1 /* $OpenBSD: hash.c,v 1.29 2016/09/21 04:38:56 guenther Exp $ */ 2 3 /*- 4 * Copyright (c) 1990, 1993, 1994 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Margo Seltzer. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 */ 34 35 #include <sys/stat.h> 36 37 #include <errno.h> 38 #include <fcntl.h> 39 #include <stdio.h> 40 #include <stdlib.h> 41 #include <string.h> 42 #include <unistd.h> 43 #ifdef DEBUG 44 #include <assert.h> 45 #endif 46 47 #include <db.h> 48 #include "hash.h" 49 #include "page.h" 50 #include "extern.h" 51 52 #define MAXIMUM(a, b) (((a) > (b)) ? (a) : (b)) 53 54 static int alloc_segs(HTAB *, int); 55 static int flush_meta(HTAB *); 56 static int hash_access(HTAB *, ACTION, DBT *, DBT *); 57 static int hash_close(DB *); 58 static int hash_delete(const DB *, const DBT *, u_int32_t); 59 static int hash_fd(const DB *); 60 static int hash_get(const DB *, const DBT *, DBT *, u_int32_t); 61 static int hash_put(const DB *, DBT *, const DBT *, u_int32_t); 62 static void *hash_realloc(SEGMENT **, int, int); 63 static int hash_seq(const DB *, DBT *, DBT *, u_int32_t); 64 static int hash_sync(const DB *, u_int32_t); 65 static int hdestroy(HTAB *); 66 static HTAB *init_hash(HTAB *, const char *, const HASHINFO *); 67 static int init_htab(HTAB *, int); 68 #if BYTE_ORDER == LITTLE_ENDIAN 69 static void swap_header(HTAB *); 70 static void swap_header_copy(HASHHDR *, HASHHDR *); 71 #endif 72 73 /* Fast arithmetic, relying on powers of 2, */ 74 #define MOD(x, y) ((x) & ((y) - 1)) 75 76 #define RETURN_ERROR(ERR, LOC) { save_errno = ERR; goto LOC; } 77 78 /* Return values */ 79 #define SUCCESS (0) 80 #define ERROR (-1) 81 #define ABNORMAL (1) 82 83 #ifdef HASH_STATISTICS 84 int hash_accesses, hash_collisions, hash_expansions, hash_overflows; 85 #endif 86 87 /************************** INTERFACE ROUTINES ***************************/ 88 /* OPEN/CLOSE */ 89 90 DB * 91 __hash_open(const char *file, int flags, int mode, 92 const HASHINFO *info, /* Special directives for create */ 93 int dflags) 94 { 95 HTAB *hashp; 96 struct stat statbuf; 97 DB *dbp; 98 int bpages, hdrsize, new_table, nsegs, save_errno; 99 100 if ((flags & O_ACCMODE) == O_WRONLY) { 101 errno = EINVAL; 102 return (NULL); 103 } 104 105 if (!(hashp = (HTAB *)calloc(1, sizeof(HTAB)))) 106 return (NULL); 107 hashp->fp = -1; 108 109 /* 110 * Even if user wants write only, we need to be able to read 111 * the actual file, so we need to open it read/write. But, the 112 * field in the hashp structure needs to be accurate so that 113 * we can check accesses. 114 */ 115 hashp->flags = flags; 116 117 if (file) { 118 if ((hashp->fp = open(file, flags | O_CLOEXEC, mode)) == -1) 119 RETURN_ERROR(errno, error0); 120 new_table = fstat(hashp->fp, &statbuf) == 0 && 121 statbuf.st_size == 0 && (flags & O_ACCMODE) != O_RDONLY; 122 } else 123 new_table = 1; 124 125 if (new_table) { 126 if (!