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 #include "inums.h"
  25 #include "ifuncs.h"
  26
  27 extern int dry_run;
  28 extern int list_only;
  29 extern int am_sender;
  30 extern int inc_recurse;
  31 extern int do_xfers;
  32 extern int link_dest;
  33 extern int preserve_acls;
  34 extern int preserve_xattrs;
  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                            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         }
 232
 233         if (atomic_create(file, fname, oldname, MAKEDEV(0, 0), sxp, statret == 0 ? DEL_FOR_FILE : 0)) {
 234                 if (itemizing) {
 235                         itemize(fname, file, ndx, statret, sxp,
 236                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 237                                 realname);
 238                 }
 239                 if (code != FNONE && INFO_GTE(NAME, 1))
 240                         rprintf(code, "%s => %s\n", fname, realname);
 241                 return 0;
 242         }
 243
 244         return -1;
 245 }
 246
 247 /* Figure out if a prior entry is still there or if we just have a
 248  * cached name for it. */
 249 static char *check_prior(struct file_struct *file, int gnum,
 250                          int *prev_ndx_p, struct file_list **flist_p)
 251 {
 252         struct file_struct *fp;
 253         struct ht_int32_node *node;
 254         int prev_ndx = F_HL_PREV(file);
 255
 256         while (1) {
 257                 struct file_list *flist;
 258                 if (prev_ndx < 0
 259                  || (flist = flist_for_ndx(prev_ndx, NULL)) == NULL)
 260                         break;
 261                 fp = flist->files[prev_ndx - flist->ndx_start];
 262                 if (!(fp->flags & FLAG_SKIP_HLINK)) {
 263                         *prev_ndx_p = prev_ndx;
 264                         *flist_p = flist;
 265                         return NULL;
 266                 }
 267                 F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
 268         }
 269
 270         if (inc_recurse
 271          && (node = hashtable_find(prior_hlinks, gnum, 0)) != NULL) {
 272                 assert(node->data != NULL);
 273                 if (CVAL(node->data, 0) != 0) {
 274                         *prev_ndx_p = -1;
 275                         *flist_p = NULL;
 276                         return node->data;
 277                 }
 278                 /* The prior file must have been skipped. */
 279                 F_HL_PREV(file) = -1;
 280         }
 281
 282         *prev_ndx_p = -1;
 283         *flist_p = NULL;
 284         return NULL;
 285 }
 286
 287 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 288  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 289 int hard_link_check(struct file_struct *file, int ndx, char *fname,
 290                     int statret, stat_x *sxp, int itemizing,
 291                     enum logcode code)
 292 {
 293         STRUCT_STAT prev_st;
 294         char namebuf[MAXPATHLEN], altbuf[MAXPATHLEN];
 295         char *realname, *prev_name;
 296         struct file_list *flist;
 297         int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
 298         int prev_ndx;
 299
 300         prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
 301
 302         if (!prev_name) {
 303                 struct file_struct *prev_file;
 304
 305                 if (!flist) {
 306                         /* The previous file was skipped, so this one is
 307                          * treated as if it were the first in its group. */
 308                         if (DEBUG_GTE(HLINK, 2)) {
 309                                 rprintf(FINFO, "hlink for %d (%s,%d): virtual first\n",
 310                                         ndx, f_name(file, NULL), gnum);
 311                         }
 312                         return 0;
 313                 }
 314
 315                 prev_file = flist->files[prev_ndx - flist->ndx_start];
 316
 317                 /* Is the previous link not complete yet? */
 318                 if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 319                         /* Is the previous link being transferred? */
 320                         if (prev_file->flags & FLAG_FILE_SENT) {
 321                                 /* Add ourselves to the list of files that will
 322                                  * be updated when the transfer completes, and
 323                                  * mark ourself as waiting for the transfer. */
 324                                 F_HL_PREV(file) = F_HL_PREV(prev_file);
 325                                 F_HL_PREV(prev_file) = ndx;
 326                                 file->flags |= FLAG_FILE_SENT;
 327                                 cur_flist->in_progress++;
 328                                 if (DEBUG_GTE(HLINK, 2)) {
 329                                         rprintf(FINFO, "hlink for %d (%s,%d): waiting for %d\n",
 330                                                 ndx, f_name(file, NULL), gnum, F_HL_PREV(file));
 331                                 }
 332                                 return 1;
 333                         }
 334                         if (DEBUG_GTE(HLINK, 2)) {
 335                                 rprintf(FINFO, "hlink for %d (%s,%d): looking for a leader\n",
