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) 2004-2022 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 alt_dest_type;
  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 void *data_when_new = "";
  52
  53 static struct hashtable *dev_tbl;
  54
  55 static struct hashtable *prior_hlinks;
  56
  57 static struct file_list *hlink_flist;
  58
  59 void init_hard_links(void)
  60 {
  61         if (am_sender || protocol_version < 30)
  62                 dev_tbl = hashtable_create(16, HT_KEY64);
  63         else if (inc_recurse)
  64                 prior_hlinks = hashtable_create(1024, HT_KEY32);
  65 }
  66
  67 struct ht_int64_node *idev_find(int64 dev, int64 ino)
  68 {
  69         static struct ht_int64_node *dev_node = NULL;
  70
  71         /* Note that some OSes have a dev == 0, so increment to avoid storing a 0. */
  72         if (!dev_node || dev_node->key != dev+1) {
  73                 /* We keep a separate hash table of inodes for every device. */
  74                 dev_node = hashtable_find(dev_tbl, dev+1, data_when_new);
  75                 if (dev_node->data == data_when_new) {
  76                         dev_node->data = hashtable_create(512, HT_KEY64);
  77                         if (DEBUG_GTE(HLINK, 3)) {
  78                                 rprintf(FINFO, "[%s] created hashtable for dev %s\n",
  79                                         who_am_i(), big_num(dev));
  80                         }
  81                 }
  82         }
  83
  84         return hashtable_find(dev_node->data, ino, (void*)-1L);
  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], (int (*)()) hlink_compare_gnum);
 121
 122         for (from = 0; from < ndx_count; from++) {
 123                 file = hlink_flist->sorted[ndx_list[from]];
 124                 gnum = F_HL_GNUM(file);
 125                 if (inc_recurse) {
 126                         node = hashtable_find(prior_hlinks, gnum, data_when_new);
 127                         if (node->data == data_when_new) {
 128                                 node->data = new_array0(char, 5);
 129                                 assert(gnum >= hlink_flist->ndx_start);
 130                                 file->flags |= FLAG_HLINK_FIRST;
 131                                 prev = -1;
 132                         } else if (CVAL(node->data, 0) == 0) {
 133                                 struct file_list *flist;
 134                                 prev = IVAL(node->data, 1);
 135                                 flist = flist_for_ndx(prev, NULL);
 136                                 if (flist)
 137                                         flist->files[prev - flist->ndx_start]->flags &= ~FLAG_HLINK_LAST;
 138                                 else {
 139                                         /* We skipped all prior files in this
 140                                          * group, so mark this as a "first". */
 141                                         file->flags |= FLAG_HLINK_FIRST;
 142                                         prev = -1;
 143                                 }
 144                         } else
 145                                 prev = -1;
 146                 } else {
 147                         file->flags |= FLAG_HLINK_FIRST;
 148                         prev = -1;
 149                 }
 150                 for ( ; from < ndx_count-1; file = file_next, gnum = gnum_next, from++) { /*SHARED ITERATOR*/
 151                         file_next = hlink_flist->sorted[ndx_list[from+1]];
 152                         gnum_next = F_HL_GNUM(file_next);
 153                         if (gnum != gnum_next)
 154                                 break;
 155                         F_HL_PREV(file) = prev;
 156                         /* The linked list uses over-the-wire ndx values. */
 157                         if (unsort_ndx)
 158                                 prev = F_NDX(file);
 159                         else
 160                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 161                 }
 162                 if (prev < 0 && !inc_recurse) {
 163                         /* Disable hard-link bit and set DONE so that
 164                          * HLINK_BUMP()-dependent values are unaffected. */
 165                         file->flags &= ~(FLAG_HLINKED | FLAG_HLINK_FIRST);
 166                         file->flags |= FLAG_HLINK_DONE;
 167                         continue;
 168                 }
 169
 170                 file->flags |= FLAG_HLINK_LAST;
 171                 F_HL_PREV(file) = prev;
 172                 if (inc_recurse && CVAL(node->data, 0) == 0) {
 173                         if (unsort_ndx)
 174                                 prev = F_NDX(file);
 175                         else
 176                                 prev = ndx_list[from] + hlink_flist->ndx_start;
 177                         SIVAL(node->data, 1, prev);
 178                 }
 179         }
 180 }
 181
 182 /* Analyze the hard-links in the file-list by creating a list of all the
 183  * items that have hlink data, sorting them, and matching up identical
 184  * values into clusters.  These will be a single linked list from last
 185  * to first when we're done. */
 186 void match_hard_links(struct file_list *flist)
 187 {
 188         if (!list_only && flist->used) {
 189                 int i, ndx_count = 0;
 190                 int32 *ndx_list;
 191
 192                 ndx_list = new_array(int32, flist->used);
