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; |
a3221d2a | 26 | extern int inplace; |
3bb400ca | 27 | extern int make_backups; |
c627d613 AT |
28 | |
29 | typedef unsigned short tag; | |
30 | ||
31 | #define TABLESIZE (1<<16) | |
0e36d9da | 32 | #define NULL_TAG ((size_t)-1) |
c627d613 AT |
33 | |
34 | static int false_alarms; | |
35 | static int tag_hits; | |
36 | static int matches; | |
a800434a | 37 | static int64 data_transfer; |
c627d613 | 38 | |
3a6a366f AT |
39 | static int total_false_alarms; |
40 | static int total_tag_hits; | |
41 | static int total_matches; | |
c627d613 | 42 | |
a800434a | 43 | extern struct stats stats; |
c627d613 AT |
44 | |
45 | struct target { | |
1e4f48d6 | 46 | tag t; |
0e36d9da | 47 | size_t i; |
c627d613 AT |
48 | }; |
49 | ||
3a6a366f | 50 | static struct target *targets; |
c627d613 | 51 | |
0e36d9da | 52 | static size_t *tag_table; |
c627d613 AT |
53 | |
54 | #define gettag2(s1,s2) (((s1) + (s2)) & 0xFFFF) | |
55 | #define gettag(sum) gettag2((sum)&0xFFFF,(sum)>>16) | |
56 | ||
57 | static int compare_targets(struct target *t1,struct target *t2) | |
58 | { | |
1e4f48d6 | 59 | return (int)t1->t - (int)t2->t; |
c627d613 AT |
60 | } |
61 | ||
62 | ||
63 | static void build_hash_table(struct sum_struct *s) | |
64 | { | |
0e36d9da | 65 | size_t i; |
c627d613 | 66 | |
1e4f48d6 | 67 | if (!tag_table) |
0e36d9da | 68 | tag_table = new_array(size_t, TABLESIZE); |
c627d613 | 69 | |
58cadc86 | 70 | targets = new_array(struct target, s->count); |
1e4f48d6 WD |
71 | if (!tag_table || !targets) |
72 | out_of_memory("build_hash_table"); | |
c627d613 | 73 | |
0e36d9da | 74 | for (i = 0; i < s->count; i++) { |
1e4f48d6 WD |
75 | targets[i].i = i; |
76 | targets[i].t = gettag(s->sums[i].sum1); | |
77 | } | |
c627d613 | 78 | |
1e4f48d6 | 79 | qsort(targets,s->count,sizeof(targets[0]),(int (*)())compare_targets); |
c627d613 | 80 | |
1e4f48d6 WD |
81 | for (i = 0; i < TABLESIZE; i++) |
82 | tag_table[i] = NULL_TAG; | |
c627d613 | 83 | |
0e36d9da | 84 | for (i = s->count; i-- > 0; ) |
1e4f48d6 | 85 | tag_table[targets[i].t] = i; |
c627d613 AT |
86 | } |
87 | ||
88 | ||
bcacc18b | 89 | static OFF_T last_match; |
c627d613 AT |
90 | |
91 | ||
f5f95a38 MP |
92 | /** |
93 | * Transmit a literal and/or match token. | |
94 | * | |
95 | * This delightfully-named function is called either when we find a | |
96 | * match and need to transmit all the unmatched data leading up to it, | |
97 | * or when we get bored of accumulating literal data and just need to | |
98 | * transmit it. As a result of this second case, it is called even if | |
99 | * we have not matched at all! | |
100 | * | |
101 | * @param i If >0, the number of a matched token. If 0, indicates we | |
102 | * have only literal data. | |
103 | **/ | |
c6e7fcb4 | 104 | static void matched(int f,struct sum_struct *s,struct map_struct *buf, |
bcacc18b | 105 | OFF_T offset,int i) |
c627d613 | 106 | { |
bcacc18b | 107 | OFF_T n = offset - last_match; |
1309d90d | 108 | OFF_T j; |
9e31c482 | 109 | |
0d0e2e93 | 110 | if (verbose > 2 && i >= 0) |
da9d12f5 | 111 | rprintf(FINFO,"match at %.0f last_match=%.0f j=%d len=%u n=%.0f\n", |
5f808dfb | 112 | (double)offset,(double)last_match,i,s->sums[i].len,(double)n); |
c627d613 | 113 | |
45f133b9 | 114 | send_token(f,i,buf,last_match,n,i<0?0:s->sums[i].len); |
e7ebc36c | 115 | data_transfer += n; |
70d794dc | 116 | |