 336                                         ndx, f_name(file, NULL), gnum);
 337                         }
 338                         return 0;
 339                 }
 340
 341                 /* There is a finished file to link with! */
 342                 if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 343                         /* The previous previous is FIRST when prev is not. */
 344                         prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
 345                         /* Update our previous pointer to point to the FIRST. */
 346                         F_HL_PREV(file) = prev_ndx;
 347                 }
 348
 349                 if (!prev_name) {
 350                         int alt_dest;
 351
 352                         assert(flist != NULL);
 353                         prev_file = flist->files[prev_ndx - flist->ndx_start];
 354                         /* F_HL_PREV() is alt_dest value when DONE && FIRST. */
 355                         alt_dest = F_HL_PREV(prev_file);
 356                         if (DEBUG_GTE(HLINK, 2)) {
 357                                 rprintf(FINFO, "hlink for %d (%s,%d): found flist match (alt %d)\n",
 358                                         ndx, f_name(file, NULL), gnum, alt_dest);
 359                         }
 360
 361                         if (alt_dest >= 0 && dry_run) {
 362                                 pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
 363                                          f_name(prev_file, NULL));
 364                                 prev_name = namebuf;
 365                                 realname = f_name(prev_file, altbuf);
 366                         } else {
 367                                 prev_name = f_name(prev_file, namebuf);
 368                                 realname = prev_name;
 369                         }
 370                 }
 371         }
 372
 373         if (DEBUG_GTE(HLINK, 2)) {
 374                 rprintf(FINFO, "hlink for %d (%s,%d): leader is %d (%s)\n",
 375                         ndx, f_name(file, NULL), gnum, prev_ndx, prev_name);
 376         }
 377
 378         if (link_stat(prev_name, &prev_st, 0) < 0) {
 379                 if (!dry_run || errno != ENOENT) {
 380                         rsyserr(FERROR_XFER, errno, "stat %s failed", full_fname(prev_name));
 381                         return -1;
 382                 }
 383                 /* A new hard-link will get a new dev & inode, so approximate
 384                  * those values in dry-run mode by zeroing them. */
 385                 memset(&prev_st, 0, sizeof prev_st);
 386         }
 387
 388         if (statret < 0 && basis_dir[0] != NULL) {
 389                 /* If we match an alt-dest item, we don't output this as a change. */
 390                 char cmpbuf[MAXPATHLEN];
 391                 stat_x alt_sx;
 392                 int j = 0;
 393                 init_stat_x(&alt_sx);
 394                 do {
 395                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 396                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 397                                 continue;
 398                         if (link_dest) {
 399                                 if (prev_st.st_dev != alt_sx.st.st_dev
 400                                  || prev_st.st_ino != alt_sx.st.st_ino)
 401                                         continue;
 402                                 statret = 1;
 403                                 if (stdout_format_has_i == 0
 404                                  || (!INFO_GTE(NAME, 2) && stdout_format_has_i < 2)) {
 405                                         itemizing = 0;
 406                                         code = FNONE;
 407                                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 408                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 409                                 }
 410                                 break;
 411                         }
 412                         if (!unchanged_file(cmpbuf, file, &alt_sx.st))
 413                                 continue;
 414                         statret = 1;
 415                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
 416                                 break;
 417                 } while (basis_dir[++j] != NULL);
 418                 if (statret == 1) {
 419                         sxp->st = alt_sx.st;
 420 #ifdef SUPPORT_ACLS
 421                         if (preserve_acls && !S_ISLNK(file->mode)) {
 422                                 free_acl(sxp);
 423                                 if (!ACL_READY(alt_sx))
 424                                         get_acl(cmpbuf, sxp);
 425                                 else {
 426                                         sxp->acc_acl = alt_sx.acc_acl;
 427                                         sxp->def_acl = alt_sx.def_acl;
 428                                         alt_sx.acc_acl = alt_sx.def_acl = NULL;
 429                                 }
 430                         }
 431 #endif
 432 #ifdef SUPPORT_XATTRS
 433                         if (preserve_xattrs) {
 434                                 free_xattr(sxp);
 435                                 if (!XATTR_READY(alt_sx))
 436                                         get_xattr(cmpbuf, sxp);
