 230                 if (verbose > 3) {
 231                         rprintf(FINFO,
 232                                 "chunk[%.0f] offset=%.0f len=%u sum1=%08lx\n",
 233                                 (double)i, (double)offset, n1,
 234                                 (unsigned long)sum1);
 235                 }
 236                 write_int(f_out, sum1);
 237                 write_buf(f_out, sum2, sum.s2length);
 238                 len -= n1;
 239                 offset += n1;
 240         }
 241
 242         if (mapbuf)
 243                 unmap_file(mapbuf);
 244 }
 245
 246
 247
 248 /*
 249  * Acts on file number @p i from @p flist, whose name is @p fname.
 250  *
 251  * First fixes up permissions, then generates checksums for the file.
 252  *
 253  * @note This comment was added later by mbp who was trying to work it
 254  * out.  It might be wrong.
 255  */
 256 static void recv_generator(char *fname, struct file_struct *file, int i,
 257                            int f_out)
 258 {
 259         int fd, f_copy;
 260         STRUCT_STAT st, partial_st;
 261         struct file_struct *back_file;
 262         int statret, stat_errno;
 263         char *fnamecmp, *partialptr, *backupptr;
 264         char fnamecmpbuf[MAXPATHLEN];
 265
 266         if (list_only)
 267                 return;
 268
 269         if (verbose > 2)
 270                 rprintf(FINFO, "recv_generator(%s,%d)\n", safe_fname(fname), i);
 271
 272         if (server_exclude_list.head
 273             && check_exclude(&server_exclude_list, fname,
 274                              S_ISDIR(file->mode)) < 0) {
 275                 if (verbose) {
 276                         rprintf(FINFO, "skipping server-excluded file \"%s\"\n",
 277                                 safe_fname(fname));
 278                 }
 279                 return;
 280         }
 281
 282         if (dry_run > 1) {
 283                 statret = -1;
 284                 stat_errno = ENOENT;
 285         } else {
 286                 statret = link_stat(fname, &st,
 287                                     keep_dirlinks && S_ISDIR(file->mode));
 288                 stat_errno = errno;
 289         }
 290
 291         if (only_existing && statret == -1 && stat_errno == ENOENT) {
 292                 /* we only want to update existing files */
 293                 if (verbose > 1) {
 294                         rprintf(FINFO, "not creating new file \"%s\"\n",
 295                                 safe_fname(fname));
 296                 }
 297                 return;
 298         }
 299
 300         if (statret == 0 && !preserve_perms
 301             && S_ISDIR(st.st_mode) == S_ISDIR(file->mode)) {
 302                 /* if the file exists already and we aren't perserving
 303                  * permissions then act as though the remote end sent
 304                  * us the file permissions we already have */
 305                 file->mode = (file->mode & ~CHMOD_BITS)
 306                            | (st.st_mode & CHMOD_BITS);
 307         }
 308
 309         if (S_ISDIR(file->mode)) {
 310                 /* The file to be received is a directory, so we need
 311                  * to prepare appropriately.  If there is already a
 312                  * file of that name and it is *not* a directory, then
 313                  * we need to delete it.  If it doesn't exist, then
 314                  * recursively create it. */
 315
 316                 if (dry_run)
 317                         return; /* TODO: causes inaccuracies -- fix */
 318                 if (statret == 0 && !S_ISDIR(st.st_mode)) {
 319                         if (robust_unlink(fname) != 0) {
 320                                 rsyserr(FERROR, errno,
 321                                         "recv_generator: unlink %s to make room for directory",
 322                                         full_fname(fname));
 323                                 return;
 324                         }
 325                         statret = -1;
 326                 }
 327                 if (statret != 0 && do_mkdir(fname,file->mode) != 0 && errno != EEXIST) {
 328                         if (!(relative_paths && errno == ENOENT
 329                             && create_directory_path(fname, orig_umask) == 0
 330                             && do_mkdir(fname, file->mode) == 0)) {
 331                                 rsyserr(FERROR, errno,
 332                                         "recv_generator: mkdir %s failed",
 333                                         full_fname(fname));
 334                         }
 335                 }
