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; |
c627d613 AT |
27 | |
28 | typedef unsigned short tag; | |
29 | ||
30 | #define TABLESIZE (1<<16) | |
0e36d9da | 31 | #define NULL_TAG ((size_t)-1) |
c627d613 AT |
32 | |
33 | static int false_alarms; | |
34 | static int tag_hits; | |
35 | static int matches; | |
a800434a | 36 | static int64 data_transfer; |
c627d613 | 37 | |
3a6a366f AT |
38 | static int total_false_alarms; |
39 | static int total_tag_hits; | |
40 | static int total_matches; | |
c627d613 | 41 | |
a800434a | 42 | extern struct stats stats; |
c627d613 AT |
43 | |
44 | struct target { | |
1e4f48d6 | 45 | tag t; |
0e36d9da | 46 | size_t i; |
c627d613 AT |
47 | }; |
48 | ||
3a6a366f | 49 | static struct target *targets; |
c627d613 | 50 | |
0e36d9da | 51 | static size_t *tag_table; |
c627d613 AT |
52 | |
53 | #define gettag2(s1,s2) (((s1) + (s2)) & 0xFFFF) | |
54 | #define gettag(sum) gettag2((sum)&0xFFFF,(sum)>>16) | |
55 | ||
56 | static int compare_targets(struct target *t1,struct target *t2) | |
57 | { | |
1e4f48d6 | 58 | return (int)t1->t - (int)t2->t; |
c627d613 AT |
59 | } |
60 | ||
61 | ||
62 | static void build_hash_table(struct sum_struct *s) | |
63 | { | |
0e36d9da | 64 | size_t i; |
c627d613 | 65 | |
1e4f48d6 | 66 | if (!tag_table) |
0e36d9da | 67 | tag_table = new_array(size_t, TABLESIZE); |
c627d613 | 68 | |
58cadc86 | 69 | targets = new_array(struct target, s->count); |
1e4f48d6 WD |
70 | if (!tag_table || !targets) |
71 | out_of_memory("build_hash_table"); | |
c627d613 | 72 | |
0e36d9da | 73 | for (i = 0; i < s->count; i++) { |
1e4f48d6 WD |
74 | targets[i].i = i; |
75 | targets[i].t = gettag(s->sums[i].sum1); | |
76 | } | |
c627d613 | 77 | |
1e4f48d6 | 78 | qsort(targets,s->count,sizeof(targets[0]),(int (*)())compare_targets); |
c627d613 | 79 | |
1e4f48d6 WD |
80 | for (i = 0; i < TABLESIZE; i++) |
81 | tag_table[i] = NULL_TAG; | |
c627d613 | 82 | |
0e36d9da | 83 | for (i = s->count; i-- > 0; ) |
1e4f48d6 | 84 | tag_table[targets[i].t] = i; |
c627d613 AT |
85 | } |
86 | ||
87 | ||
bcacc18b | 88 | static OFF_T last_match; |
c627d613 AT |
89 | |
90 | ||
f5f95a38 MP |
91 | /** |
92 | * Transmit a literal and/or match token. | |
93 | * | |
94 | * This delightfully-named function is called either when we find a | |
95 | * match and need to transmit all the unmatched data leading up to it, | |
96 | * or when we get bored of accumulating literal data and just need to | |
97 | * transmit it. As a result of this second case, it is called even if | |
98 | * we have not matched at all! | |
99 | * | |
100 | * @param i If >0, the number of a matched token. If 0, indicates we | |
101 | * have only literal data. | |
102 | **/ | |
c6e7fcb4 | 103 | static void matched(int f,struct sum_struct *s,struct map_struct *buf, |
bcacc18b | 104 | OFF_T offset,int i) |
c627d613 | 105 | { |
bcacc18b | 106 | OFF_T n = offset - last_match; |
1309d90d | 107 | OFF_T j; |
9e31c482 | 108 | |
0d0e2e93 | 109 | if (verbose > 2 && i >= 0) |
da9d12f5 | 110 | rprintf(FINFO,"match at %.0f last_match=%.0f j=%d len=%u n=%.0f\n", |
5f808dfb | 111 | (double)offset,(double)last_match,i,s->sums[i].len,(double)n); |
c627d613 | 112 | |
