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                 blength = MIN(blength, MAX_BLOCK_SIZE);
 158         }
 159
 160         if (protocol_version < 27) {
 161                 s2length = csum_length;
 162         } else if (csum_length == SUM_LENGTH) {
 163                 s2length = SUM_LENGTH;
 164         } else {
 165                 int b = BLOCKSUM_BIAS;
 166                 l = len;
 167                 while (l >>= 1) {
 168                         b += 2;
 169                 }
 170                 c = blength;
 171                 while (c >>= 1 && b) {
 172                         b--;
 173                 }
 174                 s2length = (b + 1 - 32 + 7) / 8; /* add a bit,
 175                                                   * subtract rollsum,
 176                                                   * round up
 177                                                   *    --optimize in compiler--
 178                                                   */
 179                 s2length = MAX(s2length, csum_length);
 180                 s2length = MIN(s2length, SUM_LENGTH);
 181         }
 182
 183         sum->flength    = len;
 184         sum->blength    = blength;
 185         sum->s2length   = s2length;
 186         sum->count      = (len + (blength - 1)) / blength;
 187         sum->remainder  = (len % blength);
 188
 189         if (sum->count && verbose > 2) {
 190                 rprintf(FINFO, "count=%.0f rem=%u blength=%u s2length=%d flength=%.0f\n",
 191                         (double)sum->count, sum->remainder, sum->blength,
 192                         sum->s2length, (double)sum->flength);
 193         }
 194 }
 195
 196
 197 /*
 198  * Generate and send a stream of signatures/checksums that describe a buffer
 199  *
 200  * Generate approximately one checksum every block_len bytes.
 201  */
 202 static void generate_and_send_sums(int fd, OFF_T len, int f_out)
 203 {
 204         size_t i;
 205         struct map_struct *mapbuf;
 206         struct sum_struct sum;
 207         OFF_T offset = 0;
 208
 209         sum_sizes_sqroot(&sum, len);
 210
 211         if (len > 0)
 212                 mapbuf = map_file(fd, len, MAX_MAP_SIZE, sum.blength);
 213         else
 214                 mapbuf = NULL;
 215
 216         write_sum_head(f_out, &sum);
 217
 218         for (i = 0; i < sum.count; i++) {
 219                 unsigned int n1 = MIN(len, sum.blength);
 220                 char *map = map_ptr(mapbuf, offset, n1);
 221                 uint32 sum1 = get_checksum1(map, n1);
 222                 char sum2[SUM_LENGTH];
 223
 224                 get_checksum2(map, n1, sum2);
 225
 226                 if (verbose > 3) {
 227                         rprintf(FINFO,
 228                                 "chunk[%.0f] offset=%.0f len=%u sum1=%08lx\n",
 229                                 (double)i, (double)offset, n1,
 230                                 (unsigned long)sum1);
 231                 }
 232                 write_int(f_out, sum1);
 233                 write_buf(f_out, sum2, sum.s2length);
 234                 len -= n1;
 235                 offset += n1;
 236         }
 237
 238         if (mapbuf)
 239                 unmap_file(mapbuf);
 240 }
 241
 242
 243
 244 /*
 245  * Acts on file number @p i from @p flist, whose name is @p fname.
 246  *
 247  * First fixes up permissions, then generates checksums for the file.
 248  *
 249  * @note This comment was added later by mbp who was trying to work it
 250  * out.  It might be wrong.
