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; | |
23 | extern int am_server; | |
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 AT |
29 | |
30 | typedef unsigned short tag; | |
31 | ||
32 | #define TABLESIZE (1<<16) | |
7cacd47e | 33 | #define NULL_TAG (-1) |
c627d613 AT |
34 | |
35 | static int false_alarms; | |
36 | static int tag_hits; | |
37 | static int matches; | |
a800434a | 38 | static int64 data_transfer; |
c627d613 | 39 | |
3a6a366f AT |
40 | static int total_false_alarms; |
41 | static int total_tag_hits; | |
42 | static int total_matches; | |
c627d613 | 43 | |
a800434a | 44 | extern struct stats stats; |
c627d613 AT |
45 | |
46 | struct target { | |
1e4f48d6 | 47 | tag t; |
7cacd47e | 48 | int32 i; |
c627d613 AT |
49 | }; |
50 | ||
3a6a366f | 51 | static struct target *targets; |
c627d613 | 52 | |
7cacd47e | 53 | static int32 *tag_table; |
c627d613 AT |
54 | |
55 | #define gettag2(s1,s2) (((s1) + (s2)) & 0xFFFF) | |
56 | #define gettag(sum) gettag2((sum)&0xFFFF,(sum)>>16) | |
57 | ||
58 | static int compare_targets(struct target *t1,struct target *t2) | |
59 | { | |
1e4f48d6 | 60 | return (int)t1->t - (int)t2->t; |
c627d613 AT |
61 | } |
62 | ||
63 | ||
64 | static void build_hash_table(struct sum_struct *s) | |
65 | { | |
7cacd47e | 66 | int32 i; |
c627d613 | 67 | |
1e4f48d6 | 68 | if (!tag_table) |
7cacd47e | 69 | tag_table = new_array(int32, TABLESIZE); |
c627d613 | 70 | |
58cadc86 | 71 | targets = new_array(struct target, s->count); |
1e4f48d6 WD |
72 | if (!tag_table || !targets) |
73 | out_of_memory("build_hash_table"); | |
c627d613 | 74 | |
0e36d9da | 75 | for (i = 0; i < s->count; i++) { |
1e4f48d6 WD |
76 | targets[i].i = i; |
77 | targets[i].t = gettag(s->sums[i].sum1); | |
78 | } | |
c627d613 | 79 | |
1e4f48d6 | 80 | qsort(targets,s->count,sizeof(targets[0]),(int (*)())compare_targets); |
c627d613 | 81 | |
1e4f48d6 WD |
82 | for (i = 0; i < TABLESIZE; i++) |
83 | tag_table[i] = NULL_TAG; | |
c627d613 | 84 | |
0e36d9da | 85 | for (i = s->count; i-- > 0; ) |
1e4f48d6 | 86 | tag_table[targets[i].t] = i; |
c627d613 AT |
87 | } |
88 | ||
89 | ||
bcacc18b | 90 | static OFF_T last_match; |
c627d613 AT |
91 | |
92 | ||
f5f95a38 MP |
93 | /** |
94 | * Transmit a literal and/or match token. | |
95 | * | |
96 | * This delightfully-named function is called either when we find a | |
97 | * match and need to transmit all the unmatched data leading up to it, | |
98 | * or when we get bored of accumulating literal data and just need to | |
99 | * transmit it. As a result of this second case, it is called even if | |
100 | * we have not matched at all! | |
101 | * | |
102 | * @param i If >0, the number of a matched token. If 0, indicates we | |
103 | * have only literal data. | |
104 | **/ | |
a06b419d | 105 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 106 | OFF_T offset, int32 i) |
c627d613 | 107 | { |
a06b419d WD |
108 | int32 n = offset - last_match; /* max value: block_size (int32) */ |
109 | int32 j; | |
110 | ||
111 | if (verbose > 2 && i >= 0) { | |
112 | rprintf(FINFO, | |
113 | "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n", | |
114 | (double)offset, (double)last_match, i, | |
115 | (long)s->sums[i].len, (long)n); | |
116 | } | |
c627d613 | 117 | |
a06b419d | 118 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 119 | data_transfer += n; |
70d794dc | 120 | |
