Tweak the copyright year.
[rsync.git] / 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) 2004-2019 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, 1);
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         /* Note that some OSes have a dev == 0, so increment to avoid storing a 0. */
71         if (!dev_node || dev_node->key != dev+1) {
72                 /* We keep a separate hash table of inodes for every device. */
73                 dev_node = hashtable_find(dev_tbl, dev+1, 1);
74                 if (!(tbl = dev_node->data)) {
75                         tbl = dev_node->data = hashtable_create(512, 1);
76                         if (DEBUG_GTE(HLINK, 3)) {
77                                 rprintf(FINFO,
78                                     "[%s] created hashtable for dev %s\n",
79                                     who_am_i(), big_num(dev));
80                         }
81                 }
82         } else
83                 tbl = dev_node->data;
84
85         return hashtable_find(tbl, ino, 1);
86 }
87
88 void idev_destroy(void)
89 {
90         int i;
91
92         for (i = 0; i < dev_tbl->size; i++) {
93                 struct ht_int32_node *node = HT_NODE(dev_tbl, dev_tbl->nodes, i);
94                 if (node->data)
95                         hashtable_destroy(node->data);
96         }
97
98         hashtable_destroy(dev_tbl);
99 }
100
101 static int hlink_compare_gnum(int *int1, int *int2)
102 {
103         struct file_struct *f1 = hlink_flist->sorted[*int1];
104         struct file_struct *f2 = hlink_flist->sorted[*int2];
105         int32 gnum1 = F_HL_GNUM(f1);
106         int32 gnum2 = F_HL_GNUM(f2);
107
108         if (gnum1 != gnum2)
109                 return gnum1 > gnum2 ? 1 : -1;
110
111         return *int1 > *int2 ? 1 : -1;
112 }
113
114 static void match_gnums(int32 *ndx_list, int ndx_count)
115 {
116         int32 from, prev;
117         struct file_struct *file, *file_next;
118         struct ht_int32_node *node = NULL;
119         int32 gnum, gnum_next;
120
121         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
122              (int (*)()) hlink_compare_gnum);
123
124         for (from = 0; from < ndx_count; from++) {
125                 file = hlink_flist->sorted[ndx_list[from]];
126                 gnum = F_HL_GNUM(file);
127                 if (inc_recurse) {
128                         node = hashtable_find(prior_hlinks, gnum, 1);
129                         if (!node->data) {
130                                 if (!(node->data = new_array0(char, 5)))
131                                         out_of_memory("match_gnums");
132                                 assert(gnum >= hlink_flist->ndx_start);
133                                 file->flags |= FLAG_HLINK_FIRST;
134                                 prev = -1;
135                         } else if (CVAL(node->data, 0) == 0) {
136                                 struct file_list *flist;
137                                 prev = IVAL(node->data, 1);
138                                 flist = flist_for_ndx(prev, NULL);
139                                 if (flist)
140                                         flist->files[prev - flist->ndx_start]->flags &= ~FLAG_HLINK_LAST;
141                                 else {
142                                         /* We skipped all prior files in this
143                                          * group, so mark this as a "first". */
144                                         file->flags |= FLAG_HLINK_FIRST;
145                                         prev = -1;
146                                 }
147                         } else
148                                 prev = -1;
149                 } else {
150                         file->flags |= FLAG_HLINK_FIRST;
151                         prev = -1;
152                 }
153                 for ( ; from < ndx_count-1; file = file_next, gnum = gnum_next, from++) { /*SHARED ITERATOR*/
154                         file_next = hlink_flist->sorted[ndx_list[from+1]];
155                         gnum_next = F_HL_GNUM(file_next);
156                         if (gnum != gnum_next)
157                                 break;
158                         F_HL_PREV(file) = prev;
159                         /* The linked list uses over-the-wire ndx values. */
160                         if (unsort_ndx)
161                                 prev = F_NDX(file);
162                         else
163                                 prev = ndx_list[from] + hlink_flist->ndx_start;
164                 }
165                 if (prev < 0 && !inc_recurse) {