 193
 194                 for (i = 0; i < flist->used; i++) {
 195                         if (F_IS_HLINKED(flist->sorted[i]))
 196                                 ndx_list[ndx_count++] = i;
 197                 }
 198
 199                 hlink_flist = flist;
 200
 201                 if (ndx_count)
 202                         match_gnums(ndx_list, ndx_count);
 203
 204                 free(ndx_list);
 205         }
 206         if (protocol_version < 30)
 207                 idev_destroy();
 208 }
 209
 210 static int maybe_hard_link(struct file_struct *file, int ndx,
 211                            char *fname, int statret, stat_x *sxp,
 212                            const char *oldname, STRUCT_STAT *old_stp,
 213                            const char *realname, int itemizing, enum logcode code)
 214 {
 215         if (statret == 0) {
 216                 if (sxp->st.st_dev == old_stp->st_dev
 217                  && sxp->st.st_ino == old_stp->st_ino) {
 218                         if (itemizing) {
 219                                 itemize(fname, file, ndx, statret, sxp,
 220                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 221                                         0, "");
 222                         }
 223                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 224                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 225                         file->flags |= FLAG_HLINK_DONE;
 226                         return 0;
 227                 }
 228         }
 229
 230         if (atomic_create(file, fname, NULL, oldname, MAKEDEV(0, 0), sxp, statret == 0 ? DEL_FOR_FILE : 0)) {
 231                 if (itemizing) {
 232                         itemize(fname, file, ndx, statret, sxp,
 233                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 234                                 realname);
 235                 }
 236                 if (code != FNONE && INFO_GTE(NAME, 1))
 237                         rprintf(code, "%s => %s\n", fname, realname);
 238                 return 0;
 239         }
 240
 241         return -1;
 242 }
 243
 244 /* Figure out if a prior entry is still there or if we just have a
 245  * cached name for it. */
 246 static char *check_prior(struct file_struct *file, int gnum,
 247                          int *prev_ndx_p, struct file_list **flist_p)
 248 {
 249         struct file_struct *fp;
 250         struct ht_int32_node *node;
 251         int prev_ndx = F_HL_PREV(file);
 252
 253         while (1) {
 254                 struct file_list *flist;
 255                 if (prev_ndx < 0
 256                  || (flist = flist_for_ndx(prev_ndx, NULL)) == NULL)
 257                         break;
 258                 fp = flist->files[prev_ndx - flist->ndx_start];
 259                 if (!(fp->flags & FLAG_SKIP_HLINK)) {
 260                         *prev_ndx_p = prev_ndx;
 261                         *flist_p = flist;
 262                         return NULL;
 263                 }
 264                 F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
 265         }
 266
 267         if (inc_recurse
 268          && (node = hashtable_find(prior_hlinks, gnum, NULL)) != NULL) {
 269                 assert(node->data != NULL);
 270                 if (CVAL(node->data, 0) != 0) {
 271                         *prev_ndx_p = -1;
 272                         *flist_p = NULL;
 273                         return node->data;
 274                 }
 275                 /* The prior file must have been skipped. */
 276                 F_HL_PREV(file) = -1;
 277         }
 278
 279         *prev_ndx_p = -1;
 280         *flist_p = NULL;
 281         return NULL;
 282 }
 283
 284 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 285  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 286 int hard_link_check(struct file_struct *file, int ndx, char *fname,
 287                     int statret, stat_x *sxp, int itemizing,
 288                     enum logcode code)
 289 {
 290         STRUCT_STAT prev_st;
 291         char namebuf[MAXPATHLEN], altbuf[MAXPATHLEN];
 292         char *realname, *prev_name;
 293         struct file_list *flist;
 294         int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
 295         int prev_ndx;
 296
 297         prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
 298
 299         if (!prev_name) {
 300                 struct file_struct *prev_file;
 301
 302                 if (!flist) {
 303                         /* The previous file was skipped, so this one is
 304                          * treated as if it were the first in its group. */
 305                         if (DEBUG_GTE(HLINK, 2)) {
 306                                 rprintf(FINFO, "hlink for %d (%s,%d): virtual first\n",
 307                                         ndx, f_name(file, NULL), gnum);
 308                         }
 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                                 if (DEBUG_GTE(HLINK, 2)) {
 326                                         rprintf(FINFO, "hlink for %d (%s,%d): waiting for %d\n",
 327                                                 ndx, f_name(file, NULL), gnum, F_HL_PREV(file));
 328                                 }
 329                                 return 1;
 330                         }
 331                         if (DEBUG_GTE(HLINK, 2)) {
