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   The routine that makes missing directories for files that get renamed
  31   down into a new sub-hierarchy doesn't properly handle the case where some
  32   path elements might exist but not be a dir yet.  We need to either change
  33   our stash-ahead algorithm (to not require unknown path elements) or we
  34   need to create a better path-making routine.
  35
  36   We need to never return a match from fattr_find() that has a basis
  37   file.  This will ensure that we don't try to give a renamed file to
  38   a file that can't use it, while missing out on giving it to a file
  39   that could use it.
  40
  41 based-on: 23a37ecac4bba997948fa30e72eb4aa8e317e394
  42 diff --git a/backup.c b/backup.c
  43 --- a/backup.c
  44 +++ b/backup.c
  45 @@ -162,7 +162,7 @@ char *get_backup_name(const char *fname)
  46                         int ret;
  47                         if (backup_dir_len > 1)
  48                                 backup_dir_buf[backup_dir_len-1] = '\0';
  49 -                       ret = make_path(backup_dir_buf, 0);
  50 +                       ret = make_path(backup_dir_buf, ACCESSPERMS, 0);
  51                         if (backup_dir_len > 1)
  52                                 backup_dir_buf[backup_dir_len-1] = '/';
  53                         if (ret < 0)
  54 diff --git a/compat.c b/compat.c
  55 --- a/compat.c
  56 +++ b/compat.c
  57 @@ -38,6 +38,7 @@ extern int checksum_seed;
  58  extern int basis_dir_cnt;
  59  extern int prune_empty_dirs;
  60  extern int protocol_version;
  61 +extern int detect_renamed;
  62  extern int protect_args;
  63  extern int preserve_uid;
  64  extern int preserve_gid;
  65 @@ -158,6 +159,7 @@ void set_allow_inc_recurse(void)
  66                 allow_inc_recurse = 0;
  67         else if (!am_sender
  68          && (delete_before || delete_after
  69 +         || detect_renamed
  70           || delay_updates || prune_empty_dirs))
  71                 allow_inc_recurse = 0;
  72         else if (am_server && !local_server
  73 diff --git a/delete.c b/delete.c
  74 --- a/delete.c
  75 +++ b/delete.c
  76 @@ -25,6 +25,7 @@
  77  extern int am_root;
  78  extern int make_backups;
  79  extern int max_delete;
  80 +extern int detect_renamed;
  81  extern char *backup_dir;
  82  extern char *backup_suffix;
  83  extern int backup_suffix_len;
  84 @@ -44,6 +45,8 @@ static inline int is_backup_file(char *fn)
  85   * its contents, otherwise just checks for content.  Returns DR_SUCCESS or
  86   * DR_NOT_EMPTY.  Note that fname must point to a MAXPATHLEN buffer!  (The
  87   * buffer is used for recursion, but returned unchanged.)
  88 + *
  89 + * Note: --detect-rename may use this routine with DEL_NO_DELETIONS set!
  90   */
  91  static enum delret delete_dir_contents(char *fname, uint16 flags)
  92  {
  93 @@ -63,7 +66,9 @@ static enum delret delete_dir_contents(char *fname, uint16 flags)
  94         save_filters = push_local_filters(fname, dlen);
  95
  96         non_perishable_cnt = 0;
  97 +       file_extra_cnt += SUM_EXTRA_CNT;
  98         dirlist = get_dirlist(fname, dlen, 0);
  99 +       file_extra_cnt -= SUM_EXTRA_CNT;
 100         ret = non_perishable_cnt ? DR_NOT_EMPTY : DR_SUCCESS;
 101
 102         if (!dirlist->used)
 103 @@ -103,7 +108,8 @@ static enum delret delete_dir_contents(char *fname, uint16 flags)
 104                 if (S_ISDIR(fp->mode)) {
 105                         if (delete_dir_contents(fname, flags | DEL_RECURSE) != DR_SUCCESS)
 106                                 ret = DR_NOT_EMPTY;
 107 -               }
 108 +               } else if (detect_renamed && S_ISREG(fp->mode))
 109 +                       look_for_rename(fp, fname);
 110                 if (delete_item(fname, fp->mode, flags) != DR_SUCCESS)
 111                         ret = DR_NOT_EMPTY;
 112         }
 113 @@ -126,6 +132,8 @@ static enum delret delete_dir_contents(char *fname, uint16 flags)
 114   *
 115   * Note that fbuf must point to a MAXPATHLEN buffer if the mode indicates it's
 116   * a directory! (The buffer is used for recursion, but returned unchanged.)
 117 + *
 118 + * Also note: --detect-rename may use this routine with DEL_NO_DELETIONS set!