 251  */
 252 static void recv_generator(char *fname, struct file_struct *file, int i,
 253                            int f_out)
 254 {
 255         int fd;
 256         STRUCT_STAT st;
 257         int statret, stat_errno;
 258         char *fnamecmp;
 259         char fnamecmpbuf[MAXPATHLEN];
 260
 261         if (list_only)
 262                 return;
 263
 264         if (verbose > 2)
 265                 rprintf(FINFO, "recv_generator(%s,%d)\n", safe_fname(fname), i);
 266
 267         if (server_exclude_list.head
 268             && check_exclude(&server_exclude_list, fname,
 269                              S_ISDIR(file->mode)) < 0) {
 270                 if (verbose) {
 271                         rprintf(FINFO, "skipping server-excluded file \"%s\"\n",
 272                                 safe_fname(fname));
 273                 }
 274                 return;
 275         }
 276
 277         statret = link_stat(fname, &st, keep_dirlinks && S_ISDIR(file->mode));
 278         stat_errno = errno;
 279
 280         if (only_existing && statret == -1 && stat_errno == ENOENT) {
 281                 /* we only want to update existing files */
 282                 if (verbose > 1) {
 283                         rprintf(FINFO, "not creating new file \"%s\"\n",
 284                                 safe_fname(fname));
 285                 }
 286                 return;
 287         }
 288
 289         if (statret == 0 && !preserve_perms
 290             && S_ISDIR(st.st_mode) == S_ISDIR(file->mode)) {
 291                 /* if the file exists already and we aren't perserving
 292                  * permissions then act as though the remote end sent
 293                  * us the file permissions we already have */
 294                 file->mode = (file->mode & ~CHMOD_BITS)
 295                            | (st.st_mode & CHMOD_BITS);
 296         }
 297
 298         if (S_ISDIR(file->mode)) {
 299                 /* The file to be received is a directory, so we need
 300                  * to prepare appropriately.  If there is already a
 301                  * file of that name and it is *not* a directory, then
 302                  * we need to delete it.  If it doesn't exist, then
 303                  * recursively create it. */
 304
 305                 if (dry_run)
 306                         return; /* TODO: causes inaccuracies -- fix */
 307                 if (statret == 0 && !S_ISDIR(st.st_mode)) {
 308                         if (robust_unlink(fname) != 0) {
 309                                 rsyserr(FERROR, errno,
 310                                         "recv_generator: unlink %s to make room for directory",
 311                                         full_fname(fname));
 312                                 return;
 313                         }
 314                         statret = -1;
 315                 }
 316                 if (statret != 0 && do_mkdir(fname,file->mode) != 0 && errno != EEXIST) {
 317                         if (!(relative_paths && errno == ENOENT
 318                             && create_directory_path(fname, orig_umask) == 0
 319                             && do_mkdir(fname, file->mode) == 0)) {
 320                                 rsyserr(FERROR, errno,
 321                                         "recv_generator: mkdir %s failed",
 322                                         full_fname(fname));
 323                         }
 324                 }
 325                 /* f_out is set to -1 when doing final directory-permission
 326                  * and modification-time repair. */
 327                 if (set_perms(fname, file, statret ? NULL : &st, 0)
 328                     && verbose && f_out != -1)
 329                         rprintf(FINFO, "%s/\n", safe_fname(fname));
 330                 return;
 331         }
 332
 333         if (preserve_links && S_ISLNK(file->mode)) {
 334 #if SUPPORT_LINKS
 335                 char lnk[MAXPATHLEN];
 336                 int l;
 337
 338                 if (safe_symlinks && unsafe_symlink(file->u.link, fname)) {
 339                         if (verbose) {
 340                                 rprintf(FINFO, "ignoring unsafe symlink %s -> \"%s\"\n",
 341                                         full_fname(fname), file->u.link);
 342                         }
 343                         return;
 344                 }
 345                 if (statret == 0) {
 346                         l = readlink(fname,lnk,MAXPATHLEN-1);
 347                         if (l > 0) {
 348                                 lnk[l] = 0;
 349                                 /* A link already pointing to the
 350                                  * right place -- no further action
 351                                  * required. */
 352                                 if (strcmp(lnk,file->u.link) == 0) {
 353                                         set_perms(fname, file, &st,
 354                                                   PERMS_REPORT);
 355                                         return;
 356                                 }
 357                         }
 358                         /* Not a symlink, so delete whatever's
 359                          * already there and put a new symlink
