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 #include "inums.h"
  25
  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, 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                         if (DEBUG_GTE(HLINK, 3)) {
  76                                 rprintf(FINFO,
  77                                     "[%s] created hashtable for dev %s\n",
  78                                     who_am_i(), big_num(dev));
  79                         }
  80                 }
  81         } else
  82                 tbl = dev_node->data;
  83
  84         return hashtable_find(tbl, ino, 1);
  85 }
  86
  87 void idev_destroy(void)
  88 {
  89         int i;
  90
  91         for (i = 0; i < dev_tbl->size; i++) {
  92                 struct ht_int32_node *node = HT_NODE(dev_tbl, dev_tbl->nodes, i);
  93                 if (node->data)
  94                         hashtable_destroy(node->data);
  95         }
  96
  97         hashtable_destroy(dev_tbl);
  98 }
  99
 100 static int hlink_compare_gnum(int *int1, int *int2)
 101 {
 102         struct file_struct *f1 = hlink_flist->sorted[*int1];
 103         struct file_struct *f2 = hlink_flist->sorted[*int2];
 104         int32 gnum1 = F_HL_GNUM(f1);
 105         int32 gnum2 = F_HL_GNUM(f2);
 106
 107         if (gnum1 != gnum2)
 108                 return gnum1 > gnum2 ? 1 : -1;
 109
 110         return *int1 > *int2 ? 1 : -1;
 111 }
 112
 113 static void match_gnums(int32 *ndx_list, int ndx_count)
 114 {
 115         int32 from, prev;
 116         struct file_struct *file, *file_next;
 117         struct ht_int32_node *node = NULL;
 118         int32 gnum, gnum_next;
 119
 120         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
 121              (int (*)()) hlink_compare_gnum);
 122
 123         for (from = 0; from < ndx_count; from++) {
 124                 file = hlink_flist->sorted[ndx_list[from]];
 125                 gnum = F_HL_GNUM(file);
 126                 if (inc_recurse) {
 127                         node = hashtable_find(prior_hlinks, gnum, 1);
 128                         if (!node->data) {
 129                                 if (!(node->data = new_array0(char, 5)))
 130                                         out_of_memory("match_gnums");
 131                                 assert(gnum >= hlink_flist->ndx_start);
 132                                 file->flags |= FLAG_HLINK_FIRST;
 133                                 prev = -1;
 134                         } else if (CVAL(node->data, 0) == 0) {
 135                                 struct file_list *flist;
 136                                 prev = IVAL(node->data, 1);
 137                                 flist = flist_for_ndx(prev, NULL);
 138                                 if (flist)
 139                                         flist->files[prev - flist->ndx_start]->flags &= ~FLAG_HLINK_LAST;
 140                                 else {
 141                                         /* We skipped all prior files in this
 142                                          * group, so mark this as a "first". */
 143                                         file->flags |= FLAG_HLINK_FIRST;
 144                                         prev = -1;
 145                                 }
 146                         } else
 147                                 prev = -1;
 148                 } else {
 149                         file->flags |= FLAG_HLINK_FIRST;
 150                         prev = -1;
 151                 }
 152                 for ( ; from < ndx_count-1; file = file_next, gnum = gnum_next, from++) { /*SHARED ITERATOR*/
 153                         file_next = hlink_flist->sorted[ndx_list[from+1]];
 154                         gnum_next = F_HL_GNUM(file_next);
 155                         if (gnum != gnum_next)
 156                                 break;
 157                         F_HL_PREV(file) = prev;
 158                         /* The linked list uses over-the-wire ndx values. */
 159                         if (unsort_ndx)
 160                                 prev = F_NDX(file);
 161                         else
 162                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 163                 }
 164                 if (prev < 0 && !inc_recurse) {
 165                         /* Disable hard-link bit and set DONE so that
 166                          * HLINK_BUMP()-dependent values are unaffected. */
 167                         file->flags &= ~(FLAG_HLINKED | FLAG_HLINK_FIRST);
 168                         file->flags |= FLAG_HLINK_DONE;
 169                         continue;
 170                 }
 171
 172                 file->flags |= FLAG_HLINK_LAST;
 173                 F_HL_PREV(file) = prev;
 174                 if (inc_recurse && CVAL(node->data, 0) == 0) {
 175                         if (unsort_ndx)
 176                                 prev = F_NDX(file);
 177                         else
 178                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 179                         SIVAL(node->data, 1, prev);
 180                 }
 181         }
 182 }
 183
 184 /* Analyze the hard-links in the file-list by creating a list of all the
 185  * items that have hlink data, sorting them, and matching up identical
 186  * values into clusters.  These will be a single linked list from last
 187  * to first when we're done. */
 188 void match_hard_links(struct file_list *flist)
 189 {
 190         if (!list_only && flist->used) {
 191                 int i, ndx_count = 0;
 192                 int32 *ndx_list;
 193
 194                 if (!(ndx_list = new_array(int32, flist->used)))
 195                         out_of_memory("match_hard_links");
 196
 197                 for (i = 0; i < flist->used; i++) {