166                         /* Disable hard-link bit and set DONE so that
167                          * HLINK_BUMP()-dependent values are unaffected. */
168                         file->flags &= ~(FLAG_HLINKED | FLAG_HLINK_FIRST);
169                         file->flags |= FLAG_HLINK_DONE;
170                         continue;
171                 }
172
173                 file->flags |= FLAG_HLINK_LAST;
174                 F_HL_PREV(file) = prev;
175                 if (inc_recurse && CVAL(node->data, 0) == 0) {
176                         if (unsort_ndx)
177                                 prev = F_NDX(file);
178                         else
179                                 prev = ndx_list[from] + hlink_flist->ndx_start;
180                         SIVAL(node->data, 1, prev);
181                 }
182         }
183 }
184
185 /* Analyze the hard-links in the file-list by creating a list of all the
186  * items that have hlink data, sorting them, and matching up identical
187  * values into clusters.  These will be a single linked list from last
188  * to first when we're done. */
189 void match_hard_links(struct file_list *flist)
190 {
191         if (!list_only && flist->used) {
192                 int i, ndx_count = 0;
193                 int32 *ndx_list;
194
195                 if (!(ndx_list = new_array(int32, flist->used)))
196                         out_of_memory("match_hard_links");
197
198                 for (i = 0; i < flist->used; i++) {
199                         if (F_IS_HLINKED(flist->sorted[i]))
200                                 ndx_list[ndx_count++] = i;
201                 }
202
203                 hlink_flist = flist;
204
205                 if (ndx_count)
206                         match_gnums(ndx_list, ndx_count);
207
208                 free(ndx_list);
209         }
210         if (protocol_version < 30)
211                 idev_destroy();
212 }
213
214 static int maybe_hard_link(struct file_struct *file, int ndx,
215                            char *fname, int statret, stat_x *sxp,
216                            const char *oldname, STRUCT_STAT *old_stp,
217                            const char *realname, int itemizing, enum logcode code)
218 {
219         if (statret == 0) {
220                 if (sxp->st.st_dev == old_stp->st_dev
221                  && sxp->st.st_ino == old_stp->st_ino) {
222                         if (itemizing) {
223                                 itemize(fname, file, ndx, statret, sxp,
224                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
225                                         0, "");
226                         }
227                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
228                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
229                         file->flags |= FLAG_HLINK_DONE;
230                         return 0;
231                 }
232         }
233
234         if (atomic_create(file, fname, NULL, oldname, MAKEDEV(0, 0), sxp, statret == 0 ? DEL_FOR_FILE : 0)) {
235                 if (itemizing) {
236                         itemize(fname, file, ndx, statret, sxp,
237                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
238                                 realname);
239                 }
240                 if (code != FNONE && INFO_GTE(NAME, 1))
241                         rprintf(code, "%s => %s\n", fname, realname);
242                 return 0;
243         }
244
245         return -1;
246 }
247
248 /* Figure out if a prior entry is still there or if we just have a
249  * cached name for it. */
250 static char *check_prior(struct file_struct *file, int gnum,
251                          int *prev_ndx_p, struct file_list **flist_p)
252 {
253         struct file_struct *fp;
254         struct ht_int32_node *node;
255         int prev_ndx = F_HL_PREV(file);
256
257         while (1) {
258                 struct file_list *flist;
259                 if (prev_ndx < 0
260                  || (flist = flist_for_ndx(prev_ndx, NULL)) == NULL)
261                         break;
262                 fp = flist->files[prev_ndx - flist->ndx_start];
263                 if (!(fp->flags & FLAG_SKIP_HLINK)) {
264                         *prev_ndx_p = prev_ndx;
265                         *flist_p = flist;
266                         return NULL;
267                 }
268                 F_HL_PREV(file) = prev_ndx = F_HL_PREV(fp);
269         }
270
271         if (inc_recurse
272          && (node = hashtable_find(prior_hlinks, gnum, 0)) != NULL) {
273                 assert(node->data != NULL);
