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