 336                 /* f_out is set to -1 when doing final directory-permission
 337                  * and modification-time repair. */
 338                 if (set_perms(fname, file, statret ? NULL : &st, 0)
 339                     && verbose && f_out != -1)
 340                         rprintf(FINFO, "%s/\n", safe_fname(fname));
 341                 return;
 342         }
 343
 344         if (preserve_links && S_ISLNK(file->mode)) {
 345 #if SUPPORT_LINKS
 346                 char lnk[MAXPATHLEN];
 347                 int l;
 348
 349                 if (safe_symlinks && unsafe_symlink(file->u.link, fname)) {
 350                         if (verbose) {
 351                                 rprintf(FINFO, "ignoring unsafe symlink %s -> \"%s\"\n",
 352                                         full_fname(fname), file->u.link);
 353                         }
 354                         return;
 355                 }
 356                 if (statret == 0) {
 357                         l = readlink(fname,lnk,MAXPATHLEN-1);
 358                         if (l > 0) {
 359                                 lnk[l] = 0;
 360                                 /* A link already pointing to the
 361                                  * right place -- no further action
 362                                  * required. */
 363                                 if (strcmp(lnk,file->u.link) == 0) {
 364                                         set_perms(fname, file, &st,
 365                                                   PERMS_REPORT);
 366                                         return;
 367                                 }
 368                         }
 369                         /* Not a symlink, so delete whatever's
 370                          * already there and put a new symlink
 371                          * in place. */
 372                         delete_file(fname);
 373                 }
 374                 if (do_symlink(file->u.link,fname) != 0) {
 375                         rsyserr(FERROR, errno, "symlink %s -> \"%s\" failed",
 376                                 full_fname(fname), safe_fname(file->u.link));
 377                 } else {
 378                         set_perms(fname,file,NULL,0);
 379                         if (verbose) {
 380                                 rprintf(FINFO, "%s -> %s\n", safe_fname(fname),
 381                                         safe_fname(file->u.link));
 382                         }
 383                 }
 384 #endif
 385                 return;
 386         }
 387
 388 #ifdef HAVE_MKNOD
 389         if (am_root && preserve_devices && IS_DEVICE(file->mode)) {
 390                 if (statret != 0 ||
 391                     st.st_mode != file->mode ||
 392                     st.st_rdev != file->u.rdev) {
 393                         delete_file(fname);
 394                         if (verbose > 2) {
 395                                 rprintf(FINFO,"mknod(%s,0%o,0x%x)\n",
 396                                         safe_fname(fname),
 397                                         (int)file->mode, (int)file->u.rdev);
 398                         }
 399                         if (do_mknod(fname,file->mode,file->u.rdev) != 0) {
 400                                 rsyserr(FERROR, errno, "mknod %s failed",
 401                                         full_fname(fname));
 402                         } else {
 403                                 set_perms(fname,file,NULL,0);
 404                                 if (verbose) {
 405                                         rprintf(FINFO, "%s\n",
 406                                                 safe_fname(fname));
 407                                 }
 408                         }
 409                 } else {
 410                         set_perms(fname, file, &st, PERMS_REPORT);
 411                 }
 412                 return;
 413         }
 414 #endif
 415
 416         if (preserve_hard_links && hard_link_check(file, HL_CHECK_MASTER))
 417                 return;
 418
 419         if (!S_ISREG(file->mode)) {
 420                 rprintf(FINFO, "skipping non-regular file \"%s\"\n",
 421                         safe_fname(fname));
 422                 return;
 423         }
 424
 425         fnamecmp = fname;
 426
 427         if (statret == -1 && compare_dest != NULL) {
 428                 /* try the file at compare_dest instead */
 429                 pathjoin(fnamecmpbuf, sizeof fnamecmpbuf, compare_dest, fname);
 430                 if (link_stat(fnamecmpbuf, &st, 0) == 0
 431                     && S_ISREG(st.st_mode)) {
 432 #if HAVE_LINK
 433                         if (link_dest && !dry_run) {
