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                 char fnamecmpdest[MAXPATHLEN];
  83
  84                 if (compare_dest != NULL) {
  85                         if (access(fname, 0) != 0) {
  86                                 pathjoin(fnamecmpdest, sizeof fnamecmpdest,
  87                                          compare_dest, fname);
  88                                 fname = fnamecmpdest;
  89                         }
  90                 }
  91                 file_checksum(fname,sum,st->st_size);
  92                 return memcmp(sum, file->u.sum, protocol_version < 21 ? 2
  93                                                         : MD4_SUM_LENGTH) == 0;
  94         }
  95
  96         if (size_only)
  97                 return 1;
  98
  99         if (ignore_times)
 100                 return 0;
 101
 102         return cmp_modtime(st->st_mtime, file->modtime) == 0;
 103 }
 104
 105
 106 /*
 107  * NULL sum_struct means we have no checksums
 108  */
 109 void write_sum_head(int f, struct sum_struct *sum)
 110 {
 111         static struct sum_struct null_sum;
 112
 113         if (sum == NULL)
 114                 sum = &null_sum;
 115
 116         write_int(f, sum->count);
 117         write_int(f, sum->blength);
 118         if (protocol_version >= 27)
 119                 write_int(f, sum->s2length);
 120         write_int(f, sum->remainder);
 121 }
 122
 123 /*
 124  * set (initialize) the size entries in the per-file sum_struct
 125  * calculating dynamic block and checksum sizes.
 126  *
 127  * This is only called from generate_and_send_sums() but is a separate
 128  * function to encapsulate the logic.
 129  *
 130  * The block size is a rounded square root of file length.
 131  *
 132  * The checksum size is determined according to:
 133  *     blocksum_bits = BLOCKSUM_EXP + 2*log2(file_len) - log2(block_len)
 134  * provided by Donovan Baarda which gives a probability of rsync
 135  * algorithm corrupting data and falling back using the whole md4
 136  * checksums.
 137  *
 138  * This might be made one of several selectable heuristics.
 139  */
 140
 141 static void sum_sizes_sqroot(struct sum_struct *sum, uint64 len)
 142 {
 143         unsigned int blength;
 144         int s2length;
 145         uint32 c;
 146         uint64 l;
 147
 148         if (block_size) {
 149                 blength = block_size;
 150         } else if (len <= BLOCK_SIZE * BLOCK_SIZE) {
 151                 blength = BLOCK_SIZE;
 152         } else {
 153                 l = len;
 154                 c = 1;
 155                 while (l >>= 2) {
 156                         c <<= 1;
 157                 }
 158                 blength = 0;
 159                 do {
 160                         blength |= c;
 161                         if (len < (uint64)blength * blength)
 162                                 blength &= ~c;
 163                         c >>= 1;
 164                 } while (c >= 8);       /* round to multiple of 8 */
 165                 blength = MAX(blength, BLOCK_SIZE);
 166                 blength = MIN(blength, MAX_MAP_SIZE);
 167         }
 168
 169         if (protocol_version < 27) {
 170                 s2length = csum_length;
 171         } else if (csum_length == SUM_LENGTH) {
 172                 s2length = SUM_LENGTH;
 173         } else {
 174                 int b = BLOCKSUM_BIAS;
 175                 l = len;
 176                 while (l >>= 1) {
 177                         b += 2;
 178                 }
 179                 c = blength;
 180                 while (c >>= 1 && b) {
 181                         b--;
 182                 }
 183                 s2length = (b + 1 - 32 + 7) / 8; /* add a bit,
 184                                                   * subtract rollsum,
 185                                                   * round up
 186                                                   *    --optimize in compiler--
 187                                                   */
 188                 s2length = MAX(s2length, csum_length);
 189                 s2length = MIN(s2length, SUM_LENGTH);
 190         }
 191
 192         sum->flength    = len;
 193         sum->blength    = blength;
 194         sum->s2length   = s2length;
 195         sum->count      = (len + (blength - 1)) / blength;
 196         sum->remainder  = (len % blength);
 197
 198         if (sum->count && verbose > 2) {
 199                 rprintf(FINFO, "count=%.0f rem=%u blength=%u s2length=%d flength=%.0f\n",
 200                         (double)sum->count, sum->remainder, sum->blength,
 201                         sum->s2length, (double)sum->flength);
 202         }
 203 }
 204
 205
 206 /*
 207  * Generate and send a stream of signatures/checksums that describe a buffer
 208  *
 209  * Generate approximately one checksum every block_len bytes.