274                 if (CVAL(node->data, 0) != 0) {
275                         *prev_ndx_p = -1;
276                         *flist_p = NULL;
277                         return node->data;
278                 }
279                 /* The prior file must have been skipped. */
280                 F_HL_PREV(file) = -1;
281         }
282
283         *prev_ndx_p = -1;
284         *flist_p = NULL;
285         return NULL;
286 }
287
288 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
289  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
290 int hard_link_check(struct file_struct *file, int ndx, char *fname,
291                     int statret, stat_x *sxp, int itemizing,
292                     enum logcode code)
293 {
294         STRUCT_STAT prev_st;
295         char namebuf[MAXPATHLEN], altbuf[MAXPATHLEN];
296         char *realname, *prev_name;
297         struct file_list *flist;
298         int gnum = inc_recurse ? F_HL_GNUM(file) : -1;
299         int prev_ndx;
300
301         prev_name = realname = check_prior(file, gnum, &prev_ndx, &flist);
302
303         if (!prev_name) {
304                 struct file_struct *prev_file;
305
306                 if (!flist) {
307                         /* The previous file was skipped, so this one is
308                          * treated as if it were the first in its group. */
309                         if (DEBUG_GTE(HLINK, 2)) {
310                                 rprintf(FINFO, "hlink for %d (%s,%d): virtual first\n",
311                                         ndx, f_name(file, NULL), gnum);
312                         }
313                         return 0;
314                 }
315
316                 prev_file = flist->files[prev_ndx - flist->ndx_start];
317
318                 /* Is the previous link not complete yet? */
319                 if (!(prev_file->flags & FLAG_HLINK_DONE)) {
320                         /* Is the previous link being transferred? */
321                         if (prev_file->flags & FLAG_FILE_SENT) {
322                                 /* Add ourselves to the list of files that will
323                                  * be updated when the transfer completes, and
324                                  * mark ourself as waiting for the transfer. */
325                                 F_HL_PREV(file) = F_HL_PREV(prev_file);
326                                 F_HL_PREV(prev_file) = ndx;
327                                 file->flags |= FLAG_FILE_SENT;
328                                 cur_flist->in_progress++;
329                                 if (DEBUG_GTE(HLINK, 2)) {
330                                         rprintf(FINFO, "hlink for %d (%s,%d): waiting for %d\n",
331                                                 ndx, f_name(file, NULL), gnum, F_HL_PREV(file));
332                                 }
333                                 return 1;
334                         }
335                         if (DEBUG_GTE(HLINK, 2)) {
336                                 rprintf(FINFO, "hlink for %d (%s,%d): looking for a leader\n",
337                                         ndx, f_name(file, NULL), gnum);
338                         }
339                         return 0;
340                 }
341
342                 /* There is a finished file to link with! */
343                 if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
344                         /* The previous previous is FIRST when prev is not. */
345                         prev_name = realname = check_prior(prev_file, gnum, &prev_ndx, &flist);
346                         /* Update our previous pointer to point to the FIRST. */
347                         F_HL_PREV(file) = prev_ndx;
348                 }
349
350                 if (!prev_name) {
351                         int alt_dest;
352
353                         assert(flist != NULL);
354                         prev_file = flist->files[prev_ndx - flist->ndx_start];
355                         /* F_HL_PREV() is alt_dest value when DONE && FIRST. */
356                         alt_dest = F_HL_PREV(prev_file);
357                         if (DEBUG_GTE(HLINK, 2)) {
358                                 rprintf(FINFO, "hlink for %d (%s,%d): found flist match (alt %d)\n",
359                                         ndx, f_name(file, NULL), gnum, alt_dest);
360                         }
361
362                         if (alt_dest >= 0 && dry_run) {
363                                 pathjoin(namebuf, MAXPATHLEN, basis_dir[alt_dest],
364                                          f_name(prev_file, NULL));
