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) 2004-2009 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 verbose;
  26 extern int dry_run;
  27 extern int list_only;
  28 extern int am_sender;
  29 extern int inc_recurse;
  30 extern int do_xfers;
  31 extern int link_dest;
  32 extern int preserve_acls;
  33 extern int preserve_xattrs;
  34 extern int make_backups;
  35 extern int protocol_version;
  36 extern int remove_source_files;
  37 extern int stdout_format_has_i;
  38 extern int maybe_ATTRS_REPORT;
  39 extern int unsort_ndx;
  40 extern char *basis_dir[MAX_BASIS_DIRS+1];
  41 extern struct file_list *cur_flist;
  42
  43 #ifdef SUPPORT_HARD_LINKS
  44
  45 /* Starting with protocol 30, we use a simple hashtable on the sending side
  46  * for hashing the st_dev and st_ino info.  The receiving side gets told
  47  * (via flags and a "group index") which items are hard-linked together, so
  48  * we can avoid the pool of dev+inode data.  For incremental recursion mode,
  49  * the receiver will use a ndx hash to remember old pathnames. */
  50
  51 static struct hashtable *dev_tbl;
  52
  53 static struct hashtable *prior_hlinks;
  54
  55 static struct file_list *hlink_flist;
  56
  57 void init_hard_links(void)
  58 {
  59         if (am_sender || protocol_version < 30)
  60                 dev_tbl = hashtable_create(16, 1);
  61         else if (inc_recurse)
  62                 prior_hlinks = hashtable_create(1024, 0);
  63 }
  64
  65 struct ht_int64_node *idev_find(int64 dev, int64 ino)
  66 {
  67         static struct ht_int64_node *dev_node = NULL;
  68         struct hashtable *tbl;
  69
  70         /* Note that some OSes have a dev == 0, so increment to avoid storing a 0. */
  71         if (!dev_node || dev_node->key != dev+1) {
  72                 /* We keep a separate hash table of inodes for every device. */
  73                 dev_node = hashtable_find(dev_tbl, dev+1, 1);
  74                 if (!(tbl = dev_node->data))
  75                         tbl = dev_node->data = hashtable_create(512, 1);
  76         } else
  77                 tbl = dev_node->data;
  78
  79         return hashtable_find(tbl, ino, 1);
  80 }
  81
  82 void idev_destroy(void)
  83 {
  84         int i;
  85
  86         for (i = 0; i < dev_tbl->size; i++) {
  87                 struct ht_int32_node *node = HT_NODE(dev_tbl, dev_tbl->nodes, i);
  88                 if (node->data)
  89                         hashtable_destroy(node->data);
  90         }
  91
  92         hashtable_destroy(dev_tbl);
  93 }
  94
  95 static int hlink_compare_gnum(int *int1, int *int2)
  96 {
  97         struct file_struct *f1 = hlink_flist->sorted[*int1];
  98         struct file_struct *f2 = hlink_flist->sorted[*int2];
  99         int32 gnum1 = F_HL_GNUM(f1);
 100         int32 gnum2 = F_HL_GNUM(f2);
 101
 102         if (gnum1 != gnum2)
 103                 return gnum1 > gnum2 ? 1 : -1;
 104
 105         return *int1 > *int2 ? 1 : -1;
 106 }
 107
 108 static void match_gnums(int32 *ndx_list, int ndx_count)
 109 {
 110         int32 from, prev;
 111         struct file_struct *file, *file_next;
 112         struct ht_int32_node *node = NULL;
 113         int32 gnum, gnum_next;
 114
 115         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
 116              (int (*)()) hlink_compare_gnum);
 117
 118         for (from = 0; from < ndx_count; from++) {
 119                 file = hlink_flist->sorted[ndx_list[from]];
 120                 gnum = F_HL_GNUM(file);
 121                 if (inc_recurse) {
 122                         node = hashtable_find(prior_hlinks, gnum, 1);
 123                         if (!node->data) {
 124                                 if (!(node->data = new_array0(char, 5)))
 125                                         out_of_memory("match_gnums");
 126                                 assert(gnum >= hlink_flist->ndx_start);
 127                                 file->flags |= FLAG_HLINK_FIRST;
 128                                 prev = -1;
 129                         } else if (CVAL(node->data, 0) == 0) {
 130                                 struct file_list *flist;
 131                                 prev = IVAL(node->data, 1);
 132                                 flist = flist_for_ndx(prev, NULL);
 133                                 if (flist)
 134                                         flist->files[prev - flist->ndx_start]->flags &= ~FLAG_HLINK_LAST;
 135                                 else {
 136                                         /* We skipped all prior files in this
 137                                          * group, so mark this as a "first". */
 138                                         file->flags |= FLAG_HLINK_FIRST;
 139                                         prev = -1;
