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 | |
03019e41 WD |
11 | To use this patch, run these commands for a successful build: |
12 | ||
13 | patch -p1 <patches/early-checksum.diff | |
14 | ./configure (optional if already run) | |
15 | make | |
16 | ||
9a7eef96 WD |
17 | --- old/flist.c |
18 | +++ new/flist.c | |
b5abdf47 | 19 | @@ -31,6 +31,7 @@ extern int am_daemon; |
2d2150f6 WD |
20 | extern int am_sender; |
21 | extern int do_progress; | |
22 | extern int always_checksum; | |
23 | +extern int pre_checksum; | |
24 | extern int module_id; | |
25 | extern int ignore_errors; | |
26 | extern int numeric_ids; | |
7c15d5aa WD |
27 | @@ -785,6 +786,15 @@ static struct file_struct *recv_file_ent |
28 | memcpy(bp, F_SUM(first), checksum_len); | |
29 | } else | |
30 | read_buf(f, bp, checksum_len); | |
70891d26 | 31 | + if (pre_checksum && sum_len) { |
2d2150f6 | 32 | + STRUCT_STAT st; |
bd68c3c2 | 33 | + char *fname = f_name(file, NULL); |
2d2150f6 | 34 | + if (stat(fname, &st) == 0 && st.st_size == file_length) { |
70891d26 WD |
35 | + file_checksum(fname, tmp_sum, st.st_size); |
36 | + if (memcmp(bp, tmp_sum, checksum_len) != 0) | |
2d2150f6 WD |
37 | + file->flags |= FLAG_SUM_DIFFERS; |
38 | + } | |
39 | + } | |
40 | } | |
41 | ||
9a7eef96 WD |
42 | return file; |
43 | --- old/generator.c | |
44 | +++ new/generator.c | |
1aa236e1 | 45 | @@ -70,6 +70,7 @@ extern int ignore_timeout; |
2d2150f6 WD |
46 | extern int protocol_version; |
47 | extern int fuzzy_basis; | |
48 | extern int always_checksum; | |
49 | +extern int pre_checksum; | |
d57e621e | 50 | extern int checksum_len; |
2d2150f6 WD |
51 | extern char *partial_dir; |
52 | extern char *basis_dir[]; | |
7c15d5aa | 53 | @@ -577,7 +578,8 @@ void itemize(struct file_struct *file, i |
2d2150f6 WD |
54 | |
55 | ||
56 | /* Perform our quick-check heuristic for determining if a file is unchanged. */ | |
f20eb450 WD |
57 | -int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st) |
58 | +int unchanged_file(char *fn, int fnamecmp_type, struct file_struct *file, | |
59 | + STRUCT_STAT *st) | |
2d2150f6 | 60 | { |
1aa236e1 | 61 | if (st->st_size != F_LENGTH(file)) |
2d2150f6 | 62 | return 0; |
7c15d5aa | 63 | @@ -586,6 +588,8 @@ int unchanged_file(char *fn, struct file |
2d2150f6 WD |
64 | of the file time to determine whether to sync */ |
65 | if (always_checksum && S_ISREG(st->st_mode)) { | |
66 | char sum[MD4_SUM_LENGTH]; | |
67 | + if (pre_checksum && fnamecmp_type == FNAMECMP_FNAME) | |
68 | + return !(file->flags & FLAG_SUM_DIFFERS); | |
69 | file_checksum(fn, sum, st->st_size); | |
70891d26 | 70 | return memcmp(sum, F_SUM(file), checksum_len) == 0; |
d57e621e | 71 | } |
7c15d5aa | 72 | @@ -825,7 +829,7 @@ static int try_dests_reg(struct file_str |
f20eb450 WD |
73 | match_level = 1; |
74 | /* FALL THROUGH */ | |
75 | case 1: | |
76 | - if (!unchanged_file(cmpbuf, file, stp)) | |
77 | + if (!unchanged_file(cmpbuf, 0, file, stp)) | |
78 | continue; | |
79 | best_match = j; | |
80 | match_level = 2; | |
7c15d5aa | 81 | @@ -1511,7 +1515,7 @@ static void recv_generator(char *fname, |
2d2150f6 WD |
82 | ; |
83 | else if (fnamecmp_type == FNAMECMP_FUZZY) | |
84 | ; | |
85 | - else if (unchanged_file(fnamecmp, file, &st)) { | |
86 | + else if (unchanged_file(fnamecmp, fnamecmp_type, file, &st)) { | |
d57e621e WD |
87 | if (partialptr) { |
88 | do_unlink(partialptr); | |
89 | handle_partial_dir(partialptr, PDIR_DELETE); | |
9a7eef96 WD |
90 | --- old/hlink.c |
91 | +++ new/hlink.c | |
7c15d5aa | 92 | @@ -448,7 +448,7 @@ int hard_link_check(struct file_struct * |
7b80cd0e WD |
93 | } |
94 | break; | |
95 | } | |
