Commit | Line | Data |
---|---|---|
1e4f48d6 | 1 | /* |
c627d613 AT |
2 | Copyright (C) Andrew Tridgell 1996 |
3 | Copyright (C) Paul Mackerras 1996 | |
1e4f48d6 | 4 | |
c627d613 AT |
5 | This program is free software; you can redistribute it and/or modify |
6 | it under the terms of the GNU General Public License as published by | |
7 | the Free Software Foundation; either version 2 of the License, or | |
8 | (at your option) any later version. | |
1e4f48d6 | 9 | |
c627d613 AT |
10 | This program is distributed in the hope that it will be useful, |
11 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
13 | GNU General Public License for more details. | |
1e4f48d6 | 14 | |
c627d613 AT |
15 | You should have received a copy of the GNU General Public License |
16 | along with this program; if not, write to the Free Software | |
17 | Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | |
18 | */ | |
19 | ||
20 | #include "rsync.h" | |
21 | ||
22 | extern int verbose; | |
16417f8b | 23 | extern int do_progress; |
ba582f75 | 24 | extern int checksum_seed; |
6cc11982 | 25 | extern int append_mode; |
a36ffd39 WD |
26 | |
27 | int updating_basis_file; | |
c627d613 | 28 | |
c627d613 AT |
29 | static int false_alarms; |
30 | static int tag_hits; | |
31 | static int matches; | |
a800434a | 32 | static int64 data_transfer; |
c627d613 | 33 | |
3a6a366f AT |
34 | static int total_false_alarms; |
35 | static int total_tag_hits; | |
36 | static int total_matches; | |
c627d613 | 37 | |
a800434a | 38 | extern struct stats stats; |
c627d613 | 39 | |
d9bca0c3 WD |
40 | static uint32 tablesize; |
41 | static int32 *sum_table; | |
c627d613 | 42 | |
d9bca0c3 WD |
43 | #define gettag2(s1,s2) gettag((s1) + ((s2)<<16)) |
44 | #define gettag(sum) ((sum)%tablesize) | |
c627d613 AT |
45 | |
46 | static void build_hash_table(struct sum_struct *s) | |
47 | { | |
7cacd47e | 48 | int32 i; |
d9bca0c3 WD |
49 | uint32 prior_size = tablesize; |
50 | ||
51 | /* Dynamically calculate the hash table size so that the hash load | |
52 | * for big files is about 80%. This number must be odd or s2 will | |
53 | * not be able to span the entire set. */ | |
54 | tablesize = (uint32)(s->count/8) * 10 + 11; | |
55 | if (tablesize < 65537) | |
56 | tablesize = 65537; /* a prime number */ | |
57 | if (tablesize != prior_size) { | |
58 | free(sum_table); | |
59 | sum_table = new_array(int32, tablesize); | |
60 | if (!sum_table) | |
61 | out_of_memory("build_hash_table"); | |
62 | } | |
c627d613 | 63 | |
d9bca0c3 | 64 | memset(sum_table, 0xFF, tablesize * sizeof sum_table[0]); |
c627d613 | 65 | |
0e36d9da | 66 | for (i = 0; i < s->count; i++) { |
d9bca0c3 WD |
67 | uint32 t = gettag(s->sums[i].sum1); |
68 | s->sums[i].chain = sum_table[t]; | |
69 | sum_table[t] = i; | |
1e4f48d6 | 70 | } |
c627d613 AT |
71 | } |
72 | ||
73 | ||
bcacc18b | 74 | static OFF_T last_match; |
c627d613 AT |
75 | |
76 | ||
f5f95a38 MP |
77 | /** |
78 | * Transmit a literal and/or match token. | |
79 | * | |
80 | * This delightfully-named function is called either when we find a | |
81 | * match and need to transmit all the unmatched data leading up to it, | |
82 | * or when we get bored of accumulating literal data and just need to | |
83 | * transmit it. As a result of this second case, it is called even if | |
84 | * we have not matched at all! | |
85 | * | |
86 | * @param i If >0, the number of a matched token. If 0, indicates we | |
87 | * have only literal data. | |
88 | **/ | |
a06b419d | 89 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 90 | OFF_T offset, int32 i) |
c627d613 | 91 | { |
a06b419d WD |
92 | int32 n = offset - last_match; /* max value: block_size (int32) */ |
