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, 2005, 2006 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 2 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, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.
  22  */
  23
  24 #include "rsync.h"
  25
  26 extern int verbose;
  27 extern int dry_run;
  28 extern int do_xfers;
  29 extern int link_dest;
  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 *the_file_list;
  37
  38 #ifdef SUPPORT_HARD_LINKS
  39
  40 #define HASH_LOAD_LIMIT(size) ((size)*3/4)
  41 #define FPTR(i) (the_file_list->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)) {
  62                 int req = size;
  63                 size = 32;
  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;
  92         int size = tbl->size * 2;
  93         int i;
  94
  95         old_buckets = tbl->buckets;
  96         if (!(tbl->buckets = new_array(struct idev_node, size)))
  97                 out_of_memory("ihash_create");
  98         memset(tbl->buckets, 0, size * sizeof tbl->buckets[0]);
  99         tbl->size = size;
 100         tbl->entries = 0;
 101
 102         for (i = size / 2; i-- > 0; ) {
 103                 int64 key = old_buckets[i].key;
 104                 if (key == 0)
 105                         continue;
 106                 ihash_node(tbl, key)->data = old_buckets[i].data;
 107         }
 108
 109         free(old_buckets);
 110 }
 111
 112 /* This returns the node for the indicated key, either newly created,
 113  * or already existing. */
 114 static struct idev_node *ihash_node(struct ihash_table *tbl, int64 key)
 115 {
 116         uint32 bkt;
 117
 118         if (tbl->entries > HASH_LOAD_LIMIT(tbl->size))
 119                 expand_ihash(tbl);
 120
 121 #if SIZEOF_INT64 < 8
 122         /* Based on Jenkins One-at-a-time hash. */
 123         {
 124                 uchar *keyp = (uchar*)&key;
 125                 int i;
 126
 127                 for (bkt = 0, i = 0; i < SIZEOF_INT64; i++) {
 128                         bkt += keyp[i];
 129                         bkt += (bkt << 10);
 130                         bkt ^= (bkt >> 6);
 131                 }
 132                 bkt += (bkt << 3);
 133                 bkt ^= (bkt >> 11);
 134                 bkt += (bkt << 15);
 135         }
 136 #else
 137 #define rot(x,k) (((x)<<(k)) ^ ((x)>>(32-(k))))
 138         /* Based on Jenkins hashword() from lookup3.c. */
 139         {
 140                 uint32 a, b, c;
 141
 142                 /* Set up the internal state */
 143                 a = b = c = 0xdeadbeef + (8 << 2);
 144
 145                 b += (uint32)(key >> 32);
 146                 a += (uint32)key;
 147                 c ^= b; c -= rot(b, 14);
 148                 a ^= c; a -= rot(c, 11);
 149                 b ^= a; b -= rot(a, 25);
 150                 c ^= b; c -= rot(b, 16);
 151                 a ^= c; a -= rot(c, 4);
 152                 b ^= a; b -= rot(a, 14);
 153                 c ^= b; c -= rot(b, 24);
 154                 bkt = c;
 155         }
 156 #endif
 157
 158         /* If it already exists, return it. */
 159         while (1) {
 160                 bkt &= tbl->size - 1;
 161                 if (tbl->buckets[bkt].key == key)
 162                         return &tbl->buckets[bkt];
 163                 if (tbl->buckets[bkt].key == 0)
 164                         break;
 165                 bkt++;
 166         }
 167
 168         /* Otherwise, take over this empty spot and then return it. */
 169         tbl->buckets[bkt].key = key;
 170         tbl->entries++;
 171         return &tbl->buckets[bkt];
 172 }
 173
 174 struct idev_node *idev_node(int64 dev, int64 ino)
 175 {
 176         static struct idev_node *dev_node = NULL;
 177         struct ihash_table *tbl;
 178
 179         if (!dev_node || dev_node->key != dev) {
 180                 /* We keep a separate hash table of inodes for every device. */
 181                 dev_node = ihash_node(dev_tbl, dev);
 182                 if (!(tbl = dev_node->data))
 183                         tbl = dev_node->data = ihash_create(512);
 184         } else
 185                 tbl = dev_node->data;
 186
 187         return ihash_node(tbl, ino);
 188 }
 189
 190 void idev_destroy(void)
 191 {
 192         int i;
 193
 194         for (i = 0; i < dev_tbl->size; i++) {
 195                 if (dev_tbl->buckets[i].data)
 196                         ihash_destroy(dev_tbl->buckets[i].data);
 197         }
 198
 199         ihash_destroy(dev_tbl);
 200 }
 201
 202 static int hlink_compare_gnum(int *int1, int *int2)
 203 {
 204         struct file_struct *f1 = FPTR(*int1);
 205         struct file_struct *f2 = FPTR(*int2);
 206         int gnum1 = F_HL_GNUM(f1);
 207         int gnum2 = F_HL_GNUM(f2);
 208
 209         if (gnum1 != gnum2)
 210                 return gnum1 > gnum2 ? 1 : -1;
 211
 212         return *int1 > *int2 ? 1 : -1;
 213 }
 214
 215 static void match_gnums(int32 *ndx_list, int ndx_count)
 216 {
 217         int32 from, prev;
 218         struct file_struct *file, *file_next;
 219         int gnum, gnum_next;
 220
 221         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
 222              (int (*)()) hlink_compare_gnum);
 223
 224         for (from = 0; from < ndx_count; from++) {
