Commit | Line | Data |
---|---|---|
1e4f48d6 | 1 | /* |
0f78b815 WD |
2 | * Block matching used by the file-transfer code. |
3 | * | |
4 | * Copyright (C) 1996 Andrew Tridgell | |
5 | * Copyright (C) 1996 Paul Mackerras | |
d3d07a5e | 6 | * Copyright (C) 2003-2008 Wayne Davison |
0f78b815 WD |
7 | * |
8 | * This program is free software; you can redistribute it and/or modify | |
8e41b68e WD |
9 | * it under the terms of the GNU General Public License as published by |
10 | * the Free Software Foundation; either version 3 of the License, or | |
11 | * (at your option) any later version. | |
0f78b815 WD |
12 | * |
13 | * This program is distributed in the hope that it will be useful, | |
14 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
15 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
16 | * GNU General Public License for more details. | |
17 | * | |
e7c67065 | 18 | * You should have received a copy of the GNU General Public License along |
4fd842f9 | 19 | * with this program; if not, visit the http://fsf.org website. |
0f78b815 | 20 | */ |
c627d613 AT |
21 | |
22 | #include "rsync.h" | |
23 | ||
ba582f75 | 24 | extern int checksum_seed; |
6cc11982 | 25 | extern int append_mode; |
a36ffd39 WD |
26 | |
27 | int updating_basis_file; | |
c627d613 | 28 | |
c627d613 | 29 | static int false_alarms; |
900cfcb5 | 30 | static int hash_hits; |
c627d613 | 31 | static int matches; |
a800434a | 32 | static int64 data_transfer; |
c627d613 | 33 | |
3a6a366f | 34 | static int total_false_alarms; |
900cfcb5 | 35 | static int total_hash_hits; |
3a6a366f | 36 | static int total_matches; |
c627d613 | 37 | |
a800434a | 38 | extern struct stats stats; |
c627d613 | 39 | |
a2c770dc | 40 | #define TRADITIONAL_TABLESIZE (1<<16) |
48cce779 | 41 | |
a2c770dc | 42 | static uint32 tablesize; |
68207537 | 43 | static int32 *hash_table; |
c627d613 | 44 | |
48cce779 WD |
45 | #define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF) |
46 | #define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16) | |
c627d613 | 47 | |
a2c770dc | 48 | #define BIG_SUM2HASH(sum) ((sum)%tablesize) |
48cce779 | 49 | |
a2c770dc WD |
50 | static void build_hash_table(struct sum_struct *s) |
51 | { | |
52 | static uint32 alloc_size; | |
53 | int32 i; | |
54 | ||
55 | /* Dynamically calculate the hash table size so that the hash load | |
56 | * for big files is about 80%. A number greater than the traditional | |
57 | * size must be odd or s2 will not be able to span the entire set. */ | |
58 | tablesize = (uint32)(s->count/8) * 10 + 11; | |
59 | if (tablesize < TRADITIONAL_TABLESIZE) | |
60 | tablesize = TRADITIONAL_TABLESIZE; | |
61 | if (tablesize > alloc_size || tablesize < alloc_size - 16*1024) { | |
62 | if (hash_table) | |
63 | free(hash_table); | |
64 | hash_table = new_array(int32, tablesize); | |
68207537 | 65 | if (!hash_table) |
d9bca0c3 | 66 | out_of_memory("build_hash_table"); |
a2c770dc | 67 | alloc_size = tablesize; |
d9bca0c3 | 68 | } |
c627d613 | 69 | |
a2c770dc | 70 | memset(hash_table, 0xFF, tablesize * sizeof hash_table[0]); |
4c17cdcb | 71 | |
a2c770dc WD |
72 | if (tablesize == TRADITIONAL_TABLESIZE) { |
73 | for (i = 0; i < s->count; i++) { | |
74 | uint32 t = SUM2HASH(s->sums[i].sum1); | |