 210  */
 211 static void generate_and_send_sums(int fd, OFF_T len, int f_out)
 212 {
 213         size_t i;
 214         struct map_struct *mapbuf;
 215         struct sum_struct sum;
 216         OFF_T offset = 0;
 217
 218         sum_sizes_sqroot(&sum, len);
 219
 220         if (len > 0)
 221                 mapbuf = map_file(fd, len, sum.blength);
 222         else
 223                 mapbuf = NULL;
 224
 225         write_sum_head(f_out, &sum);
 226
 227         for (i = 0; i < sum.count; i++) {
 228                 unsigned int n1 = MIN(len, sum.blength);
 229                 char *map = map_ptr(mapbuf, offset, n1);
 230                 uint32 sum1 = get_checksum1(map, n1);
 231                 char sum2[SUM_LENGTH];
 232
 233                 get_checksum2(map, n1, sum2);
 234
 235                 if (verbose > 3) {
 236                         rprintf(FINFO,
 237                                 "chunk[%.0f] offset=%.0f len=%u sum1=%08lx\n",
 238                                 (double)i, (double)offset, n1,
 239                                 (unsigned long)sum1);
 240                 }
 241                 write_int(f_out, sum1);
 242                 write_buf(f_out, sum2, sum.s2length);
 243                 len -= n1;
 244                 offset += n1;
 245         }
 246
 247         if (mapbuf)
 248                 unmap_file(mapbuf);
 249 }
 250
 251
 252
 253 /*
 254  * Acts on file number @p i from @p flist, whose name is @p fname.
 255  *
 256  * First fixes up permissions, then generates checksums for the file.
 257  *
 258  * @note This comment was added later by mbp who was trying to work it
 259  * out.  It might be wrong.
 260  */
 261 static void recv_generator(char *fname, struct file_struct *file, int i,
 262                            int f_out)
 263 {
 264         int fd;
 265         STRUCT_STAT st;
 266         int statret, stat_errno;
 267         char *fnamecmp;
 268         char fnamecmpbuf[MAXPATHLEN];
 269
 270         if (list_only)
 271                 return;
 272
 273         if (verbose > 2)
 274                 rprintf(FINFO, "recv_generator(%s,%d)\n", safe_fname(fname), i);
 275
 276         if (server_exclude_list.head
 277             && check_exclude(&server_exclude_list, fname,
 278                              S_ISDIR(file->mode)) < 0) {
 279                 if (verbose) {
 280                         rprintf(FINFO, "skipping server-excluded file \"%s\"\n",
 281                                 safe_fname(fname));
 282                 }
 283                 return;
 284         }
 285
 286         statret = link_stat(fname, &st, keep_dirlinks && S_ISDIR(file->mode));
 287         stat_errno = errno;
 288
 289         if (only_existing && statret == -1 && stat_errno == ENOENT) {
 290                 /* we only want to update existing files */
 291                 if (verbose > 1) {
 292                         rprintf(FINFO, "not creating new file \"%s\"\n",
 293                                 safe_fname(fname));
 294                 }
 295                 return;
 296         }
 297
 298         if (statret == 0 && !preserve_perms
 299             && S_ISDIR(st.st_mode) == S_ISDIR(file->mode)) {
 300                 /* if the file exists already and we aren't perserving
 301                  * permissions then act as though the remote end sent
 302                  * us the file permissions we already have */
 303                 file->mode = (file->mode & ~CHMOD_BITS)
 304                            | (st.st_mode & CHMOD_BITS);
 305         }
 306
 307         if (S_ISDIR(file->mode)) {
 308                 /* The file to be received is a directory, so we need
 309                  * to prepare appropriately.  If there is already a
 310                  * file of that name and it is *not* a directory, then
 311                  * we need to delete it.  If it doesn't exist, then
 312                  * recursively create it. */
 313
 314                 if (dry_run)
 315                         return; /* TODO: causes inaccuracies -- fix */
 316                 if (statret == 0 && !S_ISDIR(st.st_mode)) {
 317                         if (robust_unlink(fname) != 0) {
 318                                 rsyserr(FERROR, errno,
 319                                         "recv_generator: unlink %s to make room for directory",