 140                                 }
 141                         } else
 142                                 prev = -1;
 143                 } else {
 144                         file->flags |= FLAG_HLINK_FIRST;
 145                         prev = -1;
 146                 }
 147                 for ( ; from < ndx_count-1; file = file_next, gnum = gnum_next, from++) { /*SHARED ITERATOR*/
 148                         file_next = hlink_flist->sorted[ndx_list[from+1]];
 149                         gnum_next = F_HL_GNUM(file_next);
 150                         if (gnum != gnum_next)
 151                                 break;
 152                         F_HL_PREV(file) = prev;
 153                         /* The linked list uses over-the-wire ndx values. */
 154                         if (unsort_ndx)
 155                                 prev = F_NDX(file);
 156                         else
 157                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 158                 }
 159                 if (prev < 0 && !inc_recurse) {
 160                         /* Disable hard-link bit and set DONE so that
 161                          * HLINK_BUMP()-dependent values are unaffected. */
 162                         file->flags &= ~(FLAG_HLINKED | FLAG_HLINK_FIRST);
 163                         file->flags |= FLAG_HLINK_DONE;
 164                         continue;
 165                 }
 166
 167                 file->flags |= FLAG_HLINK_LAST;
 168                 F_HL_PREV(file) = prev;
 169                 if (inc_recurse && CVAL(node->data, 0) == 0) {
 170                         if (unsort_ndx)
 171                                 prev = F_NDX(file);
 172                         else
 173                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 174                         SIVAL(node->data, 1, prev);
 175                 }
 176         }
 177 }
 178
 179 /* Analyze the hard-links in the file-list by creating a list of all the
 180  * items that have hlink data, sorting them, and matching up identical
 181  * values into clusters.  These will be a single linked list from last
 182  * to first when we're done. */
 183 void match_hard_links(struct file_list *flist)
 184 {
 185         if (!list_only && flist->used) {
 186                 int i, ndx_count = 0;
 187                 int32 *ndx_list;
 188
 189                 if (!(ndx_list = new_array(int32, flist->used)))
 190                         out_of_memory("match_hard_links");
 191
 192                 for (i = 0; i < flist->used; i++) {
 193                         if (F_IS_HLINKED(flist->sorted[i]))
 194                                 ndx_list[ndx_count++] = i;
 195                 }
 196
 197                 hlink_flist = flist;
 198
 199                 if (ndx_count)
 200                         match_gnums(ndx_list, ndx_count);
 201
 202                 free(ndx_list);
 203         }
 204         if (protocol_version < 30)
 205                 idev_destroy();
 206 }
 207
 208 static int maybe_hard_link(struct file_struct *file, int ndx,
 209                            const char *fname, int statret, stat_x *sxp,
 210                            const char *oldname, STRUCT_STAT *old_stp,
 211                            const char *realname, int itemizing, enum logcode code)
 212 {
 213         if (statret == 0) {
 214                 if (sxp->st.st_dev == old_stp->st_dev
 215                  && sxp->st.st_ino == old_stp->st_ino) {
 216                         if (itemizing) {
 217                                 itemize(fname, file, ndx, statret, sxp,
 218                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 219                                         0, "");
 220                         }
 221                         if (verbose > 1 && maybe_ATTRS_REPORT)
 222                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 223                         file->flags |= FLAG_HLINK_DONE;
 224                         return 0;
 225                 }
 226                 if (make_backups > 0) {
 227                         if (!make_backup(fname))
 228                                 return -1;
 229                 } else if (robust_unlink(fname)) {
 230                         rsyserr(FERROR_XFER, errno, "unlink %s failed",
 231                                 full_fname(fname));
 232                         return -1;
 233                 }
 234         }
 235
 236         if (hard_link_one(file, fname, oldname, 0)) {
 237                 if (itemizing) {
 238                         itemize(fname, file, ndx, statret, sxp,
 239                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 240                                 realname);
 241                 }
 242                 if (code != FNONE && verbose)
