hlink.c

   1 /*
   2  * Routines to support hard-linking.
   3  *
   4  * Copyright (C) 1996 Andrew Tridgell
   5  * Copyright (C) 1996 Paul Mackerras
   6  * Copyright (C) 2002 Martin Pool <mbp@samba.org>
   7  * Copyright (C) 2004-2008 Wayne Davison
   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 3 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 along
  20  * with this program; if not, visit the http://fsf.org website.
  21  */
  22
  23 #include "rsync.h"
  24
  25 extern int dry_run;
  26 extern int list_only;
  27 extern int am_sender;
  28 extern int inc_recurse;
  29 extern int do_xfers;
  30 extern int link_dest;
  31 extern int preserve_acls;
  32 extern int make_backups;
  33 extern int protocol_version;
  34 extern int remove_source_files;
  35 extern int stdout_format_has_i;
  36 extern int maybe_ATTRS_REPORT;
  37 extern int unsort_ndx;
  38 extern char *basis_dir[MAX_BASIS_DIRS+1];
  39 extern struct file_list *cur_flist, *first_flist;
  40
  41 #ifdef SUPPORT_HARD_LINKS
  42
  43 /* Starting with protocol 30, we use a simple hashtable on the sending side
  44  * for hashing the st_dev and st_ino info.  The receiving side gets told
  45  * (via flags and a "group index") which items are hard-linked together, so
  46  * we can avoid the pool of dev+inode data.  For incremental recursion mode,
  47  * the receiver will use a ndx hash to remember old pathnames. */
  48
  49 static struct hashtable *dev_tbl;
  50
  51 static struct hashtable *prior_hlinks;
  52
  53 static struct file_list *hlink_flist;
  54
  55 void init_hard_links(void)
  56 {
  57         if (am_sender || protocol_version < 30)
  58                 dev_tbl = hashtable_create(16, SIZEOF_INT64 == 8);
  59         else if (inc_recurse)
  60                 prior_hlinks = hashtable_create(1024, 0);
  61 }
  62
  63 struct ht_int64_node *idev_find(int64 dev, int64 ino)
  64 {
  65         static struct ht_int64_node *dev_node = NULL;
  66         struct hashtable *tbl;
  67
  68         if (!dev_node || dev_node->key != dev) {
  69                 /* We keep a separate hash table of inodes for every device. */
  70                 dev_node = hashtable_find(dev_tbl, dev, 1);
  71                 if (!(tbl = dev_node->data)) {
  72                         tbl = dev_node->data = hashtable_create(512, SIZEOF_INT64 == 8);
  73                         if (DEBUG_GTE(HLINK, 3)) {
  74                                 rprintf(FINFO,
  75                                     "[%s] created hashtable for dev %s\n",
  76                                     who_am_i(), big_num(dev, 0));
  77                         }
  78                 }
  79         } else
  80                 tbl = dev_node->data;
  81
  82         return hashtable_find(tbl, ino, 1);
  83 }
  84
  85 void idev_destroy(void)
  86 {
  87         int i;
  88
  89         for (i = 0; i < dev_tbl->size; i++) {
  90                 struct ht_int32_node *node = HT_NODE(dev_tbl, dev_tbl->nodes, i);
  91                 if (node->data)
  92                         hashtable_destroy(node->data);
  93         }
  94
  95         hashtable_destroy(dev_tbl);
  96 }
  97
  98 static int hlink_compare_gnum(int *int1, int *int2)
  99 {
 100         struct file_struct *f1 = hlink_flist->sorted[*int1];
 101         struct file_struct *f2 = hlink_flist->sorted[*int2];
 102         int32 gnum1 = F_HL_GNUM(f1);
 103         int32 gnum2 = F_HL_GNUM(f2);
 104
 105         if (gnum1 != gnum2)
 106                 return gnum1 > gnum2 ? 1 : -1;
 107
 108         return *int1 > *int2 ? 1 : -1;
 109 }
 110
 111 static void match_gnums(int32 *ndx_list, int ndx_count)
 112 {
 113         int32 from, prev;
 114         struct file_struct *file, *file_next;
 115         struct ht_int32_node *node = NULL;
 116         int32 gnum, gnum_next;
 117
 118         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
 119              (int (*)()) hlink_compare_gnum);
 120
 121         for (from = 0; from < ndx_count; from++) {
 122                 file = hlink_flist->sorted[ndx_list[from]];
 123                 gnum = F_HL_GNUM(file);
 124                 if (inc_recurse) {
 125                         node = hashtable_find(prior_hlinks, gnum, 1);
 126                         if (!node->data) {
 127                                 node->data = new_array0(char, 5);
 128                                 assert(gnum >= hlink_flist->ndx_start);
 129                                 file->flags |= FLAG_HLINK_FIRST;
 130                                 prev = -1;
 131                         } else if (CVAL(node->data, 0) == 0) {
 132                                 struct file_list *flist;
 133                                 prev = IVAL(node->data, 1);
 134                                 flist = flist_for_ndx(prev);
 135                                 if (flist)
 136                                         flist->files[prev - flist->ndx_start]->flags &= ~FLAG_HLINK_LAST;
