match.c

   1 /*
   2  * Block matching used by the file-transfer code.
   3  *
   4  * Copyright (C) 1996 Andrew Tridgell
   5  * Copyright (C) 1996 Paul Mackerras
   6  * Copyright (C) 2003-2007 Wayne Davison
   7  *
   8  * This program is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 3 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * This program is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with this program; if not, visit the http://fsf.org website.
  20  */
  21
  22 #include "rsync.h"
  23
  24 extern int verbose;
  25 extern int do_progress;
  26 extern int checksum_seed;
  27 extern int append_mode;
  28
  29 int updating_basis_file;
  30
  31 static int false_alarms;
  32 static int hash_hits;
  33 static int matches;
  34 static int64 data_transfer;
  35
  36 static int total_false_alarms;
  37 static int total_hash_hits;
  38 static int total_matches;
  39
  40 extern struct stats stats;
  41
  42 #define TABLESIZE (1<<16)
  43
  44 static int32 *hash_table;
  45
  46 #define SUM2HASH2(s1,s2) (((s1) + (s2)) & 0xFFFF)
  47 #define SUM2HASH(sum) SUM2HASH2((sum)&0xFFFF,(sum)>>16)
  48
  49 static int32 build_hash_table(struct sum_struct *s, int32 start)
  50 {
  51         int32 i, end = s->count;
  52
  53         if (!hash_table) {
  54                 hash_table = new_array(int32, TABLESIZE);
  55                 if (!hash_table)
  56                         out_of_memory("build_hash_table");
  57         }
  58
  59         memset(hash_table, 0xFF, TABLESIZE * sizeof hash_table[0]);
  60
  61         if (end - start > TABLESIZE*8/10)
  62                 end = start + TABLESIZE*8/10;
  63
  64         for (i = start; i < end; i++) {
  65                 uint32 t = SUM2HASH(s->sums[i].sum1);
  66                 s->sums[i].chain = hash_table[t];
  67                 hash_table[t] = i;
  68         }
  69
  70         if (verbose > 2) {
  71                 rprintf(FINFO, "built hash table for entries %ld - %ld\n",
  72                         (long)start, (long)end - 1);
  73         }
  74
  75         return end;
  76 }
  77
  78
  79 static OFF_T last_match;
  80
  81
  82 /**
  83  * Transmit a literal and/or match token.
  84  *
  85  * This delightfully-named function is called either when we find a
  86  * match and need to transmit all the unmatched data leading up to it,
  87  * or when we get bored of accumulating literal data and just need to
  88  * transmit it.  As a result of this second case, it is called even if
  89  * we have not matched at all!
  90  *
  91  * @param i If >0, the number of a matched token.  If 0, indicates we
  92  * have only literal data.
  93  **/
  94 static void matched(int f, struct sum_struct *s, struct map_struct *buf,
  95                     OFF_T offset, int32 i)
  96 {
  97         int32 n = (int32)(offset - last_match); /* max value: block_size (int32) */
  98         int32 j;
  99
 100         if (verbose > 2 && i >= 0) {
 101                 rprintf(FINFO,
 102                         "match at %.0f last_match=%.0f j=%d len=%ld n=%ld\n",
 103                         (double)offset, (double)last_match, i,
 104                         (long)s->sums[i].len, (long)n);
 105         }
 106
 107         send_token(f, i, buf, last_match, n, i < 0 ? 0 : s->sums[i].len);
 108         data_transfer += n;
 109
 110         if (i >= 0) {
 111                 stats.matched_data += s->sums[i].len;
 112                 n += s->sums[i].len;
 113         }
 114
 115         for (j = 0; j < n; j += CHUNK_SIZE) {
 116                 int32 n1 = MIN(CHUNK_SIZE, n - j);
 117                 sum_update(map_ptr(buf, last_match + j, n1), n1);
 118         }
 119
 120         if (i >= 0)
 121                 last_match = offset + s->sums[i].len;
 122         else
 123                 last_match = offset;
 124
 125         if (buf && do_progress)
 126                 show_progress(last_match, buf->file_size);
 127 }
 128
 129
 130 static void hash_search(int f,struct sum_struct *s,
 131                         struct map_struct *buf, OFF_T len)
 132 {
 133         OFF_T offset, end, reset = 0;
 134         int32 k, want_i, backup, sum_pos = 0;
 135         char sum2[SUM_LENGTH];
 136         uint32 s1, s2, sum;
 137         int more;
 138         schar *map;
 139
 140         /* want_i is used to encourage adjacent matches, allowing the RLL
 141          * coding of the output to work more efficiently. */
 142         want_i = 0;
 143
 144         if (verbose > 2) {
 145                 rprintf(FINFO, "hash search b=%ld len=%.0f\n",
 146                         (long)s->blength, (double)len);
 147         }
 148
 149         k = (int32)MIN(len, (OFF_T)s->blength);
