1243447ae43dbf02969bd20e4494206956f22809
[rsync-patches.git] / detect-renamed.diff
1 This patch adds the --detect-renamed option which makes rsync notice files
2 that either (1) match in size & modify-time (plus the basename, if possible)
3 or (2) match in size & checksum (when --checksum was also specified) and use
4 each match as an alternate basis file to speed up the transfer.
5
6 The algorithm attempts to scan the receiving-side's files in an efficient
7 manner.  If --delete[-before] is enabled, we'll take advantage of the
8 pre-transfer delete pass to prepare any alternate-basis-file matches we
9 might find.  If --delete-before is not enabled, rsync does the rename scan
10 during the regular file-sending scan (scanning each directory right before
11 the generator starts updating files from that dir).  In this latter mode,
12 rsync might delay the updating of a file (if no alternate-basis match was
13 yet found) until the full scan of the receiving side is complete, at which
14 point any delayed files are processed.
15
16 I chose to hard-link the alternate-basis files into a ".~tmp~" subdir that
17 takes advantage of rsync's pre-existing partial-dir logic.  This uses less
18 memory than trying to keep track of the matches internally, and also allows
19 any deletions or file-updates to occur normally without interfering with
20 these alternate-basis discoveries.
21
22 To use this patch, run these commands for a successful build:
23
24     patch -p1 <patches/detect-renamed.diff
25     ./configure                                 (optional if already run)
26     make
27
28 TODO:
29
30   We need to never return a match from fattr_find() that has a basis
31   file.  This will ensure that we don't try to give a renamed file to
32   a file that can't use it, while missing out on giving it to a file
33   that could use it.
34
35 diff --git a/compat.c b/compat.c
36 --- a/compat.c
37 +++ b/compat.c
38 @@ -41,6 +41,7 @@ extern int checksum_seed;
39  extern int basis_dir_cnt;
40  extern int prune_empty_dirs;
41  extern int protocol_version;
42 +extern int detect_renamed;
43  extern int protect_args;
44  extern int preserve_uid;
45  extern int preserve_gid;
46 @@ -117,6 +118,7 @@ void set_allow_inc_recurse(void)
47                 allow_inc_recurse = 0;
48         else if (!am_sender
49          && (delete_before || delete_after
50 +         || detect_renamed
51           || delay_updates || prune_empty_dirs))
52                 allow_inc_recurse = 0;
53         else if (am_server && !local_server
54 diff --git a/flist.c b/flist.c
55 --- a/flist.c
56 +++ b/flist.c
57 @@ -63,6 +63,7 @@ extern int non_perishable_cnt;
58  extern int prune_empty_dirs;
59  extern int copy_links;
60  extern int copy_unsafe_links;
61 +extern int detect_renamed;
62  extern int protocol_version;
63  extern int sanitize_paths;
64  extern int munge_symlinks;
65 @@ -120,6 +121,8 @@ static int64 tmp_dev, tmp_ino;
66  #endif
67  static char tmp_sum[MAX_DIGEST_LEN];
68  
69 +struct file_list the_fattr_list;
70 +
71  static char empty_sum[MAX_DIGEST_LEN];
72  static int flist_count_offset; /* for --delete --progress */
73  static int dir_count = 0;
74 @@ -263,6 +266,45 @@ static int is_excluded(char *fname, int is_dir, int filter_level)
75         return 0;
76  }
77  
78 +static int fattr_compare(struct file_struct **file1, struct file_struct **file2)
79 +{
80 +       struct file_struct *f1 = *file1;
81 +       struct file_struct *f2 = *file2;
82 +       int64 len1 = F_LENGTH(f1), len2 = F_LENGTH(f2);
83 +       int diff;
84 +
85 +       if (!f1->basename || !S_ISREG(f1->mode) || !len1) {