 137                                 else {
 138                                         /* We skipped all prior files in this
 139                                          * group, so mark this as a "first". */
 140                                         file->flags |= FLAG_HLINK_FIRST;
 141                                         prev = -1;
 142                                 }
 143                         } else
 144                                 prev = -1;
 145                 } else {
 146                         file->flags |= FLAG_HLINK_FIRST;
 147                         prev = -1;
 148                 }
 149                 for ( ; from < ndx_count-1; file = file_next, gnum = gnum_next, from++) { /*SHARED ITERATOR*/
 150                         file_next = hlink_flist->sorted[ndx_list[from+1]];
 151                         gnum_next = F_HL_GNUM(file_next);
 152                         if (gnum != gnum_next)
 153                                 break;
 154                         F_HL_PREV(file) = prev;
 155                         /* The linked list uses over-the-wire ndx values. */
 156                         if (unsort_ndx)
 157                                 prev = F_NDX(file);
 158                         else
 159                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 160                 }
 161                 if (prev < 0 && !inc_recurse) {
 162                         /* Disable hard-link bit and set DONE so that
 163                          * HLINK_BUMP()-dependent values are unaffected. */
 164                         file->flags &= ~(FLAG_HLINKED | FLAG_HLINK_FIRST);
 165                         file->flags |= FLAG_HLINK_DONE;
 166                         continue;
 167                 }
 168
 169                 file->flags |= FLAG_HLINK_LAST;
 170                 F_HL_PREV(file) = prev;
 171                 if (inc_recurse && CVAL(node->data, 0) == 0) {
 172                         if (unsort_ndx)
 173                                 prev = F_NDX(file);
 174                         else
 175                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 176                         SIVAL(node->data, 1, prev);
 177                 }
 178         }
 179 }
 180
 181 /* Analyze the hard-links in the file-list by creating a list of all the
 182  * items that have hlink data, sorting them, and matching up identical
 183  * values into clusters.  These will be a single linked list from last
 184  * to first when we're done. */
 185 void match_hard_links(struct file_list *flist)
 186 {
 187         if (!list_only && flist->used) {
 188                 int i, ndx_count = 0;
 189                 int32 *ndx_list;
 190
 191                 if (!(ndx_list = new_array(int32, flist->used)))
 192                         out_of_memory("match_hard_links");
 193
 194                 for (i = 0; i < flist->used; i++) {
 195                         if (F_IS_HLINKED(flist->sorted[i]))
 196                                 ndx_list[ndx_count++] = i;
 197                 }
 198
 199                 hlink_flist = flist;
 200
 201                 if (ndx_count)
 202                         match_gnums(ndx_list, ndx_count);
 203
 204                 free(ndx_list);
 205         }
 206         if (protocol_version < 30)
 207                 idev_destroy();
 208 }
 209
 210 static int maybe_hard_link(struct file_struct *file, int ndx,
 211                            const char *fname, int statret, stat_x *sxp,
 212                            const char *oldname, STRUCT_STAT *old_stp,
 213                            const char *realname, int itemizing, enum logcode code)
 214 {
 215         if (statret == 0) {
 216                 if (sxp->st.st_dev == old_stp->st_dev
 217                  && sxp->st.st_ino == old_stp->st_ino) {
 218                         if (itemizing) {
 219                                 itemize(fname, file, ndx, statret, sxp,
 220                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 221                                         0, "");
 222                         }
 223                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 224                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 225                         file->flags |= FLAG_HLINK_DONE;
 226                         return 0;
 227                 }
 228                 if (make_backups > 0) {
 229                         if (!make_backup(fname))
 230                                 return -1;
 231                 } else if (robust_unlink(fname)) {
 232                         rsyserr(FERROR_XFER, errno, "unlink %s failed",
 233                                 full_fname(fname));
 234                         return -1;
 235                 }
 236         }
 237
 238         if (hard_link_one(file, fname, oldname, 0)) {
 239                 if (itemizing) {
