Commit | Line | Data |
---|---|---|
2d2150f6 WD |
1 | This patch changes the way the --checksum option works by having the |
2 | receiving side perform a checksum-read of every file in the file list | |
d47d38c7 | 3 | as the list is received (if the sizes are equal), marking non-matching |
2d2150f6 WD |
4 | items with a flag. The idea is that the checksum pass on the sender and |
5 | the receiver can then happen in parallel instead of having the reciever | |
6 | to its checksum pass during its normal find-the-different-files pass. | |
7 | ||
d47d38c7 WD |
8 | I have benchmarked this a little, and it appears to slow things down |
9 | for a local copy, so the old algorithm is used for local copies. | |
2d2150f6 | 10 | |
d57e621e | 11 | --- orig/flist.c 2005-11-10 16:58:36 |
2d2150f6 WD |
12 | +++ flist.c 2005-09-16 16:41:25 |
13 | @@ -36,6 +36,7 @@ extern int am_daemon; | |
14 | extern int am_sender; | |
15 | extern int do_progress; | |
16 | extern int always_checksum; | |
17 | +extern int pre_checksum; | |
18 | extern int module_id; | |
19 | extern int ignore_errors; | |
20 | extern int numeric_ids; | |
d47d38c7 | 21 | @@ -707,6 +708,16 @@ static struct file_struct *receive_file_ |
2d2150f6 WD |
22 | sum = empty_sum; |
23 | } | |
d57e621e | 24 | read_buf(f, sum, checksum_len); |
2d2150f6 WD |
25 | + if (pre_checksum) { |
26 | + char sum2[MD4_SUM_LENGTH]; | |
27 | + STRUCT_STAT st; | |
28 | + char *fname = f_name(file); | |
29 | + if (stat(fname, &st) == 0 && st.st_size == file_length) { | |
30 | + file_checksum(fname, sum2, st.st_size); | |
d57e621e | 31 | + if (memcmp(sum, sum2, checksum_len) != 0) |
2d2150f6 WD |
32 | + file->flags |= FLAG_SUM_DIFFERS; |
33 | + } | |
34 | + } | |
35 | } | |
36 | ||
37 | if (!preserve_perms) { | |
f20eb450 WD |
38 | --- orig/generator.c 2005-12-15 23:00:49 |
39 | +++ generator.c 2005-12-15 23:16:48 | |
2d2150f6 WD |
40 | @@ -69,6 +69,7 @@ extern int ignore_timeout; |
41 | extern int protocol_version; | |
42 | extern int fuzzy_basis; | |
43 | extern int always_checksum; | |
44 | +extern int pre_checksum; | |
d57e621e | 45 | extern int checksum_len; |
2d2150f6 WD |
46 | extern char *partial_dir; |
47 | extern char *basis_dir[]; | |
f20eb450 | 48 | @@ -376,7 +377,8 @@ void itemize(struct file_struct *file, i |
2d2150f6 WD |
49 | |
50 | ||
51 | /* Perform our quick-check heuristic for determining if a file is unchanged. */ | |
f20eb450 WD |
52 | -int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st) |
53 | +int unchanged_file(char *fn, int fnamecmp_type, struct file_struct *file, | |
54 | + STRUCT_STAT *st) | |
2d2150f6 WD |
55 | { |
56 | if (st->st_size != file->length) | |
57 | return 0; | |
f20eb450 | 58 | @@ -385,6 +387,8 @@ int unchanged_file(char *fn, struct file |
2d2150f6 WD |
59 | of the file time to determine whether to sync */ |
60 | if (always_checksum && S_ISREG(st->st_mode)) { | |
61 | char sum[MD4_SUM_LENGTH]; | |
62 | + if (pre_checksum && fnamecmp_type == FNAMECMP_FNAME) | |
63 | + return !(file->flags & FLAG_SUM_DIFFERS); | |
64 | file_checksum(fn, sum, st->st_size); | |
d57e621e WD |
65 | return memcmp(sum, file->u.sum, checksum_len) == 0; |
66 | } | |
f20eb450 WD |
67 | @@ -622,7 +626,7 @@ static int try_dests_reg(struct file_str |
68 | match_level = 1; | |
69 | /* FALL THROUGH */ | |
70 | case 1: | |
71 | - if (!unchanged_file(cmpbuf, file, stp)) | |
72 | + if (!unchanged_file(cmpbuf, 0, file, stp)) | |
73 | continue; | |
74 | best_match = j; | |
75 | match_level = 2; | |
76 | @@ -1160,7 +1164,7 @@ static void recv_generator(char *fname, | |
2d2150f6 WD |
77 | ; |
78 | else if (fnamecmp_type == FNAMECMP_FUZZY) | |
79 | ; | |
80 | - else if (unchanged_file(fnamecmp, file, &st)) { | |
81 | + else if (unchanged_file(fnamecmp, fnamecmp_type, file, &st)) { | |
d57e621e WD |
82 | if (partialptr) { |