86 +               if (!f2->basename || !S_ISREG(f2->mode) || !len2)
87 +                       return 0;
88 +               return 1;
89 +       }
90 +       if (!f2->basename || !S_ISREG(f2->mode) || !len2)
91 +               return -1;
92 +
93 +       /* Don't use diff for values that are longer than an int. */
94 +       if (len1 != len2)
95 +               return len1 < len2 ? -1 : 1;
96 +
97 +       if (always_checksum) {
98 +               diff = u_memcmp(F_SUM(f1), F_SUM(f2), checksum_len);
99 +               if (diff)
100 +                       return diff;
101 +       } else if (f1->modtime != f2->modtime)
102 +               return f1->modtime < f2->modtime ? -1 : 1;
103 +
104 +       diff = u_strcmp(f1->basename, f2->basename);
105 +       if (diff)
106 +               return diff;
107 +
108 +       if (f1->dirname == f2->dirname)
109 +               return 0;
110 +       if (!f1->dirname)
111 +               return -1;
112 +       if (!f2->dirname)
113 +               return 1;
114 +       return u_strcmp(f1->dirname, f2->dirname);
115 +}
116 +
117  static void send_directory(int f, struct file_list *flist,
118                            char *fbuf, int len, int flags);
119  
120 @@ -2180,6 +2222,25 @@ struct file_list *recv_file_list(int f)
121  
122         flist_sort_and_clean(flist, relative_paths);
123  
124 +       if (detect_renamed) {
125 +               int j = flist->used;
126 +               the_fattr_list.used = j;
127 +               the_fattr_list.files = new_array(struct file_struct *, j);
128 +               if (!the_fattr_list.files)
129 +                       out_of_memory("recv_file_list");
130 +               memcpy(the_fattr_list.files, flist->files,
131 +                      j * sizeof (struct file_struct *));
132 +               qsort(the_fattr_list.files, j,
133 +                     sizeof the_fattr_list.files[0], (int (*)())fattr_compare);
134 +               the_fattr_list.low = 0;
135 +               while (j-- > 0) {
136 +                       struct file_struct *fp = the_fattr_list.files[j];
137 +                       if (fp->basename && S_ISREG(fp->mode) && F_LENGTH(fp))
138 +                               break;
139 +               }
140 +               the_fattr_list.high = j;
141 +       }
142 +
143         if (protocol_version < 30) {
144                 /* Recv the io_error flag */
145                 if (ignore_errors)
146 diff --git a/generator.c b/generator.c
147 --- a/generator.c
148 +++ b/generator.c
149 @@ -81,6 +81,7 @@ extern char *basis_dir[];
150  extern int compare_dest;
151  extern int copy_dest;
152  extern int link_dest;
153 +extern int detect_renamed;
154  extern int whole_file;
155  extern int list_only;
156  extern int read_batch;
157 @@ -99,6 +100,7 @@ extern char *backup_suffix;
158  extern int backup_suffix_len;
159  extern struct file_list *cur_flist, *first_flist, *dir_flist;
160  extern struct filter_list_struct server_filter_list;
161 +extern struct file_list the_fattr_list;
162  
163  int ignore_perishable = 0;
164  int non_perishable_cnt = 0;
165 @@ -106,6 +108,7 @@ int maybe_ATTRS_REPORT = 0;
166  
167  static dev_t dev_zero;
168  static int deletion_count = 0; /* used to implement --max-delete */
169 +static int unexplored_dirs = 1;
170  static int deldelay_size = 0, deldelay_cnt = 0;
171  static char *deldelay_buf = NULL;
172  static int deldelay_fd = -1;
173 @@ -115,7 +118,7 @@ static int need_retouch_dir_times;
174  static int need_retouch_dir_perms;
175  static const char *solo_file = NULL;
176  
177 -/* For calling delete_item() and delete_dir_contents(). */