 119   */
 120  enum delret delete_item(char *fbuf, uint16 mode, uint16 flags)
 121  {
 122 @@ -153,6 +161,9 @@ enum delret delete_item(char *fbuf, uint16 mode, uint16 flags)
 123                 /* OK: try to delete the directory. */
 124         }
 125
 126 +       if (flags & DEL_NO_DELETIONS)
 127 +               return DR_SUCCESS;
 128 +
 129         if (!(flags & DEL_MAKE_ROOM) && max_delete >= 0 && stats.deleted_files >= max_delete) {
 130                 skipped_deletes++;
 131                 return DR_AT_LIMIT;
 132 diff --git a/flist.c b/flist.c
 133 --- a/flist.c
 134 +++ b/flist.c
 135 @@ -63,6 +63,7 @@ extern int non_perishable_cnt;
 136  extern int prune_empty_dirs;
 137  extern int copy_links;
 138  extern int copy_unsafe_links;
 139 +extern int detect_renamed;
 140  extern int protocol_version;
 141  extern int sanitize_paths;
 142  extern int munge_symlinks;
 143 @@ -129,6 +130,8 @@ static int64 tmp_dev = -1, tmp_ino;
 144  #endif
 145  static char tmp_sum[MAX_DIGEST_LEN];
 146
 147 +struct file_list the_fattr_list;
 148 +
 149  static char empty_sum[MAX_DIGEST_LEN];
 150  static int flist_count_offset; /* for --delete --progress */
 151  static int show_filelist_progress;
 152 @@ -275,6 +278,45 @@ static inline int is_excluded(const char *fname, int is_dir, int filter_level)
 153         return name_is_excluded(fname, is_dir ? NAME_IS_DIR : NAME_IS_FILE, filter_level);
 154  }
 155
 156 +static int fattr_compare(struct file_struct **file1, struct file_struct **file2)
 157 +{
 158 +       struct file_struct *f1 = *file1;
 159 +       struct file_struct *f2 = *file2;
 160 +       int64 len1 = F_LENGTH(f1), len2 = F_LENGTH(f2);
 161 +       int diff;
 162 +
 163 +       if (!f1->basename || !S_ISREG(f1->mode) || !len1) {
 164 +               if (!f2->basename || !S_ISREG(f2->mode) || !len2)
 165 +                       return 0;
 166 +               return 1;
 167 +       }
 168 +       if (!f2->basename || !S_ISREG(f2->mode) || !len2)
 169 +               return -1;
 170 +
 171 +       /* Don't use diff for values that are longer than an int. */
 172 +       if (len1 != len2)
 173 +               return len1 < len2 ? -1 : 1;
 174 +
 175 +       if (always_checksum) {
 176 +               diff = u_memcmp(F_SUM(f1), F_SUM(f2), flist_csum_len);
 177 +               if (diff)
 178 +                       return diff;
 179 +       } else if (f1->modtime != f2->modtime)
 180 +               return f1->modtime < f2->modtime ? -1 : 1;
 181 +
 182 +       diff = u_strcmp(f1->basename, f2->basename);
 183 +       if (diff)
 184 +               return diff;
 185 +
 186 +       if (f1->dirname == f2->dirname)
 187 +               return 0;
 188 +       if (!f1->dirname)
 189 +               return -1;
 190 +       if (!f2->dirname)
 191 +               return 1;
 192 +       return u_strcmp(f1->dirname, f2->dirname);
 193 +}
 194 +
 195  static void send_directory(int f, struct file_list *flist,
 196                            char *fbuf, int len, int flags);
 197
 198 @@ -2621,6 +2663,25 @@ struct file_list *recv_file_list(int f, int dir_ndx)
 199          * for a non-relative transfer in recv_file_entry(). */
 200         flist_sort_and_clean(flist, relative_paths);
 201
 202 +       if (detect_renamed) {
 203 +               int j = flist->used;
 204 +               the_fattr_list.used = j;
 205 +               the_fattr_list.files = new_array(struct file_struct *, j);
 206 +               if (!the_fattr_list.files)
 207 +                       out_of_memory("recv_file_list");
 208 +               memcpy(the_fattr_list.files, flist->files,
 209 +                      j * sizeof (struct file_struct *));
 210 +               qsort(the_fattr_list.files, j,
 211 +                     sizeof the_fattr_list.files[0], (int (*)())fattr_compare);
 212 +               the_fattr_list.low = 0;
 213 +               while (j-- > 0) {
 214 +                       struct file_struct *fp = the_fattr_list.files[j];
 215 +                       if (fp->basename && S_ISREG(fp->mode) && F_LENGTH(fp))
 216 +                               break;
 217 +               }
 218 +               the_fattr_list.high = j;
 219 +       }
 220 +
 221         if (protocol_version < 30) {
 222                 /* Recv the io_error flag */
 223                 int err = read_int(f);
 224 diff --git a/generator.c b/generator.c
 225 --- a/generator.c
 226 +++ b/generator.c
 227 @@ -80,6 +80,7 @@ extern char *partial_dir;
 228  extern int compare_dest;
 229  extern int copy_dest;
 230  extern int link_dest;
 231 +extern int detect_renamed;
 232  extern int whole_file;
