a7586938eda68666e2d5844ea0187de174d71589
[rsync/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) 2004-2007 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 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
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 #ifdef ICONV_OPTION
38 extern int ic_ndx;
39 #endif
40
41 #ifdef SUPPORT_HARD_LINKS
42
43 /* Starting with protocol 30, we use a simple hashtable on the sending side
44  * for hashing the st_dev and st_ino info.  The receiving side gets told
45  * (via flags and a "group index") which items are hard-linked together, so
46  * we can avoid the pool of dev+inode data. */
47
48 struct hashtable *dev_tbl;
49
50 void init_hard_links(void)
51 {
52         dev_tbl = hashtable_create(16, SIZEOF_INT64 == 8);
53 }
54
55 struct ht_int64_node *idev_find(int64 dev, int64 ino)
56 {
57         static struct ht_int64_node *dev_node = NULL;
58         struct hashtable *tbl;
59
60         if (!dev_node || dev_node->key != dev) {
61                 /* We keep a separate hash table of inodes for every device. */
62                 dev_node = hashtable_find(dev_tbl, dev, 1);
63                 if (!(tbl = dev_node->data))
64                         tbl = dev_node->data = hashtable_create(512, SIZEOF_INT64 == 8);
65         } else
66                 tbl = dev_node->data;
67
68         return hashtable_find(tbl, ino, 1);
69 }
70
71 void idev_destroy(void)
72 {
73         int i;
74
75         for (i = 0; i < dev_tbl->size; i++) {
76                 struct ht_int32_node *node = HT_NODE(dev_tbl, dev_tbl->nodes, i);
77                 if (node->data)
78                         hashtable_destroy(node->data);
79         }
80
81         hashtable_destroy(dev_tbl);
82 }
83
84 static int hlink_compare_gnum(int *int1, int *int2)
85 {
86         struct file_struct *f1 = cur_flist->sorted[*int1];
87         struct file_struct *f2 = cur_flist->sorted[*int2];
88         int32 gnum1 = F_HL_GNUM(f1);
89         int32 gnum2 = F_HL_GNUM(f2);
90
91         if (gnum1 != gnum2)
92                 return gnum1 > gnum2 ? 1 : -1;
93
94         return *int1 > *int2 ? 1 : -1;
95 }
96
97 static void match_gnums(int32 *ndx_list, int ndx_count)
98 {
99         int32 from, prev;
100         struct file_struct *file, *file_next;
101         int32 gnum, gnum_next;
102
103         qsort(ndx_list, ndx_count, sizeof ndx_list[0],
104              (int (*)()) hlink_compare_gnum);
105
106         for (from = 0; from < ndx_count; from++) {
107                 for (file = cur_flist->sorted[ndx_list[from]], gnum = F_HL_GNUM(file), prev = -1;
108                      from < ndx_count-1;
109                      file = file_next, gnum = gnum_next, from++) /*SHARED ITERATOR*/
110                 {
111                         file_next = cur_flist->sorted[ndx_list[from+1]];
112                         gnum_next = F_HL_GNUM(file_next);
113                         if (gnum != gnum_next)
114                                 break;
115                         if (prev < 0)
116                                 file->flags |= FLAG_HLINK_FIRST;
117                         F_HL_PREV(file) = prev;
118                         /* The linked list must use raw ndx values. */
119 #ifdef ICONV_OPTION
120                         if (ic_ndx)
121                                 prev = F_NDX(file);
122                         else
123 #endif
124                                 prev = ndx_list[from];
125                 }
126                 if (prev < 0)
127                         file->flags &= ~FLAG_HLINKED;
128                 else {
129                         file->flags |= FLAG_HLINK_LAST;
130                         F_HL_PREV(file) = prev;
131                 }
132         }
133 }
134
135 /* Analyze the hard-links in the file-list by creating a list of all the
136  * items that have hlink data, sorting them, and matching up identical
137  * values into clusters.  These will be a single linked list from last
138  * to first when we're done. */
139 void match_hard_links(void)
140 {
141         int i, ndx_count = 0;
142         int32 *ndx_list;
143
144         if (!(ndx_list = new_array(int32, cur_flist->used)))
145                 out_of_memory("match_hard_links");
146
147         for (i = 0; i < cur_flist->used; i++) {
148                 if (F_IS_HLINKED(cur_flist->sorted[i]))
149                         ndx_list[ndx_count++] = i;
150         }
151
152         if (ndx_count)
153                 match_gnums(ndx_list, ndx_count);
154
155         free(ndx_list);
156         if (protocol_version < 30)
157                 idev_destroy();
158 }
159
160 static int maybe_hard_link(struct file_struct *file, int ndx,
161                            const char *fname, int statret, statx *sxp,
162                            const char *oldname, STRUCT_STAT *old_stp,
163                            const char *realname, int itemizing, enum logcode code)