 243                         rprintf(code, "%s => %s\n", fname, realname);
 244                 return 0;
 245         }
 246         return -1;
 247 }
 248
 249 /* Figure out if a prior entry is still there or if we just have a
 250  * cached name for it. */
 251 static char *check_prior(struct file_struct *file, int gnum,
 252                          int *prev_ndx_p, struct file_list **flist_p)
 253 {
 254         struct file_struct *fp;
 255         struct ht_int32_node *node;
 256         int prev_ndx = F_HL_PREV(file);
 257
 258         while (1) {
 259                 struct file_list *flist;
 260                 if (prev_ndx < 0
 261                  || (flist = flist_for_ndx(prev_ndx, NULL)) == NULL)
 262                         break;
 263                 fp = flist->files[prev_ndx - flist->ndx_start];
 264                 if (!(fp->flags & FLAG_SKIP_HLINK)) {
 265                         *prev_ndx_p = prev_ndx;
 266                         *flist_p = flist;
 267                         return NULL;
 268                 }
 269                 F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
 270         }
 271
 272         if (inc_recurse
 273          && (node = hashtable_find(prior_hlinks, gnum, 0)) != NULL) {
 274                 assert(node->data != NULL);
 275                 if (CVAL(node->data, 0) != 0) {
 276                         *prev_ndx_p = -1;
 277                         *flist_p = NULL;
 278                         return node->data;
 279                 }
 280                 /* The prior file must have been skipped. */
 281                 F_HL_PREV(file) = -1;
 282         }
 283
 284         *prev_ndx_p = -1;
 285         *flist_p = NULL;
 286         return NULL;
 287 }
 288
 289 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 290  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 291 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 292                     int statret, stat_x *sxp, int itemizing,
 293                     enum logcode code)
 294 {
 295         STRUCT_STAT prev_st;
 296         char namebuf[MAXPATHLEN], altbuf[MAXPATHLEN];
 297         char *realname, *prev_name;
 298         struct file_list *flist;
 299         int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
 300         int prev_ndx;
 301
 302         prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
 303
 304         if (!prev_name) {
 305                 struct file_struct *prev_file;
 306
 307                 if (!flist) {
 308                         /* The previous file was skipped, so this one is
 309                          * treated as if it were the first in its group. */
 310                         return 0;
 311                 }
 312
 313                 prev_file = flist->files[prev_ndx - flist->ndx_start];
 314
 315                 /* Is the previous link not complete yet? */
 316                 if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 317                         /* Is the previous link being transferred? */
 318                         if (prev_file->flags & FLAG_FILE_SENT) {
 319                                 /* Add ourselves to the list of files that will
 320                                  * be updated when the transfer completes, and
 321                                  * mark ourself as waiting for the transfer. */
 322                                 F_HL_PREV(file) = F_HL_PREV(prev_file);
 323                                 F_HL_PREV(prev_file) = ndx;
 324                                 file->flags |= FLAG_FILE_SENT;
 325                                 cur_flist->in_progress++;
 326                                 return 1;
 327                         }
 328                         return 0;
 329                 }
 330
 331                 /* There is a finished file to link with! */
 332                 if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 333                         /* The previous previous is FIRST when prev is not. */
 334                         prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
 335                         assert(prev_name != NULL || flist != NULL);
 336                         /* Update our previous pointer to point to the FIRST. */
 337                         F_HL_PREV(file) = prev_ndx;
 338                 }
 339
 340                 if (!prev_name) {
 341                         int alt_dest;
 342
 343                         prev_file = flist->files[prev_ndx - flist->ndx_start];
 344                         /* F_HL_PREV() is alt_dest value when DONE && FIRST. */
 345                         alt_dest = F_HL_PREV(prev_file);
 346
 347                         if (alt_dest >= 0 && dry_run) {
 348                                 pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