 150
 151         map = (schar *)map_ptr(buf, 0, k);
 152
 153         sum = get_checksum1((char *)map, k);
 154         s1 = sum & 0xFFFF;
 155         s2 = sum >> 16;
 156         if (verbose > 3)
 157                 rprintf(FINFO, "sum=%.8x k=%ld\n", sum, (long)k);
 158
 159         offset = 0;
 160
 161         end = len + 1 - s->sums[s->count-1].len;
 162
 163         if (verbose > 3) {
 164                 rprintf(FINFO, "hash search s->blength=%ld len=%.0f count=%.0f\n",
 165                         (long)s->blength, (double)len, (double)s->count);
 166         }
 167
 168         do {
 169                 int done_csum2 = 0;
 170                 int32 i;
 171
 172                 if (offset >= reset) {
 173                         sum_pos = build_hash_table(s, sum_pos);
 174                         if (sum_pos == s->count)
 175                                 reset = len;
 176                         else
 177                                 reset = sum_pos * s->blength;
 178                 }
 179
 180                 if (verbose > 4) {
 181                         rprintf(FINFO, "offset=%.0f sum=%04x%04x\n",
 182                                 (double)offset, s2 & 0xFFFF, s1 & 0xFFFF);
 183                 }
 184
 185                 i = hash_table[SUM2HASH2(s1,s2)];
 186                 if (i < 0)
 187                         goto null_hash;
 188
 189                 sum = (s1 & 0xffff) | (s2 << 16);
 190                 hash_hits++;
 191                 do {
 192                         int32 l;
 193
 194                         if (sum != s->sums[i].sum1)
 195                                 continue;
 196
 197                         /* also make sure the two blocks are the same length */
 198                         l = (int32)MIN((OFF_T)s->blength, len-offset);
 199                         if (l != s->sums[i].len)
 200                                 continue;
 201
 202                         /* in-place: ensure chunk's offset is either >= our
 203                          * offset or that the data didn't move. */
 204                         if (updating_basis_file && s->sums[i].offset < offset
 205                             && !(s->sums[i].flags & SUMFLG_SAME_OFFSET))
 206                                 continue;
 207
 208                         if (verbose > 3) {
 209                                 rprintf(FINFO,
 210                                         "potential match at %.0f i=%ld sum=%08x\n",
 211                                         (double)offset, (long)i, sum);
 212                         }
 213
 214                         if (!done_csum2) {
 215                                 map = (schar *)map_ptr(buf,offset,l);
 216                                 get_checksum2((char *)map,l,sum2);
 217                                 done_csum2 = 1;
 218                         }
 219
 220                         if (memcmp(sum2,s->sums[i].sum2,s->s2length) != 0) {
 221                                 false_alarms++;
 222                                 continue;
 223                         }
 224
 225                         /* When updating in-place, the best possible match is
 226                          * one with an identical offset, so we prefer that over
 227                          * the following want_i optimization. */
 228                         if (updating_basis_file) {
 229                                 int32 i2;
 230                                 for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) {
 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                                 }
 246                         }
 247
 248                         /* we've found a match, but now check to see
 249                          * if want_i can hint at a better match. */
 250                         if (i != want_i && want_i < s->count
 251                             && (!updating_basis_file || s->sums[want_i].offset >= offset
 252                              || s->sums[want_i].flags & SUMFLG_SAME_OFFSET)
 253                             && sum == s->sums[want_i].sum1
 254                             && memcmp(sum2, s->sums[want_i].sum2, s->s2length) == 0) {
 255                                 /* we've found an adjacent match - the RLL coder
 256                                  * will be happy */
 257                                 i = want_i;
 258                         }
 259                     set_want_i:
 260                         want_i = i + 1;
 261
 262                         matched(f,s,buf,offset,i);
 263                         offset += s->sums[i].len - 1;
 264                         k = (int32)MIN((OFF_T)s->blength, len-offset);
