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 | |
4fd842f9 | 9 | * it under the terms of the GNU General Public License version 3 as |
ba2133d6 | 10 | * published by the Free Software Foundation. |
0f78b815 WD |
11 | * |
12 | * This program is distributed in the hope that it will be useful, | |
13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
15 | * GNU General Public License for more details. | |
16 | * | |
e7c67065 | 17 | * You should have received a copy of the GNU General Public License along |
4fd842f9 | 18 | * with this program; if not, visit the http://fsf.org website. |
0f78b815 | 19 | */ |
c627d613 AT |
20 | |
21 | #include "rsync.h" | |
22 | ||
23 | extern int verbose; | |
16417f8b | 24 | extern int do_progress; |
ba582f75 | 25 | extern int checksum_seed; |
6cc11982 | 26 | extern int append_mode; |
a36ffd39 WD |
27 | |
28 | int updating_basis_file; | |
c627d613 | 29 | |
c627d613 | 30 | static int false_alarms; |
900cfcb5 | 31 | static int hash_hits; |
c627d613 | 32 | static int matches; |
a800434a | 33 | static int64 data_transfer; |
c627d613 | 34 | |
3a6a366f | 35 | static int total_false_alarms; |
900cfcb5 | 36 | static int total_hash_hits; |
3a6a366f | 37 | static int total_matches; |
c627d613 | 38 | |
a800434a | 39 | extern struct stats stats; |
c627d613 | 40 | |
48cce779 WD |
41 | #define TABLESIZE (1<<16) |
42 | ||
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 AT |
47 | |
48 | static void build_hash_table(struct sum_struct *s) | |
49 | { | |
7cacd47e | 50 | int32 i; |
48cce779 WD |
51 | |
52 | if (!hash_table) { | |
53 | hash_table = new_array(int32, TABLESIZE); | |
68207537 | 54 | if (!hash_table) |
d9bca0c3 WD |
55 | out_of_memory("build_hash_table"); |
56 | } | |
c627d613 | 57 | |
48cce779 | 58 | memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]); |
c627d613 | 59 | |
0e36d9da | 60 | for (i = 0; i < s->count; i++) { |
fbbe9a01 | 61 | uint32 t = SUM2HASH(s->sums[i].sum1); |
68207537 WD |
62 | s->sums[i].chain = hash_table[t]; |
63 | hash_table[t] = i; | |
1e4f48d6 | 64 | } |
c627d613 AT |
65 | } |
66 | ||
67 | ||
bcacc18b | 68 | static OFF_T last_match; |
c627d613 AT |
69 | |
70 | ||
f5f95a38 MP |
71 | /** |
72 | * Transmit a literal and/or match token. | |
73 | * | |
74 | * This delightfully-named function is called either when we find a | |
75 | * match and need to transmit all the unmatched data leading up to it, | |
76 | * or when we get bored of accumulating literal data and just need to | |
77 | * transmit it. As a result of this second case, it is called even if | |
78 | * we have not matched at all! | |
79 | * | |
80 | * @param i If >0, the number of a matched token. If 0, indicates we | |
81 | * have only literal data. | |
82 | **/ | |
a06b419d | 83 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 84 | OFF_T offset, int32 i) |
c627d613 | 85 | { |
26404276 | 86 | int32 n = (int32)(offset - last_match); /* max value: block_size (int32) */ |
a06b419d WD |
87 | int32 j; |
88 | ||
89 | if (verbose > 2 && i >= 0) { | |
90 | rprintf(FINFO, | |
91 | "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n", | |
92 | (double)offset, (double)last_match, i, | |
93 | (long)s->sums[i].len, (long)n); | |
94 | } | |
c627d613 | 95 | |
a06b419d | 96 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 97 | data_transfer += n; |
70d794dc | 98 | |
a800434a AT |
99 | if (i >= 0) { |
100 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 101 | n += s->sums[i].len; |
a800434a | 102 | } |
1e4f48d6 WD |
103 | |
104 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
105 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
106 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 107 | } |
9e31c482 | 108 | |
0d0e2e93 | 109 | if (i >= 0) |
e7ebc36c | 110 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
111 | else |
112 | last_match = offset; | |
eb86d661 | 113 | |
0394e34a | 114 | if (buf && do_progress) |
4440b8aa | 115 | show_progress(last_match, buf->file_size); |
9e31c482 AT |
116 | } |
117 | ||
118 | ||
