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 | |
6 | * Copyright (C) 2003, 2004, 2005, 2006 Wayne Davison | |
7 | * | |
8 | * This program is free software; you can redistribute it and/or modify | |
9 | * it under the terms of the GNU General Public License as published by | |
10 | * the Free Software Foundation; either version 2 of the License, or | |
11 | * (at your option) any later version. | |
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 WD |
18 | * You should have received a copy of the GNU General Public License along |
19 | * with this program; if not, write to the Free Software Foundation, Inc., | |
20 | * 51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA. | |
0f78b815 | 21 | */ |
c627d613 AT |
22 | |
23 | #include "rsync.h" | |
24 | ||
25 | extern int verbose; | |
16417f8b | 26 | extern int do_progress; |
ba582f75 | 27 | extern int checksum_seed; |
6cc11982 | 28 | extern int append_mode; |
a36ffd39 WD |
29 | |
30 | int updating_basis_file; | |
c627d613 | 31 | |
c627d613 | 32 | static int false_alarms; |
900cfcb5 | 33 | static int hash_hits; |
c627d613 | 34 | static int matches; |
a800434a | 35 | static int64 data_transfer; |
c627d613 | 36 | |
3a6a366f | 37 | static int total_false_alarms; |
900cfcb5 | 38 | static int total_hash_hits; |
3a6a366f | 39 | static int total_matches; |
c627d613 | 40 | |
a800434a | 41 | extern struct stats stats; |
c627d613 | 42 | |
48cce779 WD |
43 | #define TABLESIZE (1<<16) |
44 | ||
68207537 | 45 | static int32 *hash_table; |
c627d613 | 46 | |
48cce779 WD |
47 | #define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF) |
48 | #define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16) | |
c627d613 AT |
49 | |
50 | static void build_hash_table(struct sum_struct *s) | |
51 | { | |
7cacd47e | 52 | int32 i; |
48cce779 WD |
53 | |
54 | if (!hash_table) { | |
55 | hash_table = new_array(int32, TABLESIZE); | |
68207537 | 56 | if (!hash_table) |
d9bca0c3 WD |
57 | out_of_memory("build_hash_table"); |
58 | } | |
c627d613 | 59 | |
48cce779 | 60 | memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]); |
c627d613 | 61 | |
0e36d9da | 62 | for (i = 0; i < s->count; i++) { |
fbbe9a01 | 63 | uint32 t = SUM2HASH(s->sums[i].sum1); |
68207537 WD |
64 | s->sums[i].chain = hash_table[t]; |
65 | hash_table[t] = i; | |
1e4f48d6 | 66 | } |
c627d613 AT |
67 | } |
68 | ||
69 | ||
bcacc18b | 70 | static OFF_T last_match; |
c627d613 AT |
71 | |
72 | ||
f5f95a38 MP |
73 | /** |
74 | * Transmit a literal and/or match token. | |
75 | * | |
76 | * This delightfully-named function is called either when we find a | |
77 | * match and need to transmit all the unmatched data leading up to it, | |
78 | * or when we get bored of accumulating literal data and just need to | |
79 | * transmit it. As a result of this second case, it is called even if | |
80 | * we have not matched at all! | |
81 | * | |
82 | * @param i If >0, the number of a matched token. If 0, indicates we | |
83 | * have only literal data. | |
84 | **/ | |
a06b419d | 85 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 86 | OFF_T offset, int32 i) |
c627d613 | 87 | { |
a06b419d WD |
88 | int32 n = offset - last_match; /* max value: block_size (int32) */ |
89 | int32 j; | |
90 | ||
91 | if (verbose > 2 && i >= 0) { | |
92 | rprintf(FINFO, | |
93 | "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n", | |
94 | (double)offset, (double)last_match, i, | |
95 | (long)s->sums[i].len, (long)n); | |
96 | } | |
c627d613 | 97 | |
a06b419d | 98 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 99 | data_transfer += n; |
