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; |
c627d613 AT |
26 | |
27 | typedef unsigned short tag; | |
28 | ||
29 | #define TABLESIZE (1<<16) | |
0e36d9da | 30 | #define NULL_TAG ((size_t)-1) |
c627d613 AT |
31 | |
32 | static int false_alarms; | |
33 | static int tag_hits; | |
34 | static int matches; | |
a800434a | 35 | static int64 data_transfer; |
c627d613 | 36 | |
3a6a366f AT |
37 | static int total_false_alarms; |
38 | static int total_tag_hits; | |
39 | static int total_matches; | |
c627d613 | 40 | |
a800434a | 41 | extern struct stats stats; |
c627d613 AT |
42 | |
43 | struct target { | |
1e4f48d6 | 44 | tag t; |
0e36d9da | 45 | size_t i; |
c627d613 AT |
46 | }; |
47 | ||
3a6a366f | 48 | static struct target *targets; |
c627d613 | 49 | |
0e36d9da | 50 | static size_t *tag_table; |
c627d613 AT |
51 | |
52 | #define gettag2(s1,s2) (((s1) + (s2)) & 0xFFFF) | |
53 | #define gettag(sum) gettag2((sum)&0xFFFF,(sum)>>16) | |
54 | ||
55 | static int compare_targets(struct target *t1,struct target *t2) | |
56 | { | |
1e4f48d6 | 57 | return (int)t1->t - (int)t2->t; |
c627d613 AT |
58 | } |
59 | ||
60 | ||
61 | static void build_hash_table(struct sum_struct *s) | |
62 | { | |
0e36d9da | 63 | size_t i; |
c627d613 | 64 | |
1e4f48d6 | 65 | if (!tag_table) |
0e36d9da | 66 | tag_table = new_array(size_t, TABLESIZE); |
c627d613 | 67 | |
58cadc86 | 68 | targets = new_array(struct target, s->count); |
1e4f48d6 WD |
69 | if (!tag_table || !targets) |
70 | out_of_memory("build_hash_table"); | |
c627d613 | 71 | |
0e36d9da | 72 | for (i = 0; i < s->count; i++) { |
1e4f48d6 WD |
73 | targets[i].i = i; |
74 | targets[i].t = gettag(s->sums[i].sum1); | |
75 | } | |
c627d613 | 76 | |
1e4f48d6 | 77 | qsort(targets,s->count,sizeof(targets[0]),(int (*)())compare_targets); |
c627d613 | 78 | |
1e4f48d6 WD |
79 | for (i = 0; i < TABLESIZE; i++) |
80 | tag_table[i] = NULL_TAG; | |
c627d613 | 81 | |
0e36d9da | 82 | for (i = s->count; i-- > 0; ) |
1e4f48d6 | 83 | tag_table[targets[i].t] = i; |
c627d613 AT |
84 | } |
85 | ||
86 | ||
bcacc18b | 87 | static OFF_T last_match; |
c627d613 AT |
88 | |
89 | ||
f5f95a38 MP |
90 | /** |
91 | * Transmit a literal and/or match token. | |
92 | * | |
93 | * This delightfully-named function is called either when we find a | |
94 | * match and need to transmit all the unmatched data leading up to it, | |
95 | * or when we get bored of accumulating literal data and just need to | |
96 | * transmit it. As a result of this second case, it is called even if | |
97 | * we have not matched at all! | |
98 | * | |
99 | * @param i If >0, the number of a matched token. If 0, indicates we | |
100 | * have only literal data. | |
101 | **/ | |
c6e7fcb4 | 102 | static void matched(int f,struct sum_struct *s,struct map_struct *buf, |
bcacc18b | 103 | OFF_T offset,int i) |
c627d613 | 104 | { |
bcacc18b | 105 | OFF_T n = offset - last_match; |
1309d90d | 106 | OFF_T j; |
9e31c482 | 107 | |
0d0e2e93 | 108 | if (verbose > 2 && i >= 0) |
da9d12f5 | 109 | rprintf(FINFO,"match at %.0f last_match=%.0f j=%d len=%u n=%.0f\n", |
5f808dfb | 110 | (double)offset,(double)last_match,i,s->sums[i].len,(double)n); |