c6e7fcb4 | 119 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 120 | struct map_struct *buf, OFF_T len) |
c627d613 | 121 | { |
26404276 WD |
122 | OFF_T offset, end; |
123 | int32 k, want_i, backup; | |
e7ebc36c | 124 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 125 | uint32 s1, s2, sum; |
7560c17a | 126 | int more; |
debb4505 | 127 | schar *map; |
e7ebc36c | 128 | |
a04d77bc WD |
129 | /* want_i is used to encourage adjacent matches, allowing the RLL |
130 | * coding of the output to work more efficiently. */ | |
131 | want_i = 0; | |
923fa978 | 132 | |
0e36d9da | 133 | if (verbose > 2) { |
a06b419d WD |
134 | rprintf(FINFO, "hash search b=%ld len=%.0f\n", |
135 | (long)s->blength, (double)len); | |
0e36d9da | 136 | } |
e7ebc36c | 137 | |
a06b419d | 138 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 139 | |
0e36d9da | 140 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 141 | |
e7ebc36c AT |
142 | sum = get_checksum1((char *)map, k); |
143 | s1 = sum & 0xFFFF; | |
144 | s2 = sum >> 16; | |
145 | if (verbose > 3) | |
a06b419d | 146 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 147 | |
e7ebc36c | 148 | offset = 0; |
1e4f48d6 | 149 | |
e7ebc36c | 150 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 151 | |
0e36d9da | 152 | if (verbose > 3) { |
a06b419d WD |
153 | rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n", |
154 | (long)s->blength, (double)len, (double)s->count); | |
0e36d9da | 155 | } |
1e4f48d6 | 156 | |
e7ebc36c | 157 | do { |
e7ebc36c | 158 | int done_csum2 = 0; |
d9bca0c3 | 159 | int32 i; |
1e4f48d6 | 160 | |
48cce779 WD |
161 | if (verbose > 4) { |
162 | rprintf(FINFO, "offset=%.0f sum=%04x%04x\n", | |
34f961bb | 163 | (double)offset, s2 & 0xFFFF, s1 & 0xFFFF); |
48cce779 | 164 | } |
1e4f48d6 | 165 | |
48cce779 | 166 | i = hash_table[SUM2HASH2(s1,s2)]; |
900cfcb5 WD |
167 | if (i < 0) |
168 | goto null_hash; | |
169 | ||
48cce779 | 170 | sum = (s1 & 0xffff) | (s2 << 16); |
900cfcb5 WD |
171 | hash_hits++; |
172 | do { | |
d9bca0c3 | 173 | int32 l; |
1e4f48d6 | 174 | |
0e36d9da WD |
175 | if (sum != s->sums[i].sum1) |
176 | continue; | |
1e4f48d6 | 177 | |
536541d5 | 178 | /* also make sure the two blocks are the same length */ |
a06b419d | 179 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
180 | if (l != s->sums[i].len) |
181 | continue; | |
536541d5 | 182 | |
a36ffd39 | 183 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 184 | * offset or that the data didn't move. */ |
a36ffd39 | 185 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
186 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
187 | continue; | |
188 | ||
d9bca0c3 WD |
189 | if (verbose > 3) { |
190 | rprintf(FINFO, | |
191 | "potential match at %.0f i=%ld sum=%08x\n", | |
192 | (double)offset, (long)i, sum); | |
193 | } | |
1e4f48d6 | 194 | |
e7ebc36c | 195 | if (!done_csum2) { |
debb4505 | 196 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
197 | get_checksum2((char *)map,l,sum2); |
198 | done_csum2 = 1; | |
199 | } | |
1e4f48d6 | 200 | |
fc0257c9 | 201 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
202 | false_alarms++; |
203 | continue; | |
204 | } | |
923fa978 | 205 | |
a36ffd39 | 206 | /* When updating in-place, the best possible match is |
a3221d2a WD |
207 | * one with an identical offset, so we prefer that over |
208 | * the following want_i optimization. */ | |
a36ffd39 | 209 | if (updating_basis_file) { |
d9bca0c3 WD |
210 | int32 i2; |
211 | for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) { | |
a3221d2a WD |
212 | if (s->sums[i2].offset != offset) |
213 | continue; | |
214 | if (i2 != i) { | |
215 | if (sum != s->sums[i2].sum1) | |
216 | break; | |
217 | if (memcmp(sum2, s->sums[i2].sum2, | |
218 | s->s2length) != 0) | |
219 | break; | |
220 | i = i2; | |
221 | } | |
222 | /* This chunk was at the same offset on | |
223 | * both the sender and the receiver. */ | |
224 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
225 | goto set_want_i; | |
d9bca0c3 | 226 | } |
a3221d2a WD |
227 | } |
228 | ||
923fa978 | 229 | /* we've found a match, but now check to see |
a04d77bc WD |