(hashp = init_hash(hashp, file, info))) 127 RETURN_ERROR(errno, error1); 128 } else { 129 /* Table already exists */ 130 if (info && info->hash) 131 hashp->hash = info->hash; 132 else 133 hashp->hash = __default_hash; 134 135 hdrsize = read(hashp->fp, &hashp->hdr, sizeof(HASHHDR)); 136 #if BYTE_ORDER == LITTLE_ENDIAN 137 swap_header(hashp); 138 #endif 139 if (hdrsize == -1) 140 RETURN_ERROR(errno, error1); 141 if (hdrsize != sizeof(HASHHDR)) 142 RETURN_ERROR(EFTYPE, error1); 143 /* Verify file type, versions and hash function */ 144 if (hashp->MAGIC != HASHMAGIC) 145 RETURN_ERROR(EFTYPE, error1); 146 #define OLDHASHVERSION 1 147 if (hashp->VERSION != HASHVERSION && 148 hashp->VERSION != OLDHASHVERSION) 149 RETURN_ERROR(EFTYPE, error1); 150 if (hashp->hash(CHARKEY, sizeof(CHARKEY)) != hashp->H_CHARKEY) 151 RETURN_ERROR(EFTYPE, error1); 152 /* 153 * Figure out how many segments we need. Max_Bucket is the 154 * maximum bucket number, so the number of buckets is 155 * max_bucket + 1. 156 */ 157 nsegs = (hashp->MAX_BUCKET + 1 + hashp->SGSIZE - 1) / 158 hashp->SGSIZE; 159 if (alloc_segs(hashp, nsegs)) 160 /* 161 * If alloc_segs fails, table will have been destroyed 162 * and errno will have been set. 163 */ 164 return (NULL); 165 /* Read in bitmaps */ 166 bpages = (hashp->SPARES[hashp->OVFL_POINT] + 167 (hashp->BSIZE << BYTE_SHIFT) - 1) >> 168 (hashp->BSHIFT + BYTE_SHIFT); 169 170 hashp->nmaps = bpages; 171 (void)memset(&hashp->mapp[0], 0, bpages * sizeof(u_int32_t *)); 172 } 173 174 /* Initialize Buffer Manager */ 175 if (info && info->cachesize) 176 __buf_init(hashp, info->cachesize); 177 else 178 __buf_init(hashp, DEF_BUFSIZE); 179 180 hashp->new_file = new_table; 181 hashp->save_file = file && (hashp->flags & O_RDWR); 182 hashp->cbucket = -1; 183 if (!(dbp = (DB *)malloc(sizeof(DB)))) { 184 save_errno = errno; 185 hdestroy(hashp); 186 errno = save_errno; 187 return (NULL); 188 } 189 dbp->internal = hashp; 190 dbp->close = hash_close; 191 dbp->del = hash_delete; 192 dbp->fd = hash_fd; 193 dbp->get = hash_get; 194 dbp->put = hash_put; 195 dbp->seq = hash_seq; 196 dbp->sync = hash_sync; 197 dbp->type = DB_HASH; 198 199 #ifdef DEBUG 200 (void)fprintf(stderr, 201 "%s\n%s%p\n%s%d\n%s%d\n%s%d\n%s%d\n%s%d\n%s%d\n%s%d\n%s%d\n%s%d\n%s%x\n%s%x\n%s%d\n%s%d\n", 202 "init_htab:", 203 "TABLE POINTER ", hashp, 204 "BUCKET SIZE ", hashp->BSIZE, 205 "BUCKET SHIFT ", hashp->BSHIFT, 206 "DIRECTORY SIZE ", hashp->DSIZE, 207 "SEGMENT SIZE ", hashp->SGSIZE, 208 "SEGMENT SHIFT ", hashp->SSHIFT, 209 "FILL FACTOR ", hashp->FFACTOR, 210 "MAX BUCKET ", hashp->MAX_BUCKET, 211 "OVFL POINT ", hashp->OVFL_POINT, 212 "LAST FREED ", hashp->LAST_FREED, 213 "HIGH MASK ", hashp->HIGH_MASK, 214 "LOW MASK ", hashp->LOW_MASK, 215 "NSEGS ", hashp->nsegs, 216 "NKEYS ", hashp->NKEYS); 217 #endif 218 #ifdef HASH_STATISTICS 219 hash_overflows = hash_accesses = hash_collisions = hash_expansions = 0; 220 #endif 221 return (dbp); 222 223 error1: 224 if (hashp != NULL) 225 (void)close(hashp->fp); 226 227 error0: 228 free(hashp); 229 errno = save_errno; 230 return (NULL); 231 } 