75 | s->sums[i].chain = hash_table[t]; | |
76 | hash_table[t] = i; | |
77 | } | |
78 | } else { | |
79 | for (i = 0; i < s->count; i++) { | |
80 | uint32 t = BIG_SUM2HASH(s->sums[i].sum1); | |
81 | s->sums[i].chain = hash_table[t]; | |
82 | hash_table[t] = i; | |
83 | } | |
4c17cdcb | 84 | } |
c627d613 AT |
85 | } |
86 | ||
87 | ||
bcacc18b | 88 | static OFF_T last_match; |
c627d613 AT |
89 | |
90 | ||
f5f95a38 MP |
91 | /** |
92 | * Transmit a literal and/or match token. | |
93 | * | |
94 | * This delightfully-named function is called either when we find a | |
95 | * match and need to transmit all the unmatched data leading up to it, | |
96 | * or when we get bored of accumulating literal data and just need to | |
97 | * transmit it. As a result of this second case, it is called even if | |
98 | * we have not matched at all! | |
99 | * | |
100 | * @param i If >0, the number of a matched token. If 0, indicates we | |
101 | * have only literal data. | |
102 | **/ | |
a06b419d | 103 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 104 | OFF_T offset, int32 i) |
c627d613 | 105 | { |
26404276 | 106 | int32 n = (int32)(offset - last_match); /* max value: block_size (int32) */ |
a06b419d WD |
107 | int32 j; |
108 | ||
951e826b | 109 | if (DEBUG_GTE(CHKSUM, 2) && i >= 0) { |
a06b419d | 110 | rprintf(FINFO, |
6d56efa6 WD |
111 | "match at %s last_match=%s j=%d len=%ld n=%ld\n", |
112 | big_num(offset, 0), big_num(last_match, 0), i, | |
a06b419d WD |
113 | (long)s->sums[i].len, (long)n); |
114 | } | |
c627d613 | 115 | |
a06b419d | 116 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 117 | data_transfer += n; |
70d794dc | 118 | |
a800434a AT |
119 | if (i >= 0) { |
120 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 121 | n += s->sums[i].len; |
a800434a | 122 | } |
1e4f48d6 WD |
123 | |
124 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
125 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
126 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 127 | } |
9e31c482 | 128 | |
0d0e2e93 | 129 | if (i >= 0) |
e7ebc36c | 130 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
131 | else |
132 | last_match = offset; | |
eb86d661 | 133 | |
951e826b | 134 | if (buf && INFO_GTE(PROGRESS, 1)) |
4440b8aa | 135 | show_progress(last_match, buf->file_size); |
9e31c482 AT |
136 | } |
137 | ||
138 | ||
c6e7fcb4 | 139 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 140 | struct map_struct *buf, OFF_T len) |
c627d613 | 141 | { |
a2c770dc WD |
142 | OFF_T offset, end; |
143 | int32 k, want_i, backup; | |
e7ebc36c | 144 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 145 | uint32 s1, s2, sum; |
7560c17a | 146 | int more; |
debb4505 | 147 | schar *map; |
e7ebc36c | 148 | |
a04d77bc WD |
149 | /* want_i is used to encourage adjacent matches, allowing the RLL |
150 | * coding of the output to work more efficiently. */ | |
151 | want_i = 0; | |
923fa978 | 152 | |
951e826b | 153 | if (DEBUG_GTE(CHKSUM, 2)) { |
6d56efa6 WD |
154 | rprintf(FINFO, "hash search b=%ld len=%s\n", |
155 | (long)s->blength, big_num(len, 0)); | |
0e36d9da | 156 | } |
e7ebc36c | 157 | |
a06b419d | 158 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 159 | |