a800434a AT |
117 | if (i >= 0) { |
118 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 119 | n += s->sums[i].len; |
a800434a | 120 | } |
1e4f48d6 WD |
121 | |
122 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
e7ebc36c AT |
123 | int n1 = MIN(CHUNK_SIZE,n-j); |
124 | sum_update(map_ptr(buf,last_match+j,n1),n1); | |
125 | } | |
9e31c482 | 126 | |
9e31c482 | 127 | |
0d0e2e93 | 128 | if (i >= 0) |
e7ebc36c | 129 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
130 | else |
131 | last_match = offset; | |
eb86d661 | 132 | |
16417f8b | 133 | if (buf && do_progress) { |
4440b8aa | 134 | show_progress(last_match, buf->file_size); |
eb86d661 | 135 | |
0e36d9da WD |
136 | if (i == -1) |
137 | end_progress(buf->file_size); | |
5e12ce11 | 138 | } |
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; |
0e36d9da | 146 | unsigned int k; |
a04d77bc | 147 | size_t want_i; |
e7ebc36c | 148 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 149 | uint32 s1, s2, sum; |
7560c17a | 150 | int more; |
debb4505 | 151 | schar *map; |
e7ebc36c | 152 | |
a04d77bc WD |
153 | /* want_i is used to encourage adjacent matches, allowing the RLL |
154 | * coding of the output to work more efficiently. */ | |
155 | want_i = 0; | |
923fa978 | 156 | |
0e36d9da WD |
157 | if (verbose > 2) { |
158 | rprintf(FINFO,"hash search b=%u len=%.0f\n", | |
159 | s->blength, (double)len); | |
160 | } | |
e7ebc36c | 161 | |
0e36d9da | 162 | k = MIN(len, s->blength); |
1e4f48d6 | 163 | |
0e36d9da | 164 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 165 | |
e7ebc36c AT |
166 | sum = get_checksum1((char *)map, k); |
167 | s1 = sum & 0xFFFF; | |
168 | s2 = sum >> 16; | |
169 | if (verbose > 3) | |
0e36d9da | 170 | rprintf(FINFO, "sum=%.8x k=%u\n", sum, k); |
1e4f48d6 | 171 | |
e7ebc36c | 172 | offset = 0; |
1e4f48d6 | 173 | |
e7ebc36c | 174 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 175 | |
0e36d9da WD |
176 | if (verbose > 3) { |
177 | rprintf(FINFO, "hash search s->blength=%u len=%.0f count=%.0f\n", | |
178 | s->blength, (double)len, (double)s->count); | |
179 | } | |
1e4f48d6 | 180 | |
e7ebc36c AT |
181 | do { |
182 | tag t = gettag2(s1,s2); | |
183 | int done_csum2 = 0; | |
0e36d9da | 184 | size_t j = tag_table[t]; |
1e4f48d6 | 185 | |
e7ebc36c | 186 | if (verbose > 4) |
5f808dfb | 187 | rprintf(FINFO,"offset=%.0f sum=%08x\n",(double)offset,sum); |
1e4f48d6 | 188 | |
0e36d9da | 189 | if (j == NULL_TAG) |
e7ebc36c | 190 | goto null_tag; |
e7ebc36c AT |
191 | |
192 | sum = (s1 & 0xffff) | (s2 << 16); | |
193 | tag_hits++; | |
066a844c | 194 | do { |
da9d12f5 | 195 | unsigned int l; |
0e36d9da | 196 | size_t i = targets[j].i; |
1e4f48d6 | 197 | |
0e36d9da WD |
198 | if (sum != s->sums[i].sum1) |
199 | continue; | |
1e4f48d6 | 200 | |
536541d5 | 201 | /* also make sure the two blocks are the same length */ |
c95dcb39 | 202 | l = MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
203 | if (l != s->sums[i].len) |
204 | continue; | |
536541d5 | 205 | |
a3221d2a WD |
206 | /* inplace: ensure chunk's offset is either >= our |
207 | * offset or that the data didn't move. */ | |
3bb400ca | 208 | if (inplace && !make_backups && s->sums[i].offset < offset |
a3221d2a WD |
209 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
210 | continue; | |
211 | ||
e7ebc36c | 212 | if (verbose > 3) |
0e36d9da WD |