232 233 static int 234 hash_close(DB *dbp) 235 { 236 HTAB *hashp; 237 int retval; 238 239 if (!dbp) 240 return (ERROR); 241 242 hashp = (HTAB *)dbp->internal; 243 retval = hdestroy(hashp); 244 free(dbp); 245 return (retval); 246 } 247 248 static int 249 hash_fd(const DB *dbp) 250 { 251 HTAB *hashp; 252 253 if (!dbp) 254 return (ERROR); 255 256 hashp = (HTAB *)dbp->internal; 257 if (hashp->fp == -1) { 258 errno = ENOENT; 259 return (-1); 260 } 261 return (hashp->fp); 262 } 263 264 /************************** LOCAL CREATION ROUTINES **********************/ 265 static HTAB * 266 init_hash(HTAB *hashp, const char *file, const HASHINFO *info) 267 { 268 struct stat statbuf; 269 int nelem; 270 271 nelem = 1; 272 hashp->NKEYS = 0; 273 hashp->LORDER = BYTE_ORDER; 274 hashp->BSIZE = DEF_BUCKET_SIZE; 275 hashp->BSHIFT = DEF_BUCKET_SHIFT; 276 hashp->SGSIZE = DEF_SEGSIZE; 277 hashp->SSHIFT = DEF_SEGSIZE_SHIFT; 278 hashp->DSIZE = DEF_DIRSIZE; 279 hashp->FFACTOR = DEF_FFACTOR; 280 hashp->hash = __default_hash; 281 memset(hashp->SPARES, 0, sizeof(hashp->SPARES)); 282 memset(hashp->BITMAPS, 0, sizeof (hashp->BITMAPS)); 283 284 /* Fix bucket size to be optimal for file system */ 285 if (file != NULL) { 286 if (stat(file, &statbuf)) 287 return (NULL); 288 hashp->BSIZE = statbuf.st_blksize; 289 hashp->BSHIFT = __log2(hashp->BSIZE); 290 } 291 292 if (info) { 293 if (info->bsize) { 294 /* Round pagesize up to power of 2 */ 295 hashp->BSHIFT = __log2(info->bsize); 296 hashp->BSIZE = 1 << hashp->BSHIFT; 297 if (hashp->BSIZE > MAX_BSIZE) { 298 errno = EINVAL; 299 return (NULL); 300 } 301 } 302 if (info->ffactor) 303 hashp->FFACTOR = info->ffactor; 304 if (info->hash) 305 hashp->hash = info->hash; 306 if (info->nelem) 307 nelem = info->nelem; 308 if (info->lorder) { 309 if (info->lorder != BIG_ENDIAN && 310 info->lorder != LITTLE_ENDIAN) { 311 errno = EINVAL; 312 return (NULL); 313 } 314 hashp->LORDER = info->lorder; 315 } 316 } 317 /* init_htab should destroy the table and set errno if it fails */ 318 if (init_htab(hashp, nelem)) 319 return (NULL); 320 else 321 return (hashp); 322 } 323 /* 324 * This calls alloc_segs which may run out of memory. Alloc_segs will destroy 325 * the table and set errno, so we just pass the error information along. 326 * 327 * Returns 0 on No Error 328 */ 329 static int 330 init_htab(HTAB *hashp, int nelem) 331 { 332 int nbuckets, nsegs, l2; 333 334 /* 335 * Divide number of elements by the fill factor and determine a 336 * desired number of buckets. Allocate space for the next greater 337 * power of two number of buckets. 