83 | do_unlink(partialptr); | |
84 | handle_partial_dir(partialptr, PDIR_DELETE); | |
f20eb450 WD |
85 | --- orig/hlink.c 2005-12-15 23:00:49 |
86 | +++ hlink.c 2005-12-15 23:17:31 | |
87 | @@ -210,7 +210,7 @@ int hard_link_check(struct file_struct * | |
88 | itemizing = code = 0; | |
89 | break; | |
90 | } | |
91 | - if (!unchanged_file(cmpbuf, file, &st3)) | |
92 | + if (!unchanged_file(cmpbuf, 0, file, &st3)) | |
93 | continue; | |
94 | statret = 1; | |
95 | st = &st3; | |
96 | --- orig/main.c 2005-12-15 07:55:39 | |
d47d38c7 | 97 | +++ main.c 2005-11-07 05:00:49 |
2d2150f6 WD |
98 | @@ -45,6 +45,7 @@ extern int copy_links; |
99 | extern int keep_dirlinks; | |
100 | extern int preserve_hard_links; | |
101 | extern int protocol_version; | |
102 | +extern int always_checksum; | |
103 | extern int recurse; | |
104 | extern int relative_paths; | |
105 | extern int rsync_port; | |
106 | @@ -60,8 +61,10 @@ extern char *filesfrom_host; | |
107 | extern char *rsync_path; | |
108 | extern char *shell_cmd; | |
109 | extern char *batch_name; | |
110 | +extern char curr_dir[MAXPATHLEN]; | |
111 | ||
112 | int local_server = 0; | |
113 | +int pre_checksum = 0; | |
114 | struct file_list *the_file_list; | |
115 | ||
116 | /* There's probably never more than at most 2 outstanding child processes, | |
f20eb450 | 117 | @@ -659,6 +662,7 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
118 | struct file_list *flist; |
119 | char *local_name = NULL; | |
120 | char *dir = NULL; | |
121 | + char olddir[sizeof curr_dir]; | |
122 | int save_verbose = verbose; | |
123 | ||
124 | if (filesfrom_fd >= 0) { | |
f20eb450 | 125 | @@ -702,6 +706,10 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
126 | filesfrom_fd = -1; |
127 | } | |
128 | ||
129 | + strlcpy(olddir, curr_dir, sizeof olddir); | |
d47d38c7 | 130 | + if (always_checksum && !local_server && argc > 0) |
2d2150f6 WD |
131 | + pre_checksum = push_dir(argv[0]); |
132 | + | |
133 | flist = recv_file_list(f_in); | |
134 | verbose = save_verbose; | |
135 | if (!flist) { | |
f20eb450 | 136 | @@ -710,6 +718,9 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
137 | } |
138 | the_file_list = flist; | |
139 | ||
140 | + if (pre_checksum) | |
141 | + pop_dir(olddir); | |
142 | + | |
143 | if (argc > 0) | |
144 | local_name = get_local_name(flist,argv[0]); | |
145 | ||
f20eb450 | 146 | @@ -758,6 +769,7 @@ int client_run(int f_in, int f_out, pid_ |
2d2150f6 WD |
147 | { |
148 | struct file_list *flist = NULL; | |
149 | int exit_code = 0, exit_code2 = 0; | |
150 | + char olddir[sizeof curr_dir]; | |
151 | char *local_name = NULL; | |
152 | ||
153 | cleanup_child_pid = pid; | |
f20eb450 | 154 | @@ -829,11 +841,18 @@ int client_run(int f_in, int f_out, pid_ |
2d2150f6 WD |
155 | filesfrom_fd = -1; |
156 | } | |
157 | ||
158 | + strlcpy(olddir, curr_dir, sizeof olddir); | |
d47d38c7 | 159 | + if (always_checksum && !local_server) |
2d2150f6 WD |
160 | + pre_checksum = push_dir(argv[0]); |
161 | + | |
162 | if (write_batch && !am_server) | |
163 | start_write_batch(f_in); | |
164 | flist = recv_file_list(f_in); | |
165 | the_file_list = flist; | |
166 | ||
167 | + if (pre_checksum) | |
168 | + pop_dir(olddir); | |
169 | + | |
170 | if (flist && flist->count > 0) { | |
171 | local_name = get_local_name(flist, argv[0]); | |
172 | ||
f20eb450 | 173 | --- orig/rsync.h 2005-12-15 23:00:49 |
2d2150f6 WD |
174 | +++ rsync.h 2005-09-16 16:41:26 |
175 | @@ -64,6 +64,7 @@ | |
176 | #define FLAG_DEL_HERE (1<<3) /* receiver/generator */ | |
177 | #define FLAG_SENT (1<<3) /* sender */ | |
178 | #define FLAG_HLINK_TOL (1<<4) /* receiver/generator */ | |
179 | +#define FLAG_SUM_DIFFERS (1<<5) /* receiver/generator */ | |
180 | ||
181 | /* update this if you make incompatible changes */ | |
182 | #define PROTOCOL_VERSION 29 |