 349                                          f_name(prev_file, NULL));
 350                                 prev_name = namebuf;
 351                                 realname = f_name(prev_file, altbuf);
 352                         } else {
 353                                 prev_name = f_name(prev_file, namebuf);
 354                                 realname = prev_name;
 355                         }
 356                 }
 357         }
 358
 359         if (link_stat(prev_name, &prev_st, 0) < 0) {
 360                 if (!dry_run || errno != ENOENT) {
 361                         rsyserr(FERROR_XFER, errno, "stat %s failed", full_fname(prev_name));
 362                         return -1;
 363                 }
 364                 /* A new hard-link will get a new dev & inode, so approximate
 365                  * those values in dry-run mode by zeroing them. */
 366                 memset(&prev_st, 0, sizeof prev_st);
 367         }
 368
 369         if (statret < 0 && basis_dir[0] != NULL) {
 370                 /* If we match an alt-dest item, we don't output this as a change. */
 371                 char cmpbuf[MAXPATHLEN];
 372                 stat_x alt_sx;
 373                 int j = 0;
 374 #ifdef SUPPORT_ACLS
 375                 alt_sx.acc_acl = alt_sx.def_acl = NULL;
 376 #endif
 377 #ifdef SUPPORT_XATTRS
 378                 alt_sx.xattr = NULL;
 379 #endif
 380                 do {
 381                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 382                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 383                                 continue;
 384                         if (link_dest) {
 385                                 if (prev_st.st_dev != alt_sx.st.st_dev
 386                                  || prev_st.st_ino != alt_sx.st.st_ino)
 387                                         continue;
 388                                 statret = 1;
 389                                 if (stdout_format_has_i == 0
 390                                  || (verbose < 2 && stdout_format_has_i < 2)) {
 391                                         itemizing = 0;
 392                                         code = FNONE;
 393                                         if (verbose > 1 && maybe_ATTRS_REPORT)
 394                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 395                                 }
 396                                 break;
 397                         }
 398                         if (!unchanged_file(cmpbuf, file, &alt_sx.st))
 399                                 continue;
 400                         statret = 1;
 401                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
 402                                 break;
 403                 } while (basis_dir[++j] != NULL);
 404                 if (statret == 1) {
 405                         sxp->st = alt_sx.st;
 406 #ifdef SUPPORT_ACLS
 407                         if (preserve_acls && !S_ISLNK(file->mode)) {
 408                                 free_acl(sxp);
 409                                 if (!ACL_READY(alt_sx))
 410                                         get_acl(cmpbuf, sxp);
 411                                 else {
 412                                         sxp->acc_acl = alt_sx.acc_acl;
 413                                         sxp->def_acl = alt_sx.def_acl;
 414                                         alt_sx.acc_acl = alt_sx.def_acl = NULL;
 415                                 }
 416                         }
 417 #endif
 418 #ifdef SUPPORT_XATTRS
 419                         if (preserve_xattrs) {
 420                                 free_xattr(sxp);
 421                                 if (!XATTR_READY(alt_sx))
 422                                         get_xattr(cmpbuf, sxp);
 423                                 else {
 424                                         sxp->xattr = alt_sx.xattr;
 425                                         alt_sx.xattr = NULL;
 426                                 }
 427                         }
 428 #endif
 429                 } else {
 430 #ifdef SUPPORT_ACLS
 431                         if (preserve_acls)
 432                                 free_acl(&alt_sx);
 433 #endif
 434 #ifdef SUPPORT_XATTRS
 435                         if (preserve_xattrs)
 436                                 free_xattr(&alt_sx);
 437 #endif
 438                 }
 439         }
 440
 441         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
 442                             realname, itemizing, code) < 0)
 443                 return -1;
 444
 445         if (remove_source_files == 1 && do_xfers)
 446                 send_msg_int(MSG_SUCCESS, ndx);
 447
 448         return 1;
 449 }
 450