 233  extern int list_only;
 234  extern int read_batch;
 235 @@ -98,11 +99,13 @@ extern char *tmpdir;
 236  extern char *basis_dir[MAX_BASIS_DIRS+1];
 237  extern struct file_list *cur_flist, *first_flist, *dir_flist;
 238  extern filter_rule_list filter_list, daemon_filter_list;
 239 +extern struct file_list the_fattr_list;
 240
 241  int maybe_ATTRS_REPORT = 0;
 242  int maybe_ATTRS_SET_NANO = 0;
 243
 244  static dev_t dev_zero;
 245 +static int unexplored_dirs = 1;
 246  static int deldelay_size = 0, deldelay_cnt = 0;
 247  static char *deldelay_buf = NULL;
 248  static int deldelay_fd = -1;
 249 @@ -273,13 +276,18 @@ static void do_delayed_deletions(char *delbuf)
 250   * all the --delete-WHEN options.  Note that the fbuf pointer must point to a
 251   * MAXPATHLEN buffer with the name of the directory in it (the functions we
 252   * call will append names onto the end, but the old dir value will be restored
 253 - * on exit). */
 254 -static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
 255 + * on exit).
 256 + *
 257 + * Note:  --detect-rename may use this routine with DEL_NO_DELETIONS set!
 258 + */
 259 +static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev,
 260 +                         int del_flags)
 261  {
 262         static int already_warned = 0;
 263         struct file_list *dirlist;
 264 -       char delbuf[MAXPATHLEN];
 265 -       int dlen, i;
 266 +       char *p, delbuf[MAXPATHLEN];
 267 +       unsigned remainder;
 268 +       int dlen, i, restore_dot = 0;
 269
 270         if (!fbuf) {
 271                 change_local_filter_dir(NULL, 0, 0);
 272 @@ -293,17 +301,22 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
 273                 maybe_send_keepalive(time(NULL), MSK_ALLOW_FLUSH);
 274
 275         if (io_error & IOERR_GENERAL && !ignore_errors) {
 276 -               if (already_warned)
 277 +               if (!already_warned) {
 278 +                       rprintf(FINFO,
 279 +                           "IO error encountered -- skipping file deletion\n");
 280 +                       already_warned = 1;
 281 +               }
 282 +               if (!detect_renamed)
 283                         return;
 284 -               rprintf(FINFO,
 285 -                       "IO error encountered -- skipping file deletion\n");
 286 -               already_warned = 1;
 287 -               return;
 288 +               del_flags |= DEL_NO_DELETIONS;
 289         }
 290
 291         dlen = strlen(fbuf);
 292         change_local_filter_dir(fbuf, dlen, F_DEPTH(file));
 293
 294 +       if (detect_renamed)
 295 +               unexplored_dirs--;
 296 +
 297         if (one_file_system) {
 298                 if (file->flags & FLAG_TOP_DIR)
 299                         filesystem_dev = *fs_dev;
 300 @@ -313,6 +326,14 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
 301
 302         dirlist = get_dirlist(fbuf, dlen, 0);
 303
 304 +       p = fbuf + dlen;
 305 +       if (dlen == 1 && *fbuf == '.') {
 306 +               restore_dot = 1;
 307 +               p = fbuf;
 308 +       } else if (dlen != 1 || *fbuf != '/')
 309 +               *p++ = '/';
 310 +       remainder = MAXPATHLEN - (p - fbuf);
 311 +
 312         /* If an item in dirlist is not found in flist, delete it
 313          * from the filesystem. */
 314         for (i = dirlist->used; i--; ) {
 315 @@ -325,6 +346,10 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
 316                                         f_name(fp, NULL));
 317                         continue;
 318                 }
 319 +               if (detect_renamed && S_ISREG(fp->mode)) {
 320 +                       strlcpy(p, fp->basename, remainder);
 321 +                       look_for_rename(fp, fbuf);
 322 +               }
 323                 /* Here we want to match regardless of file type.  Replacement
 324                  * of a file with one of another type is handled separately by
 325                  * a delete_item call with a DEL_MAKE_ROOM flag. */
 326 @@ -333,14 +358,19 @@ static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev)
 327                         if (!(fp->mode & S_IWUSR) && !am_root && fp->flags & FLAG_OWNED_BY_US)
 328                                 flags |= DEL_NO_UID_WRITE;
 329                         f_name(fp, delbuf);
 330 -                       if (delete_during == 2) {
 331 -                               if (!remember_delete(fp, delbuf, flags))
 332 +                       if (delete_during == 2 && !(del_flags & DEL_NO_DELETIONS)) {
 333 +                               if (!remember_delete(fp, delbuf, del_flags | flags))
