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; |
a36ffd39 WD |
26 | |
27 | int updating_basis_file; | |
c627d613 AT |
28 | |
29 | typedef unsigned short tag; | |
30 | ||
31 | #define TABLESIZE (1<<16) | |
7cacd47e | 32 | #define NULL_TAG (-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; |
7cacd47e | 47 | int32 i; |
c627d613 AT |
48 | }; |
49 | ||
3a6a366f | 50 | static struct target *targets; |
c627d613 | 51 | |
7cacd47e | 52 | static int32 *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 | { | |
7cacd47e | 65 | int32 i; |
c627d613 | 66 | |
1e4f48d6 | 67 | if (!tag_table) |
7cacd47e | 68 | tag_table = new_array(int32, 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 | **/ | |
a06b419d | 104 | static void matched(int f, struct sum_struct *s, struct map_struct *buf, |
37a56445 | 105 | OFF_T offset, int32 i) |
c627d613 | 106 | { |
a06b419d WD |
107 | int32 n = offset - last_match; /* max value: block_size (int32) */ |
108 | int32 j; | |
109 | ||
110 | if (verbose > 2 && i >= 0) { | |
111 | rprintf(FINFO, | |
112 | "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n", | |
113 | (double)offset, (double)last_match, i, | |
114 | (long)s->sums[i].len, (long)n); | |
115 | } | |
c627d613 | 116 | |
a06b419d | 117 | send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len); |
e7ebc36c | 118 | data_transfer += n; |
70d794dc | 119 | |
a800434a AT |
120 | if (i >= 0) { |
121 | stats.matched_data += s->sums[i].len; | |
e7ebc36c | 122 | n += s->sums[i].len; |
a800434a | 123 | } |
1e4f48d6 WD |
124 | |
125 | for (j = 0; j < n; j += CHUNK_SIZE) { | |
a06b419d WD |
126 | int32 n1 = MIN(CHUNK_SIZE, n - j); |
127 | sum_update(map_ptr(buf, last_match + j, n1), n1); | |
e7ebc36c | 128 | } |
9e31c482 | 129 | |
9e31c482 | 130 | |
0d0e2e93 | 131 | if (i >= 0) |
e7ebc36c | 132 | last_match = offset + s->sums[i].len; |
0d0e2e93 AT |
133 | else |
134 | last_match = offset; | |
eb86d661 | 135 | |
16417f8b | 136 | if (buf && do_progress) { |
4440b8aa | 137 | show_progress(last_match, buf->file_size); |
eb86d661 | 138 | |
0e36d9da WD |
139 | if (i == -1) |
140 | end_progress(buf->file_size); | |
5e12ce11 | 141 | } |
9e31c482 AT |
142 | } |
143 | ||
144 | ||
c6e7fcb4 | 145 | static void hash_search(int f,struct sum_struct *s, |
da9d12f5 | 146 | struct map_struct *buf, OFF_T len) |
c627d613 | 147 | { |
7560c17a | 148 | OFF_T offset, end, backup; |
7cacd47e | 149 | int32 k, want_i; |
e7ebc36c | 150 | char sum2[SUM_LENGTH]; |
1e4f48d6 | 151 | uint32 s1, s2, sum; |
7560c17a | 152 | int more; |
debb4505 | 153 | schar *map; |
e7ebc36c | 154 | |
a04d77bc WD |
155 | /* want_i is used to encourage adjacent matches, allowing the RLL |
156 | * coding of the output to work more efficiently. */ | |
157 | want_i = 0; | |
923fa978 | 158 | |
0e36d9da | 159 | if (verbose > 2) { |
a06b419d WD |
160 | rprintf(FINFO, "hash search b=%ld len=%.0f\n", |
161 | (long)s->blength, (double)len); | |
0e36d9da | 162 | } |
e7ebc36c | 163 | |
a06b419d | 164 | k = (int32)MIN(len, (OFF_T)s->blength); |
1e4f48d6 | 165 | |
0e36d9da | 166 | map = (schar *)map_ptr(buf, 0, k); |
1e4f48d6 | 167 | |
e7ebc36c AT |
168 | sum = get_checksum1((char *)map, k); |
169 | s1 = sum & 0xFFFF; | |
170 | s2 = sum >> 16; | |
171 | if (verbose > 3) | |
a06b419d | 172 | rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k); |
1e4f48d6 | 173 | |
e7ebc36c | 174 | offset = 0; |
1e4f48d6 | 175 | |
e7ebc36c | 176 | end = len + 1 - s->sums[s->count-1].len; |
