The Samba-Bugzilla – Attachment 7435 Details for
Bug 8847
detect-renamed.diff update to ensure existence of directory for partial-dir
Home
|
New
|
Browse
|
Search
|
[?]
|
Reports
|
Requests
|
Help
|
New Account
|
Log In
[x]
|
Forgot Password
Login:
[x]
[patch]
replacement for patches/detect-renamed.diff
detect-renamed.diff (text/plain), 26.47 KB, created by
J.R.
on 2012-04-05 20:52:27 UTC
(
hide
)
Description:
replacement for patches/detect-renamed.diff
Filename:
MIME Type:
Creator:
J.R.
Created:
2012-04-05 20:52:27 UTC
Size:
26.47 KB
patch
obsolete
>This patch adds the --detect-renamed option which makes rsync notice files >that either (1) match in size & modify-time (plus the basename, if possible) >or (2) match in size & checksum (when --checksum was also specified) and use >each match as an alternate basis file to speed up the transfer. > >The algorithm attempts to scan the receiving-side's files in an efficient >manner. If --delete[-before] is enabled, we'll take advantage of the >pre-transfer delete pass to prepare any alternate-basis-file matches we >might find. If --delete-before is not enabled, rsync does the rename scan >during the regular file-sending scan (scanning each directory right before >the generator starts updating files from that dir). In this latter mode, >rsync might delay the updating of a file (if no alternate-basis match was >yet found) until the full scan of the receiving side is complete, at which >point any delayed files are processed. > >I chose to hard-link the alternate-basis files into a ".~tmp~" subdir that >takes advantage of rsync's pre-existing partial-dir logic. This uses less >memory than trying to keep track of the matches internally, and also allows >any deletions or file-updates to occur normally without interfering with >these alternate-basis discoveries. > >To use this patch, run these commands for a successful build: > > patch -p1 <patches/detect-renamed.diff > ./configure (optional if already run) > make > >TODO: > > We need to never return a match from fattr_find() that has a basis > file. This will ensure that we don't try to give a renamed file to > a file that can't use it, while missing out on giving it to a file > that could use it. > >based-on: 40afd365cc8ca968fd16e161d24df5b8a8a520cc >--- compat.c.orig 2012-04-03 10:38:42.000000000 -0400 >+++ compat.c 2012-04-03 10:38:42.000000000 -0400 >@@ -41,6 +41,7 @@ extern int read_batch; > extern int delay_updates; > extern int checksum_seed; > extern int basis_dir_cnt; >+extern int detect_renamed; > extern int prune_empty_dirs; > extern int protocol_version; > extern int protect_args; >@@ -124,6 +125,7 @@ void set_allow_inc_recurse(void) > allow_inc_recurse = 0; > else if (!am_sender > && (delete_before || delete_after >+ || detect_renamed > || delay_updates || prune_empty_dirs)) > allow_inc_recurse = 0; > else if (am_server && !local_server >--- flist.c.orig 2012-04-03 10:38:42.000000000 -0400 >+++ flist.c 2012-04-03 10:38:42.000000000 -0400 >@@ -61,6 +61,7 @@ extern int non_perishable_cnt; > extern int prune_empty_dirs; > extern int copy_links; > extern int copy_unsafe_links; >+extern int detect_renamed; > extern int protocol_version; > extern int sanitize_paths; > extern int munge_symlinks; >@@ -121,6 +122,8 @@ static int64 tmp_dev = -1, tmp_ino; > #endif > static char tmp_sum[MAX_DIGEST_LEN]; > >+struct file_list the_fattr_list; >+ > static char empty_sum[MAX_DIGEST_LEN]; > static