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 | |
ba2133d6 | 6 | * Copyright (C) 2003-2007 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 | ||
24 | extern int verbose; | |
16417f8b | 25 | extern int do_progress; |
ba582f75 | 26 | extern int checksum_seed; |
6cc11982 | 27 | extern int append_mode; |
a36ffd39 WD |
28 | |
29 | int updating_basis_file; | |
c627d613 | 30 | |
c627d613 | 31 | static int false_alarms; |
900cfcb5 | 32 | static int hash_hits; |
c627d613 | 33 | static int matches; |
a800434a | 34 | static int64 data_transfer; |
c627d613 | 35 | |
3a6a366f | 36 | static int total_false_alarms; |
900cfcb5 | 37 | static int total_hash_hits; |
3a6a366f | 38 | static int total_matches; |
c627d613 | 39 | |
a800434a | 40 | extern struct stats stats; |
c627d613 | 41 | |
48cce779 WD |
42 | #define TABLESIZE (1<<16) |
43 | ||
68207537 | 44 | static int32 *hash_table; |
c627d613 | 45 | |
48cce779 WD |
46 | #define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF) |
47 | #define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16) | |
c627d613 | 48 | |
4c17cdcb | 49 | static int32 build_hash_table(struct sum_struct *s, int32 start) |
c627d613 | 50 | { |
4c17cdcb | 51 | int32 i, end = s->count; |
48cce779 WD |
52 | |
53 | if (!hash_table) { | |
54 | hash_table = new_array(int32, TABLESIZE); | |
68207537 | 55 | if (!hash_table) |
d9bca0c3 WD |
56 | out_of_memory("build_hash_table"); |
57 | } | |
c627d613 | 58 | |
48cce779 | 59 | memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]); |
c627d613 | 60 | |
4c17cdcb WD |
61 | if (end - start > TABLESIZE*8/10) |
62 | end = start + TABLESIZE*8/10; | |
63 | ||
64 | for (i = start; i < end; i++) { | |
fbbe9a01 | 65 | uint32 t = SUM2HASH(s->sums[i].sum1); |
68207537 WD |
66 | s->sums[i].chain = hash_table[t]; |
67 | hash_table[t] = i; | |
1e4f48d6 | 68 | } |
4c17cdcb WD |
69 | |
70 | if (verbose > 2) { | |
71 | rprintf(FINFO, "built hash table for entries %ld - %ld\n", | |
72 | (long)start, (long)end - 1); | |
73 | } | |
74 | ||
75 | return end; | |
c627d613 AT |
76 | } |
77 | ||
78 | ||
bcacc18b | 79 | static OFF_T last_match; |
c627d613 AT |
80 | |
81 | ||
f5f95a38 MP |
82 | /** |
83 | * Transmit a literal and/or match token. | |
84 | * | |
85 | * This delightfully-named function is called either when we find a | |
86 | * match and need to transmit all the unmatched data leading up to it, | |
87 | * or when we get bored of accumulating literal data and just need to | |
88 | * transmit it. As a result of this second case, it is called even if | |
89 | * we have not matched at all! | |
90 | * | |
91 | * @param i If >0, the number of a matched token. If 0, indicates we | |
92 | * have only literal data. | |
93 | **/ | |
a06b419d | 94 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 95 | OFF_T offset, int32 i) |
c627d613 | 96 | { |
26404276 | 97 | int32 n = (int32)(offset - last_match); /* max value: block_size (int32) */ |
a06b419d WD |
98 | int32 j; |
99 | ||
100 | if (verbose > 2 && i >= 0) { | |
101 | rprintf(FINFO, | |
102 | "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n", | |
103 | (double)offset, (double)last_match, i, | |
104 | (long)s->sums[i].len, (long)n); | |
105 | } | |
c627d613 | 106 | |
a06b419d | 107 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 108 | data_transfer += n; |
70d794dc | 109 | |
a800434a AT |
110 | if (i >= 0) { |
111 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 112 | n += s->sums[i].len; |
a800434a | 113 | } |
1e4f48d6 WD |
114 | |
115 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