178 +/* For calling delete_item(), delete_dir_contents(), and delete_in_dir(). */
179  #define DEL_NO_UID_WRITE       (1<<0) /* file/dir has our uid w/o write perm */
180  #define DEL_RECURSE            (1<<1) /* if dir, delete all contents */
181  #define DEL_DIR_IS_EMPTY       (1<<2) /* internal delete_FUNCTIONS use only */
182 @@ -124,6 +127,7 @@ static const char *solo_file = NULL;
183  #define DEL_FOR_SYMLINK        (1<<5) /* making room for a replacement symlink */
184  #define DEL_FOR_DEVICE         (1<<6) /* making room for a replacement device */
185  #define DEL_FOR_SPECIAL        (1<<7) /* making room for a replacement special */
186 +#define DEL_NO_DELETIONS       (1<<9) /* just check for renames w/o deleting */
187  
188  #define DEL_MAKE_ROOM (DEL_FOR_FILE|DEL_FOR_DIR|DEL_FOR_SYMLINK|DEL_FOR_DEVICE|DEL_FOR_SPECIAL)
189  
190 @@ -144,11 +148,121 @@ static int is_backup_file(char *fn)
191         return k > 0 && strcmp(fn+k, backup_suffix) == 0;
192  }
193  
194 +/* Search for a regular file that matches either (1) the size & modified
195 + * time (plus the basename, if possible) or (2) the size & checksum.  If
196 + * we find an exact match down to the dirname, return -1 because we found
197 + * an up-to-date file in the transfer, not a renamed file. */
198 +static int fattr_find(struct file_struct *f, char *fname)
199 +{
200 +       int low = the_fattr_list.low, high = the_fattr_list.high;
201 +       int mid, ok_match = -1, good_match = -1;
202 +       struct file_struct *fmid;
203 +       int diff;
204 +
205 +       while (low <= high) {
206 +               mid = (low + high) / 2;
207 +               fmid = the_fattr_list.files[mid];
208 +               if (F_LENGTH(fmid) != F_LENGTH(f)) {
209 +                       if (F_LENGTH(fmid) < F_LENGTH(f))
210 +                               low = mid + 1;
211 +                       else
212 +                               high = mid - 1;
213 +                       continue;
214 +               }
215 +               if (always_checksum) {
216 +                       /* We use the FLAG_FILE_SENT flag to indicate when we
217 +                        * have computed the checksum for an entry. */
218 +                       if (!(f->flags & FLAG_FILE_SENT)) {
219 +                               if (fmid->modtime == f->modtime
220 +                                && f_name_cmp(fmid, f) == 0)
221 +                                       return -1; /* assume we can't help */
222 +                               file_checksum(fname, F_SUM(f), F_LENGTH(f));
223 +                               f->flags |= FLAG_FILE_SENT;
224 +                       }
225 +                       diff = u_memcmp(F_SUM(fmid), F_SUM(f), checksum_len);
226 +                       if (diff) {
227 +                               if (diff < 0)
228 +                                       low = mid + 1;
229 +                               else
230 +                                       high = mid - 1;
231 +                               continue;
232 +                       }
233 +               } else {
234 +                       if (fmid->modtime != f->modtime) {
235 +                               if (fmid->modtime < f->modtime)
236 +                                       low = mid + 1;
237 +                               else
238 +                                       high = mid - 1;
239 +                               continue;
240 +                       }
241 +               }
242 +               ok_match = mid;
243 +               diff = u_strcmp(fmid->basename, f->basename);
244 +               if (diff == 0) {
