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) 2004-2007 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 version 2 as
  11  * published by the Free Software Foundation.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.
  21  */
  22
  23 #include "rsync.h"
  24
  25 extern int verbose;
  26 extern int dry_run;
  27 extern int do_xfers;
  28 extern int link_dest;
  29 extern int preserve_acls;
  30 extern int make_backups;
  31 extern int protocol_version;
  32 extern int remove_source_files;
  33 extern int stdout_format_has_i;
  34 extern int maybe_ATTRS_REPORT;
  35 extern char *basis_dir[];
  36 extern struct file_list *cur_flist;
  37
  38 #ifdef SUPPORT_HARD_LINKS
  39
  40 #define HASH_LOAD_LIMIT(size) ((size)*3/4)
  41 #define FPTR(i) (cur_flist->files[i])
  42
  43 struct ihash_table {
  44         int32 size;
  45         int32 entries;
  46         struct idev_node *buckets;
  47 } *dev_tbl;
  48
  49 static struct idev_node *ihash_node(struct ihash_table *tbl, int64 key);
  50
  51 /* Starting with protocol 30, we use a simple hashtable on the sending side
  52  * for hashing the st_dev and st_ino info.  The receiving side gets told
  53  * (via flags and a "group index") which items are hard-linked together, so
  54  * we can avoid the pool of dev+inode data. */
  55
  56 static struct ihash_table *ihash_create(int size)
  57 {
  58         struct ihash_table *tbl;
  59
  60         /* Pick a power of 2 that can hold the requested size. */
  61         if (size & (size-1) || size < 16) {
  62                 int req = size;
  63                 size = 16;
  64                 while (size < req)
  65                         size *= 2;
  66         }
  67
  68         if (!(tbl = new(struct ihash_table))
  69          || !(tbl->buckets = new_array(struct idev_node, size)))
  70                 out_of_memory("ihash_create");
  71         memset(tbl->buckets, 0, size * sizeof tbl->buckets[0]);
  72         tbl->size = size;
  73         tbl->entries = 0;
  74
  75         return tbl;
  76 }
  77
  78 static void ihash_destroy(struct ihash_table *tbl)
  79 {
  80         free(tbl->buckets);
  81         free(tbl);
  82 }
  83
  84 void init_hard_links(void)
  85 {
  86         dev_tbl = ihash_create(16);
  87 }
  88
  89 static void expand_ihash(struct ihash_table *tbl)
  90 {
  91         struct idev_node *old_buckets = tbl->buckets;
  92         int size = tbl->size * 2;
  93         int i;
  94
  95         if (!(tbl->buckets = new_array(struct idev_node, size)))
  96                 out_of_memory("ihash_create");
  97         memset(tbl->buckets, 0, size * sizeof (struct idev_node));
  98         tbl->size = size;
  99         tbl->entries = 0;
 100
 101         for (i = size / 2; i-- > 0; ) {
 102                 int64 key = old_buckets[i].key;
 103                 if (key == 0)
 104                         continue;
 105                 ihash_node(tbl, key)->data = old_buckets[i].data;
 106         }
 107
 108         free(old_buckets);
 109 }
 110
 111 /* This returns the node for the indicated key, either newly created,
 112  * or already existing. */
 113 static struct idev_node *ihash_node(struct ihash_table *tbl, int64 key)
 114 {
 115         uint32 bkt;
 116
 117         if (tbl->entries > HASH_LOAD_LIMIT(tbl->size))
 118                 expand_ihash(tbl);
 119
 120 #if SIZEOF_INT64 < 8
 121         /* Based on Jenkins One-at-a-time hash. */
 122         {
 123                 uchar *keyp = (uchar*)&key;
 124                 int i;
 125
 126                 for (bkt = 0, i = 0; i < SIZEOF_INT64; i++) {
 127                         bkt += keyp[i];
 128                         bkt += (bkt << 10);
 129                         bkt ^= (bkt >> 6);
 130                 }
 131                 bkt += (bkt << 3);
 132                 bkt ^= (bkt >> 11);
 133                 bkt += (bkt << 15);
 134         }
 135 #else
 136 #define rot(x,k) (((x)<<(k)) ^ ((x)>>(32-(k))))
 137         /* Based on Jenkins hashword() from lookup3.c. */
 138         {
 139                 uint32 a, b, c;
 140
 141                 /* Set up the internal state */
 142                 a = b = c = 0xdeadbeef + (8 << 2);