230 | * if want_i can hint at a better match. */ |
231 | if (i != want_i && want_i < s->count | |
a36ffd39 | 232 | && (!updating_basis_file || s->sums[want_i].offset >= offset |
a3221d2a | 233 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
234 | && sum == s->sums[want_i].sum1 |
235 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
236 | /* we've found an adjacent match - the RLL coder |
237 | * will be happy */ | |
a04d77bc | 238 | i = want_i; |
923fa978 | 239 | } |
a3221d2a | 240 | set_want_i: |
a04d77bc | 241 | want_i = i + 1; |
1e4f48d6 | 242 | |
e7ebc36c AT |
243 | matched(f,s,buf,offset,i); |
244 | offset += s->sums[i].len - 1; | |
a06b419d | 245 | k = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da | 246 | map = (schar *)map_ptr(buf, offset, k); |
e7ebc36c AT |
247 | sum = get_checksum1((char *)map, k); |
248 | s1 = sum & 0xFFFF; | |
249 | s2 = sum >> 16; | |
250 | matches++; | |
251 | break; | |
900cfcb5 | 252 | } while ((i = s->sums[i].chain) >= 0); |
1e4f48d6 | 253 | |
900cfcb5 | 254 | null_hash: |
26404276 | 255 | backup = (int32)(offset - last_match); |
7560c17a WD |
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 | { |
a0456b9c WD |
305 | char file_sum[MAX_DIGEST_LEN]; |
306 | int sum_len; | |
e7ebc36c AT |
307 | |
308 | last_match = 0; | |
309 | false_alarms = 0; | |
900cfcb5 | 310 | hash_hits = 0; |
1e4f48d6 WD |
311 | matches = 0; |
312 | data_transfer = 0; | |
e7ebc36c | 313 | |
ba582f75 | 314 | sum_init(checksum_seed); |
e7ebc36c | 315 | |
edb97721 | 316 | if (append_mode > 0) { |
6cc11982 WD |
317 | OFF_T j = 0; |
318 | for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) { | |
1f86fcf5 WD |
319 | if (buf && do_progress) |
320 | show_progress(last_match, buf->file_size); | |
6cc11982 WD |
321 | sum_update(map_ptr(buf, last_match, CHUNK_SIZE), |
322 | CHUNK_SIZE); | |
323 | last_match = j; | |
324 | } | |
325 | if (last_match < s->flength) { | |
a0456b9c | 326 | int32 n = (int32)(s->flength - last_match); |
1f86fcf5 WD |
327 | if (buf && do_progress) |
328 | show_progress(last_match, buf->file_size); | |
a0456b9c | 329 | sum_update(map_ptr(buf, last_match, n), n); |
6cc11982 WD |
330 | last_match = s->flength; |
331 | } | |
332 | s->count = 0; | |
333 | } | |
334 | ||
a06b419d | 335 | if (len > 0 && s->count > 0) { |
e7ebc36c | 336 | build_hash_table(s); |
1e4f48d6 WD |
337 | |
338 | if (verbose > 2) | |
9486289c | 339 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 340 | |
a0456b9c | 341 | hash_search(f, s, buf, len); |
1e4f48d6 WD |
342 | |
343 | if (verbose > 2) | |
9486289c | 344 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 345 | } else { |
eb86d661 AT |
346 | OFF_T j; |
347 | /* by doing this in pieces we avoid too many seeks */ | |
6cc11982 | 348 | for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
a06b419d WD |
349 | matched(f, s, buf, j, -2); |
350 | matched(f, s, buf, len, -1); | |
e7ebc36c | 351 | } |
9e31c482 | 352 | |
a0456b9c | 353 | sum_len = sum_end(file_sum); |
edecdad5 WD |
354 | /* If we had a read error, send a bad checksum. */ |
355 | if (buf && buf->status != 0) | |
356 | file_sum[0]++; | |
c627d613 | 357 | |
bc63ae3f S |
358 | if (verbose > 2) |
359 | rprintf(FINFO,"sending file_sum\n"); | |
a0456b9c | 360 | write_buf(f, file_sum, sum_len); |
c627d613 | 361 | |
e7ebc36c | 362 | if (verbose > 2) |
900cfcb5 WD |
363 | rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n", |
364 | false_alarms, hash_hits, matches); | |
1e4f48d6 | 365 | |
900cfcb5 | 366 | total_hash_hits += hash_hits; |
e7ebc36c AT |
367 | total_false_alarms += false_alarms; |
368 | total_matches += matches; | |
a800434a | 369 | stats.literal_data += data_transfer; |
c627d613 AT |
370 | } |
371 | ||
372 | void match_report(void) | |
373 | { | |
e7ebc36c AT |
374 | if (verbose <= 1) |
375 | return; | |
c627d613 | 376 | |
9486289c | 377 | rprintf(FINFO, |
900cfcb5 WD |
378 | "total: matches=%d hash_hits=%d false_alarms=%d data=%.0f\n", |
379 | total_matches, total_hash_hits, total_false_alarms, | |
a800434a | 380 | (double)stats.literal_data); |
c627d613 | 381 | } |