245 +                       good_match = mid;
246 +                       if (fmid->dirname == f->dirname)
247 +                               return -1; /* file is up-to-date */
248 +                       if (!fmid->dirname) {
249 +                               low = mid + 1;
250 +                               continue;
251 +                       }
252 +                       if (!f->dirname) {
253 +                               high = mid - 1;
254 +                               continue;
255 +                       }
256 +                       diff = u_strcmp(fmid->dirname, f->dirname);
257 +                       if (diff == 0)
258 +                               return -1; /* file is up-to-date */
259 +               }
260 +               if (diff < 0)
261 +                       low = mid + 1;
262 +               else
263 +                       high = mid - 1;
264 +       }
265 +
266 +       return good_match >= 0 ? good_match : ok_match;
267 +}
268 +
269 +static void look_for_rename(struct file_struct *file, char *fname)
270 +{
271 +       struct file_struct *fp;
272 +       char *partialptr, *fn;
273 +       STRUCT_STAT st;
274 +       int ndx;
275 +
276 +       if (!partial_dir || (ndx = fattr_find(file, fname)) < 0)
277 +               return;
278 +
279 +       fp = the_fattr_list.files[ndx];
280 +       fn = f_name(fp, NULL);
281 +       /* We don't provide an alternate-basis file if there is a basis file. */
282 +       if (link_stat(fn, &st, 0) == 0)
283 +               return;
284 +
285 +       if (!dry_run) {
286 +               if ((partialptr = partial_dir_fname(fn)) == NULL
287 +                || !handle_partial_dir(partialptr, PDIR_CREATE))
288 +                       return;
289 +               /* We only use the file if we can hard-link it into our tmp dir. */
290 +               if (link(fname, partialptr) != 0) {
291 +                       if (errno != EEXIST)
292 +                               handle_partial_dir(partialptr, PDIR_DELETE);
293 +                       return;
294 +               }
295 +       }
296 +
297 +       /* I think this falls into the -vv category with "%s is uptodate", etc. */
298 +       if (verbose > 1)
299 +               rprintf(FINFO, "found renamed: %s => %s\n", fname, fn);
300 +}
301 +
302  /* Delete a file or directory.  If DEL_RECURSE is set in the flags, this will
303   * delete recursively.
304   *
305   * Note that fbuf must point to a MAXPATHLEN buffer if the mode indicates it's
306   * a directory! (The buffer is used for recursion, but returned unchanged.)
307 + *
308 + * Also note: --detect-rename may use this routine with DEL_NO_DELETIONS set!
309   */
310  static enum delret delete_item(char *fbuf, uint16 mode, uint16 flags)
311  {
312 @@ -182,6 +296,8 @@ static enum delret delete_item(char *fbuf, uint16 mode, uint16 flags)
313                         goto check_ret;
314                 /* OK: try to delete the directory. */
315         }
316 +       if (flags & DEL_NO_DELETIONS)
317 +               return DR_SUCCESS;
318  
319         if (!(flags & DEL_MAKE_ROOM) && max_delete >= 0 && ++deletion_count > max_delete)
320                 return DR_AT_LIMIT;
321 @@ -237,6 +353,8 @@ static enum delret delete_item(char *fbuf, uint16 mode, uint16 flags)
322   * its contents, otherwise just checks for content.  Returns DR_SUCCESS or
323   * DR_NOT_EMPTY.  Note that fname must point to a MAXPATHLEN buffer!  (The
324   * buffer is used for recursion, but returned unchanged.)
325 + *
326 + * Note: --detect-rename may use this routine with DEL_NO_DELETIONS set!