c627d613 | 111 | |
45f133b9 | 112 | send_token(f,i,buf,last_match,n,i<0?0:s->sums[i].len); |
e7ebc36c | 113 | data_transfer += n; |
70d794dc | 114 | |
a800434a AT |
115 | if (i >= 0) { |
116 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 117 | n += s->sums[i].len; |
a800434a | 118 | } |
1e4f48d6 WD |
119 | |
120 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
e7ebc36c AT |
121 | int n1 = MIN(CHUNK_SIZE,n-j); |
122 | sum_update(map_ptr(buf,last_match+j,n1),n1); | |
123 | } | |
9e31c482 | 124 | |
9e31c482 | 125 | |
0d0e2e93 | 126 | if (i >= 0) |
e7ebc36c | 127 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
128 | else |
129 | last_match = offset; | |
eb86d661 | 130 | |
16417f8b | 131 | if (buf && do_progress) { |
4440b8aa | 132 | show_progress(last_match, buf->file_size); |
eb86d661 | 133 | |
0e36d9da WD |
134 | if (i == -1) |
135 | end_progress(buf->file_size); | |
5e12ce11 | 136 | } |
9e31c482 AT |
137 | } |
138 | ||
139 | ||
c6e7fcb4 | 140 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 141 | struct map_struct *buf, OFF_T len) |
c627d613 | 142 | { |
5f808dfb | 143 | OFF_T offset, end; |
0e36d9da WD |
144 | unsigned int k; |
145 | size_t last_i; | |
e7ebc36c | 146 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 147 | uint32 s1, s2, sum; |
debb4505 | 148 | schar *map; |
e7ebc36c | 149 | |
923fa978 AT |
150 | /* last_i is used to encourage adjacent matches, allowing the RLL coding of the |
151 | output to work more efficiently */ | |
0e36d9da | 152 | last_i = (size_t)-1; |
923fa978 | 153 | |
0e36d9da WD |
154 | if (verbose > 2) { |
155 | rprintf(FINFO,"hash search b=%u len=%.0f\n", | |
156 | s->blength, (double)len); | |
157 | } | |
e7ebc36c | 158 | |
0e36d9da | 159 | k = MIN(len, s->blength); |
1e4f48d6 | 160 | |
0e36d9da | 161 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 162 | |
e7ebc36c AT |
163 | sum = get_checksum1((char *)map, k); |
164 | s1 = sum & 0xFFFF; | |
165 | s2 = sum >> 16; | |
166 | if (verbose > 3) | |
0e36d9da | 167 | rprintf(FINFO, "sum=%.8x k=%u\n", sum, k); |
1e4f48d6 | 168 | |
e7ebc36c | 169 | offset = 0; |
1e4f48d6 | 170 | |
e7ebc36c | 171 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 172 | |
0e36d9da WD |
173 | if (verbose > 3) { |
174 | rprintf(FINFO, "hash search s->blength=%u len=%.0f count=%.0f\n", | |
175 | s->blength, (double)len, (double)s->count); | |
176 | } | |
1e4f48d6 | 177 | |
e7ebc36c AT |
178 | do { |
179 | tag t = gettag2(s1,s2); | |
180 | int done_csum2 = 0; | |
0e36d9da | 181 | size_t j = tag_table[t]; |
1e4f48d6 | 182 | |
e7ebc36c | 183 | if (verbose > 4) |
5f808dfb | 184 | rprintf(FINFO,"offset=%.0f sum=%08x\n",(double)offset,sum); |
1e4f48d6 | 185 | |
0e36d9da | 186 | if (j == NULL_TAG) |
e7ebc36c | 187 | goto null_tag; |
e7ebc36c AT |
188 | |
189 | sum = (s1 & 0xffff) | (s2 << 16); | |
190 | tag_hits++; | |
0e36d9da | 191 | for (; j < s->count && targets[j].t == t; j++) { |
da9d12f5 | 192 | unsigned int l; |
0e36d9da | 193 | size_t i = targets[j].i; |
1e4f48d6 | 194 | |
0e36d9da WD |
195 | if (sum != s->sums[i].sum1) |
196 | continue; | |
