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