int flist_count_offset; /* for --delete --progress */ > static int dir_count = 0; >@@ -288,6 +291,45 @@ static int is_excluded(const char *fname > return 0; > } > >+static int fattr_compare(struct file_struct **file1, struct file_struct **file2) >+{ >+ struct file_struct *f1 = *file1; >+ struct file_struct *f2 = *file2; >+ int64 len1 = F_LENGTH(f1), len2 = F_LENGTH(f2); >+ int diff; >+ >+ if (!f1->basename || !S_ISREG(f1->mode) || !len1) { >+ if (!f2->basename || !S_ISREG(f2->mode) || !len2) >+ return 0; >+ return 1; >+ } >+ if (!f2->basename || !S_ISREG(f2->mode) || !len2) >+ return -1; >+ >+ /* Don't use diff for values that are longer than an int. */ >+ if (len1 != len2) >+ return len1 < len2 ? -1 : 1; >+ >+ if (always_checksum) { >+ diff = u_memcmp(F_SUM(f1), F_SUM(f2), checksum_len); >+ if (diff) >+ return diff; >+ } else if (f1->modtime != f2->modtime) >+ return f1->modtime < f2->modtime ? -1 : 1; >+ >+ diff = u_strcmp(f1->basename, f2->basename); >+ if (diff) >+ return diff; >+ >+ if (f1->dirname == f2->dirname) >+ return 0; >+ if (!f1->dirname) >+ return -1; >+ if (!f2->dirname) >+ return 1; >+ return u_strcmp(f1->dirname, f2->dirname); >+} >+ > static void send_directory(int f, struct file_list *flist, > char *fbuf, int len, int flags); > >@@ -2451,6 +2493,25 @@ struct file_list *recv_file_list(int f) > > flist_sort_and_clean(flist, relative_paths); > >+ if (detect_renamed) { >+ int j = flist->used; >+ the_fattr_list.used = j; >+ the_fattr_list.files = new_array(struct file_struct *, j); >+ if (!the_fattr_list.files) >+ out_of_memory("recv_file_list"); >+ memcpy(the_fattr_list.files, flist->files, >+ j * sizeof (struct file_struct *)); >+ qsort(the_fattr_list.files, j, >+ sizeof the_fattr_list.files[0], (int (*)())fattr_compare); >+ the_fattr_list.low = 0; >+ while (j-- > 0) { >+ struct file_struct *fp = the_fattr_list.files[j]; >+ if (fp->basename && S_ISREG(fp->mode) && F_LENGTH(fp)) >+ break; >+ } >+ the_fattr_list.high = j; >+ } >+ > if (protocol_version < 30) { > /* Recv the io_error flag */ > if (ignore_errors) >--- generator.c.orig 2012-04-03 10:38:42.000000000 -0400 >+++ generator.c 2012-04-03 23:16:14.000000000 -0400 >@@ -79,6 +79,7 @@ extern char *basis_dir[MAX_BASIS_DIRS+1] > extern int compare_dest; > extern int copy_dest; > extern int link_dest; >+extern int detect_renamed; > extern int whole_file; > extern int list_only; > extern int read_batch; >@@ -97,6 +98,7 @@ extern char *backup_suffix; > extern int backup_suffix_len; > extern struct file_list *cur_flist, *first_flist, *dir_flist; > extern struct filter_list_struct daemon_filter_list; >+extern struct file_list the_fattr_list; > > int ignore_perishable = 0; > int non_perishable_cnt = 0; >@@ -104,6 +106,7 @@ int maybe_ATTRS_REPORT = 0; > > static dev_t dev_zero; > static int deletion_count = 0; /* used to implement --max-delete */ >+static int unexplored_dirs = 1; > static int deldelay_size = 0, deldelay_cnt = 0; > static char *deldelay_buf = NULL; > static int deldelay_fd = -1; >@@ -114,7 +117,7 @@ static int need_retouch_dir_times; > static int need_retouch_dir_perms; > static const char *solo_file = NULL; > >-/* For calling delete_item() and delete_dir_contents(). */ >+/* For calling delete_item(), delete_dir_contents(), and delete_in_dir(). */ > #define DEL_NO_UID_WRITE (1<<0) /* file/dir has our uid w/o write perm */ > #define DEL_RECURSE (1<<1) /* if dir, delete all contents */ > #define