1e4f48d6 | 177 | |
0e36d9da | 178 | if (verbose > 3) { |
a06b419d WD |
179 | rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n", |
180 | (long)s->blength, (double)len, (double)s->count); | |
0e36d9da | 181 | } |
1e4f48d6 | 182 | |
e7ebc36c AT |
183 | do { |
184 | tag t = gettag2(s1,s2); | |
185 | int done_csum2 = 0; | |
7cacd47e | 186 | int32 j = tag_table[t]; |
1e4f48d6 | 187 | |
e7ebc36c | 188 | if (verbose > 4) |
5f808dfb | 189 | rprintf(FINFO,"offset=%.0f sum=%08x\n",(double)offset,sum); |
1e4f48d6 | 190 | |
0e36d9da | 191 | if (j == NULL_TAG) |
e7ebc36c | 192 | goto null_tag; |
e7ebc36c AT |
193 | |
194 | sum = (s1 & 0xffff) | (s2 << 16); | |
195 | tag_hits++; | |
066a844c | 196 | do { |
7cacd47e | 197 | int32 l, i = targets[j].i; |
1e4f48d6 | 198 | |
0e36d9da WD |
199 | if (sum != s->sums[i].sum1) |
200 | continue; | |
1e4f48d6 | 201 | |
536541d5 | 202 | /* also make sure the two blocks are the same length */ |
a06b419d | 203 | l = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da WD |
204 | if (l != s->sums[i].len) |
205 | continue; | |
536541d5 | 206 | |
a36ffd39 | 207 | /* in-place: ensure chunk's offset is either >= our |
a3221d2a | 208 | * offset or that the data didn't move. */ |
a36ffd39 | 209 | if (updating_basis_file && s->sums[i].offset < offset |
a3221d2a WD |
210 | && !(s->sums[i].flags & SUMFLG_SAME_OFFSET)) |
211 | continue; | |
212 | ||
e7ebc36c | 213 | if (verbose > 3) |
0e36d9da WD |
214 | rprintf(FINFO,"potential match at %.0f target=%.0f %.0f sum=%08x\n", |
215 | (double)offset,(double)j,(double)i,sum); | |
1e4f48d6 | 216 | |
e7ebc36c | 217 | if (!done_csum2) { |
debb4505 | 218 | map = (schar *)map_ptr(buf,offset,l); |
e7ebc36c AT |
219 | get_checksum2((char *)map,l,sum2); |
220 | done_csum2 = 1; | |
221 | } | |
1e4f48d6 | 222 | |
fc0257c9 | 223 | if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) { |
e7ebc36c AT |
224 | false_alarms++; |
225 | continue; | |
226 | } | |
923fa978 | 227 | |
a36ffd39 | 228 | /* When updating in-place, the best possible match is |
a3221d2a WD |
229 | * one with an identical offset, so we prefer that over |
230 | * the following want_i optimization. */ | |
a36ffd39 | 231 | if (updating_basis_file) { |
a3221d2a | 232 | do { |
7cacd47e | 233 | int32 i2 = targets[j].i; |
a3221d2a WD |
234 | if (s->sums[i2].offset != offset) |
235 | continue; | |
236 | if (i2 != i) { | |
237 | if (sum != s->sums[i2].sum1) | |
238 | break; | |
239 | if (memcmp(sum2, s->sums[i2].sum2, | |
240 | s->s2length) != 0) | |
241 | break; | |
242 | i = i2; | |
243 | } | |
244 | /* This chunk was at the same offset on | |
245 | * both the sender and the receiver. */ | |
246 | s->sums[i].flags |= SUMFLG_SAME_OFFSET; | |
247 | goto set_want_i; | |
248 | } while (++j < s->count && targets[j].t == t); | |
249 | } | |
250 | ||
923fa978 | 251 | /* we've found a match, but now check to see |
a04d77bc WD |
252 | * if want_i can hint at a better match. */ |
253 | if (i != want_i && want_i < s->count | |
a36ffd39 | 254 | && (!updating_basis_file || s->sums[want_i].offset >= offset |
a3221d2a | 255 | || s->sums[want_i].flags & SUMFLG_SAME_OFFSET) |
a04d77bc WD |
256 | && sum == s->sums[want_i].sum1 |
257 | && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) { | |
5e252dea WD |
258 | /* we've found an adjacent match - the RLL coder |
259 | * will be happy */ | |
a04d77bc | 260 | i = want_i; |
923fa978 | 261 | } |