164 {
165         if (statret == 0) {
166                 if (sxp->st.st_dev == old_stp->st_dev
167                  && sxp->st.st_ino == old_stp->st_ino) {
168                         if (itemizing) {
169                                 itemize(fname, file, ndx, statret, sxp,
170                                         ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS,
171                                         0, "");
172                         }
173                         if (verbose > 1 && maybe_ATTRS_REPORT)
174                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
175                         file->flags |= FLAG_HLINK_DONE;
176                         return 0;
177                 }
178                 if (make_backups > 0) {
179                         if (!make_backup(fname))
180                                 return -1;
181                 } else if (robust_unlink(fname)) {
182                         rsyserr(FERROR, errno, "unlink %s failed",
183                                 full_fname(fname));
184                         return -1;
185                 }
186         }
187
188         if (hard_link_one(file, fname, oldname, 0)) {
189                 if (itemizing) {
190                         itemize(fname, file, ndx, statret, sxp,
191                                 ITEM_LOCAL_CHANGE | ITEM_XNAME_FOLLOWS, 0,
192                                 realname);
193                 }
194                 if (code != FNONE && verbose)
195                         rprintf(code, "%s => %s\n", fname, realname);
196                 return 0;
197         }
198         return -1;
199 }
200
201 /* Only called if FLAG_HLINKED is set and FLAG_HLINK_FIRST is not.  Returns:
202  * 0 = process the file, 1 = skip the file, -1 = error occurred. */
203 int hard_link_check(struct file_struct *file, int ndx, const char *fname,
204                     int statret, statx *sxp, int itemizing,
205                     enum logcode code)
206 {
207         STRUCT_STAT prev_st;
208         char prev_name[MAXPATHLEN], altbuf[MAXPATHLEN], *realname;
209         int alt_dest, prev_ndx = F_HL_PREV(file);
210         struct file_struct *prev_file = cur_flist->files[prev_ndx];
211
212         /* Is the previous link is not complete yet? */
213         if (!(prev_file->flags & FLAG_HLINK_DONE)) {
214                 /* Is the previous link being transferred? */
215                 if (prev_file->flags & FLAG_FILE_SENT) {
216                         /* Add ourselves to the list of files that will be
217                          * updated when the transfer completes, and mark
218                          * ourself as waiting for the transfer. */
219                         F_HL_PREV(file) = F_HL_PREV(prev_file);
220                         F_HL_PREV(prev_file) = ndx;
221                         file->flags |= FLAG_FILE_SENT;
222                         return 1;
223                 }
224                 return 0;
225         }
226
227         /* There is a finished file to link with! */
228         if (!(prev_file->flags & FLAG_HLINK_FIRST)) {
229                 /* The previous previous will be marked with FIRST. */
230                 prev_ndx = F_HL_PREV(prev_file);
231                 prev_file = cur_flist->files[prev_ndx];
232                 /* Update our previous pointer to point to the first. */
233                 F_HL_PREV(file) = prev_ndx;
234         }
235         alt_dest = F_HL_PREV(prev_file); /* alternate value when DONE && FIRST */
236         if (alt_dest >= 0 && dry_run) {
237                 pathjoin(prev_name, MAXPATHLEN, basis_dir[alt_dest],
238                          f_name(prev_file, NULL));
239                 f_name(prev_file, altbuf);
240                 realname = altbuf;
241         } else {
242                 f_name(prev_file, prev_name);
243                 realname = prev_name;
244         }
245
246         if (link_stat(prev_name, &prev_st, 0) < 0) {
247                 rsyserr(FERROR, errno, "stat %s failed",
248                         full_fname(prev_name));
249                 return -1;
250         }
251
252         if (statret < 0 && basis_dir[0] != NULL) {
253                 /* If we match an alt-dest item, we don't output this as a change. */
254                 char cmpbuf[MAXPATHLEN];
255                 statx alt_sx;
256                 int j = 0;
257 #ifdef SUPPORT_ACLS
258                 alt_sx.acc_acl = alt_sx.def_acl = NULL;
259 #endif
260                 do {
261                         pathjoin(cmpbuf, MAXPATHLEN, basis_dir[j], fname);
262                         if (link_stat(cmpbuf, &alt_sx.st, 0) < 0)
263                                 continue;
264                         if (link_dest) {
265                                 if (prev_st.st_dev != alt_sx.st.st_dev
266                                  || prev_st.st_ino != alt_sx.st.st_ino)