 334                                         break;
 335                         } else
 336 -                               delete_item(delbuf, fp->mode, flags);
 337 -               }
 338 +                               delete_item(delbuf, fp->mode, del_flags | flags);
 339 +               } else if (detect_renamed && S_ISDIR(fp->mode))
 340 +                       unexplored_dirs++;
 341         }
 342
 343 +       if (restore_dot)
 344 +               fbuf[0] = '.';
 345 +       fbuf[dlen] = '\0';
 346 +
 347         flist_free(dirlist);
 348  }
 349
 350 @@ -376,14 +406,125 @@ static void do_delete_pass(void)
 351                  || !S_ISDIR(st.st_mode))
 352                         continue;
 353
 354 -               delete_in_dir(fbuf, file, &st.st_dev);
 355 +               delete_in_dir(fbuf, file, &st.st_dev, 0);
 356         }
 357 -       delete_in_dir(NULL, NULL, &dev_zero);
 358 +       delete_in_dir(NULL, NULL, &dev_zero, 0);
 359
 360         if (INFO_GTE(FLIST, 2) && !am_server)
 361                 rprintf(FINFO, "                    \r");
 362  }
 363
 364 +/* Search for a regular file that matches either (1) the size & modified
 365 + * time (plus the basename, if possible) or (2) the size & checksum.  If
 366 + * we find an exact match down to the dirname, return -1 because we found
 367 + * an up-to-date file in the transfer, not a renamed file. */
 368 +static int fattr_find(struct file_struct *f, char *fname)
 369 +{
 370 +       int low = the_fattr_list.low, high = the_fattr_list.high;
 371 +       int mid, ok_match = -1, good_match = -1;
 372 +       struct file_struct *fmid;
 373 +       int diff;
 374 +
 375 +       while (low <= high) {
 376 +               mid = (low + high) / 2;
 377 +               fmid = the_fattr_list.files[mid];
 378 +               if (F_LENGTH(fmid) != F_LENGTH(f)) {
 379 +                       if (F_LENGTH(fmid) < F_LENGTH(f))
 380 +                               low = mid + 1;
 381 +                       else
 382 +                               high = mid - 1;
 383 +                       continue;
 384 +               }
 385 +               if (always_checksum) {
 386 +                       /* We use the FLAG_FILE_SENT flag to indicate when we
 387 +                        * have computed the checksum for an entry. */
 388 +                       if (!(f->flags & FLAG_FILE_SENT)) {
 389 +                               STRUCT_STAT st;
 390 +                               if (fmid->modtime == f->modtime
 391 +                                && f_name_cmp(fmid, f) == 0)
 392 +                                       return -1; /* assume we can't help */
 393 +                               st.st_size = F_LENGTH(f);
 394 +                               st.st_mtime = f->modtime;
 395 +                               file_checksum(fname, &st, F_SUM(f));
 396 +                               f->flags |= FLAG_FILE_SENT;
 397 +                       }
 398 +                       diff = u_memcmp(F_SUM(fmid), F_SUM(f), flist_csum_len);
 399 +                       if (diff) {
 400 +                               if (diff < 0)
 401 +                                       low = mid + 1;
 402 +                               else
 403 +                                       high = mid - 1;
 404 +                               continue;
 405 +                       }
 406 +               } else {
 407 +                       if (fmid->modtime != f->modtime) {
 408 +                               if (fmid->modtime < f->modtime)
 409 +                                       low = mid + 1;
 410 +                               else
 411 +                                       high = mid - 1;
 412 +                               continue;
 413 +                       }
 414 +               }
 415 +               ok_match = mid;
 416 +               diff = u_strcmp(fmid->basename, f->basename);
 417 +               if (diff == 0) {
 418 +                       good_match = mid;
 419 +                       if (fmid->dirname == f->dirname)
 420 +                               return -1; /* file is up-to-date */
 421 +                       if (!fmid->dirname) {
 422 +                               low = mid + 1;
 423 +                               continue;
 424 +                       }
 425 +                       if (!f->dirname) {
 426 +                               high = mid - 1;
 427 +                               continue;
 428 +                       }
 429 +                       diff = u_strcmp(fmid->dirname, f->dirname);