338 */ 339 nelem = (nelem - 1) / hashp->FFACTOR + 1; 340 341 l2 = __log2(MAXIMUM(nelem, 2)); 342 nbuckets = 1 << l2; 343 344 hashp->SPARES[l2] = l2 + 1; 345 hashp->SPARES[l2 + 1] = l2 + 1; 346 hashp->OVFL_POINT = l2; 347 hashp->LAST_FREED = 2; 348 349 /* First bitmap page is at: splitpoint l2 page offset 1 */ 350 if (__ibitmap(hashp, OADDR_OF(l2, 1), l2 + 1, 0)) 351 return (-1); 352 353 hashp->MAX_BUCKET = hashp->LOW_MASK = nbuckets - 1; 354 hashp->HIGH_MASK = (nbuckets << 1) - 1; 355 hashp->HDRPAGES = ((MAXIMUM(sizeof(HASHHDR), MINHDRSIZE) - 1) >> 356 hashp->BSHIFT) + 1; 357 358 nsegs = (nbuckets - 1) / hashp->SGSIZE + 1; 359 nsegs = 1 << __log2(nsegs); 360 361 if (nsegs > hashp->DSIZE) 362 hashp->DSIZE = nsegs; 363 return (alloc_segs(hashp, nsegs)); 364 } 365 366 /********************** DESTROY/CLOSE ROUTINES ************************/ 367 368 /* 369 * Flushes any changes to the file if necessary and destroys the hashp 370 * structure, freeing all allocated space. 371 */ 372 static int 373 hdestroy(HTAB *hashp) 374 { 375 int i, save_errno; 376 377 save_errno = 0; 378 379 #ifdef HASH_STATISTICS 380 (void)fprintf(stderr, "hdestroy: accesses %ld collisions %ld\n", 381 hash_accesses, hash_collisions); 382 (void)fprintf(stderr, "hdestroy: expansions %ld\n", 383 hash_expansions); 384 (void)fprintf(stderr, "hdestroy: overflows %ld\n", 385 hash_overflows); 386 (void)fprintf(stderr, "keys %ld maxp %d segmentcount %d\n", 387 hashp->NKEYS, hashp->MAX_BUCKET, hashp->nsegs); 388 389 for (i = 0; i < NCACHED; i++) 390 (void)fprintf(stderr, 391 "spares[%d] = %d\n", i, hashp->SPARES[i]); 392 #endif 393 /* 394 * Call on buffer manager to free buffers, and if required, 395 * write them to disk. 396 */ 397 if (__buf_free(hashp, 1, hashp->save_file)) 398 save_errno = errno; 399 if (hashp->dir) { 400 free(*hashp->dir); /* Free initial segments */ 401 /* Free extra segments */ 402 while (hashp->exsegs--) 403 free(hashp->dir[--hashp->nsegs]); 404 free(hashp->dir); 405 } 406 if (flush_meta(hashp) && !save_errno) 407 save_errno = errno; 408 /* Free Bigmaps */ 409 for (i = 0; i < hashp->nmaps; i++) 410 free(hashp->mapp[i]); 411 free(hashp->tmp_key); 412 free(hashp->tmp_buf); 413 414 if (hashp->fp != -1) 415 (void)close(hashp->fp); 416 417 free(hashp); 418 419 if (save_errno) { 420 errno = save_errno; 421 return (ERROR); 422 } 423 return (SUCCESS); 424 } 425 /* 426 * Write modified pages to disk 427 * 428 * Returns: 429 * 0 == OK 430 * -1 ERROR 431 */ 432 static int 433 hash_sync(const DB *dbp, u_int32_t flags) 434 { 435 HTAB *hashp; 436 437 if (flags != 0) { 438 errno = EINVAL; 439 return (ERROR); 440 } 441 442 if (!dbp) 443 return (ERROR); 444 445 hashp = (HTAB *)dbp->internal; 446 if (!hashp->save_file) 447 return (0); 448 if (__buf_free(hashp, 0, 1) || flush_meta(hashp)) 449 return (ERROR); 450 hashp->new_file = 0; 451 return (0); 452 } 453 454 /* 455 * Returns: 456 * 0 == OK 457 * -1 indicates that errno should be set 458 */ 459 static int 460 flush_meta(HTAB *hashp) 461 { 462 HASHHDR *whdrp; 463 #if BYTE_ORDER == LITTLE_ENDIAN 464 HASHHDR whdr; 465 #endif 466 int fp, i, wsize; 467 468 if (!hashp->save_file) 469 return (0); 470 hashp->MAGIC = HASHMAGIC; 471 hashp->VERSION = HASHVERSION; 472 hashp->H_CHARKEY = hashp->hash(CHARKEY, sizeof(CHARKEY)); 473 474 fp = hashp->fp; 475 whdrp = &hashp->hdr; 476 #if BYTE_ORDER == LITTLE_ENDIAN 477 whdrp = &whdr; 478 