116 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
117 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 118 | } |
9e31c482 | 119 | |
0d0e2e93 | 120 | if (i >= 0) |
e7ebc36c | 121 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
122 | else |
123 | last_match = offset; | |
eb86d661 | 124 | |
0394e34a | 125 | if (buf && do_progress) |
4440b8aa | 126 | show_progress(last_match, buf->file_size); |
9e31c482 AT |
127 | } |
128 | ||
129 | ||
c6e7fcb4 | 130 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 131 | struct map_struct *buf, OFF_T len) |
c627d613 | 132 | { |
4c17cdcb WD |
133 | OFF_T offset, end, reset = 0; |
134 | int32 k, want_i, backup, sum_pos = 0; | |
e7ebc36c | 135 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 136 | uint32 s1, s2, sum; |
7560c17a | 137 | int more; |
debb4505 | 138 | schar *map; |
e7ebc36c | 139 | |
a04d77bc WD |
140 | /* want_i is used to encourage adjacent matches, allowing the RLL |
141 | * coding of the output to work more efficiently. */ | |
142 | want_i = 0; | |
923fa978 | 143 | |
0e36d9da | 144 | if (verbose > 2) { |
a06b419d WD |
145 | rprintf(FINFO, "hash search b=%ld len=%.0f\n", |
146 | (long)s->blength, (double)len); | |
0e36d9da | 147 | } |
e7ebc36c | 148 | |
a06b419d | 149 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 150 | |
0e36d9da | 151 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 152 | |
e7ebc36c AT |
153 | sum = get_checksum1((char *)map, k); |
154 | s1 = sum & 0xFFFF; | |
155 | s2 = sum >> 16; | |
156 | if (verbose > 3) | |
a06b419d | 157 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 158 | |
e7ebc36c | 159 | offset = 0; |
1e4f48d6 | 160 | |
e7ebc36c | 161 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 162 | |
0e36d9da | 163 | if (verbose > 3) { |
a06b419d WD |
164 | rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n", |
165 | (long)s->blength, (double)len, (double)s->count); | |
0e36d9da | 166 | } |
1e4f48d6 | 167 | |
e7ebc36c | 168 | do { |
e7ebc36c | 169 | int done_csum2 = 0; |
d9bca0c3 | 170 | int32 i; |
1e4f48d6 | 171 | |
4c17cdcb WD |
172 | if (offset >= reset) { |
173 | sum_pos = build_hash_table(s, sum_pos); | |
5851ac2d WD |
174 | if (sum_pos == s->count) |
175 | reset = len; | |
176 | else | |
177 | reset = sum_pos * s->blength; | |
4c17cdcb WD |
178 | } |
179 | ||
48cce779 WD |
180 | if (verbose > 4) { |
181 | rprintf(FINFO, "offset=%.0f sum=%04x%04x\n", | |
34f961bb | 182 | (double)offset, s2 & 0xFFFF, s1 & 0xFFFF); |
48cce779 | 183 | } |
1e4f48d6 | 184 | |
48cce779 | 185 | i = hash_table[SUM2HASH2(s1,s2)]; |
900cfcb5 WD |
186 | if (i < 0) |
187 | goto null_hash; | |
188 | ||
48cce779 | 189 | sum = (s1 & 0xffff) | (s2 << 16); |
900cfcb5 WD |
190 | hash_hits++; |
191 | do { | |
d9bca0c3 | 192 | int32 l; |
1e4f48d6 | 193 | |
0e36d9da WD |
194 | if (sum != s->sums[i].sum1) |
195 | continue; | |
1e4f48d6 | 196 | |
536541d5 | 197 | /* also make sure the two blocks are the same length */ |
a06b419d | 198 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
199 | if (l != s->sums[i].len) |
200 | continue; | |
536541d5 | 201 | |
a36ffd39 | 202 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 203 | * offset or that the data didn't move. */ |
a36ffd39 | 204 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
205 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
206 | continue; | |
207 | ||
d9bca0c3 WD |
208 | if (verbose > 3) { |
209 | rprintf(FINFO, | |
210 | "potential match at %.0f i=%ld sum=%08x\n", | |
211 | (double)offset, (long)i, sum); | |
212 | } | |
1e4f48d6 | 213 | |