 143
 144                 b += (uint32)(key >> 32);
 145                 a += (uint32)key;
 146                 c ^= b; c -= rot(b, 14);
 147                 a ^= c; a -= rot(c, 11);
 148                 b ^= a; b -= rot(a, 25);
 149                 c ^= b; c -= rot(b, 16);
 150                 a ^= c; a -= rot(c, 4);
 151                 b ^= a; b -= rot(a, 14);
 152                 c ^= b; c -= rot(b, 24);
 153                 bkt = c;
 154         }
 155 #endif
 156
 157         /* If it already exists, return it. */
 158         while (1) {
 159                 bkt &= tbl->size - 1;
 160                 if (tbl->buckets[bkt].key == key)
 161                         return &tbl->buckets[bkt];
 162                 if (tbl->buckets[bkt].key == 0)
 163                         break;
 164                 bkt++;
 165         }
 166
 167         /* Otherwise, take over this empty spot and then return it. */
 168         tbl->buckets[bkt].key = key;
 169         tbl->entries++;
 170         return &tbl->buckets[bkt];
 171 }
 172
 173 struct idev_node *idev_node(int64 dev, int64 ino)
 174 {
 175         static struct idev_node *dev_node = NULL;
 176         struct ihash_table *tbl;
 177
 178         if (!dev_node || dev_node->key != dev) {
 179                 /* We keep a separate hash table of inodes for every device. */
 180                 dev_node = ihash_node(dev_tbl, dev);
 181                 if (!(tbl = dev_node->data))
 182                         tbl = dev_node->data = ihash_create(512);
 183         } else
 184                 tbl = dev_node->data;
 185
 186         return ihash_node(tbl, ino);
 187 }
 188
 189 void idev_destroy(void)
 190 {
 191         int i;
 192
 193         for (i = 0; i < dev_tbl->size; i++) {
 194                 if (dev_tbl->buckets[i].data)
 195                         ihash_destroy(dev_tbl->buckets[i].data);
 196         }
 197
 198         ihash_destroy(dev_tbl);
 199 }
 200
 201 static int hlink_compare_gnum(int *int1, int *int2)
 202 {
 203         struct file_struct *f1 = FPTR(*int1);
 204         struct file_struct *f2 = FPTR(*int2);
 205         int32 gnum1 = F_HL_GNUM(f1);
 206         int32 gnum2 = F_HL_GNUM(f2);
 207
 208         if (gnum1 != gnum2)
 209                 return gnum1 > gnum2 ? 1 : -1;
 210
 211         return *int1 > *int2 ? 1 : -1;
 212 }
 213
 214 static void match_gnums(int32 *ndx_list, int ndx_count)
 215 {
 216         int32 from, prev;
 217         struct file_struct *file, *file_next;
 218         int32 gnum, gnum_next;
 219
 220         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
 221              (int (*)()) hlink_compare_gnum);
 222
 223         for (from = 0; from < ndx_count; from++) {
 224                 for (file = FPTR(ndx_list[from]), gnum = F_HL_GNUM(file), prev = -1;
 225                      from < ndx_count-1;
 226                      file = file_next, gnum = gnum_next, prev = ndx_list[from++])
 227                 {
 228                         file_next = FPTR(ndx_list[from+1]);
 229                         gnum_next = F_HL_GNUM(file_next);
 230                         if (gnum != gnum_next)
 231                                 break;
 232                         if (prev < 0)
 233                                 file->flags |= FLAG_HLINK_FIRST;
 234                         F_HL_PREV(file) = prev;
 235                 }
 236                 if (prev < 0)
 237                         file->flags &= ~FLAG_HLINKED;
 238                 else {
 239                         file->flags |= FLAG_HLINK_LAST;
 240                         F_HL_PREV(file) = prev;
 241                 }
 242         }
 243 }
 244
 245 /* Analyze the hard-links in the file-list by creating a list of all the
 246  * items that have hlink data, sorting them, and matching up identical
 247  * values into clusters.  These will be a single linked list from last
 248  * to first when we're done. */
 249 void match_hard_links(void)
 250 {
 251         int i, ndx_count = 0;
 252         int32 *ndx_list;
 253
 254         if (!(ndx_list = new_array(int32, cur_flist->count)))
 255                 out_of_memory("match_hard_links");