 240                         itemize(fname, file, ndx, statret, sxp,
 241                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 242                                 realname);
 243                 }
 244                 if (code != FNONE && INFO_GTE(NAME, 1))
 245                         rprintf(code, "%s => %s\n", fname, realname);
 246                 return 0;
 247         }
 248         return -1;
 249 }
 250
 251 /* Figure out if a prior entry is still there or if we just have a
 252  * cached name for it. */
 253 static char *check_prior(struct file_struct *file, int gnum,
 254                          int *prev_ndx_p, struct file_list **flist_p)
 255 {
 256         struct file_struct *fp;
 257         struct ht_int32_node *node;
 258         int prev_ndx = F_HL_PREV(file);
 259
 260         while (1) {
 261                 struct file_list *flist;
 262                 if (prev_ndx < 0
 263                  || (flist = flist_for_ndx(prev_ndx)) == NULL)
 264                         break;
 265                 fp = flist->files[prev_ndx - flist->ndx_start];
 266                 if (!(fp->flags & FLAG_SKIP_HLINK)) {
 267                         *prev_ndx_p = prev_ndx;
 268                         *flist_p = flist;
 269                         return NULL;
 270                 }
 271                 F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
 272         }
 273
 274         if (inc_recurse
 275          && (node = hashtable_find(prior_hlinks, gnum, 0)) != NULL) {
 276                 assert(node->data != NULL);
 277                 if (CVAL(node->data, 0) != 0) {
 278                         *prev_ndx_p = -1;
 279                         *flist_p = NULL;
 280                         return node->data;
 281                 }
 282                 /* The prior file must have been skipped. */
 283                 F_HL_PREV(file) = -1;
 284         }
 285
 286         *prev_ndx_p = -1;
 287         *flist_p = NULL;
 288         return NULL;
 289 }
 290
 291 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 292  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 293 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 294                     int statret, stat_x *sxp, int itemizing,
 295                     enum logcode code)
 296 {
 297         STRUCT_STAT prev_st;
 298         char namebuf[MAXPATHLEN], altbuf[MAXPATHLEN];
 299         char *realname, *prev_name;
 300         struct file_list *flist;
 301         int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
 302         int prev_ndx;
 303
 304         prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
 305
 306         if (!prev_name) {
 307                 struct file_struct *prev_file;
 308
 309                 if (!flist) {
 310                         /* The previous file was skipped, so this one is
 311                          * treated as if it were the first in its group. */
 312                         if (DEBUG_GTE(HLINK, 2)) {
 313                                 rprintf(FINFO, "hlink for %d (%s,%d): virtual first\n",
 314                                         ndx, f_name(file, NULL), gnum);
 315                         }
 316                         return 0;
 317                 }
 318
 319                 prev_file = flist->files[prev_ndx - flist->ndx_start];
 320
 321                 /* Is the previous link not complete yet? */
 322                 if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 323                         /* Is the previous link being transferred? */
 324                         if (prev_file->flags & FLAG_FILE_SENT) {
 325                                 /* Add ourselves to the list of files that will
 326                                  * be updated when the transfer completes, and
 327                                  * mark ourself as waiting for the transfer. */
 328                                 F_HL_PREV(file) = F_HL_PREV(prev_file);
 329                                 F_HL_PREV(prev_file) = ndx;
 330                                 file->flags |= FLAG_FILE_SENT;
 331                                 cur_flist->in_progress++;
 332                                 if (DEBUG_GTE(HLINK, 2)) {
 333                                         rprintf(FINFO, "hlink for %d (%s,%d): waiting for %d\n",
 334                                                 ndx, f_name(file, NULL), gnum, F_HL_PREV(file));
 335                                 }
 336                                 return 1;
 337                         }
 338                         if (DEBUG_GTE(HLINK, 2)) {
 339                                 rprintf(FINFO, "hlink for %d (%s,%d): looking for a leader\n",
 340                                         ndx, f_name(file, NULL), gnum);
 341                         }
 342                         return 0;
 343                 }
 344
 345                 /* There is a finished file to link with! */
