xref: /netbsd-src/lib/libc/db/hash/hash_page.c (revision ce0bb6e8d2e560ecacbe865a848624f94498063b)
1 /*	$NetBSD: hash_page.c,v 1.7 1995/02/27 13:22:34 cgd 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. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *	This product includes software developed by the University of
21  *	California, Berkeley and its contributors.
22  * 4. Neither the name of the University nor the names of its contributors
23  *    may be used to endorse or promote products derived from this software
24  *    without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36  * SUCH DAMAGE.
37  */
38 
39 #if defined(LIBC_SCCS) && !defined(lint)
40 #if 0
41 static char sccsid[] = "@(#)hash_page.c	8.6 (Berkeley) 6/16/94";
42 #else
43 static char rcsid[] = "$NetBSD: hash_page.c,v 1.7 1995/02/27 13:22:34 cgd Exp $";
44 #endif
45 #endif /* LIBC_SCCS and not lint */
46 
47 /*
48  * PACKAGE:  hashing
49  *
50  * DESCRIPTION:
51  *	Page manipulation for hashing package.
52  *
53  * ROUTINES:
54  *
55  * External
56  *	__get_page
57  *	__add_ovflpage
58  * Internal
59  *	overflow_page
60  *	open_temp
61  */
62 
63 #include <sys/types.h>
64 
65 #include <errno.h>
66 #include <fcntl.h>
67 #include <signal.h>
68 #include <stdio.h>
69 #include <stdlib.h>
70 #include <string.h>
71 #include <unistd.h>
72 #ifdef DEBUG
73 #include <assert.h>
74 #endif
75 
76 #include <db.h>
77 #include "hash.h"
78 #include "page.h"
79 #include "extern.h"
80 
81 static u_int32_t	*fetch_bitmap __P((HTAB *, int));
82 static u_int32_t	 first_free __P((u_int32_t));
83 static int	 open_temp __P((HTAB *));
84 static u_int16_t	 overflow_page __P((HTAB *));
85 static void	 putpair __P((char *, const DBT *, const DBT *));
86 static void	 squeeze_key __P((u_int16_t *, const DBT *, const DBT *));
87 static int	 ugly_split
88 		    __P((HTAB *, u_int32_t, BUFHEAD *, BUFHEAD *, int, int));
89 
90 #define	PAGE_INIT(P) { \
91 	((u_int16_t *)(P))[0] = 0; \
92 	((u_int16_t *)(P))[1] = hashp->BSIZE - 3 * sizeof(u_int16_t); \
93 	((u_int16_t *)(P))[2] = hashp->BSIZE; \
94 }
95 
96 /*
97  * This is called AFTER we have verified that there is room on the page for
98  * the pair (PAIRFITS has returned true) so we go right ahead and start moving
99  * stuff on.
100  */
101 static void
102 putpair(p, key, val)
103 	char *p;
104 	const DBT *key, *val;
105 {
106 	register u_int16_t *bp, n, off;
107 
108 	bp = (u_int16_t *)p;
109 
110 	/* Enter the key first. */
111 	n = bp[0];
112 
113 	off = OFFSET(bp) - key->size;
114 	memmove(p + off, key->data, key->size);
115 	bp[++n] = off;
116 
117 	/* Now the data. */
118 	off -= val->size;
119 	memmove(p + off, val->data, val->size);
120 	bp[++n] = off;
121 
122 	/* Adjust page info. */
123 	bp[0] = n;
124 	bp[n + 1] = off - ((n + 3) * sizeof(u_int16_t));
125 	bp[n + 2] = off;
126 }
127 
128 /*
129  * Returns:
130  *	 0 OK
131  *	-1 error
132  */
133 extern int
134 __delpair(hashp, bufp, ndx)
135 	HTAB *hashp;
136 	BUFHEAD *bufp;
137 	register int ndx;
138 {
139 	register u_int16_t *bp, newoff;
140 	register int n;
141 	u_int16_t pairlen;
142 
143 	bp = (u_int16_t *)bufp->page;
144 	n = bp[0];
145 
146 	if (bp[ndx + 1] < REAL_KEY)
147 		return (__big_delete(hashp, bufp));
148 	if (ndx != 1)
149 		newoff = bp[ndx - 1];
150 	else
151 		newoff = hashp->BSIZE;
152 	pairlen = newoff - bp[ndx + 1];