 225                 for (file = FPTR(ndx_list[from]), gnum = F_HL_GNUM(file), prev = -1;
 226                      from < ndx_count-1;
 227                      file = file_next, gnum = gnum_next, prev = ndx_list[from++])
 228                 {
 229                         file_next = FPTR(ndx_list[from+1]);
 230                         gnum_next = F_HL_GNUM(file_next);
 231                         if (gnum != gnum_next)
 232                                 break;
 233                         if (prev < 0)
 234                                 file->flags |= FLAG_HLINK_FIRST;
 235                         F_HL_PREV(file) = prev;
 236                 }
 237                 if (prev < 0)
 238                         file->flags &= ~FLAG_HLINKED;
 239                 else {
 240                         file->flags |= FLAG_HLINK_LAST;
 241                         F_HL_PREV(file) = prev;
 242                 }
 243         }
 244 }
 245
 246 /* Analyze the hard-links in the file-list by creating a list of all the
 247  * items that have hlink data, sorting them, and matching up identical
 248  * values into clusters.  These will be a single linked list from last
 249  * to first when we're done. */
 250 void match_hard_links(void)
 251 {
 252         int i, ndx_count = 0;
 253         int32 *ndx_list;
 254
 255         if (!(ndx_list = new_array(int32, the_file_list->count)))
 256                 out_of_memory("match_hard_links");
 257
 258         for (i = 0; i < the_file_list->count; i++) {
 259                 if (F_IS_HLINKED(FPTR(i)))
 260                         ndx_list[ndx_count++] = i;
 261         }
 262
 263         if (ndx_count)
 264                 match_gnums(ndx_list, ndx_count);
 265
 266         free(ndx_list);
 267         if (protocol_version < 30)
 268                 idev_destroy();
 269 }
 270
 271 static int maybe_hard_link(struct file_struct *file, int ndx,
 272                            const char *fname, int statret, STRUCT_STAT *stp,
 273                            const char *oldname, STRUCT_STAT *old_stp,
 274                            const char *realname, int itemizing, enum logcode code)
 275 {
 276         if (statret == 0) {
 277                 if (stp->st_dev == old_stp->st_dev
 278                  && stp->st_ino == old_stp->st_ino) {
 279                         if (itemizing) {
 280                                 itemize(file, ndx, statret, stp,
 281                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
 282                                         0, "");
 283                         }
 284                         if (verbose > 1 && maybe_ATTRS_REPORT)
 285                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 286                         file->flags |= FLAG_HLINK_DONE;
 287                         return 0;
 288                 }
 289                 if (make_backups) {
 290                         if (!make_backup(fname))
 291                                 return -1;
 292                 } else if (robust_unlink(fname)) {
 293                         rsyserr(FERROR, errno, "unlink %s failed",
 294                                 full_fname(fname));
 295                         return -1;
 296                 }
 297         }
 298
 299         if (hard_link_one(file, fname, oldname, 0)) {
 300                 if (itemizing) {
 301                         itemize(file, ndx, statret, stp,
 302                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
 303                                 realname);
 304                 }
 305                 if (code != FNONE && verbose)
 306                         rprintf(code, "%s => %s\n", fname, realname);
 307                 return 0;
 308         }
 309         return -1;
 310 }
 311
 312 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
 313  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
 314 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
 315                     int statret, STRUCT_STAT *stp, int itemizing,
 316                     enum logcode code)
 317 {
 318         STRUCT_STAT prev_st;
 319         char prev_name[MAXPATHLEN], altbuf[MAXPATHLEN], *realname;
 320         int alt_dest, prev_ndx = F_HL_PREV(file);
 321         struct file_struct *prev_file = FPTR(prev_ndx);
 322
 323         /* Is the previous link is not complete yet? */
 324         if (!(prev_file->flags & FLAG_HLINK_DONE)) {
 325                 /* Is the previous link being transferred? */
 326                 if (prev_file->flags & FLAG_SENT) {
 327                         /* Add ourselves to the list of files that will be
 328                          * updated when the transfer completes, and mark
 329                          * ourself as waiting for the transfer. */
 330                         F_HL_PREV(file) = F_HL_PREV(prev_file);
 331                         F_HL_PREV(prev_file) = ndx;
 332                         file->flags |= FLAG_SENT;
 333                         return 1;
 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 will be marked with FIRST. */
 341                 prev_ndx = F_HL_PREV(prev_file);
 342                 prev_file = FPTR(prev_ndx);
 343                 /* Update our previous pointer to point to the first. */
 344                 F_HL_PREV(file) = prev_ndx;
 345         }