 430 +                       if (diff == 0)
 431 +                               return -1; /* file is up-to-date */
 432 +               }
 433 +               if (diff < 0)
 434 +                       low = mid + 1;
 435 +               else
 436 +                       high = mid - 1;
 437 +       }
 438 +
 439 +       return good_match >= 0 ? good_match : ok_match;
 440 +}
 441 +
 442 +void look_for_rename(struct file_struct *file, char *fname)
 443 +{
 444 +       struct file_struct *fp;
 445 +       char *partialptr, *fn;
 446 +       STRUCT_STAT st;
 447 +       int ndx;
 448 +
 449 +       if (!partial_dir || (ndx = fattr_find(file, fname)) < 0)
 450 +               return;
 451 +
 452 +       fp = the_fattr_list.files[ndx];
 453 +       fn = f_name(fp, NULL);
 454 +       /* We don't provide an alternate-basis file if there is a basis file. */
 455 +       if (link_stat(fn, &st, 0) == 0)
 456 +               return;
 457 +
 458 +       if (!dry_run) {
 459 +               if ((partialptr = partial_dir_fname(fn)) == NULL
 460 +                || !handle_partial_dir(partialptr, PDIR_CREATE))
 461 +                       return;
 462 +               /* We only use the file if we can hard-link it into our tmp dir. */
 463 +               if (link(fname, partialptr) != 0) {
 464 +                       if (errno != EEXIST)
 465 +                               handle_partial_dir(partialptr, PDIR_DELETE);
 466 +                       return;
 467 +               }
 468 +       }
 469 +
 470 +       /* I think this falls into the -vv category with "%s is uptodate", etc. */
 471 +       if (INFO_GTE(MISC, 2))
 472 +               rprintf(FINFO, "found renamed: %s => %s\n", fname, fn);
 473 +}
 474 +
 475  static inline int time_diff(STRUCT_STAT *stp, struct file_struct *file)
 476  {
 477  #ifdef ST_MTIME_NSEC
 478 @@ -1162,6 +1303,7 @@ static void list_file_entry(struct file_struct *f)
 479         }
 480  }
 481
 482 +static struct bitbag *delayed_bits = NULL;
 483  static int phase = 0;
 484  static int dflt_perms;
 485
 486 @@ -1292,7 +1434,7 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 487                          && do_stat(dn, &sx.st) < 0) {
 488                                 if (dry_run)
 489                                         goto parent_is_dry_missing;
 490 -                               if (make_path(fname, MKP_DROP_NAME | MKP_SKIP_SLASH) < 0) {
 491 +                               if (make_path(fname, ACCESSPERMS, MKP_DROP_NAME | MKP_SKIP_SLASH) < 0) {
 492                                         rsyserr(FERROR_XFER, errno,
 493                                                 "recv_generator: mkdir %s failed",
 494                                                 full_fname(dn));
 495 @@ -1445,7 +1587,7 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 496                 }
 497                 if (real_ret != 0 && do_mkdir(fname,file->mode|added_perms) < 0 && errno != EEXIST) {
 498                         if (!relative_paths || errno != ENOENT
 499 -                        || make_path(fname, MKP_DROP_NAME | MKP_SKIP_SLASH) < 0
 500 +                        || make_path(fname, ACCESSPERMS, MKP_DROP_NAME | MKP_SKIP_SLASH) < 0
 501                          || (do_mkdir(fname, file->mode|added_perms) < 0 && errno != EEXIST)) {
 502                                 rsyserr(FERROR_XFER, errno,
 503                                         "recv_generator: mkdir %s failed",
 504 @@ -1494,9 +1636,12 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 505                 }
 506                 else if (delete_during && f_out != -1 && !phase
 507                     && !(file->flags & FLAG_MISSING_DIR)) {
 508 -                       if (file->flags & FLAG_CONTENT_DIR)
 509 -                               delete_in_dir(fname, file, &real_sx.st.st_dev);
 510 -                       else
 511 +                       if (file->flags & FLAG_CONTENT_DIR) {
 512 +                               if (detect_renamed && real_ret != 0)
 513 +                                       unexplored_dirs++;
 514 +                               delete_in_dir(fname, file, &real_sx.st.st_dev,
 515 +                                             delete_during < 0 ? DEL_NO_DELETIONS : 0);
 516 +                       } else
 517                                 change_local_filter_dir(fname, strlen(fname), F_DEPTH(file));
 518                 }
 519                 prior_dir_file = file;
 520 @@ -1761,8 +1906,14 @@ static void recv_generator(char *fname, struct file_struct *file, int ndx,
 521                         goto cleanup;
 522                 }