153 
154 	if (ndx != (n - 1)) {
155 		/* Hard Case -- need to shuffle keys */
156 		register int i;
157 		register char *src = bufp->page + (int)OFFSET(bp);
158 		register char *dst = src + (int)pairlen;
159 		memmove(dst, src, bp[ndx + 1] - OFFSET(bp));
160 
161 		/* Now adjust the pointers */
162 		for (i = ndx + 2; i <= n; i += 2) {
163 			if (bp[i + 1] == OVFLPAGE) {
164 				bp[i - 2] = bp[i];
165 				bp[i - 1] = bp[i + 1];
166 			} else {
167 				bp[i - 2] = bp[i] + pairlen;
168 				bp[i - 1] = bp[i + 1] + pairlen;
169 			}
170 		}
171 	}
172 	/* Finally adjust the page data */
173 	bp[n] = OFFSET(bp) + pairlen;
174 	bp[n - 1] = bp[n + 1] + pairlen + 2 * sizeof(u_int16_t);
175 	bp[0] = n - 2;
176 	hashp->NKEYS--;
177 
178 	bufp->flags |= BUF_MOD;
179 	return (0);
180 }
181 /*
182  * Returns:
183  *	 0 ==> OK
184  *	-1 ==> Error
185  */
186 extern int
187 __split_page(hashp, obucket, nbucket)
188 	HTAB *hashp;
189 	u_int32_t obucket, nbucket;
190 {
191 	register BUFHEAD *new_bufp, *old_bufp;
192 	register u_int16_t *ino;
193 	register char *np;
194 	DBT key, val;
195 	int n, ndx, retval;
196 	u_int16_t copyto, diff, off, moved;
197 	char *op;
198 
199 	copyto = (u_int16_t)hashp->BSIZE;
200 	off = (u_int16_t)hashp->BSIZE;
201 	old_bufp = __get_buf(hashp, obucket, NULL, 0);
202 	if (old_bufp == NULL)
203 		return (-1);
204 	new_bufp = __get_buf(hashp, nbucket, NULL, 0);
205 	if (new_bufp == NULL)
206 		return (-1);
207 
208 	old_bufp->flags |= (BUF_MOD | BUF_PIN);
209 	new_bufp->flags |= (BUF_MOD | BUF_PIN);
210 
211 	ino = (u_int16_t *)(op = old_bufp->page);
212 	np = new_bufp->page;
213 
214 	moved = 0;
215 
216 	for (n = 1, ndx = 1; n < ino[0]; n += 2) {
217 		if (ino[n + 1] < REAL_KEY) {
218 			retval = ugly_split(hashp, obucket, old_bufp, new_bufp,
219 			    (int)copyto, (int)moved);
220 			old_bufp->flags &= ~BUF_PIN;
221 			new_bufp->flags &= ~BUF_PIN;
222 			return (retval);
223 
224 		}
225 		key.data = (u_char *)op + ino[n];
226 		key.size = off - ino[n];
227 
228 		if (__call_hash(hashp, key.data, key.size) == obucket) {
229 			/* Don't switch page */
230 			diff = copyto - off;
231 			if (diff) {
232 				copyto = ino[n + 1] + diff;
233 				memmove(op + copyto, op + ino[n + 1],
234 				    off - ino[n + 1]);
235 				ino[ndx] = copyto + ino[n] - ino[n + 1];
236 				ino[ndx + 1] = copyto;
237 			} else
238 				copyto = ino[n + 1];
239 			ndx += 2;
240 		} else {
241 			/* Switch page */
242 			val.data = (u_char *)op + ino[n + 1];
243 			val.size = ino[n] - ino[n + 1];
244 			putpair(np, &key, &val);
245 			moved += 2;
246 		}
247 
248 		off = ino[n + 1];
249 	}
250 
251 	/* Now clean up the page */
252 	ino[0] -= moved;
253 	FREESPACE(ino) = copyto - sizeof(u_int16_t) * (ino[0] + 3);
254 	OFFSET(ino) = copyto;
255 
256 #ifdef DEBUG3
257 	(void)fprintf(stderr, "split %d/%d\n",
258 	    ((u_int16_t *)np)[0] / 2,
259 	    ((u_int16_t *)op)[0] / 2);
260 #endif
261 	/* unpin both pages */
262 	old_bufp->flags &= ~BUF_PIN;
263 	new_bufp->flags &= ~BUF_PIN;
264 	return (0);
265 }
266 
267 /*
268  * Called when we encounter an overflow or big key/data page during split
269  * handling.  This is special cased since we have to begin checking whether
270  * the key/data pairs fit on their respective pages and because we may need
271  * overflow pages for both the old and new pages.