1e4f48d6 | 197 | |
536541d5 | 198 | /* also make sure the two blocks are the same length */ |
c95dcb39 | 199 | l = MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
200 | if (l != s->sums[i].len) |
201 | continue; | |
536541d5 | 202 | |
e7ebc36c | 203 | if (verbose > 3) |
0e36d9da WD |
204 | rprintf(FINFO,"potential match at %.0f target=%.0f %.0f sum=%08x\n", |
205 | (double)offset,(double)j,(double)i,sum); | |
1e4f48d6 | 206 | |
e7ebc36c | 207 | if (!done_csum2) { |
debb4505 | 208 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
209 | get_checksum2((char *)map,l,sum2); |
210 | done_csum2 = 1; | |
211 | } | |
1e4f48d6 | 212 | |
fc0257c9 | 213 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
214 | false_alarms++; |
215 | continue; | |
216 | } | |
923fa978 AT |
217 | |
218 | /* we've found a match, but now check to see | |
1e4f48d6 | 219 | * if last_i can hint at a better match */ |
0e36d9da WD |
220 | for (j++; j < s->count && targets[j].t == t; j++) { |
221 | size_t i2 = targets[j].i; | |
923fa978 | 222 | if (i2 == last_i + 1) { |
0e36d9da WD |
223 | if (sum != s->sums[i2].sum1) |
224 | break; | |
225 | if (memcmp(sum2,s->sums[i2].sum2,s->s2length) != 0) | |
226 | break; | |
1e4f48d6 WD |
227 | /* we've found an adjacent match - the RLL coder |
228 | * will be happy */ | |
923fa978 AT |
229 | i = i2; |
230 | break; | |
231 | } | |
232 | } | |
233 | ||
234 | last_i = i; | |
1e4f48d6 | 235 | |
e7ebc36c AT |
236 | matched(f,s,buf,offset,i); |
237 | offset += s->sums[i].len - 1; | |
0e36d9da WD |
238 | k = MIN(s->blength, len-offset); |
239 | map = (schar *)map_ptr(buf, offset, k); | |
e7ebc36c AT |
240 | sum = get_checksum1((char *)map, k); |
241 | s1 = sum & 0xFFFF; | |
242 | s2 = sum >> 16; | |
243 | matches++; | |
244 | break; | |
245 | } | |
1e4f48d6 | 246 | |
e7ebc36c AT |
247 | null_tag: |
248 | /* Trim off the first byte from the checksum */ | |
0e36d9da | 249 | map = (schar *)map_ptr(buf, offset, k+1); |
e7ebc36c AT |
250 | s1 -= map[0] + CHAR_OFFSET; |
251 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 252 | |
e7ebc36c AT |
253 | /* Add on the next byte (if there is one) to the checksum */ |
254 | if (k < (len-offset)) { | |
255 | s1 += (map[k]+CHAR_OFFSET); | |
256 | s2 += s1; | |
0e36d9da | 257 | } else |
e7ebc36c | 258 | --k; |
45f133b9 | 259 | |
5914bf15 PM |
260 | /* By matching early we avoid re-reading the |
261 | data 3 times in the case where a token | |
262 | match comes a long way after last | |
263 | match. The 3 reads are caused by the | |
264 | running match, the checksum update and the | |
265 | literal send. */ | |
1e4f48d6 WD |
266 | if (offset > last_match |
267 | && offset-last_match >= CHUNK_SIZE+s->blength | |
268 | && end-offset > CHUNK_SIZE) { | |
fc0257c9 | 269 | matched(f,s,buf,offset - s->blength, -2); |
45f133b9 | 270 | } |
e7ebc36c | 271 | } while (++offset < end); |
1e4f48d6 | 272 | |
e7ebc36c AT |
273 | matched(f,s,buf,len,-1); |
274 | map_ptr(buf,len-1,1); | |
c627d613 AT |
275 | } |
276 | ||
277 | ||
f5f95a38 MP |
278 | /** |
279 | * Scan through a origin file, looking for sections that match | |
