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 | |
9a7eef96 WD |
11 | --- old/flist.c |
12 | +++ new/flist.c | |
b5abdf47 | 13 | @@ -31,6 +31,7 @@ extern int am_daemon; |
2d2150f6 WD |
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; | |
f4e59d46 | 21 | @@ -692,6 +693,16 @@ 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; | |
bd68c3c2 | 28 | + char *fname = f_name(file, NULL); |
2d2150f6 WD |
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 | ||
9a7eef96 WD |
37 | return file; |
38 | --- old/generator.c | |
39 | +++ new/generator.c | |
f4e59d46 | 40 | @@ -71,6 +71,7 @@ extern int ignore_timeout; |
2d2150f6 WD |
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[]; | |
f4e59d46 | 48 | @@ -374,7 +375,8 @@ 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; | |
f4e59d46 | 58 | @@ -383,6 +385,8 @@ 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 | } | |
f4e59d46 | 67 | @@ -626,7 +630,7 @@ static int try_dests_reg(struct file_str |
f20eb450 WD |
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; | |
f4e59d46 | 76 | @@ -1194,7 +1198,7 @@ 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); | |
9a7eef96 WD |
85 | --- old/hlink.c |
86 | +++ new/hlink.c | |
b5abdf47 | 87 | @@ -212,7 +212,7 @@ int hard_link_check(struct file_struct * |
f20eb450 WD |
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; | |
9a7eef96 WD |
96 | --- old/main.c |
97 | +++ new/main.c | |
b5abdf47 | 98 | @@ -47,6 +47,7 @@ extern int copy_dirlinks; |
2d2150f6 WD |
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; | |
b5abdf47 | 105 | extern int sanitize_paths; |
f4e59d46 | 106 | @@ -69,6 +70,9 @@ extern char *shell_cmd; |
af7eb67a | 107 | extern char *batch_name; |
f4e59d46 | 108 | extern struct filter_list_struct server_filter_list; |
2d2150f6 | 109 | |
af7eb67a WD |
110 | +extern char curr_dir[MAXPATHLEN]; |
111 | + | |
2d2150f6 | 112 | +int pre_checksum = 0; |
d9a67109 WD |
113 | int local_server = 0; |
114 | mode_t orig_umask = 0; | |
f4e59d46 WD |
115 | struct file_list *the_file_list; |
116 | @@ -749,6 +753,7 @@ static void do_server_recv(int f_in, int | |
2d2150f6 WD |
117 | struct file_list *flist; |
118 | char *local_name = NULL; | |
119 | char *dir = NULL; | |
120 | + char olddir[sizeof curr_dir]; | |
121 | int save_verbose = verbose; | |
122 | ||
123 | if (filesfrom_fd >= 0) { | |
f4e59d46 | 124 | @@ -792,6 +797,10 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
125 | filesfrom_fd = -1; |
126 | } | |
127 | ||
128 | + strlcpy(olddir, curr_dir, sizeof olddir); | |
d47d38c7 | 129 | + if (always_checksum && !local_server && argc > 0) |
2d2150f6 WD |
130 | + pre_checksum = push_dir(argv[0]); |
131 | + | |
132 | flist = recv_file_list(f_in); | |
133 | verbose = save_verbose; | |
134 | if (!flist) { | |
f4e59d46 | 135 | @@ -800,6 +809,9 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
136 | } |
137 | the_file_list = flist; | |
138 | ||
139 | + if (pre_checksum) | |
140 | + pop_dir(olddir); | |
141 | + | |
142 | if (argc > 0) | |
143 | local_name = get_local_name(flist,argv[0]); | |
144 | ||
f4e59d46 | 145 | @@ -883,6 +895,7 @@ int client_run(int f_in, int f_out, pid_ |
2d2150f6 WD |
146 | { |
147 | struct file_list *flist = NULL; | |
148 | int exit_code = 0, exit_code2 = 0; | |
149 | + char olddir[sizeof curr_dir]; | |
150 | char *local_name = NULL; | |
151 | ||
152 | cleanup_child_pid = pid; | |
f4e59d46 | 153 | @@ -957,11 +970,18 @@ int client_run(int f_in, int f_out, pid_ |
2d2150f6 WD |
154 | filesfrom_fd = -1; |
155 | } | |
156 | ||
157 | + strlcpy(olddir, curr_dir, sizeof olddir); | |
d47d38c7 | 158 | + if (always_checksum && !local_server) |
2d2150f6 WD |
159 | + pre_checksum = push_dir(argv[0]); |
160 | + | |
161 | if (write_batch && !am_server) | |
162 | start_write_batch(f_in); | |
163 | flist = recv_file_list(f_in); | |
164 | the_file_list = flist; | |
165 | ||
166 | + if (pre_checksum) | |
167 | + pop_dir(olddir); | |
168 | + | |
169 | if (flist && flist->count > 0) { | |
170 | local_name = get_local_name(flist, argv[0]); | |
171 | ||
9a7eef96 WD |
172 | --- old/rsync.h |
173 | +++ new/rsync.h | |
2d2150f6 WD |
174 | @@ -64,6 +64,7 @@ |
175 | #define FLAG_DEL_HERE (1<<3) /* receiver/generator */ | |
2d2150f6 | 176 | #define FLAG_HLINK_TOL (1<<4) /* receiver/generator */ |
0ca6aebe WD |
177 | #define FLAG_NO_FUZZY (1<<5) /* generator */ |
178 | +#define FLAG_SUM_DIFFERS (1<<6) /* receiver/generator */ | |
2d2150f6 WD |
179 | |
180 | /* update this if you make incompatible changes */ | |
181 | #define PROTOCOL_VERSION 29 |