 332                                 rprintf(FINFO, "hlink for %d (%s,%d): looking for a leader\n",
 333                                         ndx, f_name(file, NULL), gnum);
 334                         }
 335                         return 0;
 336                 }
 337
 338                 /* There is a finished file to link with! */
 339                 if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 340                         /* The previous previous is FIRST when prev is not. */
 341                         prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
 342                         /* Update our previous pointer to point to the FIRST. */
 343                         F_HL_PREV(file) = prev_ndx;
 344                 }
 345
 346                 if (!prev_name) {
 347                         int alt_dest;
 348
 349                         assert(flist != NULL);
 350                         prev_file = flist->files[prev_ndx - flist->ndx_start];
 351                         /* F_HL_PREV() is alt_dest value when DONE && FIRST. */
 352                         alt_dest = F_HL_PREV(prev_file);
 353                         if (DEBUG_GTE(HLINK, 2)) {
 354                                 rprintf(FINFO, "hlink for %d (%s,%d): found flist match (alt %d)\n",
 355                                         ndx, f_name(file, NULL), gnum, alt_dest);
 356                         }
 357
 358                         if (alt_dest >= 0 && dry_run) {
 359                                 pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
 360                                          f_name(prev_file, NULL));
 361                                 prev_name = namebuf;
 362                                 realname = f_name(prev_file, altbuf);
 363                         } else {
 364                                 prev_name = f_name(prev_file, namebuf);
 365                                 realname = prev_name;
 366                         }
 367                 }
 368         }
 369
 370         if (DEBUG_GTE(HLINK, 2)) {
 371                 rprintf(FINFO, "hlink for %d (%s,%d): leader is %d (%s)\n",
 372                         ndx, f_name(file, NULL), gnum, prev_ndx, prev_name);
 373         }
 374
 375         if (link_stat(prev_name, &prev_st, 0) < 0) {
 376                 if (!dry_run || errno != ENOENT) {
 377                         rsyserr(FERROR_XFER, errno, "stat %s failed", full_fname(prev_name));
 378                         return -1;
 379                 }
 380                 /* A new hard-link will get a new dev & inode, so approximate
 381                  * those values in dry-run mode by zeroing them. */
 382                 memset(&prev_st, 0, sizeof prev_st);
 383         }
 384
 385         if (statret < 0 && basis_dir[0] != NULL) {
 386                 /* If we match an alt-dest item, we don't output this as a change. */
 387                 char cmpbuf[MAXPATHLEN];
 388                 stat_x alt_sx;
 389                 int j = 0;
 390                 init_stat_x(&alt_sx);
 391                 do {
 392                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 393                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 394                                 continue;
 395                         if (alt_dest_type == LINK_DEST) {
 396                                 if (prev_st.st_dev != alt_sx.st.st_dev
 397                                  || prev_st.st_ino != alt_sx.st.st_ino)
 398                                         continue;
 399                                 statret = 1;
 400                                 if (stdout_format_has_i == 0
 401                                  || (!INFO_GTE(NAME, 2) && stdout_format_has_i < 2)) {
 402                                         itemizing = 0;
 403                                         code = FNONE;
 404                                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
 405                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 406                                 }
 407                                 break;
 408                         }
 409                         if (!quick_check_ok(FT_REG, cmpbuf, file, &alt_sx.st))
 410                                 continue;
 411                         statret = 1;
 412                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
 413                                 break;
 414                 } while (basis_dir[++j] != NULL);
 415                 if (statret == 1) {
 416                         sxp->st = alt_sx.st;
 417 #ifdef SUPPORT_ACLS
 418                         if (preserve_acls && !S_ISLNK(file->mode)) {
 419                                 free_acl(sxp);
 420                                 if (!ACL_READY(alt_sx))
 421                                         get_acl(cmpbuf, sxp);
 422                                 else {
 423                                         sxp->acc_acl = alt_sx.acc_acl;
 424                                         sxp->def_acl = alt_sx.def_acl;
 425                                         alt_sx.acc_acl = alt_sx.def_acl = NULL;
 426                                 }
 427                         }
 428 #endif
 429 #ifdef SUPPORT_XATTRS
 430                         if (preserve_xattrs) {
 431                                 free_xattr(sxp);
 432                                 if (!XATTR_READY(alt_sx))
 433                                         get_xattr(cmpbuf, sxp);