93 | int32 j; | |
94 | ||
95 | if (verbose > 2 && i >= 0) { | |
96 | rprintf(FINFO, | |
97 | "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n", | |
98 | (double)offset, (double)last_match, i, | |
99 | (long)s->sums[i].len, (long)n); | |
100 | } | |
c627d613 | 101 | |
a06b419d | 102 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 103 | data_transfer += n; |
70d794dc | 104 | |
a800434a AT |
105 | if (i >= 0) { |
106 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 107 | n += s->sums[i].len; |
a800434a | 108 | } |
1e4f48d6 WD |
109 | |
110 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
111 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
112 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 113 | } |
9e31c482 | 114 | |
0d0e2e93 | 115 | if (i >= 0) |
e7ebc36c | 116 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
117 | else |
118 | last_match = offset; | |
eb86d661 | 119 | |
0394e34a | 120 | if (buf && do_progress) |
4440b8aa | 121 | show_progress(last_match, buf->file_size); |
9e31c482 AT |
122 | } |
123 | ||
124 | ||
c6e7fcb4 | 125 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 126 | struct map_struct *buf, OFF_T len) |
c627d613 | 127 | { |
7560c17a | 128 | OFF_T offset, end, backup; |
7cacd47e | 129 | int32 k, want_i; |
e7ebc36c | 130 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 131 | uint32 s1, s2, sum; |
7560c17a | 132 | int more; |
debb4505 | 133 | schar *map; |
e7ebc36c | 134 | |
a04d77bc WD |
135 | /* want_i is used to encourage adjacent matches, allowing the RLL |
136 | * coding of the output to work more efficiently. */ | |
137 | want_i = 0; | |
923fa978 | 138 | |
0e36d9da | 139 | if (verbose > 2) { |
a06b419d WD |
140 | rprintf(FINFO, "hash search b=%ld len=%.0f\n", |
141 | (long)s->blength, (double)len); | |
0e36d9da | 142 | } |
e7ebc36c | 143 | |
a06b419d | 144 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 145 | |
0e36d9da | 146 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 147 | |
e7ebc36c AT |
148 | sum = get_checksum1((char *)map, k); |
149 | s1 = sum & 0xFFFF; | |
150 | s2 = sum >> 16; | |
151 | if (verbose > 3) | |
a06b419d | 152 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 153 | |
e7ebc36c | 154 | offset = 0; |
1e4f48d6 | 155 | |
e7ebc36c | 156 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 157 | |
0e36d9da | 158 | if (verbose > 3) { |
a06b419d WD |
159 | rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n", |
160 | (long)s->blength, (double)len, (double)s->count); | |
0e36d9da | 161 | } |
1e4f48d6 | 162 | |
e7ebc36c | 163 | do { |
d9bca0c3 | 164 | uint32 t = gettag2(s1,s2); |
e7ebc36c | 165 | int done_csum2 = 0; |
d9bca0c3 | 166 | int32 i; |
1e4f48d6 | 167 | |
e7ebc36c | 168 | if (verbose > 4) |
5f808dfb | 169 | rprintf(FINFO,"offset=%.0f sum=%08x\n",(double)offset,sum); |
1e4f48d6 | 170 | |
e7ebc36c AT |
171 | sum = (s1 & 0xffff) | (s2 << 16); |
172 | tag_hits++; | |
d9bca0c3 WD |
173 | for (i = sum_table[t]; i >= 0; i = s->sums[i].chain) { |
174 | int32 l; | |
1e4f48d6 | 175 | |
0e36d9da WD |
176 | if (sum != s->sums[i].sum1) |
177 | continue; | |
1e4f48d6 | 178 | |
536541d5 | 179 | /* also make sure the two blocks are the same length */ |
a06b419d | 180 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
181 | if (l != s->sums[i].len) |
182 | continue; | |
536541d5 | 183 | |
a36ffd39 | 184 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 185 | * offset or that the data didn't move. */ |
a36ffd39 | 186 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
187 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
188 | continue; | |
189 | ||
d9bca0c3 WD |