 256
 257         for (i = 0; i < cur_flist->count; i++) {
 258                 if (F_IS_HLINKED(FPTR(i)))
 259                         ndx_list[ndx_count++] = i;
 260         }
 261
 262         if (ndx_count)
 263                 match_gnums(ndx_list, ndx_count);
 264
 265         free(ndx_list);
 266         if (protocol_version < 30)
 267                 idev_destroy();
 268 }
 269
 270 static int maybe_hard_link(struct file_struct *file, int ndx,
 271                            const char *fname, int statret, statx *sxp,
 272                            const char *oldname, STRUCT_STAT *old_stp,
 273                            const char *realname, int itemizing, enum logcode code)
 274 {
 275         if (statret == 0) {
 276                 if (sxp->st.st_dev == old_stp->st_dev
 277                  && sxp->st.st_ino == old_stp->st_ino) {
 278                         if (itemizing) {
 279                                 itemize(fname, file, ndx, statret, sxp,
 280                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 281                                         0, "");
 282                         }
 283                         if (verbose > 1 && maybe_ATTRS_REPORT)
 284                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 285                         file->flags |= FLAG_HLINK_DONE;
 286                         return 0;
 287                 }
 288                 if (make_backups > 0) {
 289                         if (!make_backup(fname))
 290                                 return -1;
 291                 } else if (robust_unlink(fname)) {
 292                         rsyserr(FERROR, errno, "unlink %s failed",
 293                                 full_fname(fname));
 294                         return -1;
 295                 }
 296         }
 297
 298         if (hard_link_one(file, fname, oldname, 0)) {
 299                 if (itemizing) {
 300                         itemize(fname, file, ndx, statret, sxp,
 301                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 302                                 realname);
 303                 }
 304                 if (code != FNONE && verbose)
 305                         rprintf(code, "%s => %s\n", fname, realname);
 306                 return 0;
 307         }
 308         return -1;
 309 }
 310
 311 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 312  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 313 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 314                     int statret, statx *sxp, int itemizing,
 315                     enum logcode code)
 316 {
 317         STRUCT_STAT prev_st;
 318         char prev_name[MAXPATHLEN], altbuf[MAXPATHLEN], *realname;
 319         int alt_dest, prev_ndx = F_HL_PREV(file);
 320         struct file_struct *prev_file = FPTR(prev_ndx);
 321
 322         /* Is the previous link is not complete yet? */
 323         if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 324                 /* Is the previous link being transferred? */
 325                 if (prev_file->flags & FLAG_FILE_SENT) {
 326                         /* Add ourselves to the list of files that will be
 327                          * updated when the transfer completes, and mark
 328                          * ourself as waiting for the transfer. */
 329                         F_HL_PREV(file) = F_HL_PREV(prev_file);
 330                         F_HL_PREV(prev_file) = ndx;
 331                         file->flags |= FLAG_FILE_SENT;
 332                         return 1;
 333                 }
 334                 return 0;
 335         }
 336
 337         /* There is a finished file to link with! */
 338         if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
 339                 /* The previous previous will be marked with FIRST. */
 340                 prev_ndx = F_HL_PREV(prev_file);
 341                 prev_file = FPTR(prev_ndx);
 342                 /* Update our previous pointer to point to the first. */
 343                 F_HL_PREV(file) = prev_ndx;
 344         }
 345         alt_dest = F_HL_PREV(prev_file); /* alternate value when DONE && FIRST */