327   */
328  static enum delret delete_dir_contents(char *fname, uint16 flags)
329  {
330 @@ -256,7 +374,9 @@ static enum delret delete_dir_contents(char *fname, uint16 flags)
331         save_filters = push_local_filters(fname, dlen);
332  
333         non_perishable_cnt = 0;
334 +       file_extra_cnt += SUM_EXTRA_CNT;
335         dirlist = get_dirlist(fname, dlen, 0);
336 +       file_extra_cnt -= SUM_EXTRA_CNT;
337         ret = non_perishable_cnt ? DR_NOT_EMPTY : DR_SUCCESS;
338  
339         if (!dirlist->used)
340 @@ -296,7 +416,8 @@ static enum delret delete_dir_contents(char *fname, uint16 flags)
341                 if (S_ISDIR(fp->mode)) {
342                         if (delete_dir_contents(fname, flags | DEL_RECURSE) != DR_SUCCESS)
343                                 ret = DR_NOT_EMPTY;
344 -               }
345 +               } else if (detect_renamed && S_ISREG(fp->mode))
346 +                       look_for_rename(fp, fname);
347                 if (delete_item(fname, fp->mode, flags) != DR_SUCCESS)
348                         ret = DR_NOT_EMPTY;
349         }
350 @@ -461,13 +582,18 @@ static void do_delayed_deletions(char *delbuf)
351   * all the --delete-WHEN options.  Note that the fbuf pointer must point to a
352   * MAXPATHLEN buffer with the name of the directory in it (the functions we
353   * call will append names onto the end, but the old dir value will be restored
354 - * on exit). */
355 -static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
356 + * on exit).
357 + *
358 + * Note:  --detect-rename may use this routine with DEL_NO_DELETIONS set!
359 + */
360 +static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev,
361 +                         int del_flags)
362  {
363         static int already_warned = 0;
364         struct file_list *dirlist;
365 -       char delbuf[MAXPATHLEN];
366 -       int dlen, i;
367 +       char *p, delbuf[MAXPATHLEN];
368 +       unsigned remainder;
369 +       int dlen, i, restore_dot = 0;
370         int save_uid_ndx = uid_ndx;
371  
372         if (!fbuf) {
373 @@ -482,17 +608,22 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
374                 maybe_send_keepalive();
375  
376         if (io_error && !ignore_errors) {
377 -               if (already_warned)
378 +               if (!already_warned) {
379 +                       rprintf(FINFO,
380 +                           "IO error encountered -- skipping file deletion\n");
381 +                       already_warned = 1;
382 +               }
383 +               if (!detect_renamed)
384                         return;
385 -               rprintf(FINFO,
386 -                       "IO error encountered -- skipping file deletion\n");
387 -               already_warned = 1;
388 -               return;
389 +               del_flags |= DEL_NO_DELETIONS;
390         }
391  
392         dlen = strlen(fbuf);
393         change_local_filter_dir(fbuf, dlen, F_DEPTH(file));
394  
395 +       if (detect_renamed)
396 +               unexplored_dirs--;
397 +
398         if (one_file_system) {
399                 if (file->flags & FLAG_TOP_DIR)
400                         filesystem_dev = *fs_dev;
401 @@ -505,6 +636,14 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
402  
403         dirlist = get_dirlist(fbuf, dlen, 0);
404  
405 +       p = fbuf + dlen;
406 +       if (dlen == 1 && *fbuf == '.') {
407 +               restore_dot = 1;
408 +               p = fbuf;
409 +       } else if (dlen != 1 || *fbuf != '/')
410 +               *p++ = '/';
411 +       remainder = MAXPATHLEN - (p - fbuf);
412 +
413         /* If an item in dirlist is not found in flist, delete it
414          * from the filesystem. */
415         for (i = dirlist->used; i--; ) {
416 @@ -517,19 +656,28 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
417                                         f_name(fp, NULL));
418                         continue;
419                 }
420 +               if (detect_renamed && S_ISREG(fp->mode)) {
421 +                       strlcpy(p, fp->basename, remainder);
422 +                       look_for_rename(fp, fbuf);