272  *
273  * The first page might be a page with regular key/data pairs in which case
274  * we have a regular overflow condition and just need to go on to the next
275  * page or it might be a big key/data pair in which case we need to fix the
276  * big key/data pair.
277  *
278  * Returns:
279  *	 0 ==> success
280  *	-1 ==> failure
281  */
282 static int
283 ugly_split(hashp, obucket, old_bufp, new_bufp, copyto, moved)
284 	HTAB *hashp;
285 	u_int32_t obucket;	/* Same as __split_page. */
286 	BUFHEAD *old_bufp, *new_bufp;
287 	int copyto;	/* First byte on page which contains key/data values. */
288 	int moved;	/* Number of pairs moved to new page. */
289 {
290 	register BUFHEAD *bufp;	/* Buffer header for ino */
291 	register u_int16_t *ino;	/* Page keys come off of */
292 	register u_int16_t *np;	/* New page */
293 	register u_int16_t *op;	/* Page keys go on to if they aren't moving */
294 
295 	BUFHEAD *last_bfp;	/* Last buf header OVFL needing to be freed */
296 	DBT key, val;
297 	SPLIT_RETURN ret;
298 	u_int16_t n, off, ov_addr, scopyto;
299 	char *cino;		/* Character value of ino */
300 
301 	bufp = old_bufp;
302 	ino = (u_int16_t *)old_bufp->page;
303 	np = (u_int16_t *)new_bufp->page;
304 	op = (u_int16_t *)old_bufp->page;
305 	last_bfp = NULL;
306 	scopyto = (u_int16_t)copyto;	/* ANSI */
307 
308 	n = ino[0] - 1;
309 	while (n < ino[0]) {
310 		if (ino[2] < REAL_KEY && ino[2] != OVFLPAGE) {
311 			if (__big_split(hashp, old_bufp,
312 			    new_bufp, bufp, bufp->addr, obucket, &ret))
313 				return (-1);
314 			old_bufp = ret.oldp;
315 			if (!old_bufp)
316 				return (-1);
317 			op = (u_int16_t *)old_bufp->page;
318 			new_bufp = ret.newp;
319 			if (!new_bufp)
320 				return (-1);
321 			np = (u_int16_t *)new_bufp->page;
322 			bufp = ret.nextp;
323 			if (!bufp)
324 				return (0);
325 			cino = (char *)bufp->page;
326 			ino = (u_int16_t *)cino;
327 			last_bfp = ret.nextp;
328 		} else if (ino[n + 1] == OVFLPAGE) {
329 			ov_addr = ino[n];
330 			/*
331 			 * Fix up the old page -- the extra 2 are the fields
332 			 * which contained the overflow information.
333 			 */
334 			ino[0] -= (moved + 2);
335 			FREESPACE(ino) =
336 			    scopyto - sizeof(u_int16_t) * (ino[0] + 3);
337 			OFFSET(ino) = scopyto;
338 
339 			bufp = __get_buf(hashp, ov_addr, bufp, 0);
340 			if (!bufp)
341 				return (-1);
342 
343 			ino = (u_int16_t *)bufp->page;
344 			n = 1;
345 			scopyto = hashp->BSIZE;
346 			moved = 0;
347 
348 			if (last_bfp)
349 				__free_ovflpage(hashp, last_bfp);
350 			last_bfp = bufp;
351 		}
352 		/* Move regular sized pairs of there are any */
353 		off = hashp->BSIZE;
354 		for (n = 1; (n < ino[0]) && (ino[n + 1] >= REAL_KEY); n += 2) {
355 			cino = (char *)ino;
356 			key.data = (u_char *)cino + ino[n];
357 			key.size = off - ino[n];
358 			val.data = (u_char *)cino + ino[n + 1];
359 			val.size = ino[n] - ino[n + 1];
360 			off = ino[n + 1];
361 
362 			if (__call_hash(hashp, key.data, key.size) == obucket) {
363 				/* Keep on old page */
364 				if (PAIRFITS(op, (&key), (&val)))
365 					putpair((char *)op, &key, &val);
366 				else {
367 					old_bufp =
368 					    __add_ovflpage(hashp, old_bufp);
369 					if (!old_bufp)
370 						return (-1);
371 					op = (u_int16_t *)old_bufp->page;
372 					putpair((char *)op, &key, &val);
373 				}
374 				old_bufp->flags |= BUF_MOD;
375 			} else {
376 				/* Move to new page */
377 				if (PAIRFITS(np, (&key), (&val)))