45f133b9 | 113 | send_token(f,i,buf,last_match,n,i<0?0:s->sums[i].len); |
e7ebc36c | 114 | data_transfer += n; |
70d794dc | 115 | |
a800434a AT |
116 | if (i >= 0) { |
117 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 118 | n += s->sums[i].len; |
a800434a | 119 | } |
1e4f48d6 WD |
120 | |
121 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
e7ebc36c AT |
122 | int n1 = MIN(CHUNK_SIZE,n-j); |
123 | sum_update(map_ptr(buf,last_match+j,n1),n1); | |
124 | } | |
9e31c482 | 125 | |
9e31c482 | 126 | |
0d0e2e93 | 127 | if (i >= 0) |
e7ebc36c | 128 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
129 | else |
130 | last_match = offset; | |
eb86d661 | 131 | |
16417f8b | 132 | if (buf && do_progress) { |
4440b8aa | 133 | show_progress(last_match, buf->file_size); |
eb86d661 | 134 | |
0e36d9da WD |
135 | if (i == -1) |
136 | end_progress(buf->file_size); | |
5e12ce11 | 137 | } |
9e31c482 AT |
138 | } |
139 | ||
140 | ||
c6e7fcb4 | 141 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 142 | struct map_struct *buf, OFF_T len) |
c627d613 | 143 | { |
7560c17a | 144 | OFF_T offset, end, backup; |
0e36d9da | 145 | unsigned int k; |
a04d77bc | 146 | size_t 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 WD |
156 | if (verbose > 2) { |
157 | rprintf(FINFO,"hash search b=%u len=%.0f\n", | |
158 | s->blength, (double)len); | |
159 | } | |
e7ebc36c | 160 | |
0e36d9da | 161 | k = MIN(len, 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) | |
0e36d9da | 169 | rprintf(FINFO, "sum=%.8x k=%u\n", sum, k); |
1e4f48d6 | 170 | |
e7ebc36c | 171 | offset = 0; |
1e4f48d6 | 172 | |
e7ebc36c | 173 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 174 | |
0e36d9da WD |
175 | if (verbose > 3) { |
176 | rprintf(FINFO, "hash search s->blength=%u len=%.0f count=%.0f\n", | |
177 | s->blength, (double)len, (double)s->count); | |
178 | } | |
1e4f48d6 | 179 | |
e7ebc36c AT |
180 | do { |
181 | tag t = gettag2(s1,s2); | |
182 | int done_csum2 = 0; | |
0e36d9da | 183 | size_t 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 { |
da9d12f5 | 194 | unsigned int l; |
0e36d9da | 195 | size_t i = targets[j].i; |
1e4f48d6 | 196 | |
0e36d9da WD |
197 | if (sum != s->sums[i].sum1) |
198 | continue; | |
1e4f48d6 | 199 | |
536541d5 | 200 | /* also make sure the two blocks are the same length */ |
c95dcb39 | 201 | l = MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
202 | if (l != s->sums[i].len) |
203 | continue; | |
536541d5 | 204 | |
a3221d2a WD |
205 | /* inplace: ensure chunk's offset is either >= our |
206 | * offset or that the data didn't move. */ | |
207 | if (inplace && s->sums[i].offset < offset | |
208 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) | |
209 | continue; | |
210 | ||
e7ebc36c | 211 | if (verbose > 3) |
0e36d9da WD |
212 | rprintf(FINFO,"potential match at %.0f target=%.0f %.0f sum=%08x\n", |
213 | (double)offset,(double)j,(double)i,sum); | |
1e4f48d6 | 214 | |
e7ebc36c | 215 | if (!done_csum2) { |
debb4505 | 216 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
217 | get_checksum2((char *)map,l,sum2); |
218 | done_csum2 = 1; | |
219 | } | |
1e4f48d6 | 220 | |
fc0257c9 | 221 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