swap_header_copy(&hashp->hdr, whdrp); 479 #endif 480 if ((wsize = pwrite(fp, whdrp, sizeof(HASHHDR), 0)) == -1) 481 return (-1); 482 else 483 if (wsize != sizeof(HASHHDR)) { 484 errno = EFTYPE; 485 hashp->err = errno; 486 return (-1); 487 } 488 for (i = 0; i < NCACHED; i++) 489 if (hashp->mapp[i]) 490 if (__put_page(hashp, (char *)hashp->mapp[i], 491 hashp->BITMAPS[i], 0, 1)) 492 return (-1); 493 return (0); 494 } 495 496 /*******************************SEARCH ROUTINES *****************************/ 497 /* 498 * All the access routines return 499 * 500 * Returns: 501 * 0 on SUCCESS 502 * 1 to indicate an external ERROR (i.e. key not found, etc) 503 * -1 to indicate an internal ERROR (i.e. out of memory, etc) 504 */ 505 static int 506 hash_get(const DB *dbp, const DBT *key, DBT *data, u_int32_t flag) 507 { 508 HTAB *hashp; 509 510 hashp = (HTAB *)dbp->internal; 511 if (flag) { 512 hashp->err = errno = EINVAL; 513 return (ERROR); 514 } 515 return (hash_access(hashp, HASH_GET, (DBT *)key, data)); 516 } 517 518 static int 519 hash_put(const DB *dbp, DBT *key, const DBT *data, u_int32_t flag) 520 { 521 HTAB *hashp; 522 523 hashp = (HTAB *)dbp->internal; 524 if (flag && flag != R_NOOVERWRITE) { 525 hashp->err = errno = EINVAL; 526 return (ERROR); 527 } 528 if ((hashp->flags & O_ACCMODE) == O_RDONLY) { 529 hashp->err = errno = EPERM; 530 return (ERROR); 531 } 532 return (hash_access(hashp, flag == R_NOOVERWRITE ? 533 HASH_PUTNEW : HASH_PUT, (DBT *)key, (DBT *)data)); 534 } 535 536 static int 537 hash_delete(const DB *dbp, const DBT *key, 538 u_int32_t flag) /* Ignored */ 539 { 540 HTAB *hashp; 541 542 hashp = (HTAB *)dbp->internal; 543 if (flag && flag != R_CURSOR) { 544 hashp->err = errno = EINVAL; 545 return (ERROR); 546 } 547 if ((hashp->flags & O_ACCMODE) == O_RDONLY) { 548 hashp->err = errno = EPERM; 549 return (ERROR); 550 } 551 return (hash_access(hashp, HASH_DELETE, (DBT *)key, NULL)); 552 } 553 554 /* 555 * Assume that hashp has been set in wrapper routine. 556 */ 557 static int 558 hash_access(HTAB *hashp, ACTION action, DBT *key, DBT *val) 559 { 560 BUFHEAD *rbufp; 561 BUFHEAD *bufp, *save_bufp; 562 u_int16_t *bp; 563 int n, ndx, off, size; 564 char *kp; 565 u_int16_t pageno; 566 567 #ifdef HASH_STATISTICS 568 hash_accesses++; 569 #endif 570 571 off = hashp->BSIZE; 572 size = key->size; 573 kp = (char *)key->data; 574 rbufp = __get_buf(hashp, __call_hash(hashp, kp, size), NULL, 0); 575 if (!rbufp) 576 return (ERROR); 577 save_bufp = rbufp; 578 579 /* Pin the bucket chain */ 580 rbufp->flags |= BUF_PIN; 581 for (bp = (u_int16_t *)rbufp->page, n = *bp++, ndx = 1; ndx < n;) 582 if (bp[1] >= REAL_KEY) { 583 /* Real key/data pair */ 584 if (size == off - *bp && 585 memcmp(kp, rbufp->page + *bp, size) == 0) 586 goto found; 587 off = bp[1]; 588 #ifdef HASH_STATISTICS 589 hash_collisions++; 590 #endif 591 bp += 2; 592 ndx += 2; 593 } else if (bp[1] == OVFLPAGE) { 594 rbufp = __get_buf(hashp, *bp, rbufp, 0); 595 if (!rbufp) { 596 save_bufp->flags &= ~BUF_PIN; 597 return (ERROR); 598 } 599 /* FOR LOOP INIT */ 600 bp = (u_int16_t *)rbufp->page; 601 n = *bp++; 602 ndx = 1; 603 off = hashp->BSIZE; 604 } else if (bp[1] < REAL_KEY) { 605 if ((ndx = 606 __find_bigpair(hashp, rbufp, ndx, kp, size)) > 0) 607 goto found; 608 if (ndx == -2) { 609 bufp = rbufp; 610 if (!