 320                                         full_fname(fname));
 321                                 return;
 322                         }
 323                         statret = -1;
 324                 }
 325                 if (statret != 0 && do_mkdir(fname,file->mode) != 0 && errno != EEXIST) {
 326                         if (!(relative_paths && errno == ENOENT
 327                             && create_directory_path(fname, orig_umask) == 0
 328                             && do_mkdir(fname, file->mode) == 0)) {
 329                                 rsyserr(FERROR, errno,
 330                                         "recv_generator: mkdir %s failed",
 331                                         full_fname(fname));
 332                         }
 333                 }
 334                 /* f_out is set to -1 when doing final directory-permission
 335                  * and modification-time repair. */
 336                 if (set_perms(fname, file, statret ? NULL : &st, 0)
 337                     && verbose && f_out != -1)
 338                         rprintf(FINFO, "%s/\n", safe_fname(fname));
 339                 return;
 340         }
 341
 342         if (preserve_links && S_ISLNK(file->mode)) {
 343 #if SUPPORT_LINKS
 344                 char lnk[MAXPATHLEN];
 345                 int l;
 346
 347                 if (safe_symlinks && unsafe_symlink(file->u.link, fname)) {
 348                         if (verbose) {
 349                                 rprintf(FINFO, "ignoring unsafe symlink %s -> \"%s\"\n",
 350                                         full_fname(fname), file->u.link);
 351                         }
 352                         return;
 353                 }
 354                 if (statret == 0) {
 355                         l = readlink(fname,lnk,MAXPATHLEN-1);
 356                         if (l > 0) {
 357                                 lnk[l] = 0;
 358                                 /* A link already pointing to the
 359                                  * right place -- no further action
 360                                  * required. */
 361                                 if (strcmp(lnk,file->u.link) == 0) {
 362                                         set_perms(fname, file, &st,
 363                                                   PERMS_REPORT);
 364                                         return;
 365                                 }
 366                         }
 367                         /* Not a symlink, so delete whatever's
 368                          * already there and put a new symlink
 369                          * in place. */
 370                         delete_file(fname);
 371                 }
 372                 if (do_symlink(file->u.link,fname) != 0) {
 373                         rsyserr(FERROR, errno, "symlink %s -> \"%s\" failed",
 374                                 full_fname(fname), safe_fname(file->u.link));
 375                 } else {
 376                         set_perms(fname,file,NULL,0);
 377                         if (verbose) {
 378                                 rprintf(FINFO, "%s -> %s\n", safe_fname(fname),
 379                                         safe_fname(file->u.link));
 380                         }
 381                 }
 382 #endif
 383                 return;
 384         }
 385
 386 #ifdef HAVE_MKNOD
 387         if (am_root && preserve_devices && IS_DEVICE(file->mode)) {
 388                 if (statret != 0 ||
 389                     st.st_mode != file->mode ||
 390                     st.st_rdev != file->u.rdev) {
 391                         delete_file(fname);
 392                         if (verbose > 2) {
 393                                 rprintf(FINFO,"mknod(%s,0%o,0x%x)\n",
 394                                         safe_fname(fname),
 395                                         (int)file->mode, (int)file->u.rdev);
 396                         }
 397                         if (do_mknod(fname,file->mode,file->u.rdev) != 0) {
 398                                 rsyserr(FERROR, errno, "mknod %s failed",
 399                                         full_fname(fname));
 400                         } else {
 401                                 set_perms(fname,file,NULL,0);
 402                                 if (verbose) {