423 +               }
424                 if (flist_find(cur_flist, fp) < 0) {
425                         int flags = DEL_RECURSE;
426                         if (!(fp->mode & S_IWUSR) && !am_root && (uid_t)F_OWNER(fp) == our_uid)
427                                 flags |= DEL_NO_UID_WRITE;
428                         f_name(fp, delbuf);
429 -                       if (delete_during == 2) {
430 -                               if (!remember_delete(fp, delbuf, flags))
431 +                       if (delete_during == 2 && !(del_flags & DEL_NO_DELETIONS)) {
432 +                               if (!remember_delete(fp, delbuf, del_flags | flags))
433                                         break;
434                         } else
435 -                               delete_item(delbuf, fp->mode, flags);
436 -               }
437 +                               delete_item(delbuf, fp->mode, del_flags | flags);
438 +               } else if (detect_renamed && S_ISDIR(fp->mode))
439 +                       unexplored_dirs++;
440         }
441  
442 +       if (restore_dot)
443 +               fbuf[0] = '.';
444 +       fbuf[dlen] = '\0';
445 +
446         flist_free(dirlist);
447  
448         if (!save_uid_ndx) {
449 @@ -564,9 +712,9 @@ static void do_delete_pass(void)
450                  || !S_ISDIR(st.st_mode))
451                         continue;
452  
453 -               delete_in_dir(fbuf, file, &st.st_dev);
454 +               delete_in_dir(fbuf, file, &st.st_dev, 0);
455         }
456 -       delete_in_dir(NULL, NULL, &dev_zero);
457 +       delete_in_dir(NULL, NULL, &dev_zero, 0);
458  
459         if (do_progress && !am_server)
460                 rprintf(FINFO, "                    \r");
461 @@ -1199,6 +1347,7 @@ static void list_file_entry(struct file_struct *f)
462         }
463  }
464  
465 +static struct bitbag *delayed_bits = NULL;
466  static int phase = 0;
467  static int dflt_perms;
468  
469 @@ -1449,8 +1598,12 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
470                         }
471                 }
472                 else if (delete_during && f_out != -1 && !phase && dry_run < 2
473 -                   && (file->flags & FLAG_CONTENT_DIR))
474 -                       delete_in_dir(fname, file, &real_sx.st.st_dev);
475 +                   && (file->flags & FLAG_CONTENT_DIR)) {
476 +                       if (detect_renamed && real_ret != 0)
477 +                               unexplored_dirs++;
478 +                       delete_in_dir(fname, file, &real_sx.st.st_dev,
479 +                                     delete_during < 0 ? DEL_NO_DELETIONS : 0);
480 +               }
481                 goto cleanup;
482         }
483  
484 @@ -1728,8 +1881,14 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
485                         goto cleanup;
486                 }
487  #endif
488 -               if (stat_errno == ENOENT)
489 +               if (stat_errno == ENOENT) {
490 +                       if (detect_renamed && unexplored_dirs > 0
491 +                        && F_LENGTH(file)) {
492 +                               bitbag_set_bit(delayed_bits, ndx);
493 +                               return;
494 +                       }
495                         goto notify_others;
496 +               }
497                 rsyserr(FERROR_XFER, stat_errno, "recv_generator: failed to stat %s",
498                         full_fname(fname));
499                 goto cleanup;
500 @@ -2072,6 +2231,12 @@ void generate_files(int f_out, const char *local_name)
501         if (verbose > 2)
502                 rprintf(FINFO, "generator starting pid=%ld\n", (long)getpid());
503  
504 +       if (detect_renamed) {
505 +               delayed_bits = bitbag_create(cur_flist->used);
506 +               if (!delete_before && !delete_during)
507 +                       delete_during = -1;
508 +       }
509 +
510         if (delete_before && !solo_file && cur_flist->used > 0)
511                 do_delete_pass();
512         if (delete_during == 2) {
513 @@ -2082,7 +2247,7 @@ void generate_files(int f_out, const char *local_name)