 346                 if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 347                         /* The previous previous is FIRST when prev is not. */
 348                         prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
 349                         /* Update our previous pointer to point to the FIRST. */
 350                         F_HL_PREV(file) = prev_ndx;
 351                 }
 352
 353                 if (!prev_name) {
 354                         int alt_dest;
 355
 356                         assert(flist != NULL);
 357                         prev_file = flist->files[prev_ndx - flist->ndx_start];
 358                         /* F_HL_PREV() is alt_dest value when DONE && FIRST. */
 359                         alt_dest = F_HL_PREV(prev_file);
 360                         if (DEBUG_GTE(HLINK, 2)) {
 361                                 rprintf(FINFO, "hlink for %d (%s,%d): found flist match (alt %d)\n",
 362                                         ndx, f_name(file, NULL), gnum, alt_dest);
 363                         }
 364
 365                         if (alt_dest >= 0 && dry_run) {
 366                                 pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
 367                                          f_name(prev_file, NULL));
 368                                 prev_name = namebuf;
 369                                 realname = f_name(prev_file, altbuf);
 370                         } else {
 371                                 prev_name = f_name(prev_file, namebuf);
 372                                 realname = prev_name;
 373                         }
 374                 }
 375         }
 376
 377         if (DEBUG_GTE(HLINK, 2)) {
 378                 rprintf(FINFO, "hlink for %d (%s,%d): leader is %d (%s)\n",
 379                         ndx, f_name(file, NULL), gnum, prev_ndx, prev_name);
 380         }
 381
 382         if (link_stat(prev_name, &prev_st, 0) < 0) {
 383                 rsyserr(FERROR_XFER, errno, "stat %s failed",
 384                         full_fname(prev_name));
 385                 return -1;
 386         }
 387
 388         if (statret < 0 && basis_dir[0] != NULL) {
 389                 /* If we match an alt-dest item, we don't output this as a change. */
 390                 char cmpbuf[MAXPATHLEN];
 391                 stat_x alt_sx;
 392                 int j = 0;
 393 #ifdef SUPPORT_ACLS
 394                 alt_sx.acc_acl = alt_sx.def_acl = NULL;
 395 #endif
 396                 do {
 397                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 398                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 399                                 continue;
 400                         if (link_dest) {
 401                                 if (prev_st.st_dev != alt_sx.st.st_dev
 402                                  || prev_st.st_ino != alt_sx.st.st_ino)
 403                                         continue;
 404                                 statret = 1;
 405                                 if (stdout_format_has_i == 0
 406                                  || (!INFO_GTE(NAME, 2) && stdout_format_has_i < 2)) {
 407                                         itemizing = 0;
 408                                         code = FNONE;
 409                                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 410                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 411                                 }
 412                                 break;
 413                         }
 414                         if (!unchanged_file(cmpbuf, file, &alt_sx.st))
 415                                 continue;
 416                         statret = 1;
 417                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
 418                                 break;
 419                 } while (basis_dir[++j] != NULL);
 420                 if (statret == 1) {
 421                         sxp->st = alt_sx.st;
 422 #ifdef SUPPORT_ACLS
 423                         if (preserve_acls && !S_ISLNK(file->mode)) {
 424                                 if (!ACL_READY(*sxp))
 425                                         get_acl(cmpbuf, sxp);
 426                                 else {
 427                                         sxp->acc_acl = alt_sx.acc_acl;
 428                                         sxp->def_acl = alt_sx.def_acl;
 429                                 }
 430                         }
 431 #endif
 432                 }
 433 #ifdef SUPPORT_ACLS
 434                 else if (preserve_acls)
 435                         free_acl(&alt_sx);
 436 #endif
 437         }
 438
 439         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
 440                             realname, itemizing, code) < 0)
 441                 return -1;
 442
 443         if (remove_source_files == 1 && do_xfers)