267                                         continue;
268                                 statret = 1;
269                                 if (verbose < 2 || !stdout_format_has_i) {
270                                         itemizing = 0;
271                                         code = FNONE;
272                                         if (verbose > 1 && maybe_ATTRS_REPORT)
273                                                 rprintf(FCLIENT, "%s is uptodate\n", fname);
274                                 }
275                                 break;
276                         }
277                         if (!unchanged_file(cmpbuf, file, &alt_sx.st))
278                                 continue;
279                         statret = 1;
280                         if (unchanged_attrs(cmpbuf, file, &alt_sx))
281                                 break;
282                 } while (basis_dir[++j] != NULL);
283                 if (statret == 1) {
284                         sxp->st = alt_sx.st;
285 #ifdef SUPPORT_ACLS
286                         if (preserve_acls && !S_ISLNK(file->mode)) {
287                                 if (!ACL_READY(*sxp))
288                                         get_acl(cmpbuf, sxp);
289                                 else {
290                                         sxp->acc_acl = alt_sx.acc_acl;
291                                         sxp->def_acl = alt_sx.def_acl;
292                                 }
293                         }
294 #endif
295                 }
296 #ifdef SUPPORT_ACLS
297                 else if (preserve_acls)
298                         free_acl(&alt_sx);
299 #endif
300         }
301
302         if (maybe_hard_link(file, ndx, fname, statret, sxp, prev_name, &prev_st,
303                             realname, itemizing, code) < 0)
304                 return -1;
305
306         if (remove_source_files == 1 && do_xfers)
307                 send_msg_int(MSG_SUCCESS, ndx);
308
309         return 1;
310 }
311
312 int hard_link_one(struct file_struct *file, const char *fname,
313                   const char *oldname, int terse)
314 {
315         if (do_link(oldname, fname) < 0) {
316                 enum logcode code;
317                 if (terse) {
318                         if (!verbose)
319                                 return -1;
320                         code = FINFO;
321                 } else
322                         code = FERROR;
323                 rsyserr(code, errno, "link %s => %s failed",
324                         full_fname(fname), oldname);
325                 return 0;
326         }
327
328         file->flags |= FLAG_HLINK_DONE;
329
330         return 1;
331 }
332
333 void finish_hard_link(struct file_struct *file, const char *fname,
334                       STRUCT_STAT *stp, int itemizing, enum logcode code,
335                       int alt_dest)
336 {
337         statx prev_sx;
338         STRUCT_STAT st;
339         char alt_name[MAXPATHLEN], *prev_name;
340         const char *our_name;
341         int prev_statret, ndx, prev_ndx = F_HL_PREV(file);
342
343         if (stp == NULL && prev_ndx >= 0) {
344                 if (link_stat(fname, &st, 0) < 0) {
345                         rsyserr(FERROR, errno, "stat %s failed",
346                                 full_fname(fname));
347                         return;
348                 }
349                 stp = &st;
350         }
351
352         /* FIRST combined with DONE means we were the first to get done. */
353         file->flags |= FLAG_HLINK_FIRST | FLAG_HLINK_DONE;
354         F_HL_PREV(file) = alt_dest;
355         if (alt_dest >= 0 && dry_run) {
356                 pathjoin(alt_name, MAXPATHLEN, basis_dir[alt_dest],
357                          f_name(file, NULL));
358                 our_name = alt_name;
359         } else
360                 our_name = fname;
361
362 #ifdef SUPPORT_ACLS
363         prev_sx.acc_acl = prev_sx.def_acl = NULL;
364 #endif
365
366         while ((ndx = prev_ndx) >= 0) {
367                 int val;
368                 file = cur_flist->files[ndx];
369                 file->flags = (file->flags & ~FLAG_HLINK_FIRST) | FLAG_HLINK_DONE;
370                 prev_ndx = F_HL_PREV(file);
371                 prev_name = f_name(file, NULL);
372                 prev_statret = link_stat(prev_name, &prev_sx.st, 0);
373                 val = maybe_hard_link(file, ndx, prev_name, prev_statret, &prev_sx,
374                                       our_name, stp, fname, itemizing, code);
375 #ifdef SUPPORT_ACLS
376                 if (preserve_acls)
377                         free_acl(&prev_sx);
378 #endif
379                 if (val < 0)
380                         continue;
381                 if (remove_source_files == 1 && do_xfers)
382                         send_msg_int(MSG_SUCCESS, ndx);
383         }
384 }
385 #endif