 360                          * in place. */
 361                         delete_file(fname);
 362                 }
 363                 if (do_symlink(file->u.link,fname) != 0) {
 364                         rsyserr(FERROR, errno, "symlink %s -> \"%s\" failed",
 365                                 full_fname(fname), safe_fname(file->u.link));
 366                 } else {
 367                         set_perms(fname,file,NULL,0);
 368                         if (verbose) {
 369                                 rprintf(FINFO, "%s -> %s\n", safe_fname(fname),
 370                                         safe_fname(file->u.link));
 371                         }
 372                 }
 373 #endif
 374                 return;
 375         }
 376
 377 #ifdef HAVE_MKNOD
 378         if (am_root && preserve_devices && IS_DEVICE(file->mode)) {
 379                 if (statret != 0 ||
 380                     st.st_mode != file->mode ||
 381                     st.st_rdev != file->u.rdev) {
 382                         delete_file(fname);
 383                         if (verbose > 2) {
 384                                 rprintf(FINFO,"mknod(%s,0%o,0x%x)\n",
 385                                         safe_fname(fname),
 386                                         (int)file->mode, (int)file->u.rdev);
 387                         }
 388                         if (do_mknod(fname,file->mode,file->u.rdev) != 0) {
 389                                 rsyserr(FERROR, errno, "mknod %s failed",
 390                                         full_fname(fname));
 391                         } else {
 392                                 set_perms(fname,file,NULL,0);
 393                                 if (verbose) {
 394                                         rprintf(FINFO, "%s\n",
 395                                                 safe_fname(fname));
 396                                 }
 397                         }
 398                 } else {
 399                         set_perms(fname, file, &st, PERMS_REPORT);
 400                 }
 401                 return;
 402         }
 403 #endif
 404
 405         if (preserve_hard_links && hard_link_check(file, HL_CHECK_MASTER))
 406                 return;
 407
 408         if (!S_ISREG(file->mode)) {
 409                 rprintf(FINFO, "skipping non-regular file \"%s\"\n",
 410                         safe_fname(fname));
 411                 return;
 412         }
 413
 414         fnamecmp = fname;
 415
 416         if (statret == -1 && compare_dest != NULL) {
 417                 /* try the file at compare_dest instead */
 418                 pathjoin(fnamecmpbuf, sizeof fnamecmpbuf, compare_dest, fname);
 419                 if (link_stat(fnamecmpbuf, &st, 0) == 0
 420                     && S_ISREG(st.st_mode)) {
 421 #if HAVE_LINK
 422                         if (link_dest && !dry_run) {
 423                                 if (do_link(fnamecmpbuf, fname) < 0) {
 424                                         if (verbose) {
 425                                                 rsyserr(FINFO, errno,
 426                                                         "link %s => %s",
 427                                                         fnamecmpbuf,
 428                                                         safe_fname(fname));
 429                                         }
 430                                         fnamecmp = fnamecmpbuf;
 431                                 }
 432                         } else
 433 #endif
 434                                 fnamecmp = fnamecmpbuf;
 435                         statret = 0;
 436                 }
 437         }
 438
 439         if (statret == 0 && !S_ISREG(st.st_mode)) {
 440                 if (delete_file(fname) != 0)
 441                         return;
 442                 statret = -1;
 443                 stat_errno = ENOENT;
 444         }
 445
 446         if (statret == -1) {
 447                 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 448                         return;
 449                 if (stat_errno == ENOENT) {
 450                         write_int(f_out,i);
 451                         if (!dry_run && !read_batch)
 452                                 write_sum_head(f_out, NULL);
 453                 } else if (verbose > 1) {
 454                         rsyserr(FERROR, stat_errno,
 455                                 "recv_generator: failed to stat %s",
 456                                 full_fname(fname));
 457                 }
 458                 return;
 459         }
 460
 461         if (opt_ignore_existing && fnamecmp == fname) {
 462                 if (verbose > 1)
 463                         rprintf(FINFO, "%s exists\n", safe_fname(fname));
 464                 return;
 465         }
 466
 467         if (update_only && fnamecmp == fname
 468             && cmp_modtime(st.st_mtime, file->modtime) > 0) {
 469                 if (verbose > 1)
 470                         rprintf(FINFO, "%s is newer\n", safe_fname(fname));
 471                 return;
 472         }
 473
 474         if (skip_file(fnamecmp, file, &st)) {
 475                 if (fnamecmp == fname)