222 | false_alarms++; |
223 | continue; | |
224 | } | |
923fa978 | 225 | |
a3221d2a WD |
226 | /* If inplace is enabled, the best possible match is |
227 | * one with an identical offset, so we prefer that over | |
228 | * the following want_i optimization. */ | |
229 | if (inplace) { | |
230 | do { | |
231 | size_t i2 = targets[j].i; | |
232 | if (s->sums[i2].offset != offset) | |
233 | continue; | |
234 | if (i2 != i) { | |
235 | if (sum != s->sums[i2].sum1) | |
236 | break; | |
237 | if (memcmp(sum2, s->sums[i2].sum2, | |
238 | s->s2length) != 0) | |
239 | break; | |
240 | i = i2; | |
241 | } | |
242 | /* This chunk was at the same offset on | |
243 | * both the sender and the receiver. */ | |
244 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
245 | goto set_want_i; | |
246 | } while (++j < s->count && targets[j].t == t); | |
247 | } | |
248 | ||
923fa978 | 249 | /* we've found a match, but now check to see |
a04d77bc WD |
250 | * if want_i can hint at a better match. */ |
251 | if (i != want_i && want_i < s->count | |
a3221d2a WD |
252 | && (!inplace || s->sums[want_i].offset >= offset |
253 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) | |
a04d77bc WD |
254 | && sum == s->sums[want_i].sum1 |
255 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
256 | /* we've found an adjacent match - the RLL coder |
257 | * will be happy */ | |
a04d77bc | 258 | i = want_i; |
923fa978 | 259 | } |
a3221d2a | 260 | set_want_i: |
a04d77bc | 261 | want_i = i + 1; |
1e4f48d6 | 262 | |
e7ebc36c AT |
263 | matched(f,s,buf,offset,i); |
264 | offset += s->sums[i].len - 1; | |
0e36d9da WD |
265 | k = MIN(s->blength, len-offset); |
266 | map = (schar *)map_ptr(buf, offset, k); | |
e7ebc36c AT |
267 | sum = get_checksum1((char *)map, k); |
268 | s1 = sum & 0xFFFF; | |
269 | s2 = sum >> 16; | |
270 | matches++; | |
271 | break; | |
066a844c | 272 | } while (++j < s->count && targets[j].t == t); |
1e4f48d6 | 273 | |
e7ebc36c | 274 | null_tag: |
7560c17a WD |
275 | backup = offset - last_match; |
276 | /* We sometimes read 1 byte prior to last_match... */ | |
277 | if (backup < 0) | |
278 | backup = 0; | |
279 | ||
e7ebc36c | 280 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
281 | more = offset + k < len; |
282 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
283 | + backup; | |
e7ebc36c AT |
284 | s1 -= map[0] + CHAR_OFFSET; |
285 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 286 | |
e7ebc36c | 287 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
288 | if (more) { |
289 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 290 | s2 += s1; |
0e36d9da | 291 | } else |
e7ebc36c | 292 | --k; |
45f133b9 | 293 | |
5914bf15 PM |
294 | /* By matching early we avoid re-reading the |
295 | data 3 times in the case where a token | |
296 | match comes a long way after last | |
297 | match. The 3 reads are caused by the | |
298 | running match, the checksum update and the | |
299 | literal send. */ | |
7560c17a WD |
300 | if (backup >= CHUNK_SIZE + s->blength |
301 | && end - offset > CHUNK_SIZE) { | |
fc0257c9 | 302 | matched(f,s,buf,offset - s->blength, -2); |
45f133b9 | 303 | } |