a800434a AT |
121 | if (i >= 0) { |
122 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 123 | n += s->sums[i].len; |
a800434a | 124 | } |
1e4f48d6 WD |
125 | |
126 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
127 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
128 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 129 | } |
9e31c482 | 130 | |
9e31c482 | 131 | |
0d0e2e93 | 132 | if (i >= 0) |
e7ebc36c | 133 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
134 | else |
135 | last_match = offset; | |
eb86d661 | 136 | |
0394e34a | 137 | if (buf && do_progress) |
4440b8aa | 138 | show_progress(last_match, buf->file_size); |
9e31c482 AT |
139 | } |
140 | ||
141 | ||
c6e7fcb4 | 142 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 143 | struct map_struct *buf, OFF_T len) |
c627d613 | 144 | { |
7560c17a | 145 | OFF_T offset, end, backup; |
7cacd47e | 146 | int32 k, want_i; |
e7ebc36c | 147 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 148 | uint32 s1, s2, sum; |
7560c17a | 149 | int more; |
debb4505 | 150 | schar *map; |
e7ebc36c | 151 | |
a04d77bc WD |
152 | /* want_i is used to encourage adjacent matches, allowing the RLL |
153 | * coding of the output to work more efficiently. */ | |
154 | want_i = 0; | |
923fa978 | 155 | |
0e36d9da | 156 | if (verbose > 2) { |
a06b419d WD |
157 | rprintf(FINFO, "hash search b=%ld len=%.0f\n", |
158 | (long)s->blength, (double)len); | |
0e36d9da | 159 | } |
e7ebc36c | 160 | |
a06b419d | 161 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 162 | |
0e36d9da | 163 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 164 | |
e7ebc36c AT |
165 | sum = get_checksum1((char *)map, k); |
166 | s1 = sum & 0xFFFF; | |
167 | s2 = sum >> 16; | |
168 | if (verbose > 3) | |
a06b419d | 169 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 170 | |
e7ebc36c | 171 | offset = 0; |
1e4f48d6 | 172 | |
e7ebc36c | 173 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 174 | |
0e36d9da | 175 | if (verbose > 3) { |
a06b419d WD |
176 | rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n", |
177 | (long)s->blength, (double)len, (double)s->count); | |
0e36d9da | 178 | } |
1e4f48d6 | 179 | |
e7ebc36c AT |
180 | do { |
181 | tag t = gettag2(s1,s2); | |
182 | int done_csum2 = 0; | |
7cacd47e | 183 | int32 j = tag_table[t]; |
1e4f48d6 | 184 | |
e7ebc36c | 185 | if (verbose > 4) |
5f808dfb | 186 | rprintf(FINFO,"offset=%.0f sum=%08x\n",(double)offset,sum); |
1e4f48d6 | 187 | |
0e36d9da | 188 | if (j == NULL_TAG) |
e7ebc36c | 189 | goto null_tag; |
e7ebc36c AT |
190 | |
191 | sum = (s1 & 0xffff) | (s2 << 16); | |
192 | tag_hits++; | |
066a844c | 193 | do { |
7cacd47e | 194 | int32 l, i = targets[j].i; |
1e4f48d6 | 195 | |
0e36d9da WD |
196 | if (sum != s->sums[i].sum1) |
197 | continue; | |
1e4f48d6 | 198 | |
536541d5 | 199 | /* also make sure the two blocks are the same length */ |
a06b419d | 200 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
201 | if (l != s->sums[i].len) |
202 | continue; | |
536541d5 | 203 | |
a36ffd39 | 204 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 205 | * offset or that the data didn't move. */ |
a36ffd39 | 206 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
207 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
208 | continue; | |
209 | ||
e7ebc36c | 210 | if (verbose > 3) |
0e36d9da WD |