365                                 prev_name = namebuf;
366                                 realname = f_name(prev_file, altbuf);
367                         } else {
368                                 prev_name = f_name(prev_file, namebuf);
369                                 realname = prev_name;
370                         }
371                 }
372         }
373
374         if (DEBUG_GTE(HLINK, 2)) {
375                 rprintf(FINFO, "hlink for %d (%s,%d): leader is %d (%s)\n",
376                         ndx, f_name(file, NULL), gnum, prev_ndx, prev_name);
377         }
378
379         if (link_stat(prev_name, &prev_st, 0) < 0) {
380                 if (!dry_run || errno != ENOENT) {
381                         rsyserr(FERROR_XFER, errno, "stat %s failed", full_fname(prev_name));
382                         return -1;
383                 }
384                 /* A new hard-link will get a new dev & inode, so approximate
385                  * those values in dry-run mode by zeroing them. */
386                 memset(&prev_st, 0, sizeof prev_st);
387         }
388
389         if (statret < 0 && basis_dir[0] != NULL) {
390                 /* If we match an alt-dest item, we don't output this as a change. */
391                 char cmpbuf[MAXPATHLEN];
392                 stat_x alt_sx;
393                 int j = 0;
394                 init_stat_x(&alt_sx);
395                 do {
396                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
397                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
398                                 continue;
399                         if (link_dest) {
400                                 if (prev_st.st_dev != alt_sx.st.st_dev
401                                  || prev_st.st_ino != alt_sx.st.st_ino)
402                                         continue;
403                                 statret = 1;
404                                 if (stdout_format_has_i == 0
405                                  || (!INFO_GTE(NAME, 2) && stdout_format_has_i < 2)) {
406                                         itemizing = 0;
407                                         code = FNONE;
408                                         if (INFO_GTE(NAME, 2) && maybe_ATTRS_REPORT)
409                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
410                                 }
411                                 break;
412                         }
413                         if (!unchanged_file(cmpbuf, file, &alt_sx.st))
414                                 continue;
415                         statret = 1;
416                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
417                                 break;
418                 } while (basis_dir[++j] != NULL);
419                 if (statret == 1) {
420                         sxp->st = alt_sx.st;
421 #ifdef SUPPORT_ACLS
422                         if (preserve_acls && !S_ISLNK(file->mode)) {
423                                 free_acl(sxp);
424                                 if (!ACL_READY(alt_sx))
425                                         get_acl(cmpbuf, sxp);
426                                 else {
427                                         sxp->acc_acl = alt_sx.acc_acl;
428                                         sxp->def_acl = alt_sx.def_acl;
429                                         alt_sx.acc_acl = alt_sx.def_acl = NULL;
430                                 }
431                         }
432 #endif
433 #ifdef SUPPORT_XATTRS
434                         if (preserve_xattrs) {
435                                 free_xattr(sxp);
436                                 if (!XATTR_READY(alt_sx))
437                                         get_xattr(cmpbuf, sxp);
438                                 else {
439                                         sxp->xattr = alt_sx.xattr;
440                                         alt_sx.xattr = NULL;
441                                 }
442                         }
443 #endif
444                 } else
445                         free_stat_x(&alt_sx);
446         }
447
448         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
449                             realname, itemizing, code) < 0)
450                 return -1;
451
452         if (remove_source_files == 1 && do_xfers)
453                 send_msg_int(MSG_SUCCESS, ndx);
454
455         return 1;
456 }
457
458 int hard_link_one(struct file_struct *file, const char *fname,
459                   const char *oldname, int terse)
460 {
461         if (do_link(oldname, fname) < 0) {