 198                         if (F_IS_HLINKED(flist->sorted[i]))
 199                                 ndx_list[ndx_count++] = i;
 200                 }
 201
 202                 hlink_flist = flist;
 203
 204                 if (ndx_count)
 205                         match_gnums(ndx_list, ndx_count);
 206
 207                 free(ndx_list);
 208         }
 209         if (protocol_version < 30)
 210                 idev_destroy();
 211 }
 212
 213 static int maybe_hard_link(struct file_struct *file, int ndx,
 214                            const char *fname, int statret, stat_x *sxp,
 215                            const char *oldname, STRUCT_STAT *old_stp,
 216                            const char *realname, int itemizing, enum logcode code)
 217 {
 218         if (statret == 0) {
 219                 if (sxp->st.st_dev == old_stp->st_dev
 220                  && sxp->st.st_ino == old_stp->st_ino) {
 221                         if (itemizing) {
 222                                 itemize(fname, file, ndx, statret, sxp,
 223                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 224                                         0, "");
 225                         }
 226                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 227                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 228                         file->flags |= FLAG_HLINK_DONE;
 229                         return 0;
 230                 }
 231                 if (make_backups > 0) {
 232                         if (!make_backup(fname))
 233                                 return -1;
 234                 } else if (robust_unlink(fname)) {
 235                         rsyserr(FERROR_XFER, errno, "unlink %s failed",
 236                                 full_fname(fname));
 237                         return -1;
 238                 }
 239         }
 240
 241         if (hard_link_one(file, fname, oldname, 0)) {
 242                 if (itemizing) {
 243                         itemize(fname, file, ndx, statret, sxp,
 244                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 245                                 realname);
 246                 }
 247                 if (code != FNONE && INFO_GTE(NAME, 1))
 248                         rprintf(code, "%s => %s\n", fname, realname);
 249                 return 0;
 250         }
 251         return -1;
 252 }
 253
 254 /* Figure out if a prior entry is still there or if we just have a
 255  * cached name for it. */
 256 static char *check_prior(struct file_struct *file, int gnum,
 257                          int *prev_ndx_p, struct file_list **flist_p)
 258 {
 259         struct file_struct *fp;
 260         struct ht_int32_node *node;
 261         int prev_ndx = F_HL_PREV(file);
 262
 263         while (1) {
 264                 struct file_list *flist;
 265                 if (prev_ndx < 0
 266                  || (flist = flist_for_ndx(prev_ndx, NULL)) == NULL)
 267                         break;
 268                 fp = flist->files[prev_ndx - flist->ndx_start];
 269                 if (!(fp->flags & FLAG_SKIP_HLINK)) {
 270                         *prev_ndx_p = prev_ndx;
 271                         *flist_p = flist;
 272                         return NULL;
 273                 }
 274                 F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
 275         }
 276
 277         if (inc_recurse
 278          && (node = hashtable_find(prior_hlinks, gnum, 0)) != NULL) {
 279                 assert(node->data != NULL);
 280                 if (CVAL(node->data, 0) != 0) {
 281                         *prev_ndx_p = -1;
 282                         *flist_p = NULL;
 283                         return node->data;
 284                 }
 285                 /* The prior file must have been skipped. */
 286                 F_HL_PREV(file) = -1;
 287         }
 288
 289         *prev_ndx_p = -1;
 290         *flist_p = NULL;
 291         return NULL;
 292 }
 293
 294 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 295  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 296 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 297                     int statret, stat_x *sxp, int itemizing,
 298                     enum logcode code)
 299 {
 300         STRUCT_STAT prev_st;
 301         char namebuf[MAXPATHLEN], altbuf[MAXPATHLEN];
 302         char *realname, *prev_name;
 303         struct file_list *flist;
 304         int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
 305         int prev_ndx;
 306
 307         prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
 308
 309         if (!prev_name) {
 310                 struct file_struct *prev_file;
 311
 312                 if (!flist) {
 313                         /* The previous file was skipped, so this one is
 314                          * treated as if it were the first in its group. */
 315                         if (DEBUG_GTE(HLINK, 2)) {
 316                                 rprintf(FINFO, "hlink for %d (%s,%d): virtual first\n",
 317                                         ndx, f_name(file, NULL), gnum);
 318                         }
 319                         return 0;
 320                 }
 321
 322                 prev_file = flist->files[prev_ndx - flist->ndx_start];
 323
 324                 /* Is the previous link not complete yet? */
 325                 if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 326                         /* Is the previous link being transferred? */
 327                         if (prev_file->flags & FLAG_FILE_SENT) {
 328                                 /* Add ourselves to the list of files that will
 329                                  * be updated when the transfer completes, and