 434                                 else {
 435                                         sxp->xattr = alt_sx.xattr;
 436                                         alt_sx.xattr = NULL;
 437                                 }
 438                         }
 439 #endif
 440                 } else
 441                         free_stat_x(&alt_sx);
 442         }
 443
 444         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
 445                             realname, itemizing, code) < 0)
 446                 return -1;
 447
 448         if (remove_source_files == 1 && do_xfers)
 449                 send_msg_success(fname, ndx);
 450
 451         return 1;
 452 }
 453
 454 int hard_link_one(struct file_struct *file, const char *fname,
 455                   const char *oldname, int terse)
 456 {
 457         if (do_link(oldname, fname) < 0) {
 458                 enum logcode code;
 459                 if (terse) {
 460                         if (!INFO_GTE(NAME, 1))
 461                                 return 0;
 462                         code = FINFO;
 463                 } else
 464                         code = FERROR_XFER;
 465                 rsyserr(code, errno, "link %s => %s failed",
 466                         full_fname(fname), oldname);
 467                 return 0;
 468         }
 469
 470         file->flags |= FLAG_HLINK_DONE;
 471
 472         return 1;
 473 }
 474
 475 void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
 476                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 477                       int alt_dest)
 478 {
 479         stat_x prev_sx;
 480         STRUCT_STAT st;
 481         char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
 482         const char *our_name;
 483         struct file_list *flist;
 484         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 485
 486         if (stp == NULL && prev_ndx >= 0) {
 487                 if (link_stat(fname, &st, 0) < 0 && !dry_run) {
 488                         rsyserr(FERROR_XFER, errno, "stat %s failed",
 489                                 full_fname(fname));
 490                         return;
 491                 }
 492                 stp = &st;
 493         }
 494
 495         /* FIRST combined with DONE means we were the first to get done. */
 496         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 497         F_HL_PREV(file) = alt_dest;
 498         if (alt_dest >= 0 && dry_run) {
 499                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 500                          f_name(file, NULL));
 501                 our_name = alt_name;
 502         } else
 503                 our_name = fname;
 504
 505         init_stat_x(&prev_sx);
 506
 507         while ((ndx = prev_ndx) >= 0) {
 508                 int val;
 509                 flist = flist_for_ndx(ndx, "finish_hard_link");
 510                 file = flist->files[ndx - flist->ndx_start];
 511                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 512                 prev_ndx = F_HL_PREV(file);
 513                 F_HL_PREV(file) = fin_ndx;
 514                 prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
 515                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 516                                       our_name, stp, fname, itemizing, code);
 517                 flist->in_progress--;
 518                 free_stat_x(&prev_sx);
 519                 if (val < 0)
 520                         continue;
 521                 if (remove_source_files == 1 && do_xfers)
 522                         send_msg_success(fname, ndx);
 523         }
 524
 525         if (inc_recurse) {
 526                 int gnum = F_HL_GNUM(file);
 527                 struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, NULL);
 528                 if (node == NULL) {
 529                         rprintf(FERROR, "Unable to find a hlink node for %d (%s)\n", gnum, f_name(file, prev_name));
 530                         exit_cleanup(RERR_MESSAGEIO);
 531                 }
 532                 if (node->data == NULL) {
 533                         rprintf(FERROR, "Hlink node data for %d is NULL (%s)\n", gnum, f_name(file, prev_name));
 534                         exit_cleanup(RERR_MESSAGEIO);
 535                 }
 536                 if (CVAL(node->data, 0) != 0) {
 537                         rprintf(FERROR, "Hlink node data for %d already has path=%s (%s)\n",
 538                                 gnum, (char*)node->data, f_name(file, prev_name));
 539                         exit_cleanup(RERR_MESSAGEIO);
 540                 }
 541                 free(node->data);
 542                 node->data = strdup(our_name);
 543         }
 544 }
 545
 546 int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
 547 {
 548         struct file_list *flist;
 549         int prev_ndx;
 550
 551         file->flags |= FLAG_SKIP_HLINK;
 552         if (!(file->flags & FLAG_HLINK_LAST))
 553                 return -1;
 554
 555         check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
 556         if (prev_ndx >= 0) {
 557                 file = flist->files[prev_ndx - flist->ndx_start];
 558                 if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
 559                         return -1;
 560                 file->flags |= FLAG_HLINK_LAST;
 561                 *flist_p = flist;
 562         }
 563
 564         return prev_ndx;
 565 }
 566 #endif