e7ebc36c | 304 | } while (++offset < end); |
1e4f48d6 | 305 | |
e7ebc36c AT |
306 | matched(f,s,buf,len,-1); |
307 | map_ptr(buf,len-1,1); | |
c627d613 AT |
308 | } |
309 | ||
310 | ||
f5f95a38 MP |
311 | /** |
312 | * Scan through a origin file, looking for sections that match | |
313 | * checksums from the generator, and transmit either literal or token | |
314 | * data. | |
315 | * | |
538ba24f MP |
316 | * Also calculates the MD4 checksum of the whole file, using the md |
317 | * accumulator. This is transmitted with the file as protection | |
318 | * against corruption on the wire. | |
319 | * | |
f5f95a38 MP |
320 | * @param s Checksums received from the generator. If <tt>s->count == |
321 | * 0</tt>, then there are actually no checksums for this file. | |
322 | * | |
323 | * @param len Length of the file to send. | |
324 | **/ | |
325 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 326 | { |
e7ebc36c AT |
327 | char file_sum[MD4_SUM_LENGTH]; |
328 | ||
329 | last_match = 0; | |
330 | false_alarms = 0; | |
331 | tag_hits = 0; | |
1e4f48d6 WD |
332 | matches = 0; |
333 | data_transfer = 0; | |
e7ebc36c | 334 | |
ba582f75 | 335 | sum_init(checksum_seed); |
e7ebc36c AT |
336 | |
337 | if (len > 0 && s->count>0) { | |
338 | build_hash_table(s); | |
1e4f48d6 WD |
339 | |
340 | if (verbose > 2) | |
9486289c | 341 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 342 | |
e7ebc36c | 343 | hash_search(f,s,buf,len); |
1e4f48d6 WD |
344 | |
345 | if (verbose > 2) | |
9486289c | 346 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 347 | } else { |
eb86d661 AT |
348 | OFF_T j; |
349 | /* by doing this in pieces we avoid too many seeks */ | |
1e4f48d6 | 350 | for (j = 0; j < len-CHUNK_SIZE; j += CHUNK_SIZE) { |
eb86d661 AT |
351 | int n1 = MIN(CHUNK_SIZE,(len-CHUNK_SIZE)-j); |
352 | matched(f,s,buf,j+n1,-2); | |
353 | } | |
e7ebc36c AT |
354 | matched(f,s,buf,len,-1); |
355 | } | |
9e31c482 | 356 | |
e7ebc36c | 357 | sum_end(file_sum); |
edecdad5 WD |
358 | /* If we had a read error, send a bad checksum. */ |
359 | if (buf && buf->status != 0) | |
360 | file_sum[0]++; | |
c627d613 | 361 | |
bc63ae3f S |
362 | if (verbose > 2) |
363 | rprintf(FINFO,"sending file_sum\n"); | |
364 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
c627d613 | 365 | |
e7ebc36c AT |
366 | if (targets) { |
367 | free(targets); | |
368 | targets=NULL; | |
369 | } | |
1e4f48d6 | 370 | |
e7ebc36c | 371 | if (verbose > 2) |
9486289c | 372 | rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", |
e7ebc36c | 373 | false_alarms, tag_hits, matches); |
1e4f48d6 | 374 | |
e7ebc36c AT |
375 | total_tag_hits += tag_hits; |
376 | total_false_alarms += false_alarms; | |
377 | total_matches += matches; | |
a800434a | 378 | stats.literal_data += data_transfer; |
c627d613 AT |
379 | } |
380 | ||
381 | void match_report(void) | |
382 | { | |
e7ebc36c AT |
383 | if (verbose <= 1) |
384 | return; | |
c627d613 | 385 | |
9486289c | 386 | rprintf(FINFO, |
a800434a | 387 | "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", |
e7ebc36c | 388 | total_matches,total_tag_hits, |
a800434a AT |
389 | total_false_alarms, |
390 | (double)stats.literal_data); | |
c627d613 | 391 | } |