 523  #endif
 524 -               if (stat_errno == ENOENT)
 525 +               if (stat_errno == ENOENT) {
 526 +                       if (detect_renamed && unexplored_dirs > 0
 527 +                        && F_LENGTH(file)) {
 528 +                               bitbag_set_bit(delayed_bits, ndx);
 529 +                               return;
 530 +                       }
 531                         goto notify_others;
 532 +               }
 533                 rsyserr(FERROR_XFER, stat_errno, "recv_generator: failed to stat %s",
 534                         full_fname(fname));
 535                 goto cleanup;
 536 @@ -2226,6 +2377,12 @@ void generate_files(int f_out, const char *local_name)
 537         if (DEBUG_GTE(GENR, 1))
 538                 rprintf(FINFO, "generator starting pid=%d\n", (int)getpid());
 539
 540 +       if (detect_renamed) {
 541 +               delayed_bits = bitbag_create(cur_flist->used);
 542 +               if (!delete_before && !delete_during)
 543 +                       delete_during = -1;
 544 +       }
 545 +
 546         if (delete_before && !solo_file && cur_flist->used > 0)
 547                 do_delete_pass();
 548         if (delete_during == 2) {
 549 @@ -2236,7 +2393,7 @@ void generate_files(int f_out, const char *local_name)
 550         }
 551         info_levels[INFO_FLIST] = info_levels[INFO_PROGRESS] = 0;
 552
 553 -       if (append_mode > 0 || whole_file < 0)
 554 +       if (append_mode > 0 || detect_renamed || whole_file < 0)
 555                 whole_file = 0;
 556         if (DEBUG_GTE(FLIST, 1)) {
 557                 rprintf(FINFO, "delta-transmission %s\n",
 558 @@ -2272,7 +2429,7 @@ void generate_files(int f_out, const char *local_name)
 559                                                 dirdev = MAKEDEV(DEV_MAJOR(devp), DEV_MINOR(devp));
 560                                         } else
 561                                                 dirdev = MAKEDEV(0, 0);
 562 -                                       delete_in_dir(fbuf, fp, &dirdev);
 563 +                                       delete_in_dir(fbuf, fp, &dirdev, 0);
 564                                 } else
 565                                         change_local_filter_dir(fbuf, strlen(fbuf), F_DEPTH(fp));
 566                         }
 567 @@ -2319,7 +2476,21 @@ void generate_files(int f_out, const char *local_name)
 568         } while ((cur_flist = cur_flist->next) != NULL);
 569
 570         if (delete_during)
 571 -               delete_in_dir(NULL, NULL, &dev_zero);
 572 +               delete_in_dir(NULL, NULL, &dev_zero, 0);
 573 +       if (detect_renamed) {
 574 +               if (delete_during < 0)
 575 +                       delete_during = 0;
 576 +               detect_renamed = 0;
 577 +
 578 +               for (i = -1; (i = bitbag_next_bit(delayed_bits, i)) >= 0; ) {
 579 +                       struct file_struct *file = cur_flist->files[i];
 580 +                       if (local_name)
 581 +                               strlcpy(fbuf, local_name, sizeof fbuf);
 582 +                       else
 583 +                               f_name(file, fbuf);
 584 +                       recv_generator(fbuf, file, i, itemizing, code, f_out);
 585 +               }
 586 +       }
 587         phase++;
 588         if (DEBUG_GTE(GENR, 1))
 589                 rprintf(FINFO, "generate_files phase=%d\n", phase);
 590 diff --git a/options.c b/options.c
 591 --- a/options.c
 592 +++ b/options.c
 593 @@ -86,6 +86,7 @@ int am_server = 0;
 594  int am_sender = 0;
 595  int am_starting_up = 1;
 596  int relative_paths = -1;
 597 +int detect_renamed = 0;
 598  int implied_dirs = 1;
 599  int missing_args = 0; /* 0 = FERROR_XFER, 1 = ignore, 2 = delete */
 600  int numeric_ids = 0;
 601 @@ -845,6 +846,7 @@ void usage(enum logcode F)
 602    rprintf(F," -@, --modify-window=NUM     set the accuracy for mod-time comparisons\n");
 603    rprintf(F," -T, --temp-dir=DIR          create temporary files in directory DIR\n");
 604    rprintf(F," -y, --fuzzy                 find similar file for basis if no dest file\n");
 605 +  rprintf(F,"     --detect-renamed        try to find renamed files to speed up the transfer\n");
 606    rprintf(F,"     --compare-dest=DIR      also compare destination files relative to DIR\n");
 607    rprintf(F,"     --copy-dest=DIR         ... and include copies of unchanged files\n");
 608    rprintf(F,"     --link-dest=DIR         hardlink to files in DIR when unchanged\n");
 609 @@ -1059,6 +1061,7 @@ static struct poptOption long_options[] = {