280 | * checksums from the generator, and transmit either literal or token | |
281 | * data. | |
282 | * | |
538ba24f MP |
283 | * Also calculates the MD4 checksum of the whole file, using the md |
284 | * accumulator. This is transmitted with the file as protection | |
285 | * against corruption on the wire. | |
286 | * | |
f5f95a38 MP |
287 | * @param s Checksums received from the generator. If <tt>s->count == |
288 | * 0</tt>, then there are actually no checksums for this file. | |
289 | * | |
290 | * @param len Length of the file to send. | |
291 | **/ | |
292 | void match_sums(int f, struct sum_struct *s, struct map_struct *buf, OFF_T len) | |
c627d613 | 293 | { |
e7ebc36c | 294 | char file_sum[MD4_SUM_LENGTH]; |
64c3523a | 295 | extern int write_batch; |
e7ebc36c AT |
296 | |
297 | last_match = 0; | |
298 | false_alarms = 0; | |
299 | tag_hits = 0; | |
1e4f48d6 WD |
300 | matches = 0; |
301 | data_transfer = 0; | |
e7ebc36c | 302 | |
ba582f75 | 303 | sum_init(checksum_seed); |
e7ebc36c AT |
304 | |
305 | if (len > 0 && s->count>0) { | |
306 | build_hash_table(s); | |
1e4f48d6 WD |
307 | |
308 | if (verbose > 2) | |
9486289c | 309 | rprintf(FINFO,"built hash table\n"); |
1e4f48d6 | 310 | |
e7ebc36c | 311 | hash_search(f,s,buf,len); |
1e4f48d6 WD |
312 | |
313 | if (verbose > 2) | |
9486289c | 314 | rprintf(FINFO,"done hash search\n"); |
e7ebc36c | 315 | } else { |
eb86d661 AT |
316 | OFF_T j; |
317 | /* by doing this in pieces we avoid too many seeks */ | |
1e4f48d6 | 318 | for (j = 0; j < len-CHUNK_SIZE; j += CHUNK_SIZE) { |
eb86d661 AT |
319 | int n1 = MIN(CHUNK_SIZE,(len-CHUNK_SIZE)-j); |
320 | matched(f,s,buf,j+n1,-2); | |
321 | } | |
e7ebc36c AT |
322 | matched(f,s,buf,len,-1); |
323 | } | |
9e31c482 | 324 | |
e7ebc36c | 325 | sum_end(file_sum); |
edecdad5 WD |
326 | /* If we had a read error, send a bad checksum. */ |
327 | if (buf && buf->status != 0) | |
328 | file_sum[0]++; | |
c627d613 | 329 | |
bc63ae3f S |
330 | if (verbose > 2) |
331 | rprintf(FINFO,"sending file_sum\n"); | |
332 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
64c3523a | 333 | if (write_batch) |
bc63ae3f | 334 | write_batch_delta_file(file_sum, MD4_SUM_LENGTH); |
c627d613 | 335 | |
e7ebc36c AT |
336 | if (targets) { |
337 | free(targets); | |
338 | targets=NULL; | |
339 | } | |
1e4f48d6 | 340 | |
e7ebc36c | 341 | if (verbose > 2) |
9486289c | 342 | rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", |
e7ebc36c | 343 | false_alarms, tag_hits, matches); |
1e4f48d6 | 344 | |
e7ebc36c AT |
345 | total_tag_hits += tag_hits; |
346 | total_false_alarms += false_alarms; | |
347 | total_matches += matches; | |
a800434a | 348 | stats.literal_data += data_transfer; |
c627d613 AT |
349 | } |
350 | ||
351 | void match_report(void) | |
352 | { | |
e7ebc36c AT |
353 | if (verbose <= 1) |
354 | return; | |
c627d613 | 355 | |
9486289c | 356 | rprintf(FINFO, |
a800434a | 357 | "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", |
e7ebc36c | 358 | total_matches,total_tag_hits, |
a800434a AT |
359 | total_false_alarms, |
360 | (double)stats.literal_data); | |
c627d613 | 361 | } |