 451 int hard_link_one(struct file_struct *file, const char *fname,
 452                   const char *oldname, int terse)
 453 {
 454         if (do_link(oldname, fname) < 0) {
 455                 enum logcode code;
 456                 if (terse) {
 457                         if (!verbose)
 458                                 return 0;
 459                         code = FINFO;
 460                 } else
 461                         code = FERROR_XFER;
 462                 rsyserr(code, errno, "link %s => %s failed",
 463                         full_fname(fname), oldname);
 464                 return 0;
 465         }
 466
 467         file->flags |= FLAG_HLINK_DONE;
 468
 469         return 1;
 470 }
 471
 472 void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 473                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 474                       int alt_dest)
 475 {
 476         stat_x prev_sx;
 477         STRUCT_STAT st;
 478         char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
 479         const char *our_name;
 480         struct file_list *flist;
 481         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 482
 483         if (stp == NULL && prev_ndx >= 0) {
 484                 if (link_stat(fname, &st, 0) < 0) {
 485                         rsyserr(FERROR_XFER, errno, "stat %s failed",
 486                                 full_fname(fname));
 487                         return;
 488                 }
 489                 stp = &st;
 490         }
 491
 492         /* FIRST combined with DONE means we were the first to get done. */
 493         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 494         F_HL_PREV(file) = alt_dest;
 495         if (alt_dest >= 0 && dry_run) {
 496                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 497                          f_name(file, NULL));
 498                 our_name = alt_name;
 499         } else
 500                 our_name = fname;
 501
 502 #ifdef SUPPORT_ACLS
 503         prev_sx.acc_acl = prev_sx.def_acl = NULL;
 504 #endif
 505 #ifdef SUPPORT_XATTRS
 506         prev_sx.xattr = NULL;
 507 #endif
 508
 509         while ((ndx = prev_ndx) >= 0) {
 510                 int val;
 511                 flist = flist_for_ndx(ndx, "finish_hard_link");
 512                 file = flist->files[ndx - flist->ndx_start];
 513                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 514                 prev_ndx = F_HL_PREV(file);
 515                 F_HL_PREV(file) = fin_ndx;
 516                 prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
 517                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 518                                       our_name, stp, fname, itemizing, code);
 519                 flist->in_progress--;
 520 #ifdef SUPPORT_ACLS
 521                 if (preserve_acls)
 522                         free_acl(&prev_sx);
 523 #endif
 524 #ifdef SUPPORT_XATTRS
 525                 if (preserve_xattrs)
 526                         free_xattr(&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                 if (node == NULL) {
 538                         rprintf(FERROR, "Unable to find a hlink node for %d (%s)\n", gnum, f_name(file, prev_name));
 539                         exit_cleanup(RERR_MESSAGEIO);
 540                 }
 541                 if (node->data == NULL) {
 542                         rprintf(FERROR, "Hlink node data for %d is NULL (%s)\n", gnum, f_name(file, prev_name));
 543                         exit_cleanup(RERR_MESSAGEIO);
 544                 }
 545                 if (CVAL(node->data, 0) != 0) {
 546                         rprintf(FERROR, "Hlink node data for %d already has path=%s (%s)\n",
 547                                 gnum, (char*)node->data, f_name(file, prev_name));
 548                         exit_cleanup(RERR_MESSAGEIO);
 549                 }
 550                 free(node->data);
 551                 if (!(node->data = strdup(our_name)))
 552                         out_of_memory("finish_hard_link");
 553         }
 554 }
 555
 556 int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
 557 {
 558         struct file_list *flist;
 559         int prev_ndx;
 560
 561         file->flags |= FLAG_SKIP_HLINK;
 562         if (!(file->flags & FLAG_HLINK_LAST))
 563                 return -1;
 564
 565         check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
 566         if (prev_ndx >= 0) {
 567                 file = flist->files[prev_ndx - flist->ndx_start];
 568                 if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
 569                         return -1;
 570                 file->flags |= FLAG_HLINK_LAST;
 571                 *flist_p = flist;
 572         }
 573
 574         return prev_ndx;
 575 }
 576 #endif