 346         if (alt_dest >= 0 && dry_run) {
 347                 pathjoin(prev_name, MAXPATHLEN, basis_dir[alt_dest],
 348                          f_name(prev_file, NULL));
 349                 f_name(prev_file, altbuf);
 350                 realname = altbuf;
 351         } else {
 352                 f_name(prev_file, prev_name);
 353                 realname = prev_name;
 354         }
 355
 356         if (link_stat(prev_name, &prev_st, 0) < 0) {
 357                 rsyserr(FERROR, errno, "stat %s failed",
 358                         full_fname(prev_name));
 359                 return -1;
 360         }
 361
 362         if (statret < 0 && basis_dir[0] != NULL) {
 363                 /* If we match an alt-dest item, we don't output this as a change. */
 364                 char cmpbuf[MAXPATHLEN];
 365                 statx alt_sx;
 366                 int j = 0;
 367 #ifdef SUPPORT_ACLS
 368                 alt_sx.acc_acl = alt_sx.def_acl = NULL;
 369 #endif
 370                 do {
 371                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 372                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
 373                                 continue;
 374                         if (link_dest) {
 375                                 if (prev_st.st_dev != alt_sx.st.st_dev
 376                                  || prev_st.st_ino != alt_sx.st.st_ino)
 377                                         continue;
 378                                 statret = 1;
 379                                 if (verbose < 2 || !stdout_format_has_i) {
 380                                         itemizing = 0;
 381                                         code = FNONE;
 382                                         if (verbose > 1 && 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 (!verbose)
 429                                 return -1;
 430                         code = FINFO;
 431                 } else
 432                         code = FERROR;
 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,
 444                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 445                       int alt_dest)
 446 {
 447         statx prev_sx;
 448         STRUCT_STAT st;
 449         char alt_name[MAXPATHLEN], *prev_name;
 450         const char *our_name;
 451         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 452
 453         if (stp == NULL && prev_ndx >= 0) {
 454                 if (link_stat(fname, &st, 0) < 0) {
 455                         rsyserr(FERROR, errno, "stat %s failed",
 456                                 full_fname(fname));
 457                         return;
 458                 }
 459                 stp = &st;
 460         }
 461
 462         /* FIRST combined with DONE means we were the first to get done. */
 463         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 464         F_HL_PREV(file) = alt_dest;
 465         if (alt_dest >= 0 && dry_run) {
 466                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 467                          f_name(file, NULL));
 468                 our_name = alt_name;
 469         } else
 470                 our_name = fname;
 471
 472 #ifdef SUPPORT_ACLS
 473         prev_sx.acc_acl = prev_sx.def_acl = NULL;
 474 #endif
 475
 476         while ((ndx = prev_ndx) >= 0) {
 477                 int val;
 478                 file = FPTR(ndx);
 479                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 480                 prev_ndx = F_HL_PREV(file);
 481                 prev_name = f_name(file, NULL);
 482                 prev_statret = link_stat(prev_name, &prev_sx.st, 0);
 483                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
 484                                       our_name, stp, fname, itemizing, code);
 485 #ifdef SUPPORT_ACLS
 486                 if (preserve_acls)
 487                         free_acl(&prev_sx);
 488 #endif
 489                 if (val < 0)
 490                         continue;
 491                 if (remove_source_files == 1 && do_xfers)
 492                         send_msg_int(MSG_SUCCESS, ndx);
 493         }
 494 }
 495 #endif