 265                         map = (schar *)map_ptr(buf, offset, k);
 266                         sum = get_checksum1((char *)map, k);
 267                         s1 = sum & 0xFFFF;
 268                         s2 = sum >> 16;
 269                         matches++;
 270                         break;
 271                 } while ((i = s->sums[i].chain) >= 0);
 272
 273           null_hash:
 274                 backup = (int32)(offset - last_match);
 275                 /* We sometimes read 1 byte prior to last_match... */
 276                 if (backup < 0)
 277                         backup = 0;
 278
 279                 /* Trim off the first byte from the checksum */
 280                 more = offset + k < len;
 281                 map = (schar *)map_ptr(buf, offset - backup, k + more + backup)
 282                     + backup;
 283                 s1 -= map[0] + CHAR_OFFSET;
 284                 s2 -= k * (map[0]+CHAR_OFFSET);
 285
 286                 /* Add on the next byte (if there is one) to the checksum */
 287                 if (more) {
 288                         s1 += map[k] + CHAR_OFFSET;
 289                         s2 += s1;
 290                 } else
 291                         --k;
 292
 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. */
 299                 if (backup >= s->blength+CHUNK_SIZE && end-offset > CHUNK_SIZE)
 300                         matched(f, s, buf, offset - s->blength, -2);
 301         } while (++offset < end);
 302
 303         matched(f, s, buf, len, -1);
 304         map_ptr(buf, len-1, 1);
 305 }
 306
 307
 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  *
 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  *
 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)
 323 {
 324         char file_sum[MAX_DIGEST_LEN];
 325         int sum_len;
 326
 327         last_match = 0;
 328         false_alarms = 0;
 329         hash_hits = 0;
 330         matches = 0;
 331         data_transfer = 0;
 332
 333         sum_init(checksum_seed);
 334
 335         if (append_mode > 0) {
 336                 if (append_mode == 2) {
 337                         OFF_T j = 0;
 338                         for (j = CHUNK_SIZE; j < s->flength; j += CHUNK_SIZE) {
 339                                 if (buf && do_progress)
 340                                         show_progress(last_match, buf->file_size);
 341                                 sum_update(map_ptr(buf, last_match, CHUNK_SIZE),
 342                                            CHUNK_SIZE);
 343                                 last_match = j;
 344                         }
 345                         if (last_match < s->flength) {
 346                                 int32 n = (int32)(s->flength - last_match);
 347                                 if (buf && do_progress)
 348                                         show_progress(last_match, buf->file_size);
 349                                 sum_update(map_ptr(buf, last_match, n), n);
 350                         }
 351                 }
 352                 last_match = s->flength;
 353                 s->count = 0;
 354         }
 355
 356         if (len > 0 && s->count > 0) {
 357                 hash_search(f, s, buf, len);
 358
 359                 if (verbose > 2)
 360                         rprintf(FINFO,"done hash search\n");
 361         } else {
 362                 OFF_T j;
 363                 /* by doing this in pieces we avoid too many seeks */
 364                 for (j = last_match + CHUNK_SIZE; j < len; j += CHUNK_SIZE)
 365                         matched(f, s, buf, j, -2);
 366                 matched(f, s, buf, len, -1);
 367         }
 368
 369         sum_len = sum_end(file_sum);
 370         /* If we had a read error, send a bad checksum. */
 371         if (buf && buf->status != 0)
 372                 file_sum[0]++;
 373
 374         if (verbose > 2)
 375                 rprintf(FINFO,"sending file_sum\n");
 376         write_buf(f, file_sum, sum_len);
 377
 378         if (verbose > 2)
 379                 rprintf(FINFO, "false_alarms=%d hash_hits=%d matches=%d\n",
 380                         false_alarms, hash_hits, matches);
 381
 382         total_hash_hits += hash_hits;
 383         total_false_alarms += false_alarms;
 384         total_matches += matches;
 385         stats.literal_data += data_transfer;
 386 }
 387
 388 void match_report(void)
 389 {
 390         if (verbose <= 1)
 391                 return;
 392
 393         rprintf(FINFO,
 394                 "total: matches=%d  hash_hits=%d  false_alarms=%d data=%.0f\n",
 395                 total_matches, total_hash_hits, total_false_alarms,
 396                 (double)stats.literal_data);
 397 }