(pageno = 611 __find_last_page(hashp, &bufp))) { 612 ndx = 0; 613 rbufp = bufp; 614 break; /* FOR */ 615 } 616 rbufp = __get_buf(hashp, pageno, bufp, 0); 617 if (!rbufp) { 618 save_bufp->flags &= ~BUF_PIN; 619 return (ERROR); 620 } 621 /* FOR LOOP INIT */ 622 bp = (u_int16_t *)rbufp->page; 623 n = *bp++; 624 ndx = 1; 625 off = hashp->BSIZE; 626 } else { 627 save_bufp->flags &= ~BUF_PIN; 628 return (ERROR); 629 } 630 } 631 632 /* Not found */ 633 switch (action) { 634 case HASH_PUT: 635 case HASH_PUTNEW: 636 if (__addel(hashp, rbufp, key, val)) { 637 save_bufp->flags &= ~BUF_PIN; 638 return (ERROR); 639 } else { 640 save_bufp->flags &= ~BUF_PIN; 641 return (SUCCESS); 642 } 643 case HASH_GET: 644 case HASH_DELETE: 645 default: 646 save_bufp->flags &= ~BUF_PIN; 647 return (ABNORMAL); 648 } 649 650 found: 651 switch (action) { 652 case HASH_PUTNEW: 653 save_bufp->flags &= ~BUF_PIN; 654 return (ABNORMAL); 655 case HASH_GET: 656 bp = (u_int16_t *)rbufp->page; 657 if (bp[ndx + 1] < REAL_KEY) { 658 if (__big_return(hashp, rbufp, ndx, val, 0)) 659 return (ERROR); 660 } else { 661 val->data = (u_char *)rbufp->page + (int)bp[ndx + 1]; 662 val->size = bp[ndx] - bp[ndx + 1]; 663 } 664 break; 665 case HASH_PUT: 666 if ((__delpair(hashp, rbufp, ndx)) || 667 (__addel(hashp, rbufp, key, val))) { 668 save_bufp->flags &= ~BUF_PIN; 669 return (ERROR); 670 } 671 break; 672 case HASH_DELETE: 673 if (__delpair(hashp, rbufp, ndx)) 674 return (ERROR); 675 break; 676 default: 677 abort(); 678 } 679 save_bufp->flags &= ~BUF_PIN; 680 return (SUCCESS); 681 } 682 683 static int 684 hash_seq(const DB *dbp, DBT *key, DBT *data, u_int32_t flag) 685 { 686 u_int32_t bucket; 687 BUFHEAD *bufp; 688 HTAB *hashp; 689 u_int16_t *bp, ndx; 690 691 hashp = (HTAB *)dbp->internal; 692 if (flag && flag != R_FIRST && flag != R_NEXT) { 693 hashp->err = errno = EINVAL; 694 return (ERROR); 695 } 696 #ifdef HASH_STATISTICS 697 hash_accesses++; 698 #endif 699 if ((hashp->cbucket < 0) || (flag == R_FIRST)) { 700 hashp->cbucket = 0; 701 hashp->cndx = 1; 702 hashp->cpage = NULL; 703 } 704 next_bucket: 705 for (bp = NULL; !bp || !bp[0]; ) { 706 if (!(bufp = hashp->cpage)) { 707 for (bucket = hashp->cbucket; 708 bucket <= hashp->MAX_BUCKET; 709 bucket++, hashp->cndx = 1) { 710 bufp = __get_buf(hashp, bucket, NULL, 0); 711 if (!bufp) 712 return (ERROR); 713 hashp->cpage = bufp; 714 bp = (u_int16_t *)bufp->page; 715 if (bp[0]) 716 break; 717 } 718 hashp->cbucket = bucket; 719 if (hashp->cbucket > hashp->MAX_BUCKET) { 720 hashp->cbucket = -1; 721 return (ABNORMAL); 722 } 723 } else { 724 bp = (u_int16_t *)hashp->cpage->page; 725 if (flag == R_NEXT) { 726 hashp->cndx += 2; 727 if (hashp->cndx > bp[0]) { 728 hashp->cpage = NULL; 729 hashp->cbucket++; 730 hashp->cndx = 1; 731 goto next_bucket; 732 } 733 } 734 } 735 736 #ifdef DEBUG 737 assert(bp); 738 assert(bufp); 739 #endif 740 while (bp[hashp->cndx + 1] == OVFLPAGE) { 741 bufp = hashp->cpage = 742 __get_buf(hashp, bp[hashp->cndx], bufp, 0); 743 if (!bufp) 744 return (ERROR); 745 bp = (u_int16_t *)(bufp->page); 746 hashp->cndx = 1; 747 } 748 if (!bp[0]) { 749 hashp->cpage = NULL; 750 ++hashp->cbucket; 751 } 752 } 753 ndx = hashp->cndx; 754 if (bp[ndx + 1] < REAL_KEY) { 755 if (__big_keydata(hashp, bufp, key, data, 1)) 756 return (ERROR); 757 } else { 758 if (hashp->cpage == 0) 759 return (ERROR); 760 key->data = (u_char *)hashp->cpage->page + bp[ndx]; 761 key->size = (ndx > 1 ? bp[ndx - 1] : hashp->BSIZE) - bp[ndx]; 762 data->data = (u_char *)hashp->cpage->page + bp[ndx + 1]; 763 data->size = bp[ndx] - bp[ndx + 1]; 764 } 765 return (SUCCESS); 766 } 767 768 /********************************* UTILITIES ************************/ 769 770 /* 771 * Returns: 772 * 0 ==> OK 773 * -1 ==> Error 774 */ 775 int 776 __expand_table(HTAB *hashp) 777 { 778 u_int32_t old_bucket, new_bucket; 779 int dirsize, new_segnum, spare_ndx; 780 781 #ifdef HASH_STATISTICS 782 hash_expansions++; 783 #endif 784 new_bucket = ++hashp->MAX_BUCKET; 785 old_bucket = (hashp->MAX_BUCKET & hashp->LOW_MASK); 786 787 new_segnum = new_bucket >> hashp->SSHIFT; 788 789 /* Check if we need a new segment */ 790 if (new_segnum >= hashp->nsegs) { 791 /* Check if we need to expand directory */ 792 if (new_segnum >= hashp->DSIZE) { 793 /* Reallocate directory */ 794 dirsize = hashp->DSIZE * sizeof(SEGMENT *); 795 if (!hash_realloc(&hashp->dir, dirsize, dirsize << 1)) 796 return (-1); 797 hashp->DSIZE = dirsize << 1; 798 } 799 if ((hashp->dir[new_segnum] = 800 (SEGMENT)calloc(hashp->SGSIZE, sizeof(SEGMENT))) == NULL) 801 return (-1); 802 hashp->exsegs++; 803 hashp->nsegs++; 804 } 805 /* 806 * If the split point is increasing (MAX_BUCKET's log base 2 807 * * increases), we need to copy the current contents of the spare 808 * split bucket to the next bucket. 809 */ 810 spare_ndx = __log2(hashp->MAX_BUCKET + 1); 811 if (spare_ndx > hashp->OVFL_POINT) { 812 hashp->SPARES[spare_ndx] = hashp->SPARES[hashp->OVFL_POINT]; 813 hashp->OVFL_POINT = spare_ndx; 814 } 815 816 if (new_bucket > hashp->HIGH_MASK) { 817 /* Starting a new doubling */ 818 hashp->LOW_MASK = hashp->HIGH_MASK; 819 hashp->HIGH_MASK = new_bucket | hashp->LOW_MASK; 820 } 821 /* Relocate records to the new bucket */ 822 return (__split_page(hashp, old_bucket, new_bucket)); 823 } 824 825 /* 826 * If realloc guarantees that the pointer is not destroyed if the realloc 827 * fails, then this routine can go away. 828 */ 829 static void * 830 hash_realloc(SEGMENT **p_ptr, int oldsize, int newsize) 831 { 832 void *p; 833 834 if ((p = malloc(newsize))) { 835 memmove(p, *p_ptr, oldsize); 836 memset((char *)p + oldsize, 0, newsize - oldsize); 837 free(*p_ptr); 838 *p_ptr = p; 839 } 840 return (p); 841 } 842 843 u_int32_t 844 __call_hash(HTAB *hashp, char *k, int len) 845 { 846 int n, bucket; 847 848 n = hashp->hash(k, len); 849 bucket = n & hashp->HIGH_MASK; 850 if (bucket > hashp->MAX_BUCKET) 851 bucket = bucket & hashp->LOW_MASK; 852 return (bucket); 853 } 854 855 /* 856 * Allocate segment table. On error, destroy the table and set errno. 