 444                 send_msg_int(MSG_SUCCESS, ndx);
 445
 446         return 1;
 447 }
 448
 449 int hard_link_one(struct file_struct *file, const char *fname,
 450                   const char *oldname, int terse)
 451 {
 452         if (do_link(oldname, fname) < 0) {
 453                 enum logcode code;
 454                 if (terse) {
 455                         if (!INFO_GTE(NAME, 1))
 456                                 return 0;
 457                         code = FINFO;
 458                 } else
 459                         code = FERROR_XFER;
 460                 rsyserr(code, errno, "link %s => %s failed",
 461                         full_fname(fname), oldname);
 462                 return 0;
 463         }
 464
 465         file->flags |= FLAG_HLINK_DONE;
 466
 467         return 1;
 468 }
 469
 470 void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 471                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 472                       int alt_dest)
 473 {
 474         stat_x prev_sx;
 475         STRUCT_STAT st;
 476         char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
 477         const char *our_name;
 478         struct file_list *flist;
 479         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 480
 481         if (stp == NULL && prev_ndx >= 0) {
 482                 if (link_stat(fname, &st, 0) < 0) {
 483                         rsyserr(FERROR_XFER, errno, "stat %s failed",
 484                                 full_fname(fname));
 485                         return;
 486                 }
 487                 stp = &st;
 488         }
 489
 490         /* FIRST combined with DONE means we were the first to get done. */
 491         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 492         F_HL_PREV(file) = alt_dest;
 493         if (alt_dest >= 0 && dry_run) {
 494                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 495                          f_name(file, NULL));
 496                 our_name = alt_name;
 497         } else
 498                 our_name = fname;
 499
 500 #ifdef SUPPORT_ACLS
 501         prev_sx.acc_acl = prev_sx.def_acl = NULL;
 502 #endif
 503
 504         while ((ndx = prev_ndx) >= 0) {
 505                 int val;
 506                 flist = flist_for_ndx(ndx);
 507                 if (flist == NULL) {
 508                         int start1 = first_flist ? first_flist->ndx_start : 0;
 509                         int start2 = first_flist ? first_flist->prev->ndx_start : 0;
 510                         int used = first_flist ? first_flist->prev->used : 0;
 511                         rprintf(FERROR,
 512                                 "File index not found: %d (%d - %d)\n",
 513                                 ndx, start1 - 1, start2 + used - 1);
 514                         exit_cleanup(RERR_PROTOCOL);
 515                 }
 516                 file = flist->files[ndx - flist->ndx_start];
 517                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 518                 prev_ndx = F_HL_PREV(file);
 519                 F_HL_PREV(file) = fin_ndx;
 520                 prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
 521                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 522                                       our_name, stp, fname, itemizing, code);
 523                 flist->in_progress--;
 524 #ifdef SUPPORT_ACLS
 525                 if (preserve_acls)
 526                         free_acl(&prev_sx);
 527 #endif
 528                 if (val < 0)
 529                         continue;
 530                 if (remove_source_files == 1 && do_xfers)
 531                         send_msg_int(MSG_SUCCESS, ndx);
 532         }
 533
 534         if (inc_recurse) {
 535                 int gnum = F_HL_GNUM(file);
 536                 struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, 0);
 537                 assert(node != NULL && node->data != NULL);
 538                 assert(CVAL(node->data, 0) == 0);
 539                 free(node->data);
 540                 if (!(node->data = strdup(our_name)))
 541                         out_of_memory("finish_hard_link");
 542         }
 543 }
 544
 545 int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
 546 {
 547         struct file_list *flist;
 548         int prev_ndx;
 549
 550         file->flags |= FLAG_SKIP_HLINK;
 551         if (!(file->flags & FLAG_HLINK_LAST))
 552                 return -1;
 553
 554         check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
 555         if (prev_ndx >= 0) {
 556                 file = flist->files[prev_ndx - flist->ndx_start];
 557                 if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
 558                         return -1;
 559                 file->flags |= FLAG_HLINK_LAST;
 560                 *flist_p = flist;
 561         }
 562
 563         return prev_ndx;
 564 }
 565 #endif