 346         alt_dest = F_HL_PREV(prev_file); /* alternate value when DONE && FIRST */
 347         if (alt_dest >= 0 && dry_run) {
 348                 pathjoin(prev_name, MAXPATHLEN, basis_dir[alt_dest],
 349                          f_name(prev_file, NULL));
 350                 f_name(prev_file, altbuf);
 351                 realname = altbuf;
 352         } else {
 353                 f_name(prev_file, prev_name);
 354                 realname = prev_name;
 355         }
 356
 357         if (link_stat(prev_name, &prev_st, 0) < 0) {
 358                 rsyserr(FERROR, errno, "stat %s failed",
 359                         full_fname(prev_name));
 360                 return -1;
 361         }
 362
 363         if (statret < 0 && basis_dir[0] != NULL) {
 364                 /* If we match an alt-dest item, we don't output this as a change. */
 365                 char cmpbuf[MAXPATHLEN];
 366                 STRUCT_STAT alt_st;
 367                 int j = 0;
 368                 do {
 369                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
 370                         if (link_stat(cmpbuf, &alt_st, 0) < 0)
 371                                 continue;
 372                         if (link_dest) {
 373                                 if (prev_st.st_dev != alt_st.st_dev
 374                                  || prev_st.st_ino != alt_st.st_ino)
 375                                         continue;
 376                                 statret = 1;
 377                                 *stp = alt_st;
 378                                 if (verbose < 2 || !stdout_format_has_i) {
 379                                         itemizing = 0;
 380                                         code = FNONE;
 381                                         if (verbose > 1 && maybe_ATTRS_REPORT)
 382                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
 383                                 }
 384                                 break;
 385                         }
 386                         if (!unchanged_file(cmpbuf, file, &alt_st))
 387                                 continue;
 388                         statret = 1;
 389                         *stp = alt_st;
 390                         if (unchanged_attrs(file, &alt_st))
 391                                 break;
 392                 } while (basis_dir[++j] != NULL);
 393         }
 394
 395         if (maybe_hard_link(file, ndx, fname, statret, stp, prev_name, &prev_st,
 396                             realname, itemizing, code) < 0)
 397                 return -1;
 398
 399         if (remove_source_files == 1 && do_xfers)
 400                 send_msg_int(MSG_SUCCESS, ndx);
 401
 402         return 1;
 403 }
 404
 405 int hard_link_one(struct file_struct *file, const char *fname,
 406                   const char *oldname, int terse)
 407 {
 408         if (do_link(oldname, fname) < 0) {
 409                 enum logcode code;
 410                 if (terse) {
 411                         if (!verbose)
 412                                 return -1;
 413                         code = FINFO;
 414                 } else
 415                         code = FERROR;
 416                 rsyserr(code, errno, "link %s => %s failed",
 417                         full_fname(fname), oldname);
 418                 return 0;
 419         }
 420
 421         file->flags |= FLAG_HLINK_DONE;
 422
 423         return 1;
 424 }
 425
 426 void finish_hard_link(struct file_struct *file, const char *fname,
 427                       STRUCT_STAT *stp, int itemizing, enum logcode code,
 428                       int alt_dest)
 429 {
 430         STRUCT_STAT st, prev_st;
 431         char alt_name[MAXPATHLEN], *prev_name;
 432         const char *our_name;
 433         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
 434
 435         if (stp == NULL && prev_ndx >= 0) {
 436                 if (link_stat(fname, &st, 0) < 0) {
 437                         rsyserr(FERROR, errno, "stat %s failed",
 438                                 full_fname(fname));
 439                         return;
 440                 }
 441                 stp = &st;
 442         }
 443
 444         /* FIRST combined with DONE means we were the first to get done. */
 445         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
 446         F_HL_PREV(file) = alt_dest;
 447         if (alt_dest >= 0 && dry_run) {
 448                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
 449                          f_name(file, NULL));
 450                 our_name = alt_name;
 451         } else
 452                 our_name = fname;
 453
 454         while ((ndx = prev_ndx) >= 0) {
 455                 file = FPTR(ndx);
 456                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
 457                 prev_ndx = F_HL_PREV(file);
 458                 prev_name = f_name(file, NULL);
 459                 prev_statret = link_stat(prev_name, &prev_st, 0);
 460                 if (maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_st,
 461                                     our_name, stp, fname, itemizing, code) < 0)
 462                         continue;
 463                 if (remove_source_files == 1 && do_xfers)
 464                         send_msg_int(MSG_SUCCESS, ndx);
 465         }
 466 }
 467 #endif