 403                                         rprintf(FINFO, "%s\n",
 404                                                 safe_fname(fname));
 405                                 }
 406                         }
 407                 } else {
 408                         set_perms(fname, file, &st, PERMS_REPORT);
 409                 }
 410                 return;
 411         }
 412 #endif
 413
 414         if (preserve_hard_links && hard_link_check(file, HL_CHECK_MASTER))
 415                 return;
 416
 417         if (!S_ISREG(file->mode)) {
 418                 rprintf(FINFO, "skipping non-regular file \"%s\"\n",
 419                         safe_fname(fname));
 420                 return;
 421         }
 422
 423         fnamecmp = fname;
 424
 425         if (statret == -1 && compare_dest != NULL) {
 426                 /* try the file at compare_dest instead */
 427                 pathjoin(fnamecmpbuf, sizeof fnamecmpbuf, compare_dest, fname);
 428                 if (link_stat(fnamecmpbuf, &st, 0) == 0
 429                     && S_ISREG(st.st_mode)) {
 430 #if HAVE_LINK
 431                         if (link_dest && !dry_run) {
 432                                 if (do_link(fnamecmpbuf, fname) < 0) {
 433                                         if (verbose) {
 434                                                 rsyserr(FINFO, errno,
 435                                                         "link %s => %s",
 436                                                         fnamecmpbuf,
 437                                                         safe_fname(fname));
 438                                         }
 439                                         fnamecmp = fnamecmpbuf;
 440                                 }
 441                         } else
 442 #endif
 443                                 fnamecmp = fnamecmpbuf;
 444                         statret = 0;
 445                 }
 446         }
 447
 448         if (statret == 0 && !S_ISREG(st.st_mode)) {
 449                 if (delete_file(fname) != 0)
 450                         return;
 451                 statret = -1;
 452                 stat_errno = ENOENT;
 453         }
 454
 455         if (statret == -1) {
 456                 if (preserve_hard_links && hard_link_check(file, HL_SKIP))
 457                         return;
 458                 if (stat_errno == ENOENT) {
 459                         write_int(f_out,i);
 460                         if (!dry_run && !read_batch)
 461                                 write_sum_head(f_out, NULL);
 462                 } else if (verbose > 1) {
 463                         rsyserr(FERROR, stat_errno,
 464                                 "recv_generator: failed to stat %s",
 465                                 full_fname(fname));
 466                 }
 467                 return;
 468         }
 469
 470         if (opt_ignore_existing && fnamecmp == fname) {
 471                 if (verbose > 1)
 472                         rprintf(FINFO, "%s exists\n", safe_fname(fname));
 473                 return;
 474         }
 475
 476         if (update_only && fnamecmp == fname
 477             && cmp_modtime(st.st_mtime, file->modtime) > 0) {
 478                 if (verbose > 1)
 479                         rprintf(FINFO, "%s is newer\n", safe_fname(fname));
 480                 return;
 481         }
 482
 483         if (skip_file(fname, file, &st)) {
 484                 if (fnamecmp == fname)
 485                         set_perms(fname, file, &st, PERMS_REPORT);
 486                 return;
 487         }
 488
 489         if (dry_run || read_batch) {
 490                 write_int(f_out,i);
 491                 return;
 492         }
 493
 494         if (whole_file > 0) {
 495                 write_int(f_out,i);
 496                 write_sum_head(f_out, NULL);
 497                 return;
 498         }
 499
 500         if (partial_dir) {
 501                 STRUCT_STAT st2;
 502                 char *partialptr = partial_dir_fname(fname);
 503                 if (partialptr && link_stat(partialptr, &st2, 0) == 0
 504                     && S_ISREG(st2.st_mode)) {
 505                         st = st2;
 506                         fnamecmp = partialptr;
 507                 }
 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 }