190 | if (verbose > 3) { |
191 | rprintf(FINFO, | |
192 | "potential match at %.0f i=%ld sum=%08x\n", | |
193 | (double)offset, (long)i, sum); | |
194 | } | |
1e4f48d6 | 195 | |
e7ebc36c | 196 | if (!done_csum2) { |
debb4505 | 197 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
198 | get_checksum2((char *)map,l,sum2); |
199 | done_csum2 = 1; | |
200 | } | |
1e4f48d6 | 201 | |
fc0257c9 | 202 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
203 | false_alarms++; |
204 | continue; | |
205 | } | |
923fa978 | 206 | |
a36ffd39 | 207 | /* When updating in-place, the best possible match is |
a3221d2a WD |
208 | * one with an identical offset, so we prefer that over |
209 | * the following want_i optimization. */ | |
a36ffd39 | 210 | if (updating_basis_file) { |
d9bca0c3 WD |
211 | int32 i2; |
212 | for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) { | |
a3221d2a WD |
213 | if (s->sums[i2].offset != offset) |
214 | continue; | |
215 | if (i2 != i) { | |
216 | if (sum != s->sums[i2].sum1) | |
217 | break; | |
218 | if (memcmp(sum2, s->sums[i2].sum2, | |
219 | s->s2length) != 0) | |
220 | break; | |
221 | i = i2; | |
222 | } | |
223 | /* This chunk was at the same offset on | |
224 | * both the sender and the receiver. */ | |
225 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
226 | goto set_want_i; | |
d9bca0c3 | 227 | } |
a3221d2a WD |
228 | } |
229 | ||
923fa978 | 230 | /* we've found a match, but now check to see |
a04d77bc WD |
231 | * if want_i can hint at a better match. */ |
232 | if (i != want_i && want_i < s->count | |
a36ffd39 | 233 | && (!updating_basis_file || s->sums[want_i].offset >= offset |
a3221d2a | 234 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
235 | && sum == s->sums[want_i].sum1 |
236 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
237 | /* we've found an adjacent match - the RLL coder |
238 | * will be happy */ | |
a04d77bc | 239 | i = want_i; |
923fa978 | 240 | } |
a3221d2a | 241 | set_want_i: |
a04d77bc | 242 | want_i = i + 1; |
1e4f48d6 | 243 | |
e7ebc36c AT |
244 | matched(f,s,buf,offset,i); |
245 | offset += s->sums[i].len - 1; | |
a06b419d | 246 | k = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da | 247 | map = (schar *)map_ptr(buf, offset, k); |
e7ebc36c AT |
248 | sum = get_checksum1((char *)map, k); |
249 | s1 = sum & 0xFFFF; | |
250 | s2 = sum >> 16; | |
251 | matches++; | |
252 | break; | |
d9bca0c3 | 253 | } |
1e4f48d6 | 254 | |
7560c17a WD |
255 | backup = offset - last_match; |
256 | /* We sometimes read 1 byte prior to last_match... */ | |
257 | if (backup < 0) | |
258 | backup = 0; | |
259 | ||
e7ebc36c | 260 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
261 | more = offset + k < len; |
262 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
263 | + backup; | |
e7ebc36c AT |
264 | s1 -= map[0] + CHAR_OFFSET; |
265 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 266 | |
e7ebc36c | 267 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
268 | if (more) { |
269 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 270 | s2 += s1; |
0e36d9da | 271 | } else |
e7ebc36c | 272 | --k; |
45f133b9 | 273 | |
5914bf15 PM |
274 | /* By matching early we avoid re-reading the |
275 | data 3 times in the case where a token | |
276 | match comes a long way after last | |
277 | match. The 3 reads are caused by the | |
278 | running match, the checksum update and the | |
279 | literal send. */ | |
a06b419d WD |
280 | if (backup >= s->blength+CHUNK_SIZE && end-offset > CHUNK_SIZE) |
281 | matched(f, s, buf, offset - s->blength, -2); | |
e7ebc36c | 282 | } while (++offset < end); |
1e4f48d6 | 283 | |
a06b419d WD |
284 | matched(f, s, buf, len, -1); |