 330                                  * mark ourself as waiting for the transfer. */
 331                                 F_HL_PREV(file) = F_HL_PREV(prev_file);
 332                                 F_HL_PREV(prev_file) = ndx;
 333                                 file->flags |= FLAG_FILE_SENT;
 334                                 cur_flist->in_progress++;
 335                                 if (DEBUG_GTE(HLINK, 2)) {
 336                                         rprintf(FINFO, "hlink for %d (%s,%d): waiting for %d\n",
 337                                                 ndx, f_name(file, NULL), gnum, F_HL_PREV(file));
 338                                 }
 339                                 return 1;
 340                         }
 341                         if (DEBUG_GTE(HLINK, 2)) {
 342                                 rprintf(FINFO, "hlink for %d (%s,%d): looking for a leader\n",
 343                                         ndx, f_name(file, NULL), gnum);
 344                         }
 345                         return 0;
 346                 }
 347
 348                 /* There is a finished file to link with! */
 349                 if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 350                         /* The previous previous is FIRST when prev is not. */
 351                         prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
 352                         /* Update our previous pointer to point to the FIRST. */
 353                         F_HL_PREV(file) = prev_ndx;
 354                 }
 355
 356                 if (!prev_name) {
 357                         int alt_dest;
 358
 359                         assert(flist != NULL);
 360                         prev_file = flist->files[prev_ndx - flist->ndx_start];
 361                         /* F_HL_PREV() is alt_dest value when DONE && FIRST. */
 362                         alt_dest = F_HL_PREV(prev_file);
 363                         if (DEBUG_GTE(HLINK, 2)) {
 364                                 rprintf(FINFO, "hlink for %d (%s,%d): found flist match (alt %d)\n",
 365                                         ndx, f_name(file, NULL), gnum, alt_dest);
 366                         }
 367
 368                         if (alt_dest >= 0 && dry_run) {
 369                                 pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
 370                                          f_name(prev_file, NULL));
 371                                 prev_name = namebuf;
 372                                 realname = f_name(prev_file, altbuf);
 373                         } else {
 374                                 prev_name = f_name(prev_file, namebuf);
 375                                 realname = prev_name;
 376                         }
 377                 }
 378         }
 379
 380         if (DEBUG_GTE(HLINK, 2)) {
 381                 rprintf(FINFO, "hlink for %d (%s,%d): leader is %d (%s)\n",
 382                         ndx, f_name(file, NULL), gnum, prev_ndx, prev_name);
 383         }
 384
 385         if (link_stat(prev_name, &prev_st, 0) < 0) {
 386                 rsyserr(FERROR_XFER, errno, "stat %s failed",
 387                         full_fname(prev_name));
 388                 return -1;
 389         }
 390
 391         if (statret < 0 && basis_dir[0] != NULL) {
 392                 /* If we match an alt-dest item, we don't output this as a change. */
 393                 char cmpbuf[MAXPATHLEN];
 394                 stat_x alt_sx;
 395                 int j = 0;
 396 #ifdef SUPPORT_ACLS
 397                 alt_sx.acc_acl = alt_sx.def_acl = NULL;
 398 #endif
 399 #ifdef SUPPORT_XATTRS
 400                 alt_sx.xattr = NULL;
 401 #endif
 402                 do {
 403                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 404                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 405                                 continue;
 406                         if (link_dest) {
 407                                 if (prev_st.st_dev != alt_sx.st.st_dev
 408                                  || prev_st.st_ino != alt_sx.st.st_ino)
 409                                         continue;
 410                                 statret = 1;
 411                                 if (stdout_format_has_i == 0
 412                                  || (!INFO_GTE(NAME, 2) && stdout_format_has_i < 2)) {
 413                                         itemizing = 0;
 414                                         code = FNONE;
 415                                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 416                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 417                                 }
 418                                 break;
 419                         }
 420                         if (!unchanged_file(cmpbuf, file, &alt_sx.st))
 421                                 continue;
 422                         statret = 1;
 423                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
 424                                 break;
 425                 } while (basis_dir[++j] != NULL);
 426                 if (statret == 1) {
 427                         sxp->st = alt_sx.st;
 428 #ifdef SUPPORT_ACLS
 429                         if (preserve_acls && !S_ISLNK(file->mode)) {
 430                                 free_acl(sxp);
 431                                 if (!ACL_READY(alt_sx))
 432                                         get_acl(cmpbuf, sxp);
 433                                 else {
 434                                         sxp->acc_acl = alt_sx.acc_acl;
 435                                         sxp->def_acl = alt_sx.def_acl;
 436                                         alt_sx.acc_acl = alt_sx.def_acl = NULL;