378 					putpair((char *)np, &key, &val);
379 				else {
380 					new_bufp =
381 					    __add_ovflpage(hashp, new_bufp);
382 					if (!new_bufp)
383 						return (-1);
384 					np = (u_int16_t *)new_bufp->page;
385 					putpair((char *)np, &key, &val);
386 				}
387 				new_bufp->flags |= BUF_MOD;
388 			}
389 		}
390 	}
391 	if (last_bfp)
392 		__free_ovflpage(hashp, last_bfp);
393 	return (0);
394 }
395 
396 /*
397  * Add the given pair to the page
398  *
399  * Returns:
400  *	0 ==> OK
401  *	1 ==> failure
402  */
403 extern int
404 __addel(hashp, bufp, key, val)
405 	HTAB *hashp;
406 	BUFHEAD *bufp;
407 	const DBT *key, *val;
408 {
409 	register u_int16_t *bp, *sop;
410 	int do_expand;
411 
412 	bp = (u_int16_t *)bufp->page;
413 	do_expand = 0;
414 	while (bp[0] && (bp[2] < REAL_KEY || bp[bp[0]] < REAL_KEY))
415 		/* Exception case */
416 		if (bp[2] == FULL_KEY_DATA && bp[0] == 2)
417 			/* This is the last page of a big key/data pair
418 			   and we need to add another page */
419 			break;
420 		else if (bp[2] < REAL_KEY && bp[bp[0]] != OVFLPAGE) {
421 			bufp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
422 			if (!bufp)
423 				return (-1);
424 			bp = (u_int16_t *)bufp->page;
425 		} else
426 			/* Try to squeeze key on this page */
427 			if (FREESPACE(bp) > PAIRSIZE(key, val)) {
428 				squeeze_key(bp, key, val);
429 				return (0);
430 			} else {
431 				bufp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
432 				if (!bufp)
433 					return (-1);
434 				bp = (u_int16_t *)bufp->page;
435 			}
436 
437 	if (PAIRFITS(bp, key, val))
438 		putpair(bufp->page, key, val);
439 	else {
440 		do_expand = 1;
441 		bufp = __add_ovflpage(hashp, bufp);
442 		if (!bufp)
443 			return (-1);
444 		sop = (u_int16_t *)bufp->page;
445 
446 		if (PAIRFITS(sop, key, val))
447 			putpair((char *)sop, key, val);
448 		else
449 			if (__big_insert(hashp, bufp, key, val))
450 				return (-1);
451 	}
452 	bufp->flags |= BUF_MOD;
453 	/*
454 	 * If the average number of keys per bucket exceeds the fill factor,
455 	 * expand the table.
456 	 */
457 	hashp->NKEYS++;
458 	if (do_expand ||
459 	    (hashp->NKEYS / (hashp->MAX_BUCKET + 1) > hashp->FFACTOR))
460 		return (__expand_table(hashp));
461 	return (0);
462 }
463 
464 /*
465  *
466  * Returns:
467  *	pointer on success
468  *	NULL on error
469  */
470 extern BUFHEAD *
471 __add_ovflpage(hashp, bufp)
472 	HTAB *hashp;
473 	BUFHEAD *bufp;
474 {
475 	register u_int16_t *sp;
476 	u_int16_t ndx, ovfl_num;
477 #ifdef DEBUG1
478 	int tmp1, tmp2;
479 #endif
480 	sp = (u_int16_t *)bufp->page;
481 
482 	/* Check if we are dynamically determining the fill factor */
483 	if (hashp->FFACTOR == DEF_FFACTOR) {
484 		hashp->FFACTOR = sp[0] >> 1;
485 		if (hashp->FFACTOR < MIN_FFACTOR)
486 			hashp->FFACTOR = MIN_FFACTOR;
487 	}
488 	bufp->flags |= BUF_MOD;
489 	ovfl_num = overflow_page(hashp);
490 #ifdef DEBUG1
491 	tmp1 = bufp->addr;
492 	tmp2 = bufp->ovfl ? bufp->ovfl->addr : 0;
493 #endif
494 	if (!ovfl_num || !(bufp->ovfl = __get_buf(hashp, ovfl_num, bufp, 1)))
495 		return (NULL);
496 	bufp->ovfl->flags |= BUF_MOD;
497 #ifdef DEBUG1
498 	(void)fprintf(stderr, "ADDOVFLPAGE: %d->ovfl was %d is now %d\n",
499 	    tmp1, tmp2, bufp->ovfl->addr);
500 #endif
501 	ndx = sp[0];
502 	/*
503 	 * Since a pair is allocated on a page only if there's room to add
504 	 * an overflow page, we know that the OVFL information will fit on
505 	 * the page.