211 | rprintf(FINFO,"potential match at %.0f target=%.0f %.0f sum=%08x\n", |
212 | (double)offset,(double)j,(double)i,sum); | |
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) { |
a3221d2a | 229 | do { |
7cacd47e | 230 | int32 i2 = targets[j].i; |
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; | |
245 | } while (++j < s->count && targets[j].t == t); | |
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; | |
066a844c | 271 | } while (++j < s->count && targets[j].t == t); |
1e4f48d6 | 272 | |
e7ebc36c | 273 | null_tag: |
7560c17a WD |
274 | backup = offset - last_match; |
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 | { |
e7ebc36c AT |
324 | char file_sum[MD4_SUM_LENGTH]; |
325 | ||
326 | last_match = 0; | |
327 | false_alarms = 0; | |
328 | tag_hits = 0; | |
1e4f48d6 WD |
329 | matches = 0; |
330 | data_transfer = 0; | |
e7ebc36c | 331 | |
ba582f75 | 332 | sum_init(checksum_seed); |
e7ebc36c | 333 | |
6cc11982 WD |
334 | if (append_mode) { |
335 | OFF_T j = 0; | |
336 | for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) { | |
337 | sum_update(map_ptr(buf, last_match, CHUNK_SIZE), | |
338 | CHUNK_SIZE); | |
339 | last_match = j; | |
340 | } | |
341 | if (last_match < s->flength) { | |
342 | int32 len = s->flength - last_match; | |
343 | sum_update(map_ptr(buf, last_match, len), len); | |
344 | last_match = s->flength; | |
345 | } | |
346 | s->count = 0; | |
347 | } | |
348 | ||
a06b419d | 349 | if (len > 0 && s->count > 0) { |
e7ebc36c | 350 | build_hash_table(s); |
1e4f48d6 WD |
351 | |
352 | if (verbose > 2) | |
9486289c | 353 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 354 | |
e7ebc36c | 355 | hash_search(f,s,buf,len); |
1e4f48d6 WD |
356 | |
357 | if (verbose > 2) | |
9486289c | 358 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 359 | } else { |
eb86d661 AT |
360 | OFF_T j; |
361 | /* by doing this in pieces we avoid too many seeks */ | |
6cc11982 | 362 | for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
a06b419d WD |
363 | matched(f, s, buf, j, -2); |
364 | matched(f, s, buf, len, -1); | |
e7ebc36c | 365 | } |
9e31c482 | 366 | |
e7ebc36c | 367 | sum_end(file_sum); |
edecdad5 WD |
368 | /* If we had a read error, send a bad checksum. */ |
369 | if (buf && buf->status != 0) | |
370 | file_sum[0]++; | |
c627d613 | 371 | |
bc63ae3f S |
372 | if (verbose > 2) |
373 | rprintf(FINFO,"sending file_sum\n"); | |
374 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
c627d613 | 375 | |
e7ebc36c AT |
376 | if (targets) { |
377 | free(targets); | |
378 | targets=NULL; | |
379 | } | |
1e4f48d6 | 380 | |
e7ebc36c | 381 | if (verbose > 2) |
9486289c | 382 | rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", |
e7ebc36c | 383 | false_alarms, tag_hits, matches); |
1e4f48d6 | 384 | |
e7ebc36c AT |
385 | total_tag_hits += tag_hits; |
386 | total_false_alarms += false_alarms; | |
387 | total_matches += matches; | |
a800434a | 388 | stats.literal_data += data_transfer; |
c627d613 AT |
389 | } |
390 | ||
391 | void match_report(void) | |
392 | { | |
e7ebc36c AT |
393 | if (verbose <= 1) |
394 | return; | |
c627d613 | 395 | |
9486289c | 396 | rprintf(FINFO, |
a800434a | 397 | "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", |
e7ebc36c | 398 | total_matches,total_tag_hits, |
a800434a AT |
399 | total_false_alarms, |
400 | (double)stats.literal_data); | |
c627d613 | 401 | } |