70d794dc | 100 | |
a800434a AT |
101 | if (i >= 0) { |
102 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 103 | n += s->sums[i].len; |
a800434a | 104 | } |
1e4f48d6 WD |
105 | |
106 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
107 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
108 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 109 | } |
9e31c482 | 110 | |
0d0e2e93 | 111 | if (i >= 0) |
e7ebc36c | 112 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
113 | else |
114 | last_match = offset; | |
eb86d661 | 115 | |
0394e34a | 116 | if (buf && do_progress) |
4440b8aa | 117 | show_progress(last_match, buf->file_size); |
9e31c482 AT |
118 | } |
119 | ||
120 | ||
c6e7fcb4 | 121 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 122 | struct map_struct *buf, OFF_T len) |
c627d613 | 123 | { |
7560c17a | 124 | OFF_T offset, end, backup; |
7cacd47e | 125 | int32 k, want_i; |
e7ebc36c | 126 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 127 | uint32 s1, s2, sum; |
7560c17a | 128 | int more; |
debb4505 | 129 | schar *map; |
e7ebc36c | 130 | |
a04d77bc WD |
131 | /* want_i is used to encourage adjacent matches, allowing the RLL |
132 | * coding of the output to work more efficiently. */ | |
133 | want_i = 0; | |
923fa978 | 134 | |
0e36d9da | 135 | if (verbose > 2) { |
a06b419d WD |
136 | rprintf(FINFO, "hash search b=%ld len=%.0f\n", |
137 | (long)s->blength, (double)len); | |
0e36d9da | 138 | } |
e7ebc36c | 139 | |
a06b419d | 140 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 141 | |
0e36d9da | 142 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 143 | |
e7ebc36c AT |
144 | sum = get_checksum1((char *)map, k); |
145 | s1 = sum & 0xFFFF; | |
146 | s2 = sum >> 16; | |
147 | if (verbose > 3) | |
a06b419d | 148 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 149 | |
e7ebc36c | 150 | offset = 0; |
1e4f48d6 | 151 | |
e7ebc36c | 152 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 153 | |
0e36d9da | 154 | if (verbose > 3) { |
a06b419d WD |
155 | rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n", |
156 | (long)s->blength, (double)len, (double)s->count); | |
0e36d9da | 157 | } |
1e4f48d6 | 158 | |
e7ebc36c | 159 | do { |
e7ebc36c | 160 | int done_csum2 = 0; |
d9bca0c3 | 161 | int32 i; |
1e4f48d6 | 162 | |
48cce779 WD |
163 | if (verbose > 4) { |
164 | rprintf(FINFO, "offset=%.0f sum=%04x%04x\n", | |
34f961bb | 165 | (double)offset, s2 & 0xFFFF, s1 & 0xFFFF); |
48cce779 | 166 | } |
1e4f48d6 | 167 | |
48cce779 | 168 | i = hash_table[SUM2HASH2(s1,s2)]; |
900cfcb5 WD |
169 | if (i < 0) |
170 | goto null_hash; | |
171 | ||
48cce779 | 172 | sum = (s1 & 0xffff) | (s2 << 16); |
900cfcb5 WD |
173 | hash_hits++; |
174 | do { | |
d9bca0c3 | 175 | int32 l; |
1e4f48d6 | 176 | |
0e36d9da WD |
177 | if (sum != s->sums[i].sum1) |
178 | continue; | |
1e4f48d6 | 179 | |
536541d5 | 180 | /* also make sure the two blocks are the same length */ |
a06b419d | 181 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
182 | if (l != s->sums[i].len) |
183 | continue; | |
536541d5 | 184 | |
a36ffd39 | 185 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 186 | * offset or that the data didn't move. */ |
a36ffd39 | 187 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
188 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
189 | continue; | |
190 | ||
d9bca0c3 WD |
191 | if (verbose > 3) { |
192 | rprintf(FINFO, | |
193 | "potential match at %.0f i=%ld sum=%08x\n", | |
194 | (double)offset, (long)i, sum); | |
195 | } | |
1e4f48d6 | 196 | |
e7ebc36c | 197 | if (!done_csum2) { |