506 	 */
507 	sp[ndx + 4] = OFFSET(sp);
508 	sp[ndx + 3] = FREESPACE(sp) - OVFLSIZE;
509 	sp[ndx + 1] = ovfl_num;
510 	sp[ndx + 2] = OVFLPAGE;
511 	sp[0] = ndx + 2;
512 #ifdef HASH_STATISTICS
513 	hash_overflows++;
514 #endif
515 	return (bufp->ovfl);
516 }
517 
518 /*
519  * Returns:
520  *	 0 indicates SUCCESS
521  *	-1 indicates FAILURE
522  */
523 extern int
524 __get_page(hashp, p, bucket, is_bucket, is_disk, is_bitmap)
525 	HTAB *hashp;
526 	char *p;
527 	u_int32_t bucket;
528 	int is_bucket, is_disk, is_bitmap;
529 {
530 	register int fd, page, size;
531 	int rsize;
532 	u_int16_t *bp;
533 
534 	fd = hashp->fp;
535 	size = hashp->BSIZE;
536 
537 	if ((fd == -1) || !is_disk) {
538 		PAGE_INIT(p);
539 		return (0);
540 	}
541 	if (is_bucket)
542 		page = BUCKET_TO_PAGE(bucket);
543 	else
544 		page = OADDR_TO_PAGE(bucket);
545 	if ((lseek(fd, (off_t)page << hashp->BSHIFT, SEEK_SET) == -1) ||
546 	    ((rsize = read(fd, p, size)) == -1))
547 		return (-1);
548 	bp = (u_int16_t *)p;
549 	if (!rsize)
550 		bp[0] = 0;	/* We hit the EOF, so initialize a new page */
551 	else
552 		if (rsize != size) {
553 			errno = EFTYPE;
554 			return (-1);
555 		}
556 	if (!is_bitmap && !bp[0]) {
557 		PAGE_INIT(p);
558 	} else
559 		if (hashp->LORDER != BYTE_ORDER) {
560 			register int i, max;
561 
562 			if (is_bitmap) {
563 				max = hashp->BSIZE >> 2; /* divide by 4 */
564 				for (i = 0; i < max; i++)
565 					M_32_SWAP(((int *)p)[i]);
566 			} else {
567 				M_16_SWAP(bp[0]);
568 				max = bp[0] + 2;
569 				for (i = 1; i <= max; i++)
570 					M_16_SWAP(bp[i]);
571 			}
572 		}
573 	return (0);
574 }
575 
576 /*
577  * Write page p to disk
578  *
579  * Returns:
580  *	 0 ==> OK
581  *	-1 ==>failure
582  */
583 extern int
584 __put_page(hashp, p, bucket, is_bucket, is_bitmap)
585 	HTAB *hashp;
586 	char *p;
587 	u_int32_t bucket;
588 	int is_bucket, is_bitmap;
589 {
590 	register int fd, page, size;
591 	int wsize;
592 
593 	size = hashp->BSIZE;
594 	if ((hashp->fp == -1) && open_temp(hashp))
595 		return (-1);
596 	fd = hashp->fp;
597 
598 	if (hashp->LORDER != BYTE_ORDER) {
599 		register int i;
600 		register int max;
601 
602 		if (is_bitmap) {
603 			max = hashp->BSIZE >> 2;	/* divide by 4 */
604 			for (i = 0; i < max; i++)
605 				M_32_SWAP(((int *)p)[i]);
606 		} else {
607 			max = ((u_int16_t *)p)[0] + 2;
608 			for (i = 0; i <= max; i++)
609 				M_16_SWAP(((u_int16_t *)p)[i]);
610 		}
611 	}
612 	if (is_bucket)
613 		page = BUCKET_TO_PAGE(bucket);
614 	else
615 		page = OADDR_TO_PAGE(bucket);
616 	if ((lseek(fd, (off_t)page << hashp->BSHIFT, SEEK_SET) == -1) ||
617 	    ((wsize = write(fd, p, size)) == -1))
618 		/* Errno is set */
619 		return (-1);
620 	if (wsize != size) {
621 		errno = EFTYPE;
622 		return (-1);
623 	}
624 	return (0);
625 }
626 
627 #define BYTE_MASK	((1 << INT_BYTE_SHIFT) -1)
628 /*
629  * Initialize a new bitmap page.  Bitmap pages are left in memory
630  * once they are read in.