857 * 858 * Returns 0 on success 859 */ 860 static int 861 alloc_segs(HTAB *hashp, int nsegs) 862 { 863 int i; 864 SEGMENT store; 865 866 int save_errno; 867 868 if ((hashp->dir = 869 (SEGMENT *)calloc(hashp->DSIZE, sizeof(SEGMENT *))) == NULL) { 870 save_errno = errno; 871 (void)hdestroy(hashp); 872 errno = save_errno; 873 return (-1); 874 } 875 hashp->nsegs = nsegs; 876 if (nsegs == 0) 877 return (0); 878 /* Allocate segments */ 879 if ((store = (SEGMENT)calloc(nsegs << hashp->SSHIFT, 880 sizeof(SEGMENT))) == NULL) { 881 save_errno = errno; 882 (void)hdestroy(hashp); 883 errno = save_errno; 884 return (-1); 885 } 886 for (i = 0; i < nsegs; i++) 887 hashp->dir[i] = &store[i << hashp->SSHIFT]; 888 return (0); 889 } 890 891 #if BYTE_ORDER == LITTLE_ENDIAN 892 /* 893 * Hashp->hdr needs to be byteswapped. 894 */ 895 static void 896 swap_header_copy(HASHHDR *srcp, HASHHDR *destp) 897 { 898 int i; 899 900 P_32_COPY(srcp->magic, destp->magic); 901 P_32_COPY(srcp->version, destp->version); 902 P_32_COPY(srcp->lorder, destp->lorder); 903 P_32_COPY(srcp->bsize, destp->bsize); 904 P_32_COPY(srcp->bshift, destp->bshift); 905 P_32_COPY(srcp->dsize, destp->dsize); 906 P_32_COPY(srcp->ssize, destp->ssize); 907 P_32_COPY(srcp->sshift, destp->sshift); 908 P_32_COPY(srcp->ovfl_point, destp->ovfl_point); 909 P_32_COPY(srcp->last_freed, destp->last_freed); 910 P_32_COPY(srcp->max_bucket, destp->max_bucket); 911 P_32_COPY(srcp->high_mask, destp->high_mask); 912 P_32_COPY(srcp->low_mask, destp->low_mask); 913 P_32_COPY(srcp->ffactor, destp->ffactor); 914 P_32_COPY(srcp->nkeys, destp->nkeys); 915 P_32_COPY(srcp->hdrpages, destp->hdrpages); 916 P_32_COPY(srcp->h_charkey, destp->h_charkey); 917 for (i = 0; i < NCACHED; i++) { 918 P_32_COPY(srcp->spares[i], destp->spares[i]); 919 P_16_COPY(srcp->bitmaps[i], destp->bitmaps[i]); 920 } 921 } 922 923 static void 924 swap_header(HTAB *hashp) 925 { 926 HASHHDR *hdrp; 927 int i; 928 929 hdrp = &hashp->hdr; 930 931 M_32_SWAP(hdrp->magic); 932 M_32_SWAP(hdrp->version); 933 M_32_SWAP(hdrp->lorder); 934 M_32_SWAP(hdrp->bsize); 935 M_32_SWAP(hdrp->bshift); 936 M_32_SWAP(hdrp->dsize); 937 M_32_SWAP(hdrp->ssize); 938 M_32_SWAP(hdrp->sshift); 939 M_32_SWAP(hdrp->ovfl_point); 940 M_32_SWAP(hdrp->last_freed); 941 M_32_SWAP(hdrp->max_bucket); 942 M_32_SWAP(hdrp->high_mask); 943 M_32_SWAP(hdrp->low_mask); 944 M_32_SWAP(hdrp->ffactor); 945 M_32_SWAP(hdrp->nkeys); 946 M_32_SWAP(hdrp->hdrpages); 947 M_32_SWAP(hdrp->h_charkey); 948 for (i = 0; i < NCACHED; i++) { 949 M_32_SWAP(hdrp->spares[i]); 950 M_16_SWAP(hdrp->bitmaps[i]); 951 } 952 } 953 #endif 954