0e36d9da | 160 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 161 | |
e7ebc36c AT |
162 | sum = get_checksum1((char *)map, k); |
163 | s1 = sum & 0xFFFF; | |
164 | s2 = sum >> 16; | |
951e826b | 165 | if (DEBUG_GTE(CHKSUM, 3)) |
a06b419d | 166 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 167 | |
e7ebc36c | 168 | offset = 0; |
1e4f48d6 | 169 | |
e7ebc36c | 170 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 171 | |
951e826b | 172 | if (DEBUG_GTE(CHKSUM, 3)) { |
6d56efa6 WD |
173 | rprintf(FINFO, "hash search s->blength=%ld len=%s count=%s\n", |
174 | (long)s->blength, big_num(len, 0), big_num(s->count, 0)); | |
0e36d9da | 175 | } |
1e4f48d6 | 176 | |
e7ebc36c | 177 | do { |
e7ebc36c | 178 | int done_csum2 = 0; |
d9bca0c3 | 179 | int32 i; |
1e4f48d6 | 180 | |
951e826b | 181 | if (DEBUG_GTE(CHKSUM, 4)) { |
6d56efa6 WD |
182 | rprintf(FINFO, "offset=%s sum=%04x%04x\n", |
183 | big_num(offset, 0), s2 & 0xFFFF, s1 & 0xFFFF); | |
48cce779 | 184 | } |
1e4f48d6 | 185 | |
a2c770dc WD |
186 | if (tablesize == TRADITIONAL_TABLESIZE) { |
187 | if ((i = hash_table[SUM2HASH2(s1,s2)]) < 0) | |
188 | goto null_hash; | |
189 | sum = (s1 & 0xffff) | (s2 << 16); | |
190 | } else { | |
191 | sum = (s1 & 0xffff) | (s2 << 16); | |
192 | if ((i = hash_table[BIG_SUM2HASH(sum)]) < 0) | |
193 | goto null_hash; | |
194 | } | |
900cfcb5 WD |
195 | |
196 | hash_hits++; | |
197 | do { | |
d9bca0c3 | 198 | int32 l; |
1e4f48d6 | 199 | |
0e36d9da WD |
200 | if (sum != s->sums[i].sum1) |
201 | continue; | |
1e4f48d6 | 202 | |
536541d5 | 203 | /* also make sure the two blocks are the same length */ |
a06b419d | 204 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
205 | if (l != s->sums[i].len) |
206 | continue; | |
536541d5 | 207 | |
a36ffd39 | 208 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 209 | * offset or that the data didn't move. */ |
a36ffd39 | 210 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
211 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
212 | continue; | |
213 | ||
951e826b | 214 | if (DEBUG_GTE(CHKSUM, 3)) { |
d9bca0c3 | 215 | rprintf(FINFO, |
6d56efa6 WD |
216 | "potential match at %s i=%ld sum=%08x\n", |
217 | big_num(offset, 0), (long)i, sum); | |
d9bca0c3 | 218 | } |
1e4f48d6 | 219 | |
e7ebc36c | 220 | if (!done_csum2) { |
debb4505 | 221 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
222 | get_checksum2((char *)map,l,sum2); |
223 | done_csum2 = 1; | |
224 | } | |
1e4f48d6 | 225 | |
fc0257c9 | 226 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
227 | false_alarms++; |
228 | continue; | |
229 | } | |
923fa978 | 230 | |
a36ffd39 | 231 | /* When updating in-place, the best possible match is |
a3221d2a WD |
232 | * one with an identical offset, so we prefer that over |
233 | * the following want_i optimization. */ | |
a36ffd39 | 234 | if (updating_basis_file) { |
d9bca0c3 WD |
235 | int32 i2; |
236 | for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) { | |
a3221d2a WD |
237 | if (s->sums[i2].offset != offset) |
238 | continue; | |
239 | if (i2 != i) { | |
240 | if (sum != s->sums[i2].sum1) | |
241 | break; | |
242 | if (memcmp(sum2, s->sums[i2].sum2, | |
243 | s->s2length) != 0) | |
244 | break; | |
245 | i = i2; | |
246 | } | |