631  */
632 extern int
633 __ibitmap(hashp, pnum, nbits, ndx)
634 	HTAB *hashp;
635 	int pnum, nbits, ndx;
636 {
637 	u_int32_t *ip;
638 	int clearbytes, clearints;
639 
640 	if ((ip = (u_int32_t *)malloc(hashp->BSIZE)) == NULL)
641 		return (1);
642 	hashp->nmaps++;
643 	clearints = ((nbits - 1) >> INT_BYTE_SHIFT) + 1;
644 	clearbytes = clearints << INT_TO_BYTE;
645 	(void)memset((char *)ip, 0, clearbytes);
646 	(void)memset(((char *)ip) + clearbytes, 0xFF,
647 	    hashp->BSIZE - clearbytes);
648 	ip[clearints - 1] = ALL_SET << (nbits & BYTE_MASK);
649 	SETBIT(ip, 0);
650 	hashp->BITMAPS[ndx] = (u_int16_t)pnum;
651 	hashp->mapp[ndx] = ip;
652 	return (0);
653 }
654 
655 static u_int32_t
656 first_free(map)
657 	u_int32_t map;
658 {
659 	register u_int32_t i, mask;
660 
661 	mask = 0x1;
662 	for (i = 0; i < BITS_PER_MAP; i++) {
663 		if (!(mask & map))
664 			return (i);
665 		mask = mask << 1;
666 	}
667 	return (i);
668 }
669 
670 static u_int16_t
671 overflow_page(hashp)
672 	HTAB *hashp;
673 {
674 	register u_int32_t *freep;
675 	register int max_free, offset, splitnum;
676 	u_int16_t addr;
677 	int bit, first_page, free_bit, free_page, i, in_use_bits, j;
678 #ifdef DEBUG2
679 	int tmp1, tmp2;
680 #endif
681 	splitnum = hashp->OVFL_POINT;
682 	max_free = hashp->SPARES[splitnum];
683 
684 	free_page = (max_free - 1) >> (hashp->BSHIFT + BYTE_SHIFT);
685 	free_bit = (max_free - 1) & ((hashp->BSIZE << BYTE_SHIFT) - 1);
686 
687 	/* Look through all the free maps to find the first free block */
688 	first_page = hashp->LAST_FREED >>(hashp->BSHIFT + BYTE_SHIFT);
689 	for ( i = first_page; i <= free_page; i++ ) {
690 		if (!(freep = (u_int32_t *)hashp->mapp[i]) &&
691 		    !(freep = fetch_bitmap(hashp, i)))
692 			return (NULL);
693 		if (i == free_page)
694 			in_use_bits = free_bit;
695 		else
696 			in_use_bits = (hashp->BSIZE << BYTE_SHIFT) - 1;
697 
698 		if (i == first_page) {
699 			bit = hashp->LAST_FREED &
700 			    ((hashp->BSIZE << BYTE_SHIFT) - 1);
701 			j = bit / BITS_PER_MAP;
702 			bit = bit & ~(BITS_PER_MAP - 1);
703 		} else {
704 			bit = 0;
705 			j = 0;
706 		}
707 		for (; bit <= in_use_bits; j++, bit += BITS_PER_MAP)
708 			if (freep[j] != ALL_SET)
709 				goto found;
710 	}
711 
712 	/* No Free Page Found */
713 	hashp->LAST_FREED = hashp->SPARES[splitnum];
714 	hashp->SPARES[splitnum]++;
715 	offset = hashp->SPARES[splitnum] -
716 	    (splitnum ? hashp->SPARES[splitnum - 1] : 0);
717 
718 #define	OVMSG	"HASH: Out of overflow pages.  Increase page size\n"
719 	if (offset > SPLITMASK) {
720 		if (++splitnum >= NCACHED) {
721 			(void)write(STDERR_FILENO, OVMSG, sizeof(OVMSG) - 1);
722 			return (NULL);
723 		}
724 		hashp->OVFL_POINT = splitnum;
725 		hashp->SPARES[splitnum] = hashp->SPARES[splitnum-1];
726 		hashp->SPARES[splitnum-1]--;
727 		offset = 1;
728 	}
729 
730 	/* Check if we need to allocate a new bitmap page */