 476                         set_perms(fname, file, &st, PERMS_REPORT);
 477                 return;
 478         }
 479
 480         if (dry_run || read_batch) {
 481                 write_int(f_out,i);
 482                 return;
 483         }
 484
 485         if (whole_file > 0) {
 486                 write_int(f_out,i);
 487                 write_sum_head(f_out, NULL);
 488                 return;
 489         }
 490
 491         if (partial_dir) {
 492                 STRUCT_STAT st2;
 493                 char *partialptr = partial_dir_fname(fname);
 494                 if (partialptr && link_stat(partialptr, &st2, 0) == 0
 495                     && S_ISREG(st2.st_mode)) {
 496                         st = st2;
 497                         fnamecmp = partialptr;
 498                 }
 499         }
 500
 501         /* open the file */
 502         fd = do_open(fnamecmp, O_RDONLY, 0);
 503
 504         if (fd == -1) {
 505                 rsyserr(FERROR, errno, "failed to open %s, continuing",
 506                         full_fname(fnamecmp));
 507                 /* pretend the file didn't exist */
 508                 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 509                         return;
 510                 write_int(f_out,i);
 511                 write_sum_head(f_out, NULL);
 512                 return;
 513         }
 514
 515         if (verbose > 3) {
 516                 rprintf(FINFO, "gen mapped %s of size %.0f\n",
 517                         safe_fname(fnamecmp), (double)st.st_size);
 518         }
 519
 520         if (verbose > 2)
 521                 rprintf(FINFO, "generating and sending sums for %d\n", i);
 522
 523         write_int(f_out,i);
 524         generate_and_send_sums(fd, st.st_size, f_out);
 525
 526         close(fd);
 527 }
 528
 529
 530 void generate_files(int f_out, struct file_list *flist, char *local_name)
 531 {
 532         int i;
 533         int phase = 0;
 534         char fbuf[MAXPATHLEN];
 535
 536         if (verbose > 2) {
 537                 rprintf(FINFO, "generator starting pid=%ld count=%d\n",
 538                         (long)getpid(), flist->count);
 539         }
 540
 541         if (verbose >= 2) {
 542                 rprintf(FINFO,
 543                         whole_file > 0
 544                         ? "delta-transmission disabled for local transfer or --whole-file\n"
 545                         : "delta transmission enabled\n");
 546         }
 547
 548         /* we expect to just sit around now, so don't exit on a
 549            timeout. If we really get a timeout then the other process should
 550            exit */
 551         io_timeout = 0;
 552
 553         for (i = 0; i < flist->count; i++) {
 554                 struct file_struct *file = flist->files[i];
 555                 struct file_struct copy;
 556
 557                 if (!file->basename)
 558                         continue;
 559                 /* we need to ensure that any directories we create have writeable
 560                    permissions initially so that we can create the files within
 561                    them. This is then fixed after the files are transferred */
 562                 if (!am_root && S_ISDIR(file->mode) && !(file->mode & S_IWUSR)) {
 563                         copy = *file;
 564                         /* XXX: Could this be causing a problem on SCO?  Perhaps their
 565                          * handling of permissions is strange? */
 566                         copy.mode |= S_IWUSR; /* user write */
 567                         file = &copy;
 568                 }
 569
 570                 recv_generator(local_name ? local_name : f_name_to(file, fbuf),
 571                                file, i, f_out);
 572         }
 573
 574         phase++;
 575         csum_length = SUM_LENGTH;
 576         ignore_times = 1;
 577
 578         if (verbose > 2)
 579                 rprintf(FINFO,"generate_files phase=%d\n",phase);
 580
 581         write_int(f_out, -1);
 582
 583         /* files can cycle through the system more than once
 584          * to catch initial checksum errors */
 585         while ((i = get_redo_num()) != -1) {
 586                 struct file_struct *file = flist->files[i];
 587                 recv_generator(local_name ? local_name : f_name_to(file, fbuf),
 588                                file, i, f_out);
 589         }
 590
 591         phase++;
 592         if (verbose > 2)
 593                 rprintf(FINFO,"generate_files phase=%d\n",phase);
 594
 595         write_int(f_out, -1);
 596
 597         if (preserve_hard_links)
 598                 do_hard_links();
 599
 600         /* now we need to fix any directory permissions that were
 601          * modified during the transfer */
 602         for (i = 0; i < flist->count; i++) {
 603                 struct file_struct *file = flist->files[i];
 604                 if (!file->basename || !S_ISDIR(file->mode))
 605                         continue;
 606                 recv_generator(local_name ? local_name : f_name(file),
 607                                file, i, -1);
 608         }
 609
 610         if (verbose > 2)
 611                 rprintf(FINFO,"generate_files finished\n");
 612 }