DEL_DIR_IS_EMPTY (1<<2) /* internal delete_FUNCTIONS use only */ >@@ -123,6 +126,7 @@ static const char *solo_file = NULL; > #define DEL_FOR_SYMLINK (1<<5) /* making room for a replacement symlink */ > #define DEL_FOR_DEVICE (1<<6) /* making room for a replacement device */ > #define DEL_FOR_SPECIAL (1<<7) /* making room for a replacement special */ >+#define DEL_NO_DELETIONS (1<<9) /* just check for renames w/o deleting */ > > #define DEL_MAKE_ROOM (DEL_FOR_FILE|DEL_FOR_DIR|DEL_FOR_SYMLINK|DEL_FOR_DEVICE|DEL_FOR_SPECIAL) > >@@ -147,11 +151,121 @@ static int is_backup_file(char *fn) > return k > 0 && strcmp(fn+k, backup_suffix) == 0; > } > >+/* Search for a regular file that matches either (1) the size & modified >+ * time (plus the basename, if possible) or (2) the size & checksum. If >+ * we find an exact match down to the dirname, return -1 because we found >+ * an up-to-date file in the transfer, not a renamed file. */ >+static int fattr_find(struct file_struct *f, char *fname) >+{ >+ int low = the_fattr_list.low, high = the_fattr_list.high; >+ int mid, ok_match = -1, good_match = -1; >+ struct file_struct *fmid; >+ int diff; >+ >+ while (low <= high) { >+ mid = (low + high) / 2; >+ fmid = the_fattr_list.files[mid]; >+ if (F_LENGTH(fmid) != F_LENGTH(f)) { >+ if (F_LENGTH(fmid) < F_LENGTH(f)) >+ low = mid + 1; >+ else >+ high = mid - 1; >+ continue; >+ } >+ if (always_checksum) { >+ /* We use the FLAG_FILE_SENT flag to indicate when we >+ * have computed the checksum for an entry. */ >+ if (!(f->flags & FLAG_FILE_SENT)) { >+ if (fmid->modtime == f->modtime >+ && f_name_cmp(fmid, f) == 0) >+ return -1; /* assume we can't help */ >+ file_checksum(fname, F_SUM(f), F_LENGTH(f)); >+ f->flags |= FLAG_FILE_SENT; >+ } >+ diff = u_memcmp(F_SUM(fmid), F_SUM(f), checksum_len); >+ if (diff) { >+ if (diff < 0) >+ low = mid + 1; >+ else >+ high = mid - 1; >+ continue; >+ } >+ } else { >+ if (fmid->modtime != f->modtime) { >+ if (fmid->modtime < f->modtime) >+ low = mid + 1; >+ else >+ high = mid - 1; >+ continue; >+ } >+ } >+ ok_match = mid; >+ diff = u_strcmp(fmid->basename, f->basename); >+ if (diff == 0) { >+ good_match = mid; >+ if (fmid->dirname == f->dirname) >+ return -1; /* file is up-to-date */ >+ if (!fmid->dirname) { >+ low = mid + 1; >+ continue; >+ } >+ if (!f->dirname) { >+ high = mid - 1; >+ continue; >+ } >+ diff = u_strcmp(fmid->dirname, f->dirname); >+ if (diff == 0) >+ return -1; /* file is up-to-date */ >+ } >+ if (diff < 0) >+ low = mid + 1; >+ else >+ high = mid - 1; >+ } >+ >+ return good_match >= 0 ? good_match : ok_match; >+} >+ >+static void look_for_rename(struct file_struct *file, char *fname) >+{ >+ struct file_struct *fp; >+ char *partialptr, *fn; >+ STRUCT_STAT st; >+ int ndx; >+ >+ if (!partial_dir || (ndx = fattr_find(file, fname)) < 0) >+ return; >+ >+ fp = the_fattr_list.files[ndx]; >+ fn = f_name(fp, NULL); >+ /* We don't provide an alternate-basis file if there is a basis file. */ >+ if (link_stat(fn, &st, 0) == 0) >+ return; >+ >+ if (!dry_run) { >+ if ((partialptr = partial_dir_fname(fn)) == NULL >+ || !handle_partial_dir(partialptr, PDIR_CREATE)) >+ return; >+ /* We only use the file if we can hard-link it into our tmp dir. */ >+ if (link(fname, partialptr) != 0) { >+ if (errno != EEXIST) >+ handle_partial_dir(partialptr, PDIR_DELETE); >+ return; >+ } >+ } >+ >+ /* I think this falls into the -vv category with "%s is uptodate", etc. */ >+ if (verbose > 1) >+ rprintf(FINFO, "found renamed: %s => %s\n", fname, fn); >+} >+ > /* Delete a file or directory. If DEL_RECURSE is set in the flags, this will > * delete recursively. > * > * Note that fbuf must point to a MAXPATHLEN buffer if the mode indicates it's > * a directory! (The buffer is used for recursion, but returned unchanged.) >+ * >+ * Also note: --detect-rename may use this routine with DEL_NO_DELETIONS set! > */ > static enum delret delete_item(char *fbuf, uint16 mode, uint16 flags) > { >@@ -178,6 +292,8 @@ static enum delret delete_item(char *fbu > goto check_ret; > /* OK: try to delete the directory. */ > } >+ if (flags & DEL_NO_DELETIONS) >+ return DR_SUCCESS; > > if (!(flags & DEL_MAKE_ROOM) && max_delete >= 0 && ++deletion_count > max_delete) > return DR_AT_LIMIT; >@@ -233,6 +349,8 @@ static enum delret delete_item(char *fbu > * its contents, otherwise just checks for content. Returns DR_SUCCESS or > * DR_NOT_EMPTY. Note that fname must point to a MAXPATHLEN buffer! (The > * buffer is used for recursion, but returned unchanged.) >+ * >+ * Note: --detect-rename may use this routine with DEL_NO_DELETIONS set! > */ > static enum delret delete_dir_contents(char *fname, uint16 flags) > { >@@ -252,7 +370,9 @@ static enum delret delete_dir_contents(c > save_filters = push_local_filters(fname, dlen); > > non_perishable_cnt = 0; >+ file_extra_cnt += SUM_EXTRA_CNT; > dirlist = get_dirlist(fname, dlen, 0); >+ file_extra_cnt -= SUM_EXTRA_CNT; > ret = non_perishable_cnt ? DR_NOT_EMPTY : DR_SUCCESS; > > if (!dirlist->used) >@@ -292,7 +412,8 @@ static enum delret delete_dir_contents(c > if (S_ISDIR(fp->mode)) { > if (delete_dir_contents(fname, flags | DEL_RECURSE) != DR_SUCCESS) > ret = DR_NOT_EMPTY; >- } >+ } else if (detect_renamed && S_ISREG(fp->mode)) >+ look_for_rename(fp, fname); > if (delete_item(fname, fp->mode, flags) != DR_SUCCESS) > ret = DR_NOT_EMPTY; > } >@@ -457,13 +578,18 @@ static void do_delayed_deletions(char *d > * all the --delete-WHEN options. Note that the fbuf pointer must point to a > * MAXPATHLEN buffer with the name of the directory in it (the functions we > * call will append names onto the end, but the old dir value will be restored >- * on exit). */ >-static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev) >+ * on exit). >+ * >+ * Note: --detect-rename may use this routine with DEL_NO_DELETIONS set! >+ */ >+static void delete_in_dir(char *fbuf, struct file_struct *file, dev_t *fs_dev, >+ int del_flags) > { > static int already_warned = 0; > struct file_list *dirlist; >- char delbuf[MAXPATHLEN]; >- int dlen, i; >+ char *p, delbuf[MAXPATHLEN]; >+ unsigned remainder; >+ int dlen, i, restore_dot = 0; > > if (!fbuf) { > change_local_filter_dir(NULL, 0, 0); >@@ -477,17 +603,22 @@ static void delete_in_dir(char *fbuf, st > maybe_send_keepalive(); > > if (io_error && !ignore_errors) { >- if (already_warned) >+ if (!already_warned) { >+ rprintf(FINFO, >+ "IO error encountered -- skipping file deletion\n"); >+ already_warned = 1; >+ } >+ if (!detect_renamed) > return; >- rprintf(FINFO, >- "IO error encountered -- skipping file deletion\n"); >- already_warned = 1; >- return; >+ del_flags |= DEL_NO_DELETIONS; > } > > dlen = strlen(fbuf); > change_local_filter_dir(fbuf, dlen, F_DEPTH(file)); > >+ if (detect_renamed) >+ unexplored_dirs--; >+ > if (one_file_system) { > if (file->flags & FLAG_TOP_DIR) > filesystem_dev = *fs_dev; >@@ -497,6 +628,14 @@ static void delete_in_dir(char *fbuf, st > > dirlist = get_dirlist(fbuf, dlen, 0); > >+ p = fbuf + dlen; >+ if (dlen == 1 && *fbuf == '.') { >+ restore_dot = 1; >+ p = fbuf; >+ } else if (dlen != 1 || *fbuf != '/') >+ *p++ = '/'; >+ remainder = MAXPATHLEN - (p - fbuf); >+ > /* If an item in dirlist is not found in flist, delete it > * from the filesystem. */ > for (i = dirlist->used; i--; ) { >@@ -509,6 +648,10 @@ static void delete_in_dir(char *fbuf, st > f_name(fp, NULL)); > continue; > } >+ if (detect_renamed && S_ISREG(fp->mode)) { >+ strlcpy(p, fp->basename, remainder); >+ look_for_rename(fp, fbuf); >+ } > /* Here we want to match regardless of file type. Replacement > * of a file with one of another type is handled separately by > * a delete_item call with a DEL_MAKE_ROOM flag. */ >@@ -517,14 +660,19 @@ static void delete_in_dir(char *fbuf, st > if (!(fp->mode & S_IWUSR) && !am_root && fp->flags & FLAG_OWNED_BY_US) > flags |= DEL_NO_UID_WRITE; > f_name(fp, delbuf); >- if (delete_during == 2) { >- if (!remember_delete(fp, delbuf, flags)) >+ if (delete_during == 2 && !(del_flags & DEL_NO_DELETIONS)) { >+ if (!remember_delete(fp, delbuf, del_flags | flags)) > break; > } else >- delete_item(delbuf, fp->mode, flags); >- } >+ delete_item(delbuf, fp->mode, del_flags | flags); >+ } else if (detect_renamed && S_ISDIR(fp->mode)) >+ unexplored_dirs++; > } > >+ if (restore_dot) >+ fbuf[0] = '.'; >+ fbuf[dlen] = '\0'; >+ > flist_free(dirlist); > } > >@@ -557,9 +705,9 @@ static void do_delete_pass(void) > || !S_ISDIR(st.st_mode)) > continue; > >- delete_in_dir(fbuf, file, &st.st_dev); >+ delete_in_dir(fbuf, file, &st.st_dev, 0); > } >- delete_in_dir(NULL, NULL, &dev_zero); >+ delete_in_dir(NULL, NULL, &dev_zero, 0); > > if (do_progress && !am_server) > rprintf(FINFO, " \r"); >@@ -1278,6 +1426,7 @@ static void list_file_entry(struct file_ > } > } > >+static struct bitbag *delayed_bits = NULL; > static int phase = 0; > static int dflt_perms; > >@@ -1567,9 +1716,12 @@ static void recv_generator(char *fname, > } > else if (delete_during && f_out != -1 && !phase > && !(file->flags & FLAG_MISSING_DIR)) { >- if (file->flags & FLAG_CONTENT_DIR) >- delete_in_dir(fname, file, &real_sx.st.st_dev); >- else >+ if (file->flags & FLAG_CONTENT_DIR) { >+ if (detect_renamed && real_ret != 0) >+ unexplored_dirs++; >+ delete_in_dir(fname, file, &real_sx.st.st_dev, >+ delete_during < 0 ? DEL_NO_DELETIONS : 0); >+ } else > change_local_filter_dir(fname, strlen(fname), F_DEPTH(file)); > } > goto cleanup; >@@ -1853,8 +2005,14 @@ static void recv_generator(char *fname, > goto cleanup; > } > #endif >- if (stat_errno == ENOENT) >+ if (stat_errno == ENOENT) { >+ if (detect_renamed && unexplored_dirs > 0 >+ && F_LENGTH(file)) { >+ bitbag_set_bit(delayed_bits, ndx); >+ return; >+ } > goto notify_others; >+ } > rsyserr(FERROR_XFER, stat_errno, "recv_generator: failed to stat %s", > full_fname(fname)); > goto cleanup; >@@ -2263,6 +2421,12 @@ void generate_files(int f_out, const cha > if (verbose > 2) > rprintf(FINFO, "generator starting pid=%ld\n", (long)getpid()); > >+ if (detect_renamed) { >+ delayed_bits = bitbag_create(cur_flist->used); >+ if (!delete_before && !delete_during) >+ delete_during = -1; >+ } >+ > if (delete_before && !solo_file && cur_flist->used > 0) > do_delete_pass(); > if (delete_during == 2) { >@@ -2273,7 +2437,7 @@ void generate_files(int f_out, const cha > } > do_progress = 0; > >- if (append_mode > 0 || whole_file < 0) >+ if (append_mode > 0 || detect_renamed || whole_file < 0) > whole_file = 0; > if (verbose >= 2) { > rprintf(FINFO, "delta-transmission %s\n", >@@ -2315,7 +2479,7 @@ void generate_files(int f_out, const cha > dirdev = MAKEDEV(DEV_MAJOR(devp), DEV_MINOR(devp)); > } else > dirdev = MAKEDEV(0, 0); >- delete_in_dir(fbuf, fp, &dirdev); >+ delete_in_dir(fbuf, fp, &dirdev, 0); > } else > change_local_filter_dir(fbuf, strlen(fbuf), F_DEPTH(fp)); > } >@@ -2362,7 +2526,21 @@ void generate_files(int f_out, const cha > } while ((cur_flist = cur_flist->next) != NULL); > > if (delete_during) >- delete_in_dir(NULL, NULL, &dev_zero); >+ delete_in_dir(NULL, NULL, &dev_zero, 0); >+ if (detect_renamed) { >+ if (delete_during < 0) >+ delete_during = 0; >+ detect_renamed = 0; >+ >+ for (i = -1; (i = bitbag_next_bit(delayed_bits, i)) >= 0; ) { >+ struct file_struct *file = cur_flist->files[i]; >+ if (local_name) >+ strlcpy(fbuf, local_name, sizeof fbuf); >+ else >+ f_name(file, fbuf); >+ recv_generator(fbuf, file, i, itemizing, code, f_out); >+ } >+ } > phase++; > if (verbose > 2) > rprintf(FINFO, "generate_files phase=%d\n", phase); >--- options.c.orig 2012-04-03 10:38:42.000000000 -0400 >+++ options.c 2012-04-03 10:38:42.000000000 -0400 >@@ -82,6 +82,7 @@ int am_sender = 0; > int am_starting_up = 1; > int relative_paths = -1; > int implied_dirs = 1; >+int detect_renamed = 0; > int numeric_ids = 0; > int allow_8bit_chars = 0; > int force_delete = 0; >@@ -392,6 +393,7 @@ void usage(enum logcode F) > rprintf(F," --modify-window=NUM compare mod-times with reduced accuracy\n"); > rprintf(F," -T, --temp-dir=DIR create temporary files in directory DIR\n"); > rprintf(F," -y, --fuzzy find similar file for basis if no dest file\n"); >+ rprintf(F," --detect-renamed try to find renamed files to speed up the transfer\n"); > rprintf(F," --compare-dest=DIR also compare destination files relative to DIR\n"); > rprintf(F," --copy-dest=DIR ... and include copies of unchanged files\n"); > rprintf(F," --link-dest=DIR hardlink to files in DIR when unchanged\n"); >@@ -582,6 +584,7 @@ static struct poptOption long_options[] > {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 }, > {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 }, > {"link-dest", 0, POPT_ARG_STRING, 0, OPT_LINK_DEST, 0, 0 }, >+ {"detect-renamed", 0, POPT_ARG_NONE, &detect_renamed, 0, 0, 0 }, > {"fuzzy", 'y', POPT_ARG_VAL, &fuzzy_basis, 1, 0, 0 }, > {"no-fuzzy", 0, POPT_ARG_VAL, &fuzzy_basis, 0, 0, 0 }, > {"no-y", 0, POPT_ARG_VAL, &fuzzy_basis, 0, 0, 0 }, >@@ -1606,7 +1609,7 @@ int parse_arguments(int *argc_p, const c > inplace = 1; > } > >- if (delay_updates && !partial_dir) >+ if ((delay_updates || detect_renamed) && !partial_dir) > partial_dir = tmp_partialdir; > > if (inplace) { >@@ -1615,6 +1618,7 @@ int parse_arguments(int *argc_p, const c > snprintf(err_buf, sizeof err_buf, > "--%s cannot be used with --%s\n", > append_mode ? "append" : "inplace", >+ detect_renamed ? "detect-renamed" : > delay_updates ? "delay-updates" : "partial-dir"); > return 0; > } >@@ -1979,6 +1983,8 @@ void server_options(char **args, int *ar > args[ac++] = "--super"; > if (size_only) > args[ac++] = "--size-only"; >+ if (detect_renamed) >+ args[ac++] = "--detect-renamed"; > } else { > if (skip_compress) { > if (asprintf(&arg, "--skip-compress=%s", skip_compress) < 0) >--- rsync.yo.orig 2012-04-03 10:38:42.000000000 -0400 >+++ rsync.yo 2012-04-03 10:38:42.000000000 -0400 >@@ -403,6 +403,7 @@ to the detailed description below for a > --modify-window=NUM compare mod-times with reduced accuracy > -T, --temp-dir=DIR create temporary files in directory DIR > -y, --fuzzy find similar file for basis if no dest file >+ --detect-renamed try to find renamed files to speed the xfer > --compare-dest=DIR also compare received files relative to DIR > --copy-dest=DIR ... and include copies of unchanged files > --link-dest=DIR hardlink to files in DIR when unchanged >@@ -1597,6 +1598,21 @@ Note that the use of the bf(--delete) op > fuzzy-match files, so either use bf(--delete-after) or specify some > filename exclusions if you need to prevent this. > >+dit(bf(--detect-renamed)) With this option, for each new source file >+(call it em(src/S)), rsync looks for a file em(dest/D) anywhere in the >+destination that passes the quick check with em(src/S). If such a em(dest/D) >+is found, rsync uses it as an alternate basis for transferring em(S). The >+idea is that if em(src/S) was renamed from em(src/D) (as opposed to em(src/S) >+passing the quick check with em(dest/D) by coincidence), the delta-transfer >+algorithm will find that all the data matches between em(src/S) and em(dest/D), >+and the transfer will be really fast. >+ >+By default, alternate-basis files are hard-linked into a directory named >+".~tmp~" in each file's destination directory, but if you've specified >+the bf(--partial-dir) option, that directory will be used instead. These >+potential alternate-basis files will be removed as the transfer progresses. >+This option conflicts with bf(--inplace) and bf(--append). >+ > dit(bf(--compare-dest=DIR)) This option instructs rsync to use em(DIR) on > the destination machine as an additional hierarchy to compare destination > files against doing transfers (if the files are missing in the destination >--- syscall.c.orig 2011-02-21 14:32:51.000000000 -0500 >+++ syscall.c 2012-04-04 11:01:33.000000000 -0400 >@@ -217,6 +217,22 @@ void trim_trailing_slashes(char *name) > } > } > >+int do_mkdir_path(char *fname, mode_t mode) >+{ >+ char fnametmp[MAXPATHLEN], *fnametmpptr; >+ STRUCT_STAT st; >+ >+ if (fname) { >+ strcpy(fnametmp, fname); >+ if ((fnametmpptr = strrchr(fnametmp, '/')) != NULL) { >+ *fnametmpptr = '\0'; >+ if (do_stat(fnametmp, &st) < 0) >+ do_mkdir_path(fnametmp, mode); >+ } >+ } >+ return do_mkdir(fname, mode); >+} >+ > int do_mkdir(char *fname, mode_t mode) > { > if (dry_run) return 0; >--- util.c.orig 2012-04-03 10:38:42.000000000 -0400 >+++ util.c 2012-04-03 23:15:12.000000000 -0400 >@@ -1172,7 +1172,7 @@ int handle_partial_dir(const char *fname > } > statret = -1; > } >- if (statret < 0 && do_mkdir(dir, 0700) < 0) { >+ if (statret < 0 && do_mkdir_path(dir, 0700) < 0) { > *fn = '/'; > return 0; > } >@@ -1183,6 +1183,32 @@ int handle_partial_dir(const char *fname > return 1; > } > >+/* We need to supply our own strcmp function for file list comparisons >+ * to ensure that signed/unsigned usage is consistent between machines. */ >+int u_strcmp(const char *p1, const char *p2) >+{ >+ for ( ; *p1; p1++, p2++) { >+ if (*p1 != *p2) >+ break; >+ } >+ >+ return (int)*(uchar*)p1 - (int)*(uchar*)p2; >+} >+ >+/* We need a memcmp function compares unsigned-byte values. */ >+int u_memcmp(const void *p1, const void *p2, size_t len) >+{ >+ const uchar *u1 = p1; >+ const uchar *u2 = p2; >+ >+ while (len--) { >+ if (*u1 != *u2) >+ return (int)*u1 - (int)*u2; >+ } >+ >+ return 0; >+} >+ > /* Determine if a symlink points outside the current directory tree. > * This is considered "unsafe" because e.g. when mirroring somebody > * else's machine it might allow them to establish a symlink to >--- proto.h.orig 2012-04-03 10:38:42.000000000 -0400 >+++ proto.h 2012-04-03 23:29:23.000000000 -0400 >@@ -305,6 +305,7 @@ int do_chmod(const char *path, mode_t mo > int do_rename(const char *fname1, const char *fname2); > int do_ftruncate(int fd, OFF_T size); > void trim_trailing_slashes(char *name); >+int do_mkdir_path(char *fname, mode_t mode); > int do_mkdir(char *fname, mode_t mode); > int do_mkstemp(char *template, mode_t perms); > int do_stat(const char *fname, STRUCT_STAT *st); >@@ -362,6 +363,8 @@ char *normalize_path(char *path, BOOL fo > char *full_fname(const char *fn); > char *partial_dir_fname(const char *fname); > int handle_partial_dir(const char *fname, int create); >+int u_strcmp(const char *p1, const char *p2); >+int u_memcmp(const void *p1, const void *p2, size_t len); > int unsafe_symlink(const char *dest, const char *src); > char *human_num(int64 num); > char *human_dnum(double dnum, int decimal_digits); >--- rsync.1.orig 2012-04-03 10:38:42.000000000 -0400 >+++ rsync.1 2012-04-03 21:48:49.000000000 -0400 >@@ -479,6 +479,7 @@ to the detailed description below for a > \-\-modify\-window=NUM compare mod\-times with reduced accuracy > \-T, \-\-temp\-dir=DIR create temporary files in directory DIR > \-y, \-\-fuzzy find similar file for basis if no dest file >+ \-\-detect\-renamed try to find renamed files to speed the xfer > \-\-compare\-dest=DIR also compare received files relative to DIR > \-\-copy\-dest=DIR ... and include copies of unchanged files > \-\-link\-dest=DIR hardlink to files in DIR when unchanged >@@ -1828,6 +1829,22 @@ Note that the use of the \fB\-\-delete\f > fuzzy\-match files, so either use \fB\-\-delete\-after\fP or specify some > filename exclusions if you need to prevent this. > .IP >+.IP "\fB\-\-detect\-renamed\fP" >+With this option, for each new source file >+(call it \fIsrc/S\fP), rsync looks for a file \fIdest/D\fP anywhere in the >+destination that passes the quick check with \fIsrc/S\fP. If such a \fIdest/D\fP >+is found, rsync uses it as an alternate basis for transferring \fIS\fP. The >+idea is that if \fIsrc/S\fP was renamed from \fIsrc/D\fP (as opposed to \fIsrc/S\fP >+passing the quick check with \fIdest/D\fP by coincidence), the delta\-transfer >+algorithm will find that all the data matches between \fIsrc/S\fP and \fIdest/D\fP, >+and the transfer will be really fast. >+.IP >+By default, alternate\-basis files are hard\-linked into a directory named >+\(dq\&.~tmp~\(dq\& in each file\(cq\&s destination directory, but if you\(cq\&ve specified >+the \fB\-\-partial\-dir\fP option, that directory will be used instead. These >+potential alternate\-basis files will be removed as the transfer progresses. >+This option conflicts with \fB\-\-inplace\fP and \fB\-\-append\fP. >+.IP > .IP "\fB\-\-compare\-dest=DIR\fP" > This option instructs rsync to use \fIDIR\fP on > the destination machine as an additional hierarchy to compare destination
You cannot view the attachment while viewing its details because your browser does not support IFRAMEs.
View the attachment on a separate page
.
View Attachment As Raw
Actions:
View
Attachments on
bug 8847
: 7435