213 | rprintf(FINFO,"potential match at %.0f target=%.0f %.0f sum=%08x\n", |
214 | (double)offset,(double)j,(double)i,sum); | |
1e4f48d6 | 215 | |
e7ebc36c | 216 | if (!done_csum2) { |
debb4505 | 217 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
218 | get_checksum2((char *)map,l,sum2); |
219 | done_csum2 = 1; | |
220 | } | |
1e4f48d6 | 221 | |
fc0257c9 | 222 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
223 | false_alarms++; |
224 | continue; | |
225 | } | |
923fa978 | 226 | |
a3221d2a WD |
227 | /* If inplace is enabled, the best possible match is |
228 | * one with an identical offset, so we prefer that over | |
229 | * the following want_i optimization. */ | |
3bb400ca | 230 | if (inplace && !make_backups) { |
a3221d2a WD |
231 | do { |
232 | size_t i2 = targets[j].i; | |
233 | if (s->sums[i2].offset != offset) | |
234 | continue; | |
235 | if (i2 != i) { | |
236 | if (sum != s->sums[i2].sum1) | |
237 | break; | |
238 | if (memcmp(sum2, s->sums[i2].sum2, | |
239 | s->s2length) != 0) | |
240 | break; | |
241 | i = i2; | |
242 | } | |
243 | /* This chunk was at the same offset on | |
244 | * both the sender and the receiver. */ | |
245 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
246 | goto set_want_i; | |
247 | } while (++j < s->count && targets[j].t == t); | |
248 | } | |
249 | ||
923fa978 | 250 | /* we've found a match, but now check to see |
a04d77bc WD |
251 | * if want_i can hint at a better match. */ |
252 | if (i != want_i && want_i < s->count | |
3bb400ca | 253 | && (!inplace || make_backups || s->sums[want_i].offset >= offset |
a3221d2a | 254 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
255 | && sum == s->sums[want_i].sum1 |
256 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
257 | /* we've found an adjacent match - the RLL coder |
258 | * will be happy */ | |
a04d77bc | 259 | i = want_i; |
923fa978 | 260 | } |
a3221d2a | 261 | set_want_i: |
a04d77bc | 262 | want_i = i + 1; |
1e4f48d6 | 263 | |
e7ebc36c AT |
264 | matched(f,s,buf,offset,i); |
265 | offset += s->sums[i].len - 1; | |
0e36d9da WD |
266 | k = MIN(s->blength, len-offset); |
267 | map = (schar *)map_ptr(buf, offset, k); | |
e7ebc36c AT |
268 | sum = get_checksum1((char *)map, k); |
269 | s1 = sum & 0xFFFF; | |
270 | s2 = sum >> 16; | |
271 | matches++; | |
272 | break; | |
066a844c | 273 | } while (++j < s->count && targets[j].t == t); |
1e4f48d6 | 274 | |
e7ebc36c | 275 | null_tag: |
7560c17a WD |
276 | backup = offset - last_match; |
277 | /* We sometimes read 1 byte prior to last_match... */ | |
278 | if (backup < 0) | |
279 | backup = 0; | |
280 | ||
e7ebc36c | 281 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
282 | more = offset + k < len; |
283 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
284 | + backup; | |
e7ebc36c AT |
285 | s1 -= map[0] + CHAR_OFFSET; |
286 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 287 | |
e7ebc36c | 288 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
289 | if (more) { |
290 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 291 | s2 += s1; |
0e36d9da | 292 | } else |
e7ebc36c | 293 | --k; |
45f133b9 | 294 | |
5914bf15 PM |
295 | /* By matching early we avoid re-reading the |
296 | data 3 times in the case where a token | |
297 | match comes a long way after last | |
298 | match. The 3 reads are caused by the | |
299 | running match, the checksum update and the | |
300 | literal send. */ | |
