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