 437                                 else {
 438                                         sxp->xattr = alt_sx.xattr;
 439                                         alt_sx.xattr = NULL;
 440                                 }
 441                         }
 442 #endif
 443                 } else {
 444 #ifdef SUPPORT_ACLS
 445                         if (preserve_acls)
 446                                 free_acl(&alt_sx);
 447 #endif
 448 #ifdef SUPPORT_XATTRS
 449                         if (preserve_xattrs)
 450                                 free_xattr(&alt_sx);
 451 #endif
 452                 }
 453         }
 454
 455         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
 456                             realname, itemizing, code) < 0)
 457                 return -1;
 458
 459         if (remove_source_files == 1 && do_xfers)
 460                 send_msg_int(MSG_SUCCESS, ndx);
 461
 462         return 1;
 463 }
 464
 465 int hard_link_one(struct file_struct *file, const char *fname,
 466                   const char *oldname, int terse)
 467 {
 468         if (do_link(oldname, fname) < 0) {
 469                 enum logcode code;
 470                 if (terse) {
 471                         if (!INFO_GTE(NAME, 1))
 472                                 return 0;
 473                         code = FINFO;
 474                 } else
 475                         code = FERROR_XFER;
 476                 rsyserr(code, errno, "link %s => %s failed",
 477                         full_fname(fname), oldname);
 478                 return 0;
 479         }
 480
 481         file->flags |= FLAG_HLINK_DONE;
 482
 483         return 1;
 484 }
 485
 486 void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 487                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 488                       int alt_dest)
 489 {
 490         stat_x prev_sx;
 491         STRUCT_STAT st;
 492         char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
 493         const char *our_name;
 494         struct file_list *flist;
 495         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 496
 497         if (stp == NULL && prev_ndx >= 0) {
 498                 if (link_stat(fname, &st, 0) < 0) {
 499                         rsyserr(FERROR_XFER, errno, "stat %s failed",
 500                                 full_fname(fname));
 501                         return;
 502                 }
 503                 stp = &st;
 504         }
 505
 506         /* FIRST combined with DONE means we were the first to get done. */
 507         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 508         F_HL_PREV(file) = alt_dest;
 509         if (alt_dest >= 0 && dry_run) {
 510                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 511                          f_name(file, NULL));
 512                 our_name = alt_name;
 513         } else
 514                 our_name = fname;
 515
 516         init_stat_x(&prev_sx);
 517
 518         while ((ndx = prev_ndx) >= 0) {
 519                 int val;
 520                 flist = flist_for_ndx(ndx, "finish_hard_link");
 521                 file = flist->files[ndx - flist->ndx_start];
 522                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 523                 prev_ndx = F_HL_PREV(file);
 524                 F_HL_PREV(file) = fin_ndx;
 525                 prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
 526                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 527                                       our_name, stp, fname, itemizing, code);
 528                 flist->in_progress--;
 529 #ifdef SUPPORT_ACLS
 530                 if (preserve_acls)
 531                         free_acl(&prev_sx);
 532 #endif
 533 #ifdef SUPPORT_XATTRS
 534                 if (preserve_xattrs)
 535                         free_xattr(&prev_sx);
 536 #endif
 537                 if (val < 0)
 538                         continue;
 539                 if (remove_source_files == 1 && do_xfers)
 540                         send_msg_int(MSG_SUCCESS, ndx);
 541         }
 542
 543         if (inc_recurse) {
 544                 int gnum = F_HL_GNUM(file);
 545                 struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, 0);
 546                 assert(node != NULL && node->data != NULL);
 547                 assert(CVAL(node->data, 0) == 0);
 548                 free(node->data);
 549                 if (!(node->data = strdup(our_name)))
 550                         out_of_memory("finish_hard_link");
 551         }
 552 }
 553
 554 int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
 555 {
 556         struct file_list *flist;
 557         int prev_ndx;
 558
 559         file->flags |= FLAG_SKIP_HLINK;
 560         if (!(file->flags & FLAG_HLINK_LAST))
 561                 return -1;
 562
 563         check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
 564         if (prev_ndx >= 0) {
 565                 file = flist->files[prev_ndx - flist->ndx_start];
 566                 if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
 567                         return -1;
 568                 file->flags |= FLAG_HLINK_LAST;
 569                 *flist_p = flist;
 570         }
 571
 572         return prev_ndx;
 573 }
 574 #endif