e7ebc36c | 214 | if (!done_csum2) { |
debb4505 | 215 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
216 | get_checksum2((char *)map,l,sum2); |
217 | done_csum2 = 1; | |
218 | } | |
1e4f48d6 | 219 | |
fc0257c9 | 220 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
221 | false_alarms++; |
222 | continue; | |
223 | } | |
923fa978 | 224 | |
a36ffd39 | 225 | /* When updating in-place, the best possible match is |
a3221d2a WD |
226 | * one with an identical offset, so we prefer that over |
227 | * the following want_i optimization. */ | |
a36ffd39 | 228 | if (updating_basis_file) { |
d9bca0c3 WD |
229 | int32 i2; |
230 | for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) { | |
a3221d2a WD |
231 | if (s->sums[i2].offset != offset) |
232 | continue; | |
233 | if (i2 != i) { | |
234 | if (sum != s->sums[i2].sum1) | |
235 | break; | |
236 | if (memcmp(sum2, s->sums[i2].sum2, | |
237 | s->s2length) != 0) | |
238 | break; | |
239 | i = i2; | |
240 | } | |
241 | /* This chunk was at the same offset on | |
242 | * both the sender and the receiver. */ | |
243 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
244 | goto set_want_i; | |
d9bca0c3 | 245 | } |
a3221d2a WD |
246 | } |
247 | ||
923fa978 | 248 | /* we've found a match, but now check to see |
a04d77bc WD |
249 | * if want_i can hint at a better match. */ |
250 | if (i != want_i && want_i < s->count | |
a36ffd39 | 251 | && (!updating_basis_file || s->sums[want_i].offset >= offset |
a3221d2a | 252 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
253 | && sum == s->sums[want_i].sum1 |
254 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
255 | /* we've found an adjacent match - the RLL coder |
256 | * will be happy */ | |
a04d77bc | 257 | i = want_i; |
923fa978 | 258 | } |
a3221d2a | 259 | set_want_i: |
a04d77bc | 260 | want_i = i + 1; |
1e4f48d6 | 261 | |
e7ebc36c AT |
262 | matched(f,s,buf,offset,i); |
263 | offset += s->sums[i].len - 1; | |
a06b419d | 264 | k = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da | 265 | map = (schar *)map_ptr(buf, offset, k); |
e7ebc36c AT |
266 | sum = get_checksum1((char *)map, k); |
267 | s1 = sum & 0xFFFF; | |
268 | s2 = sum >> 16; | |
269 | matches++; | |
270 | break; | |
900cfcb5 | 271 | } while ((i = s->sums[i].chain) >= 0); |
1e4f48d6 | 272 | |
900cfcb5 | 273 | null_hash: |
26404276 | 274 | backup = (int32)(offset - last_match); |
7560c17a WD |
275 | /* We sometimes read 1 byte prior to last_match... */ |
276 | if (backup < 0) | |
277 | backup = 0; | |
278 | ||
e7ebc36c | 279 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
280 | more = offset + k < len; |
281 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
282 | + backup; | |
e7ebc36c AT |
283 | s1 -= map[0] + CHAR_OFFSET; |
284 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 285 | |
e7ebc36c | 286 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
287 | if (more) { |
288 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 289 | s2 += s1; |
0e36d9da | 290 | } else |
e7ebc36c | 291 | --k; |
45f133b9 | 292 | |
5914bf15 PM |
293 | /* By matching early we avoid re-reading the |
294 | data 3 times in the case where a token | |
295 | match comes a long way after last | |
296 | match. The 3 reads are caused by the | |
297 | running match, the checksum update and the | |
298 | literal send. */ | |
a06b419d WD |
299 | if (backup >= s->blength+CHUNK_SIZE && end-offset > CHUNK_SIZE) |
300 | matched(f, s, buf, offset - s->blength, -2); | |
e7ebc36c | 301 | } while (++offset < end); |
1e4f48d6 | 302 | |
a06b419d WD |
303 | matched(f, s, buf, len, -1); |