 434                                 if (do_link(fnamecmpbuf, fname) < 0) {
 435                                         if (verbose) {
 436                                                 rsyserr(FINFO, errno,
 437                                                         "link %s => %s",
 438                                                         fnamecmpbuf,
 439                                                         safe_fname(fname));
 440                                         }
 441                                         fnamecmp = fnamecmpbuf;
 442                                 }
 443                         } else
 444 #endif
 445                                 fnamecmp = fnamecmpbuf;
 446                         statret = 0;
 447                 }
 448         }
 449
 450         if (statret == 0 && !S_ISREG(st.st_mode)) {
 451                 if (delete_file(fname) != 0)
 452                         return;
 453                 statret = -1;
 454                 stat_errno = ENOENT;
 455         }
 456
 457         if (partial_dir && (partialptr = partial_dir_fname(fname))
 458             && link_stat(partialptr, &partial_st, 0) == 0
 459             && S_ISREG(partial_st.st_mode)) {
 460                 if (statret == -1)
 461                         goto prepare_to_open;
 462         } else
 463                 partialptr = NULL;
 464
 465         if (statret == -1) {
 466                 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 467                         return;
 468                 if (stat_errno == ENOENT) {
 469                         write_int(f_out,i);
 470                         if (!dry_run && !read_batch)
 471                                 write_sum_head(f_out, NULL);
 472                 } else if (verbose > 1) {
 473                         rsyserr(FERROR, stat_errno,
 474                                 "recv_generator: failed to stat %s",
 475                                 full_fname(fname));
 476                 }
 477                 return;
 478         }
 479
 480         if (opt_ignore_existing && fnamecmp == fname) {
 481                 if (verbose > 1)
 482                         rprintf(FINFO, "%s exists\n", safe_fname(fname));
 483                 return;
 484         }
 485
 486         if (update_only && fnamecmp == fname
 487             && cmp_modtime(st.st_mtime, file->modtime) > 0) {
 488                 if (verbose > 1)
 489                         rprintf(FINFO, "%s is newer\n", safe_fname(fname));
 490                 return;
 491         }
 492
 493         if (skip_file(fnamecmp, file, &st)) {
 494                 if (fnamecmp == fname)
 495                         set_perms(fname, file, &st, PERMS_REPORT);
 496                 return;
 497         }
 498
 499 prepare_to_open:
 500         if (dry_run || read_batch) {
 501                 write_int(f_out,i);
 502                 return;
 503         }
 504
 505         if (whole_file > 0) {
 506                 write_int(f_out,i);
 507                 write_sum_head(f_out, NULL);
 508                 return;
 509         }
 510
 511         if (partialptr) {
 512                 st = partial_st;
 513                 fnamecmp = partialptr;
 514         }
 515
 516         /* open the file */
 517         fd = do_open(fnamecmp, O_RDONLY, 0);
 518
 519         if (fd == -1) {
 520                 rsyserr(FERROR, errno, "failed to open %s, continuing",
 521                         full_fname(fnamecmp));
 522             pretend_missing:
 523                 /* pretend the file didn't exist */
 524                 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 525                         return;
 526                 write_int(f_out,i);
 527                 write_sum_head(f_out, NULL);
 528                 return;
 529         }
 530
 531         if (inplace && make_backups) {
 532                 if (!(backupptr = get_backup_name(fname))) {
 533                         close(fd);
 534                         return;
 535                 }
 536                 if (!(back_file = make_file(fname, NULL, NO_EXCLUDES))) {
 537                         close(fd);
 538                         goto pretend_missing;
 539                 }
 540                 if (robust_unlink(backupptr) && errno != ENOENT) {
 541                         rsyserr(FERROR, errno, "unlink %s",
 542                                 full_fname(backupptr));
 543                         free(back_file);
 544                         close(fd);
 545                         return;
 546                 }
 547                 if ((f_copy = do_open(backupptr,