debb4505 | 198 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
199 | get_checksum2((char *)map,l,sum2); |
200 | done_csum2 = 1; | |
201 | } | |
1e4f48d6 | 202 | |
fc0257c9 | 203 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
204 | false_alarms++; |
205 | continue; | |
206 | } | |
923fa978 | 207 | |
a36ffd39 | 208 | /* When updating in-place, the best possible match is |
a3221d2a WD |
209 | * one with an identical offset, so we prefer that over |
210 | * the following want_i optimization. */ | |
a36ffd39 | 211 | if (updating_basis_file) { |
d9bca0c3 WD |
212 | int32 i2; |
213 | for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) { | |
a3221d2a WD |
214 | if (s->sums[i2].offset != offset) |
215 | continue; | |
216 | if (i2 != i) { | |
217 | if (sum != s->sums[i2].sum1) | |
218 | break; | |
219 | if (memcmp(sum2, s->sums[i2].sum2, | |
220 | s->s2length) != 0) | |
221 | break; | |
222 | i = i2; | |
223 | } | |
224 | /* This chunk was at the same offset on | |
225 | * both the sender and the receiver. */ | |
226 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
227 | goto set_want_i; | |
d9bca0c3 | 228 | } |
a3221d2a WD |
229 | } |
230 | ||
923fa978 | 231 | /* we've found a match, but now check to see |
a04d77bc WD |
232 | * if want_i can hint at a better match. */ |
233 | if (i != want_i && want_i < s->count | |
a36ffd39 | 234 | && (!updating_basis_file || s->sums[want_i].offset >= offset |
a3221d2a | 235 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
236 | && sum == s->sums[want_i].sum1 |
237 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
238 | /* we've found an adjacent match - the RLL coder |
239 | * will be happy */ | |
a04d77bc | 240 | i = want_i; |
923fa978 | 241 | } |
a3221d2a | 242 | set_want_i: |
a04d77bc | 243 | want_i = i + 1; |
1e4f48d6 | 244 | |
e7ebc36c AT |
245 | matched(f,s,buf,offset,i); |
246 | offset += s->sums[i].len - 1; | |
a06b419d | 247 | k = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da | 248 | map = (schar *)map_ptr(buf, offset, k); |
e7ebc36c AT |
249 | sum = get_checksum1((char *)map, k); |
250 | s1 = sum & 0xFFFF; | |
251 | s2 = sum >> 16; | |
252 | matches++; | |
253 | break; | |
900cfcb5 | 254 | } while ((i = s->sums[i].chain) >= 0); |
1e4f48d6 | 255 | |
900cfcb5 | 256 | null_hash: |
7560c17a WD |
257 | backup = offset - last_match; |
258 | /* We sometimes read 1 byte prior to last_match... */ | |
259 | if (backup < 0) | |
260 | backup = 0; | |
261 | ||
e7ebc36c | 262 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
263 | more = offset + k < len; |
264 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
265 | + backup; | |
e7ebc36c AT |
266 | s1 -= map[0] + CHAR_OFFSET; |
267 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 268 | |
e7ebc36c | 269 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
270 | if (more) { |
271 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 272 | s2 += s1; |
0e36d9da | 273 | } else |
e7ebc36c | 274 | --k; |
45f133b9 | 275 | |
5914bf15 PM |
276 | /* By matching early we avoid re-reading the |
277 | data 3 times in the case where a token | |
278 | match comes a long way after last | |
279 | match. The 3 reads are caused by the | |
280 | running match, the checksum update and the | |
281 | literal send. */ | |
a06b419d WD |
282 | if (backup >= s->blength+CHUNK_SIZE && end-offset > CHUNK_SIZE) |
283 | matched(f, s, buf, offset - s->blength, -2); | |
e7ebc36c | 284 | } while (++offset < end); |
1e4f48d6 | 285 | |
a06b419d WD |
286 | matched(f, s, buf, len, -1); |
287 | map_ptr(buf, len-1, 1); | |
c627d613 AT |
288 | } |