 437                                 }
 438                         }
 439 #endif
 440 #ifdef SUPPORT_XATTRS
 441                         if (preserve_xattrs) {
 442                                 free_xattr(sxp);
 443                                 if (!XATTR_READY(alt_sx))
 444                                         get_xattr(cmpbuf, sxp);
 445                                 else {
 446                                         sxp->xattr = alt_sx.xattr;
 447                                         alt_sx.xattr = NULL;
 448                                 }
 449                         }
 450 #endif
 451                 } else {
 452 #ifdef SUPPORT_ACLS
 453                         if (preserve_acls)
 454                                 free_acl(&alt_sx);
 455 #endif
 456 #ifdef SUPPORT_XATTRS
 457                         if (preserve_xattrs)
 458                                 free_xattr(&alt_sx);
 459 #endif
 460                 }
 461         }
 462
 463         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
 464                             realname, itemizing, code) < 0)
 465                 return -1;
 466
 467         if (remove_source_files == 1 && do_xfers)
 468                 send_msg_int(MSG_SUCCESS, ndx);
 469
 470         return 1;
 471 }
 472
 473 int hard_link_one(struct file_struct *file, const char *fname,
 474                   const char *oldname, int terse)
 475 {
 476         if (do_link(oldname, fname) < 0) {
 477                 enum logcode code;
 478                 if (terse) {
 479                         if (!INFO_GTE(NAME, 1))
 480                                 return 0;
 481                         code = FINFO;
 482                 } else
 483                         code = FERROR_XFER;
 484                 rsyserr(code, errno, "link %s => %s failed",
 485                         full_fname(fname), oldname);
 486                 return 0;
 487         }
 488
 489         file->flags |= FLAG_HLINK_DONE;
 490
 491         return 1;
 492 }
 493
 494 void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 495                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 496                       int alt_dest)
 497 {
 498         stat_x prev_sx;
 499         STRUCT_STAT st;
 500         char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
 501         const char *our_name;
 502         struct file_list *flist;
 503         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 504
 505         if (stp == NULL && prev_ndx >= 0) {
 506                 if (link_stat(fname, &st, 0) < 0) {
 507                         rsyserr(FERROR_XFER, errno, "stat %s failed",
 508                                 full_fname(fname));
 509                         return;
 510                 }
 511                 stp = &st;
 512         }
 513
 514         /* FIRST combined with DONE means we were the first to get done. */
 515         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 516         F_HL_PREV(file) = alt_dest;
 517         if (alt_dest >= 0 && dry_run) {
 518                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 519                          f_name(file, NULL));
 520                 our_name = alt_name;
 521         } else
 522                 our_name = fname;
 523
 524 #ifdef SUPPORT_ACLS
 525         prev_sx.acc_acl = prev_sx.def_acl = NULL;
 526 #endif
 527 #ifdef SUPPORT_XATTRS
 528         prev_sx.xattr = NULL;
 529 #endif
 530
 531         while ((ndx = prev_ndx) >= 0) {
 532                 int val;
 533                 flist = flist_for_ndx(ndx, "finish_hard_link");
 534                 file = flist->files[ndx - flist->ndx_start];
 535                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 536                 prev_ndx = F_HL_PREV(file);
 537                 F_HL_PREV(file) = fin_ndx;
 538                 prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
 539                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 540                                       our_name, stp, fname, itemizing, code);
 541                 flist->in_progress--;
 542 #ifdef SUPPORT_ACLS
 543                 if (preserve_acls)
 544                         free_acl(&prev_sx);
 545 #endif
 546 #ifdef SUPPORT_XATTRS
 547                 if (preserve_xattrs)
 548                         free_xattr(&prev_sx);
 549 #endif
 550                 if (val < 0)
 551                         continue;
 552                 if (remove_source_files == 1 && do_xfers)
 553                         send_msg_int(MSG_SUCCESS, ndx);
 554         }
 555
 556         if (inc_recurse) {
 557                 int gnum = F_HL_GNUM(file);
 558                 struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, 0);
 559                 assert(node != NULL && node->data != NULL);
 560                 assert(CVAL(node->data, 0) == 0);
 561                 free(node->data);
 562                 if (!(node->data = strdup(our_name)))
 563                         out_of_memory("finish_hard_link");
 564         }
 565 }
 566
 567 int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
 568 {
 569         struct file_list *flist;
 570         int prev_ndx;
 571
 572         file->flags |= FLAG_SKIP_HLINK;
 573         if (!(file->flags & FLAG_HLINK_LAST))
 574                 return -1;
 575
 576         check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
 577         if (prev_ndx >= 0) {
 578                 file = flist->files[prev_ndx - flist->ndx_start];
 579                 if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
 580                         return -1;
 581                 file->flags |= FLAG_HLINK_LAST;
 582                 *flist_p = flist;
 583         }
 584
 585         return prev_ndx;
 586 }
 587 #endif