462                 enum logcode code;
463                 if (terse) {
464                         if (!INFO_GTE(NAME, 1))
465                                 return 0;
466                         code = FINFO;
467                 } else
468                         code = FERROR_XFER;
469                 rsyserr(code, errno, "link %s => %s failed",
470                         full_fname(fname), oldname);
471                 return 0;
472         }
473
474         file->flags |= FLAG_HLINK_DONE;
475
476         return 1;
477 }
478
479 void finish_hard_link(struct file_struct *file, const char *fname, int fin_ndx,
480                       STRUCT_STAT *stp, int itemizing, enum logcode code,
481                       int alt_dest)
482 {
483         stat_x prev_sx;
484         STRUCT_STAT st;
485         char prev_name[MAXPATHLEN], alt_name[MAXPATHLEN];
486         const char *our_name;
487         struct file_list *flist;
488         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
489
490         if (stp == NULL && prev_ndx >= 0) {
491                 if (link_stat(fname, &st, 0) < 0 && !dry_run) {
492                         rsyserr(FERROR_XFER, errno, "stat %s failed",
493                                 full_fname(fname));
494                         return;
495                 }
496                 stp = &st;
497         }
498
499         /* FIRST combined with DONE means we were the first to get done. */
500         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
501         F_HL_PREV(file) = alt_dest;
502         if (alt_dest >= 0 && dry_run) {
503                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
504                          f_name(file, NULL));
505                 our_name = alt_name;
506         } else
507                 our_name = fname;
508
509         init_stat_x(&prev_sx);
510
511         while ((ndx = prev_ndx) >= 0) {
512                 int val;
513                 flist = flist_for_ndx(ndx, "finish_hard_link");
514                 file = flist->files[ndx - flist->ndx_start];
515                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
516                 prev_ndx = F_HL_PREV(file);
517                 F_HL_PREV(file) = fin_ndx;
518                 prev_statret = link_stat(f_name(file, prev_name), &prev_sx.st, 0);
519                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
520                                       our_name, stp, fname, itemizing, code);
521                 flist->in_progress--;
522                 free_stat_x(&prev_sx);
523                 if (val < 0)
524                         continue;
525                 if (remove_source_files == 1 && do_xfers)
526                         send_msg_int(MSG_SUCCESS, ndx);
527         }
528
529         if (inc_recurse) {
530                 int gnum = F_HL_GNUM(file);
531                 struct ht_int32_node *node = hashtable_find(prior_hlinks, gnum, 0);
532                 if (node == NULL) {
533                         rprintf(FERROR, "Unable to find a hlink node for %d (%s)\n", gnum, f_name(file, prev_name));
534                         exit_cleanup(RERR_MESSAGEIO);
535                 }
536                 if (node->data == NULL) {
537                         rprintf(FERROR, "Hlink node data for %d is NULL (%s)\n", gnum, f_name(file, prev_name));
538                         exit_cleanup(RERR_MESSAGEIO);
539                 }
540                 if (CVAL(node->data, 0) != 0) {
541                         rprintf(FERROR, "Hlink node data for %d already has path=%s (%s)\n",
542                                 gnum, (char*)node->data, f_name(file, prev_name));
543                         exit_cleanup(RERR_MESSAGEIO);
544                 }
545                 free(node->data);
546                 if (!(node->data = strdup(our_name)))
547                         out_of_memory("finish_hard_link");
548         }
549 }
550
551 int skip_hard_link(struct file_struct *file, struct file_list **flist_p)
552 {
553         struct file_list *flist;
554         int prev_ndx;
555
556         file->flags |= FLAG_SKIP_HLINK;
557         if (!(file->flags & FLAG_HLINK_LAST))
558                 return -1;
559
560         check_prior(file, F_HL_GNUM(file), &prev_ndx, &flist);
561         if (prev_ndx >= 0) {
562                 file = flist->files[prev_ndx - flist->ndx_start];
563                 if (file->flags & (FLAG_HLINK_DONE|FLAG_FILE_SENT))
564                         return -1;
565                 file->flags |= FLAG_HLINK_LAST;
566                 *flist_p = flist;
567         }
568
569         return prev_ndx;
570 }
571 #endif