7560c17a WD |
301 | if (backup >= CHUNK_SIZE + s->blength |
302 | && end - offset > CHUNK_SIZE) { | |
fc0257c9 | 303 | matched(f,s,buf,offset - s->blength, -2); |
45f133b9 | 304 | } |
e7ebc36c | 305 | } while (++offset < end); |
1e4f48d6 | 306 | |
e7ebc36c AT |
307 | matched(f,s,buf,len,-1); |
308 | map_ptr(buf,len-1,1); | |
c627d613 AT |
309 | } |
310 | ||
311 | ||
f5f95a38 MP |
312 | /** |
313 | * Scan through a origin file, looking for sections that match | |
314 | * checksums from the generator, and transmit either literal or token | |
315 | * data. | |
316 | * | |
538ba24f MP |
317 | * Also calculates the MD4 checksum of the whole file, using the md |
318 | * accumulator. This is transmitted with the file as protection | |
319 | * against corruption on the wire. | |
320 | * | |
f5f95a38 MP |
321 | * @param s Checksums received from the generator. If <tt>s->count == |
322 | * 0</tt>, then there are actually no checksums for this file. | |
323 | * | |
324 | * @param len Length of the file to send. | |
325 | **/ | |
326 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 327 | { |
e7ebc36c AT |
328 | char file_sum[MD4_SUM_LENGTH]; |
329 | ||
330 | last_match = 0; | |
331 | false_alarms = 0; | |
332 | tag_hits = 0; | |
1e4f48d6 WD |
333 | matches = 0; |
334 | data_transfer = 0; | |
e7ebc36c | 335 | |
ba582f75 | 336 | sum_init(checksum_seed); |
e7ebc36c AT |
337 | |
338 | if (len > 0 && s->count>0) { | |
339 | build_hash_table(s); | |
1e4f48d6 WD |
340 | |
341 | if (verbose > 2) | |
9486289c | 342 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 343 | |
e7ebc36c | 344 | hash_search(f,s,buf,len); |
1e4f48d6 WD |
345 | |
346 | if (verbose > 2) | |
9486289c | 347 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 348 | } else { |
eb86d661 AT |
349 | OFF_T j; |
350 | /* by doing this in pieces we avoid too many seeks */ | |
1e4f48d6 | 351 | for (j = 0; j < len-CHUNK_SIZE; j += CHUNK_SIZE) { |
eb86d661 AT |
352 | int n1 = MIN(CHUNK_SIZE,(len-CHUNK_SIZE)-j); |
353 | matched(f,s,buf,j+n1,-2); | |
354 | } | |
e7ebc36c AT |
355 | matched(f,s,buf,len,-1); |
356 | } | |
9e31c482 | 357 | |
e7ebc36c | 358 | sum_end(file_sum); |
edecdad5 WD |
359 | /* If we had a read error, send a bad checksum. */ |
360 | if (buf && buf->status != 0) | |
361 | file_sum[0]++; | |
c627d613 | 362 | |
bc63ae3f S |
363 | if (verbose > 2) |
364 | rprintf(FINFO,"sending file_sum\n"); | |
365 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
c627d613 | 366 | |
e7ebc36c AT |
367 | if (targets) { |
368 | free(targets); | |
369 | targets=NULL; | |
370 | } | |
1e4f48d6 | 371 | |
e7ebc36c | 372 | if (verbose > 2) |
9486289c | 373 | rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", |
e7ebc36c | 374 | false_alarms, tag_hits, matches); |
1e4f48d6 | 375 | |
e7ebc36c AT |
376 | total_tag_hits += tag_hits; |
377 | total_false_alarms += false_alarms; | |
378 | total_matches += matches; | |
a800434a | 379 | stats.literal_data += data_transfer; |
c627d613 AT |
380 | } |
381 | ||
382 | void match_report(void) | |
383 | { | |
e7ebc36c AT |
384 | if (verbose <= 1) |
385 | return; | |
c627d613 | 386 | |
9486289c | 387 | rprintf(FINFO, |
a800434a | 388 | "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", |
e7ebc36c | 389 | total_matches,total_tag_hits, |
a800434a AT |
390 | total_false_alarms, |
391 | (double)stats.literal_data); | |
c627d613 | 392 | } |