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
3 as the list is received (if the sizes are equal), marking non-matching
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.
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.
11 --- orig/flist.c 2005-11-10 16:58:36
12 +++ flist.c 2005-09-16 16:41:25
13 @@ -36,6 +36,7 @@ extern int am_daemon;
15 extern int do_progress;
16 extern int always_checksum;
17 +extern int pre_checksum;
19 extern int ignore_errors;
20 extern int numeric_ids;
21 @@ -707,6 +708,16 @@ static struct file_struct *receive_file_
24 read_buf(f, sum, checksum_len);
26 + char sum2[MD4_SUM_LENGTH];
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);
31 + if (memcmp(sum, sum2, checksum_len) != 0)
32 + file->flags |= FLAG_SUM_DIFFERS;
37 if (!preserve_perms) {
38 --- orig/generator.c 2005-11-10 16:58:36
39 +++ generator.c 2005-11-10 03:16:39
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;
45 extern int checksum_len;
46 extern char *partial_dir;
47 extern char *basis_dir[];
48 @@ -375,7 +376,8 @@ void itemize(struct file_struct *file, i
51 /* Perform our quick-check heuristic for determining if a file is unchanged. */
52 -static int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
53 +static int unchanged_file(char *fn, int fnamecmp_type, struct file_struct *file,
56 if (st->st_size != file->length)
58 @@ -384,6 +386,8 @@ static int unchanged_file(char *fn, stru
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);
65 return memcmp(sum, file->u.sum, checksum_len) == 0;
67 @@ -940,7 +944,7 @@ static void recv_generator(char *fname,
71 - if (!unchanged_file(fnamecmpbuf, file, &st))
72 + if (!unchanged_file(fnamecmpbuf, 0, file, &st))
76 @@ -1068,7 +1072,7 @@ static void recv_generator(char *fname,
78 else if (fnamecmp_type == FNAMECMP_FUZZY)
80 - else if (unchanged_file(fnamecmp, file, &st)) {
81 + else if (unchanged_file(fnamecmp, fnamecmp_type, file, &st)) {
83 do_unlink(partialptr);
84 handle_partial_dir(partialptr, PDIR_DELETE);
85 --- orig/main.c 2005-11-10 16:58:36
86 +++ main.c 2005-11-07 05:00:49
87 @@ -45,6 +45,7 @@ extern int copy_links;
88 extern int keep_dirlinks;
89 extern int preserve_hard_links;
90 extern int protocol_version;
91 +extern int always_checksum;
93 extern int relative_paths;
94 extern int rsync_port;
95 @@ -60,8 +61,10 @@ extern char *filesfrom_host;
96 extern char *rsync_path;
97 extern char *shell_cmd;
98 extern char *batch_name;
99 +extern char curr_dir[MAXPATHLEN];
101 int local_server = 0;
102 +int pre_checksum = 0;
103 struct file_list *the_file_list;
105 /* There's probably never more than at most 2 outstanding child processes,
106 @@ -633,6 +636,7 @@ static void do_server_recv(int f_in, int
107 struct file_list *flist;
108 char *local_name = NULL;
110 + char olddir[sizeof curr_dir];
111 int save_verbose = verbose;
113 if (filesfrom_fd >= 0) {
114 @@ -676,6 +680,10 @@ static void do_server_recv(int f_in, int
118 + strlcpy(olddir, curr_dir, sizeof olddir);
119 + if (always_checksum && !local_server && argc > 0)
120 + pre_checksum = push_dir(argv[0]);
122 flist = recv_file_list(f_in);
123 verbose = save_verbose;
125 @@ -684,6 +692,9 @@ static void do_server_recv(int f_in, int
127 the_file_list = flist;
133 local_name = get_local_name(flist,argv[0]);
135 @@ -732,6 +743,7 @@ int client_run(int f_in, int f_out, pid_
137 struct file_list *flist = NULL;
138 int exit_code = 0, exit_code2 = 0;
139 + char olddir[sizeof curr_dir];
140 char *local_name = NULL;
142 cleanup_child_pid = pid;
143 @@ -803,11 +815,18 @@ int client_run(int f_in, int f_out, pid_
147 + strlcpy(olddir, curr_dir, sizeof olddir);
148 + if (always_checksum && !local_server)
149 + pre_checksum = push_dir(argv[0]);
151 if (write_batch && !am_server)
152 start_write_batch(f_in);
153 flist = recv_file_list(f_in);
154 the_file_list = flist;
159 if (flist && flist->count > 0) {
160 local_name = get_local_name(flist, argv[0]);
162 --- orig/rsync.h 2005-10-14 18:45:50
163 +++ rsync.h 2005-09-16 16:41:26
165 #define FLAG_DEL_HERE (1<<3) /* receiver/generator */
166 #define FLAG_SENT (1<<3) /* sender */
167 #define FLAG_HLINK_TOL (1<<4) /* receiver/generator */
168 +#define FLAG_SUM_DIFFERS (1<<5) /* receiver/generator */
170 /* update this if you make incompatible changes */
171 #define PROTOCOL_VERSION 29