 548                     O_WRONLY | O_CREAT | O_TRUNC | O_EXCL, 0600)) < 0) {
 549                         rsyserr(FERROR, errno, "open %s",
 550                                 full_fname(backupptr));
 551                         free(back_file);
 552                         close(fd);
 553                         return;
 554                 }
 555         } else {
 556                 backupptr = NULL;
 557                 back_file = NULL;
 558                 f_copy = -1;
 559         }
 560
 561         if (verbose > 3) {
 562                 rprintf(FINFO, "gen mapped %s of size %.0f\n",
 563                         safe_fname(fnamecmp), (double)st.st_size);
 564         }
 565
 566         if (verbose > 2)
 567                 rprintf(FINFO, "generating and sending sums for %d\n", i);
 568
 569         write_int(f_out,i);
 570         generate_and_send_sums(fd, st.st_size, f_out, f_copy);
 571
 572         if (f_copy >= 0) {
 573                 close(f_copy);
 574                 set_perms(backupptr, back_file, NULL, 0);
 575                 if (verbose > 1)
 576                         rprintf(FINFO, "backed up %s to %s\n", fname, backupptr);
 577                 free(back_file);
 578         }
 579
 580         close(fd);
 581 }
 582
 583
 584 void generate_files(int f_out, struct file_list *flist, char *local_name)
 585 {
 586         int i;
 587         int phase = 0;
 588         char fbuf[MAXPATHLEN];
 589
 590         if (verbose > 2) {
 591                 rprintf(FINFO, "generator starting pid=%ld count=%d\n",
 592                         (long)getpid(), flist->count);
 593         }
 594
 595         if (verbose >= 2) {
 596                 rprintf(FINFO,
 597                         whole_file > 0
 598                         ? "delta-transmission disabled for local transfer or --whole-file\n"
 599                         : "delta transmission enabled\n");
 600         }
 601
 602         /* we expect to just sit around now, so don't exit on a
 603            timeout. If we really get a timeout then the other process should
 604            exit */
 605         io_timeout = 0;
 606
 607         for (i = 0; i < flist->count; i++) {
 608                 struct file_struct *file = flist->files[i];
 609                 struct file_struct copy;
 610
 611                 if (!file->basename)
 612                         continue;
 613                 /* we need to ensure that any directories we create have writeable
 614                    permissions initially so that we can create the files within
 615                    them. This is then fixed after the files are transferred */
 616                 if (!am_root && S_ISDIR(file->mode) && !(file->mode & S_IWUSR)) {
 617                         copy = *file;
 618                         /* XXX: Could this be causing a problem on SCO?  Perhaps their
 619                          * handling of permissions is strange? */
 620                         copy.mode |= S_IWUSR; /* user write */
 621                         file = &copy;
 622                 }
 623
 624                 recv_generator(local_name ? local_name : f_name_to(file, fbuf),
 625                                file, i, f_out);
 626         }
 627
 628         phase++;
 629         csum_length = SUM_LENGTH;
 630         ignore_times = 1;
 631
 632         if (verbose > 2)
 633                 rprintf(FINFO,"generate_files phase=%d\n",phase);
 634
 635         write_int(f_out, -1);
 636
 637         /* files can cycle through the system more than once
 638          * to catch initial checksum errors */
 639         while ((i = get_redo_num()) != -1) {
 640                 struct file_struct *file = flist->files[i];
 641                 recv_generator(local_name ? local_name : f_name_to(file, fbuf),
 642                                file, i, f_out);
 643         }
 644
 645         phase++;
 646         if (verbose > 2)
 647                 rprintf(FINFO,"generate_files phase=%d\n",phase);
 648
 649         write_int(f_out, -1);
 650
 651         if (preserve_hard_links)
 652                 do_hard_links();
 653
 654         /* now we need to fix any directory permissions that were
 655          * modified during the transfer */
 656         for (i = 0; i < flist->count; i++) {
 657                 struct file_struct *file = flist->files[i];
 658                 if (!file->basename || !S_ISDIR(file->mode))
 659                         continue;
 660                 recv_generator(local_name ? local_name : f_name(file),
 661                                file, i, -1);
 662         }
 663
 664         if (verbose > 2)
 665                 rprintf(FINFO,"generate_files finished\n");
 666 }