247 | /* This chunk was at the same offset on | |
248 | * both the sender and the receiver. */ | |
249 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
250 | goto set_want_i; | |
d9bca0c3 | 251 | } |
a3221d2a WD |
252 | } |
253 | ||
923fa978 | 254 | /* we've found a match, but now check to see |
a04d77bc WD |
255 | * if want_i can hint at a better match. */ |
256 | if (i != want_i && want_i < s->count | |
a36ffd39 | 257 | && (!updating_basis_file || s->sums[want_i].offset >= offset |
a3221d2a | 258 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
259 | && sum == s->sums[want_i].sum1 |
260 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
261 | /* we've found an adjacent match - the RLL coder |
262 | * will be happy */ | |
a04d77bc | 263 | i = want_i; |
923fa978 | 264 | } |
a3221d2a | 265 | set_want_i: |
a04d77bc | 266 | want_i = i + 1; |
1e4f48d6 | 267 | |
e7ebc36c AT |
268 | matched(f,s,buf,offset,i); |
269 | offset += s->sums[i].len - 1; | |
a06b419d | 270 | k = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da | 271 | map = (schar *)map_ptr(buf, offset, k); |
e7ebc36c AT |
272 | sum = get_checksum1((char *)map, k); |
273 | s1 = sum & 0xFFFF; | |
274 | s2 = sum >> 16; | |
275 | matches++; | |
276 | break; | |
900cfcb5 | 277 | } while ((i = s->sums[i].chain) >= 0); |
1e4f48d6 | 278 | |
900cfcb5 | 279 | null_hash: |
26404276 | 280 | backup = (int32)(offset - last_match); |
7560c17a WD |
281 | /* We sometimes read 1 byte prior to last_match... */ |
282 | if (backup < 0) | |
283 | backup = 0; | |
284 | ||
e7ebc36c | 285 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
286 | more = offset + k < len; |
287 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
288 | + backup; | |
e7ebc36c AT |
289 | s1 -= map[0] + CHAR_OFFSET; |
290 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 291 | |
e7ebc36c | 292 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
293 | if (more) { |
294 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 295 | s2 += s1; |
0e36d9da | 296 | } else |
e7ebc36c | 297 | --k; |
45f133b9 | 298 | |
5914bf15 PM |
299 | /* By matching early we avoid re-reading the |
300 | data 3 times in the case where a token | |
301 | match comes a long way after last | |
302 | match. The 3 reads are caused by the | |
303 | running match, the checksum update and the | |
304 | literal send. */ | |
a06b419d WD |
305 | if (backup >= s->blength+CHUNK_SIZE && end-offset > CHUNK_SIZE) |
306 | matched(f, s, buf, offset - s->blength, -2); | |
e7ebc36c | 307 | } while (++offset < end); |
1e4f48d6 | 308 | |
a06b419d WD |
309 | matched(f, s, buf, len, -1); |
310 | map_ptr(buf, len-1, 1); | |
c627d613 AT |
311 | } |
312 | ||
313 | ||
f5f95a38 MP |
314 | /** |
315 | * Scan through a origin file, looking for sections that match | |
316 | * checksums from the generator, and transmit either literal or token | |
317 | * data. | |
318 | * | |
538ba24f MP |
319 | * Also calculates the MD4 checksum of the whole file, using the md |
320 | * accumulator. This is transmitted with the file as protection | |
321 | * against corruption on the wire. | |
322 | * | |
f5f95a38 MP |
323 | * @param s Checksums received from the generator. If <tt>s->count == |
324 | * 0</tt>, then there are actually no checksums for this file. | |