304 | map_ptr(buf, len-1, 1); | |
c627d613 AT |
305 | } |
306 | ||
307 | ||
f5f95a38 MP |
308 | /** |
309 | * Scan through a origin file, looking for sections that match | |
310 | * checksums from the generator, and transmit either literal or token | |
311 | * data. | |
312 | * | |
538ba24f MP |
313 | * Also calculates the MD4 checksum of the whole file, using the md |
314 | * accumulator. This is transmitted with the file as protection | |
315 | * against corruption on the wire. | |
316 | * | |
f5f95a38 MP |
317 | * @param s Checksums received from the generator. If <tt>s->count == |
318 | * 0</tt>, then there are actually no checksums for this file. | |
319 | * | |
320 | * @param len Length of the file to send. | |
321 | **/ | |
322 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 323 | { |
a0456b9c WD |
324 | char file_sum[MAX_DIGEST_LEN]; |
325 | int sum_len; | |
e7ebc36c AT |
326 | |
327 | last_match = 0; | |
328 | false_alarms = 0; | |
900cfcb5 | 329 | hash_hits = 0; |
1e4f48d6 WD |
330 | matches = 0; |
331 | data_transfer = 0; | |
e7ebc36c | 332 | |
ba582f75 | 333 | sum_init(checksum_seed); |
e7ebc36c | 334 | |
edb97721 | 335 | if (append_mode > 0) { |
936fa865 WD |
336 | if (append_mode == 2) { |
337 | OFF_T j = 0; | |
338 | for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) { | |
339 | if (buf && do_progress) | |
340 | show_progress(last_match, buf->file_size); | |
341 | sum_update(map_ptr(buf, last_match, CHUNK_SIZE), | |
342 | CHUNK_SIZE); | |
343 | last_match = j; | |
344 | } | |
345 | if (last_match < s->flength) { | |
346 | int32 n = (int32)(s->flength - last_match); | |
347 | if (buf && do_progress) | |
348 | show_progress(last_match, buf->file_size); | |
349 | sum_update(map_ptr(buf, last_match, n), n); | |
350 | } | |
6cc11982 | 351 | } |
936fa865 | 352 | last_match = s->flength; |
6cc11982 WD |
353 | s->count = 0; |
354 | } | |
355 | ||
a06b419d | 356 | if (len > 0 && s->count > 0) { |
a0456b9c | 357 | hash_search(f, s, buf, len); |
1e4f48d6 WD |
358 | |
359 | if (verbose > 2) | |
9486289c | 360 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 361 | } else { |
eb86d661 AT |
362 | OFF_T j; |
363 | /* by doing this in pieces we avoid too many seeks */ | |
6cc11982 | 364 | for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
a06b419d WD |
365 | matched(f, s, buf, j, -2); |
366 | matched(f, s, buf, len, -1); | |
e7ebc36c | 367 | } |
9e31c482 | 368 | |
a0456b9c | 369 | sum_len = sum_end(file_sum); |
edecdad5 WD |
370 | /* If we had a read error, send a bad checksum. */ |
371 | if (buf && buf->status != 0) | |
372 | file_sum[0]++; | |
c627d613 | 373 | |
bc63ae3f S |
374 | if (verbose > 2) |
375 | rprintf(FINFO,"sending file_sum\n"); | |
a0456b9c | 376 | write_buf(f, file_sum, sum_len); |
c627d613 | 377 | |
e7ebc36c | 378 | if (verbose > 2) |
900cfcb5 WD |
379 | rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n", |
380 | false_alarms, hash_hits, matches); | |
1e4f48d6 | 381 | |
900cfcb5 | 382 | total_hash_hits += hash_hits; |
e7ebc36c AT |
383 | total_false_alarms += false_alarms; |
384 | total_matches += matches; | |
a800434a | 385 | stats.literal_data += data_transfer; |
c627d613 AT |
386 | } |
387 | ||
388 | void match_report(void) | |
389 | { | |
e7ebc36c AT |
390 | if (verbose <= 1) |
391 | return; | |
c627d613 | 392 | |
9486289c | 393 | rprintf(FINFO, |
900cfcb5 WD |
394 | "total: matches=%d hash_hits=%d false_alarms=%d data=%.0f\n", |
395 | total_matches, total_hash_hits, total_false_alarms, | |
a800434a | 396 | (double)stats.literal_data); |
c627d613 | 397 | } |