285 | map_ptr(buf, len-1, 1); | |
c627d613 AT |
286 | } |
287 | ||
288 | ||
f5f95a38 MP |
289 | /** |
290 | * Scan through a origin file, looking for sections that match | |
291 | * checksums from the generator, and transmit either literal or token | |
292 | * data. | |
293 | * | |
538ba24f MP |
294 | * Also calculates the MD4 checksum of the whole file, using the md |
295 | * accumulator. This is transmitted with the file as protection | |
296 | * against corruption on the wire. | |
297 | * | |
f5f95a38 MP |
298 | * @param s Checksums received from the generator. If <tt>s->count == |
299 | * 0</tt>, then there are actually no checksums for this file. | |
300 | * | |
301 | * @param len Length of the file to send. | |
302 | **/ | |
303 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 304 | { |
e7ebc36c AT |
305 | char file_sum[MD4_SUM_LENGTH]; |
306 | ||
307 | last_match = 0; | |
308 | false_alarms = 0; | |
309 | tag_hits = 0; | |
1e4f48d6 WD |
310 | matches = 0; |
311 | data_transfer = 0; | |
e7ebc36c | 312 | |
ba582f75 | 313 | sum_init(checksum_seed); |
e7ebc36c | 314 | |
6cc11982 WD |
315 | if (append_mode) { |
316 | OFF_T j = 0; | |
317 | for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) { | |
1f86fcf5 WD |
318 | if (buf && do_progress) |
319 | show_progress(last_match, buf->file_size); | |
6cc11982 WD |
320 | sum_update(map_ptr(buf, last_match, CHUNK_SIZE), |
321 | CHUNK_SIZE); | |
322 | last_match = j; | |
323 | } | |
324 | if (last_match < s->flength) { | |
325 | int32 len = s->flength - last_match; | |
1f86fcf5 WD |
326 | if (buf && do_progress) |
327 | show_progress(last_match, buf->file_size); | |
6cc11982 WD |
328 | sum_update(map_ptr(buf, last_match, len), len); |
329 | last_match = s->flength; | |
330 | } | |
331 | s->count = 0; | |
332 | } | |
333 | ||
a06b419d | 334 | if (len > 0 && s->count > 0) { |
e7ebc36c | 335 | build_hash_table(s); |
1e4f48d6 WD |
336 | |
337 | if (verbose > 2) | |
9486289c | 338 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 339 | |
e7ebc36c | 340 | hash_search(f,s,buf,len); |
1e4f48d6 WD |
341 | |
342 | if (verbose > 2) | |
9486289c | 343 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 344 | } else { |
eb86d661 AT |
345 | OFF_T j; |
346 | /* by doing this in pieces we avoid too many seeks */ | |
6cc11982 | 347 | for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
a06b419d WD |
348 | matched(f, s, buf, j, -2); |
349 | matched(f, s, buf, len, -1); | |
e7ebc36c | 350 | } |
9e31c482 | 351 | |
e7ebc36c | 352 | sum_end(file_sum); |
edecdad5 WD |
353 | /* If we had a read error, send a bad checksum. */ |
354 | if (buf && buf->status != 0) | |
355 | file_sum[0]++; | |
c627d613 | 356 | |
bc63ae3f S |
357 | if (verbose > 2) |
358 | rprintf(FINFO,"sending file_sum\n"); | |
359 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
c627d613 | 360 | |
e7ebc36c | 361 | if (verbose > 2) |
9486289c | 362 | rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", |
e7ebc36c | 363 | false_alarms, tag_hits, matches); |
1e4f48d6 | 364 | |
e7ebc36c AT |
365 | total_tag_hits += tag_hits; |
366 | total_false_alarms += false_alarms; | |
367 | total_matches += matches; | |
a800434a | 368 | stats.literal_data += data_transfer; |
c627d613 AT |
369 | } |
370 | ||
371 | void match_report(void) | |
372 | { | |
e7ebc36c AT |
373 | if (verbose <= 1) |
374 | return; | |
c627d613 | 375 | |
9486289c | 376 | rprintf(FINFO, |
a800434a | 377 | "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", |
e7ebc36c | 378 | total_matches,total_tag_hits, |
a800434a AT |
379 | total_false_alarms, |
380 | (double)stats.literal_data); | |
c627d613 | 381 | } |