514         }
515         do_progress = 0;
516  
517 -       if (append_mode > 0 || whole_file < 0)
518 +       if (append_mode > 0 || detect_renamed || whole_file < 0)
519                 whole_file = 0;
520         if (verbose >= 2) {
521                 rprintf(FINFO, "delta-transmission %s\n",
522 @@ -2120,7 +2285,7 @@ void generate_files(int f_out, const char *local_name)
523                                                 dirdev = MAKEDEV(DEV_MAJOR(devp), DEV_MINOR(devp));
524                                         } else
525                                                 dirdev = MAKEDEV(0, 0);
526 -                                       delete_in_dir(f_name(fp, fbuf), fp, &dirdev);
527 +                                       delete_in_dir(f_name(fp, fbuf), fp, &dirdev, 0);
528                                 }
529                         }
530                 }
531 @@ -2163,7 +2328,21 @@ void generate_files(int f_out, const char *local_name)
532         } while ((cur_flist = cur_flist->next) != NULL);
533  
534         if (delete_during)
535 -               delete_in_dir(NULL, NULL, &dev_zero);
536 +               delete_in_dir(NULL, NULL, &dev_zero, 0);
537 +       if (detect_renamed) {
538 +               if (delete_during < 0)
539 +                       delete_during = 0;
540 +               detect_renamed = 0;
541 +
542 +               for (i = -1; (i = bitbag_next_bit(delayed_bits, i)) >= 0; ) {
543 +                       struct file_struct *file = cur_flist->files[i];
544 +                       if (local_name)
545 +                               strlcpy(fbuf, local_name, sizeof fbuf);
546 +                       else
547 +                               f_name(file, fbuf);
548 +                       recv_generator(fbuf, file, i, itemizing, code, f_out);
549 +               }
550 +       }
551         phase++;
552         if (verbose > 2)
553                 rprintf(FINFO, "generate_files phase=%d\n", phase);
554 diff --git a/options.c b/options.c
555 --- a/options.c
556 +++ b/options.c
557 @@ -81,6 +81,7 @@ int am_generator = 0;
558  int am_starting_up = 1;
559  int relative_paths = -1;
560  int implied_dirs = 1;
561 +int detect_renamed = 0;
562  int numeric_ids = 0;
563  int allow_8bit_chars = 0;
564  int force_delete = 0;
565 @@ -390,6 +391,7 @@ void usage(enum logcode F)
566    rprintf(F,"     --modify-window=NUM     compare mod-times with reduced accuracy\n");
567    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
568    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
569 +  rprintf(F,"     --detect-renamed        try to find renamed files to speed up the transfer\n");
570    rprintf(F,"     --compare-dest=DIR      also compare destination files relative to DIR\n");
571    rprintf(F,"     --copy-dest=DIR         ... and include copies of unchanged files\n");
572    rprintf(F,"     --link-dest=DIR         hardlink to files in DIR when unchanged\n");
573 @@ -570,6 +572,7 @@ static struct poptOption long_options[] = {
574    {"compare-dest",     0,  POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
575    {"copy-dest",        0,  POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
576    {"link-dest",        0,  POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 },
577 +  {"detect-renamed",   0,  POPT_ARG_NONE,   &detect_renamed, 0, 0, 0 },
578    {"fuzzy",           'y', POPT_ARG_NONE,   &fuzzy_basis, 0, 0, 0 },
579    {"compress",        'z', POPT_ARG_NONE,   0, 'z', 0, 0 },
580    {"no-compress",      0,  POPT_ARG_VAL,    &do_compression, 0, 0, 0 },
581 @@ -1564,7 +1567,7 @@ int parse_arguments(int *argc_p, const char ***argv_p, int frommain)
582                 inplace = 1;
583         }
584  
585 -       if (delay_updates && !partial_dir)
586 +       if ((delay_updates || detect_renamed) && !partial_dir)
587                 partial_dir = tmp_partialdir;
588  
589         if (inplace) {
590 @@ -1573,6 +1576,7 @@ int parse_arguments(int *argc_p, const char ***argv_p, int frommain)
591                         snprintf(err_buf, sizeof err_buf,