325 | * | |
326 | * @param len Length of the file to send. | |
327 | **/ | |
328 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 329 | { |
a0456b9c WD |
330 | char file_sum[MAX_DIGEST_LEN]; |
331 | int sum_len; | |
e7ebc36c AT |
332 | |
333 | last_match = 0; | |
334 | false_alarms = 0; | |
900cfcb5 | 335 | hash_hits = 0; |
1e4f48d6 WD |
336 | matches = 0; |
337 | data_transfer = 0; | |
e7ebc36c | 338 | |
ba582f75 | 339 | sum_init(checksum_seed); |
e7ebc36c | 340 | |
edb97721 | 341 | if (append_mode > 0) { |
936fa865 WD |
342 | if (append_mode == 2) { |
343 | OFF_T j = 0; | |
344 | for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) { | |
951e826b | 345 | if (buf && INFO_GTE(PROGRESS, 1)) |
936fa865 WD |
346 | show_progress(last_match, buf->file_size); |
347 | sum_update(map_ptr(buf, last_match, CHUNK_SIZE), | |
348 | CHUNK_SIZE); | |
349 | last_match = j; | |
350 | } | |
351 | if (last_match < s->flength) { | |
352 | int32 n = (int32)(s->flength - last_match); | |
951e826b | 353 | if (buf && INFO_GTE(PROGRESS, 1)) |
936fa865 WD |
354 | show_progress(last_match, buf->file_size); |
355 | sum_update(map_ptr(buf, last_match, n), n); | |
356 | } | |
6cc11982 | 357 | } |
936fa865 | 358 | last_match = s->flength; |
6cc11982 WD |
359 | s->count = 0; |
360 | } | |
361 | ||
a06b419d | 362 | if (len > 0 && s->count > 0) { |
a2c770dc WD |
363 | build_hash_table(s); |
364 | ||
951e826b | 365 | if (DEBUG_GTE(CHKSUM, 2)) |
a2c770dc WD |
366 | rprintf(FINFO,"built hash table\n"); |
367 | ||
a0456b9c | 368 | hash_search(f, s, buf, len); |
1e4f48d6 | 369 | |
951e826b | 370 | if (DEBUG_GTE(CHKSUM, 2)) |
9486289c | 371 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 372 | } else { |
eb86d661 AT |
373 | OFF_T j; |
374 | /* by doing this in pieces we avoid too many seeks */ | |
6cc11982 | 375 | for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
a06b419d WD |
376 | matched(f, s, buf, j, -2); |
377 | matched(f, s, buf, len, -1); | |
e7ebc36c | 378 | } |
9e31c482 | 379 | |
a0456b9c | 380 | sum_len = sum_end(file_sum); |
edecdad5 WD |
381 | /* If we had a read error, send a bad checksum. */ |
382 | if (buf && buf->status != 0) | |
383 | file_sum[0]++; | |
c627d613 | 384 | |
951e826b | 385 | if (DEBUG_GTE(CHKSUM, 2)) |
bc63ae3f | 386 | rprintf(FINFO,"sending file_sum\n"); |
a0456b9c | 387 | write_buf(f, file_sum, sum_len); |
c627d613 | 388 | |
951e826b | 389 | if (DEBUG_GTE(CHKSUM, 2)) |
900cfcb5 WD |
390 | rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n", |
391 | false_alarms, hash_hits, matches); | |
1e4f48d6 | 392 | |
900cfcb5 | 393 | total_hash_hits += hash_hits; |
e7ebc36c AT |
394 | total_false_alarms += false_alarms; |
395 | total_matches += matches; | |
a800434a | 396 | stats.literal_data += data_transfer; |
c627d613 AT |
397 | } |
398 | ||
399 | void match_report(void) | |
400 | { | |
951e826b | 401 | if (!DEBUG_GTE(CHKSUM, 1)) |
e7ebc36c | 402 | return; |
c627d613 | 403 | |
9486289c | 404 | rprintf(FINFO, |
6d56efa6 | 405 | "total: matches=%d hash_hits=%d false_alarms=%d data=%s\n", |
900cfcb5 | 406 | total_matches, total_hash_hits, total_false_alarms, |
6d56efa6 | 407 | big_num(stats.literal_data, 0)); |
c627d613 | 408 | } |