289 | ||
290 | ||
f5f95a38 MP |
291 | /** |
292 | * Scan through a origin file, looking for sections that match | |
293 | * checksums from the generator, and transmit either literal or token | |
294 | * data. | |
295 | * | |
538ba24f MP |
296 | * Also calculates the MD4 checksum of the whole file, using the md |
297 | * accumulator. This is transmitted with the file as protection | |
298 | * against corruption on the wire. | |
299 | * | |
f5f95a38 MP |
300 | * @param s Checksums received from the generator. If <tt>s->count == |
301 | * 0</tt>, then there are actually no checksums for this file. | |
302 | * | |
303 | * @param len Length of the file to send. | |
304 | **/ | |
305 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 306 | { |
e7ebc36c AT |
307 | char file_sum[MD4_SUM_LENGTH]; |
308 | ||
309 | last_match = 0; | |
310 | false_alarms = 0; | |
900cfcb5 | 311 | hash_hits = 0; |
1e4f48d6 WD |
312 | matches = 0; |
313 | data_transfer = 0; | |
e7ebc36c | 314 | |
ba582f75 | 315 | sum_init(checksum_seed); |
e7ebc36c | 316 | |
6cc11982 WD |
317 | if (append_mode) { |
318 | OFF_T j = 0; | |
319 | for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) { | |
1f86fcf5 WD |
320 | if (buf && do_progress) |
321 | show_progress(last_match, buf->file_size); | |
6cc11982 WD |
322 | sum_update(map_ptr(buf, last_match, CHUNK_SIZE), |
323 | CHUNK_SIZE); | |
324 | last_match = j; | |
325 | } | |
326 | if (last_match < s->flength) { | |
327 | int32 len = s->flength - last_match; | |
1f86fcf5 WD |
328 | if (buf && do_progress) |
329 | show_progress(last_match, buf->file_size); | |
6cc11982 WD |
330 | sum_update(map_ptr(buf, last_match, len), len); |
331 | last_match = s->flength; | |
332 | } | |
333 | s->count = 0; | |
334 | } | |
335 | ||
a06b419d | 336 | if (len > 0 && s->count > 0) { |
e7ebc36c | 337 | build_hash_table(s); |
1e4f48d6 WD |
338 | |
339 | if (verbose > 2) | |
9486289c | 340 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 341 | |
e7ebc36c | 342 | hash_search(f,s,buf,len); |
1e4f48d6 WD |
343 | |
344 | if (verbose > 2) | |
9486289c | 345 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 346 | } else { |
eb86d661 AT |
347 | OFF_T j; |
348 | /* by doing this in pieces we avoid too many seeks */ | |
6cc11982 | 349 | for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
a06b419d WD |
350 | matched(f, s, buf, j, -2); |
351 | matched(f, s, buf, len, -1); | |
e7ebc36c | 352 | } |
9e31c482 | 353 | |
e7ebc36c | 354 | sum_end(file_sum); |
edecdad5 WD |
355 | /* If we had a read error, send a bad checksum. */ |
356 | if (buf && buf->status != 0) | |
357 | file_sum[0]++; | |
c627d613 | 358 | |
bc63ae3f S |
359 | if (verbose > 2) |
360 | rprintf(FINFO,"sending file_sum\n"); | |
361 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
c627d613 | 362 | |
e7ebc36c | 363 | if (verbose > 2) |
900cfcb5 WD |
364 | rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n", |
365 | false_alarms, hash_hits, matches); | |
1e4f48d6 | 366 | |
900cfcb5 | 367 | total_hash_hits += hash_hits; |
e7ebc36c AT |
368 | total_false_alarms += false_alarms; |
369 | total_matches += matches; | |
a800434a | 370 | stats.literal_data += data_transfer; |
c627d613 AT |
371 | } |
372 | ||
373 | void match_report(void) | |
374 | { | |
e7ebc36c AT |
375 | if (verbose <= 1) |
376 | return; | |
c627d613 | 377 | |
9486289c | 378 | rprintf(FINFO, |
900cfcb5 WD |
379 | "total: matches=%d hash_hits=%d false_alarms=%d data=%.0f\n", |
380 | total_matches, total_hash_hits, total_false_alarms, | |
a800434a | 381 | (double)stats.literal_data); |
c627d613 | 382 | } |