592                                  "--%s cannot be used with --%s\n",
593                                  append_mode ? "append" : "inplace",
594 +                                detect_renamed ? "detect-renamed" :
595                                  delay_updates ? "delay-updates" : "partial-dir");
596                         return 0;
597                 }
598 @@ -1919,6 +1923,8 @@ void server_options(char **args, int *argc_p)
599                         args[ac++] = "--super";
600                 if (size_only)
601                         args[ac++] = "--size-only";
602 +               if (detect_renamed)
603 +                       args[ac++] = "--detect-renamed";
604         } else {
605                 if (skip_compress) {
606                         if (asprintf(&arg, "--skip-compress=%s", skip_compress) < 0)
607 diff --git a/rsync.yo b/rsync.yo
608 --- a/rsync.yo
609 +++ b/rsync.yo
610 @@ -385,6 +385,7 @@ to the detailed description below for a complete description.  verb(
611       --modify-window=NUM     compare mod-times with reduced accuracy
612   -T, --temp-dir=DIR          create temporary files in directory DIR
613   -y, --fuzzy                 find similar file for basis if no dest file
614 +     --detect-renamed        try to find renamed files to speed the xfer
615       --compare-dest=DIR      also compare received files relative to DIR
616       --copy-dest=DIR         ... and include copies of unchanged files
617       --link-dest=DIR         hardlink to files in DIR when unchanged
618 @@ -1467,6 +1468,21 @@ Note that the use of the bf(--delete) option might get rid of any potential
619  fuzzy-match files, so either use bf(--delete-after) or specify some
620  filename exclusions if you need to prevent this.
621  
622 +dit(bf(--detect-renamed)) With this option, for each new source file
623 +(call it em(src/S)), rsync looks for a file em(dest/D) anywhere in the
624 +destination that passes the quick check with em(src/S).  If such a em(dest/D)
625 +is found, rsync uses it as an alternate basis for transferring em(S).  The
626 +idea is that if em(src/S) was renamed from em(src/D) (as opposed to em(src/S)
627 +passing the quick check with em(dest/D) by coincidence), the delta-transfer
628 +algorithm will find that all the data matches between em(src/S) and em(dest/D),
629 +and the transfer will be really fast.
630 +
631 +By default, alternate-basis files are hard-linked into a directory named
632 +".~tmp~" in each file's destination directory, but if you've specified
633 +the bf(--partial-dir) option, that directory will be used instead.  These
634 +potential alternate-basis files will be removed as the transfer progresses.
635 +This option conflicts with bf(--inplace) and bf(--append).
636 +
637  dit(bf(--compare-dest=DIR)) This option instructs rsync to use em(DIR) on
638  the destination machine as an additional hierarchy to compare destination
639  files against doing transfers (if the files are missing in the destination
640 diff --git a/util.c b/util.c
641 --- a/util.c
642 +++ b/util.c
643 @@ -1022,6 +1022,32 @@ int handle_partial_dir(const char *fname, int create)
644         return 1;
645  }
646  
647 +/* We need to supply our own strcmp function for file list comparisons
648 + * to ensure that signed/unsigned usage is consistent between machines. */
649 +int u_strcmp(const char *p1, const char *p2)
650 +{
651 +        for ( ; *p1; p1++, p2++) {
652 +               if (*p1 != *p2)
653 +                       break;
654 +       }
655 +
656 +       return (int)*(uchar*)p1 - (int)*(uchar*)p2;
657 +}
658 +
659 +/* We need a memcmp function compares unsigned-byte values. */
660 +int u_memcmp(const void *p1, const void *p2, size_t len)
661 +{
662 +       const uchar *u1 = p1;
663 +       const uchar *u2 = p2;
664 +
665 +       while (len--) {
666 +               if (*u1 != *u2)
667 +                       return (int)*u1 - (int)*u2;
668 +       }
669 +
670 +       return 0;
671 +}
672 +
673  /**
674   * Determine if a symlink points outside the current directory tree.
675   * This is considered "unsafe" because e.g. when mirroring somebody