 610    {"compare-dest",     0,  POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
 611    {"copy-dest",        0,  POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
 612    {"link-dest",        0,  POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 },
 613 +  {"detect-renamed",   0,  POPT_ARG_NONE,   &detect_renamed, 0, 0, 0 },
 614    {"fuzzy",           'y', POPT_ARG_NONE,   0, 'y', 0, 0 },
 615    {"no-fuzzy",         0,  POPT_ARG_VAL,    &fuzzy_basis, 0, 0, 0 },
 616    {"no-y",             0,  POPT_ARG_VAL,    &fuzzy_basis, 0, 0, 0 },
 617 @@ -2458,7 +2461,7 @@ int parse_arguments(int *argc_p, const char ***argv_p)
 618                 inplace = 1;
 619         }
 620
 621 -       if (delay_updates && !partial_dir)
 622 +       if ((delay_updates || detect_renamed) && !partial_dir)
 623                 partial_dir = tmp_partialdir;
 624
 625         if (inplace) {
 626 @@ -2467,6 +2470,7 @@ int parse_arguments(int *argc_p, const char ***argv_p)
 627                         snprintf(err_buf, sizeof err_buf,
 628                                  "--%s cannot be used with --%s\n",
 629                                  append_mode ? "append" : "inplace",
 630 +                                detect_renamed ? "detect-renamed" :
 631                                  delay_updates ? "delay-updates" : "partial-dir");
 632                         return 0;
 633                 }
 634 @@ -2868,6 +2872,8 @@ void server_options(char **args, int *argc_p)
 635                         args[ac++] = "--super";
 636                 if (size_only)
 637                         args[ac++] = "--size-only";
 638 +               if (detect_renamed)
 639 +                       args[ac++] = "--detect-renamed";
 640                 if (do_stats)
 641                         args[ac++] = "--stats";
 642         } else {
 643 diff --git a/receiver.c b/receiver.c
 644 --- a/receiver.c
 645 +++ b/receiver.c
 646 @@ -217,7 +217,7 @@ int open_tmpfile(char *fnametmp, const char *fname, struct file_struct *file)
 647          * information should have been previously transferred, but that may
 648          * not be the case with -R */
 649         if (fd == -1 && relative_paths && errno == ENOENT
 650 -        && make_path(fnametmp, MKP_SKIP_SLASH | MKP_DROP_NAME) == 0) {
 651 +        && make_path(fnametmp, ACCESSPERMS, MKP_SKIP_SLASH | MKP_DROP_NAME) == 0) {
 652                 /* Get back to name with XXXXXX in it. */
 653                 get_tmpname(fnametmp, fname, False);
 654                 fd = do_mkstemp(fnametmp, (file->mode|added_perms) & INITACCESSPERMS);
 655 diff --git a/rsync.h b/rsync.h
 656 --- a/rsync.h
 657 +++ b/rsync.h
 658 @@ -264,7 +264,7 @@ enum msgcode {
 659  #define NDX_DEL_STATS -3
 660  #define NDX_FLIST_OFFSET -101
 661
 662 -/* For calling delete_item() and delete_dir_contents(). */
 663 +/* For calling delete_item(), delete_dir_contents(), and delete_in_dir(). */
 664  #define DEL_NO_UID_WRITE       (1<<0) /* file/dir has our uid w/o write perm */
 665  #define DEL_RECURSE            (1<<1) /* if dir, delete all contents */
 666  #define DEL_DIR_IS_EMPTY       (1<<2) /* internal delete_FUNCTIONS use only */
 667 @@ -274,6 +274,7 @@ enum msgcode {
 668  #define DEL_FOR_DEVICE         (1<<6) /* making room for a replacement device */
 669  #define DEL_FOR_SPECIAL        (1<<7) /* making room for a replacement special */
 670  #define DEL_FOR_BACKUP         (1<<8) /* the delete is for a backup operation */
 671 +#define DEL_NO_DELETIONS       (1<<9) /* just check for renames w/o deleting */
 672
 673  #define DEL_MAKE_ROOM (DEL_FOR_FILE|DEL_FOR_DIR|DEL_FOR_SYMLINK|DEL_FOR_DEVICE|DEL_FOR_SPECIAL)
 674
 675 diff --git a/rsync.yo b/rsync.yo
 676 --- a/rsync.yo
 677 +++ b/rsync.yo
 678 @@ -426,6 +426,7 @@ to the detailed description below for a complete description.  verb(
 679   -@, --modify-window=NUM     set the accuracy for mod-time comparisons
 680   -T, --temp-dir=DIR          create temporary files in directory DIR
 681   -y, --fuzzy                 find similar file for basis if no dest file
 682 +     --detect-renamed        try to find renamed files to speed the xfer
 683       --compare-dest=DIR      also compare received files relative to DIR
 684       --copy-dest=DIR         ... and include copies of unchanged files
 685       --link-dest=DIR         hardlink to files in DIR when unchanged
 686 @@ -1975,6 +1976,21 @@ Note that the use of the bf(--delete) option might get rid of any potential
 687  fuzzy-match files, so either use bf(--delete-after) or specify some
 688  filename exclusions if you need to prevent this.