a3221d2a | 262 | set_want_i: |
a04d77bc | 263 | want_i = i + 1; |
1e4f48d6 | 264 | |
e7ebc36c AT |
265 | matched(f,s,buf,offset,i); |
266 | offset += s->sums[i].len - 1; | |
a06b419d | 267 | k = (int32)MIN((OFF_T)s->blength, len-offset); |
0e36d9da | 268 | map = (schar *)map_ptr(buf, offset, k); |
e7ebc36c AT |
269 | sum = get_checksum1((char *)map, k); |
270 | s1 = sum & 0xFFFF; | |
271 | s2 = sum >> 16; | |
272 | matches++; | |
273 | break; | |
066a844c | 274 | } while (++j < s->count && targets[j].t == t); |
1e4f48d6 | 275 | |
e7ebc36c | 276 | null_tag: |
7560c17a WD |
277 | backup = offset - last_match; |
278 | /* We sometimes read 1 byte prior to last_match... */ | |
279 | if (backup < 0) | |
280 | backup = 0; | |
281 | ||
e7ebc36c | 282 | /* Trim off the first byte from the checksum */ |
7560c17a WD |
283 | more = offset + k < len; |
284 | map = (schar *)map_ptr(buf, offset - backup, k + more + backup) | |
285 | + backup; | |
e7ebc36c AT |
286 | s1 -= map[0] + CHAR_OFFSET; |
287 | s2 -= k * (map[0]+CHAR_OFFSET); | |
1e4f48d6 | 288 | |
e7ebc36c | 289 | /* Add on the next byte (if there is one) to the checksum */ |
7560c17a WD |
290 | if (more) { |
291 | s1 += map[k] + CHAR_OFFSET; | |
e7ebc36c | 292 | s2 += s1; |
0e36d9da | 293 | } else |
e7ebc36c | 294 | --k; |
45f133b9 | 295 | |
5914bf15 PM |
296 | /* By matching early we avoid re-reading the |
297 | data 3 times in the case where a token | |
298 | match comes a long way after last | |
299 | match. The 3 reads are caused by the | |
300 | running match, the checksum update and the | |
301 | literal send. */ | |
a06b419d WD |
302 | if (backup >= s->blength+CHUNK_SIZE && end-offset > CHUNK_SIZE) |
303 | matched(f, s, buf, offset - s->blength, -2); | |
e7ebc36c | 304 | } while (++offset < end); |
1e4f48d6 | 305 | |
a06b419d WD |
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 | 336 | |
a06b419d | 337 | if (len > 0 && s->count > 0) { |
e7ebc36c | 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 */ | |
a06b419d WD |
350 | for (j = CHUNK_SIZE; j < len; j += CHUNK_SIZE) |
351 | matched(f, s, buf, j, -2); | |
352 | matched(f, s, buf, len, -1); | |
e7ebc36c | 353 | } |
9e31c482 | 354 | |
e7ebc36c | 355 | sum_end(file_sum); |
edecdad5 WD |
356 | /* If we had a read error, send a bad checksum. */ |
357 | if (buf && buf->status != 0) | |
358 | file_sum[0]++; | |
c627d613 | 359 | |
bc63ae3f S |
360 | if (verbose > 2) |
361 | rprintf(FINFO,"sending file_sum\n"); | |
362 | write_buf(f,file_sum,MD4_SUM_LENGTH); | |
c627d613 | 363 | |
e7ebc36c AT |
364 | if (targets) { |
365 | free(targets); | |
366 | targets=NULL; | |
367 | } | |
1e4f48d6 | 368 | |
e7ebc36c | 369 | if (verbose > 2) |
9486289c | 370 | rprintf(FINFO, "false_alarms=%d tag_hits=%d matches=%d\n", |
e7ebc36c | 371 | false_alarms, tag_hits, matches); |
1e4f48d6 | 372 | |
e7ebc36c AT |
373 | total_tag_hits += tag_hits; |
374 | total_false_alarms += false_alarms; | |
375 | total_matches += matches; | |
a800434a | 376 | stats.literal_data += data_transfer; |
c627d613 AT |
377 | } |
378 | ||
379 | void match_report(void) | |
380 | { | |
e7ebc36c AT |
381 | if (verbose <= 1) |
382 | return; | |
c627d613 | 383 | |
9486289c | 384 | rprintf(FINFO, |
a800434a | 385 | "total: matches=%d tag_hits=%d false_alarms=%d data=%.0f\n", |
e7ebc36c | 386 | total_matches,total_tag_hits, |
a800434a AT |
387 | total_false_alarms, |
388 | (double)stats.literal_data); | |
c627d613 | 389 | } |