96 | - if (!unchanged_file(cmpbuf, file, &alt_st)) | |
97 | + if (!unchanged_file(cmpbuf, 0, file, &alt_st)) | |
98 | continue; | |
99 | statret = 1; | |
100 | *stp = alt_st; | |
9a7eef96 WD |
101 | --- old/main.c |
102 | +++ new/main.c | |
b5abdf47 | 103 | @@ -47,6 +47,7 @@ extern int copy_dirlinks; |
2d2150f6 WD |
104 | extern int keep_dirlinks; |
105 | extern int preserve_hard_links; | |
106 | extern int protocol_version; | |
107 | +extern int always_checksum; | |
108 | extern int recurse; | |
109 | extern int relative_paths; | |
b5abdf47 | 110 | extern int sanitize_paths; |
60a8bf36 WD |
111 | @@ -71,6 +72,9 @@ extern char *batch_name; |
112 | extern char curr_dir[MAXPATHLEN]; | |
f4e59d46 | 113 | extern struct filter_list_struct server_filter_list; |
2d2150f6 | 114 | |
af7eb67a WD |
115 | +extern char curr_dir[MAXPATHLEN]; |
116 | + | |
2d2150f6 | 117 | +int pre_checksum = 0; |
d9a67109 | 118 | int local_server = 0; |
60a8bf36 | 119 | int new_root_dir = 0; |
d9a67109 | 120 | mode_t orig_umask = 0; |
7b80cd0e | 121 | @@ -788,6 +792,7 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
122 | struct file_list *flist; |
123 | char *local_name = NULL; | |
124 | char *dir = NULL; | |
125 | + char olddir[sizeof curr_dir]; | |
126 | int save_verbose = verbose; | |
127 | ||
128 | if (filesfrom_fd >= 0) { | |
7c15d5aa WD |
129 | @@ -831,6 +836,10 @@ static void do_server_recv(int f_in, int |
130 | filesfrom_fd = -1; | |
131 | } | |
2d2150f6 WD |
132 | |
133 | + strlcpy(olddir, curr_dir, sizeof olddir); | |
d47d38c7 | 134 | + if (always_checksum && !local_server && argc > 0) |
de565f59 | 135 | + pre_checksum = push_dir(argv[0], 0); |
2d2150f6 WD |
136 | + |
137 | flist = recv_file_list(f_in); | |
138 | verbose = save_verbose; | |
139 | if (!flist) { | |
7c15d5aa | 140 | @@ -839,6 +848,9 @@ static void do_server_recv(int f_in, int |
2d2150f6 WD |
141 | } |
142 | the_file_list = flist; | |
143 | ||
144 | + if (pre_checksum) | |
145 | + pop_dir(olddir); | |
146 | + | |
147 | if (argc > 0) | |
148 | local_name = get_local_name(flist,argv[0]); | |
149 | ||
7c15d5aa | 150 | @@ -918,6 +930,7 @@ int client_run(int f_in, int f_out, pid_ |
2d2150f6 WD |
151 | { |
152 | struct file_list *flist = NULL; | |
153 | int exit_code = 0, exit_code2 = 0; | |
154 | + char olddir[sizeof curr_dir]; | |
155 | char *local_name = NULL; | |
156 | ||
157 | cleanup_child_pid = pid; | |
7c15d5aa WD |
158 | @@ -992,11 +1005,18 @@ int client_run(int f_in, int f_out, pid_ |
159 | filesfrom_fd = -1; | |
160 | } | |
2d2150f6 WD |
161 | |
162 | + strlcpy(olddir, curr_dir, sizeof olddir); | |
d47d38c7 | 163 | + if (always_checksum && !local_server) |
de565f59 | 164 | + pre_checksum = push_dir(argv[0], 0); |
2d2150f6 WD |
165 | + |
166 | if (write_batch && !am_server) | |
167 | start_write_batch(f_in); | |
168 | flist = recv_file_list(f_in); | |
169 | the_file_list = flist; | |
170 | ||
171 | + if (pre_checksum) | |
172 | + pop_dir(olddir); | |
173 | + | |
174 | if (flist && flist->count > 0) { | |
175 | local_name = get_local_name(flist, argv[0]); | |
176 | ||
9a7eef96 WD |
177 | --- old/rsync.h |
178 | +++ new/rsync.h | |
7c15d5aa | 179 | @@ -68,6 +68,7 @@ |
1aa236e1 | 180 | #define FLAG_HLINK_LAST (1<<7) /* receiver/generator */ |
7b80cd0e WD |
181 | #define FLAG_HLINK_DONE (1<<8) /* receiver/generator */ |
182 | #define FLAG_LENGTH64 (1<<9) /* sender/receiver/generator */ | |
183 | +#define FLAG_SUM_DIFFERS (1<<10)/* receiver/generator */ | |
1aa236e1 | 184 | |
7b80cd0e WD |
185 | #define BITS_SET(val,bits) (((val) & (bits)) == (bits)) |
186 | #define BITS_SETnUNSET(val,onbits,offbits) (((val) & ((onbits)|(offbits))) == (onbits)) |