 689
 690 +dit(bf(--detect-renamed)) With this option, for each new source file
 691 +(call it em(src/S)), rsync looks for a file em(dest/D) anywhere in the
 692 +destination that passes the quick check with em(src/S).  If such a em(dest/D)
 693 +is found, rsync uses it as an alternate basis for transferring em(S).  The
 694 +idea is that if em(src/S) was renamed from em(src/D) (as opposed to em(src/S)
 695 +passing the quick check with em(dest/D) by coincidence), the delta-transfer
 696 +algorithm will find that all the data matches between em(src/S) and em(dest/D),
 697 +and the transfer will be really fast.
 698 +
 699 +By default, alternate-basis files are hard-linked into a directory named
 700 +".~tmp~" in each file's destination directory, but if you've specified
 701 +the bf(--partial-dir) option, that directory will be used instead.  These
 702 +potential alternate-basis files will be removed as the transfer progresses.
 703 +This option conflicts with bf(--inplace) and bf(--append).
 704 +
 705  dit(bf(--compare-dest=DIR)) This option instructs rsync to use em(DIR) on
 706  the destination machine as an additional hierarchy to compare destination
 707  files against doing transfers (if the files are missing in the destination
 708 diff --git a/util.c b/util.c
 709 --- a/util.c
 710 +++ b/util.c
 711 @@ -182,7 +182,7 @@ int set_times(const char *fname, STRUCT_STAT *stp)
 712  /* Create any necessary directories in fname.  Any missing directories are
 713   * created with default permissions.  Returns < 0 on error, or the number
 714   * of directories created. */
 715 -int make_path(char *fname, int flags)
 716 +int make_path(char *fname, mode_t mode, int flags)
 717  {
 718         char *end, *p;
 719         int ret = 0;
 720 @@ -213,7 +213,7 @@ int make_path(char *fname, int flags)
 721                                 else
 722                                         errno = ENOTDIR;
 723                         }
 724 -               } else if (do_mkdir(fname, ACCESSPERMS) == 0) {
 725 +               } else if (do_mkdir(fname, mode) == 0) {
 726                         ret++;
 727                         break;
 728                 }
 729 @@ -252,7 +252,7 @@ int make_path(char *fname, int flags)
 730                 p += strlen(p);
 731                 if (ret < 0) /* Skip mkdir on error, but keep restoring the path. */
 732                         continue;
 733 -               if (do_mkdir(fname, ACCESSPERMS) < 0)
 734 +               if (do_mkdir(fname, mode) < 0)
 735                         ret = -ret - 1;
 736                 else
 737                         ret++;
 738 @@ -1172,6 +1172,32 @@ char *normalize_path(char *path, BOOL force_newbuf, unsigned int *len_ptr)
 739         return path;
 740  }
 741
 742 +/* We need to supply our own strcmp function for file list comparisons
 743 + * to ensure that signed/unsigned usage is consistent between machines. */
 744 +int u_strcmp(const char *p1, const char *p2)
 745 +{
 746 +        for ( ; *p1; p1++, p2++) {
 747 +               if (*p1 != *p2)
 748 +                       break;
 749 +       }
 750 +
 751 +       return (int)*(uchar*)p1 - (int)*(uchar*)p2;
 752 +}
 753 +
 754 +/* We need a memcmp function compares unsigned-byte values. */
 755 +int u_memcmp(const void *p1, const void *p2, size_t len)
 756 +{
 757 +       const uchar *u1 = p1;
 758 +       const uchar *u2 = p2;
 759 +
 760 +       while (len--) {
 761 +               if (*u1 != *u2)
 762 +                       return (int)*u1 - (int)*u2;
 763 +       }
 764 +
 765 +       return 0;
 766 +}
 767 +
 768  /**
 769   * Return a quoted string with the full pathname of the indicated filename.
 770   * The string " (in MODNAME)" may also be appended.  The returned pointer
 771 @@ -1265,7 +1291,7 @@ int handle_partial_dir(const char *fname, int create)
 772                         }
 773                         statret = -1;
 774                 }
 775 -               if (statret < 0 && do_mkdir(dir, 0700) < 0) {
 776 +               if (statret < 0 && make_path(dir, 0700, 0) < 0) {
 777                         *fn = '/';
 778                         return 0;
 779                 }