731 	if (free_bit == (hashp->BSIZE << BYTE_SHIFT) - 1) {
732 		free_page++;
733 		if (free_page >= NCACHED) {
734 			(void)write(STDERR_FILENO, OVMSG, sizeof(OVMSG) - 1);
735 			return (NULL);
736 		}
737 		/*
738 		 * This is tricky.  The 1 indicates that you want the new page
739 		 * allocated with 1 clear bit.  Actually, you are going to
740 		 * allocate 2 pages from this map.  The first is going to be
741 		 * the map page, the second is the overflow page we were
742 		 * looking for.  The init_bitmap routine automatically, sets
743 		 * the first bit of itself to indicate that the bitmap itself
744 		 * is in use.  We would explicitly set the second bit, but
745 		 * don't have to if we tell init_bitmap not to leave it clear
746 		 * in the first place.
747 		 */
748 		if (__ibitmap(hashp, (int)OADDR_OF(splitnum, offset),
749 		    1, free_page))
750 			return (NULL);
751 		hashp->SPARES[splitnum]++;
752 #ifdef DEBUG2
753 		free_bit = 2;
754 #endif
755 		offset++;
756 		if (offset > SPLITMASK) {
757 			if (++splitnum >= NCACHED) {
758 				(void)write(STDERR_FILENO, OVMSG,
759 				    sizeof(OVMSG) - 1);
760 				return (NULL);
761 			}
762 			hashp->OVFL_POINT = splitnum;
763 			hashp->SPARES[splitnum] = hashp->SPARES[splitnum-1];
764 			hashp->SPARES[splitnum-1]--;
765 			offset = 0;
766 		}
767 	} else {
768 		/*
769 		 * Free_bit addresses the last used bit.  Bump it to address
770 		 * the first available bit.
771 		 */
772 		free_bit++;
773 		SETBIT(freep, free_bit);
774 	}
775 
776 	/* Calculate address of the new overflow page */
777 	addr = OADDR_OF(splitnum, offset);
778 #ifdef DEBUG2
779 	(void)fprintf(stderr, "OVERFLOW_PAGE: ADDR: %d BIT: %d PAGE %d\n",
780 	    addr, free_bit, free_page);
781 #endif
782 	return (addr);
783 
784 found:
785 	bit = bit + first_free(freep[j]);
786 	SETBIT(freep, bit);
787 #ifdef DEBUG2
788 	tmp1 = bit;
789 	tmp2 = i;
790 #endif
791 	/*
792 	 * Bits are addressed starting with 0, but overflow pages are addressed
793 	 * beginning at 1. Bit is a bit addressnumber, so we need to increment
794 	 * it to convert it to a page number.
795 	 */
796 	bit = 1 + bit + (i * (hashp->BSIZE << BYTE_SHIFT));
797 	if (bit >= hashp->LAST_FREED)
798 		hashp->LAST_FREED = bit - 1;
799 
800 	/* Calculate the split number for this page */
801 	for (i = 0; (i < splitnum) && (bit > hashp->SPARES[i]); i++);
802 	offset = (i ? bit - hashp->SPARES[i - 1] : bit);
803 	if (offset >= SPLITMASK)
804 		return (NULL);	/* Out of overflow pages */
805 	addr = OADDR_OF(i, offset);
806 #ifdef DEBUG2
807 	(void)fprintf(stderr, "OVERFLOW_PAGE: ADDR: %d BIT: %d PAGE %d\n",
808 	    addr, tmp1, tmp2);
809 #endif
810 
811 	/* Allocate and return the overflow page */
812 	return (addr);
813 }
814 
815 /*
816  * Mark this overflow page as free.
817  */
818 extern void
819 __free_ovflpage(hashp, obufp)
820 	HTAB *hashp;
821 	BUFHEAD *obufp;
822 {
823 	register u_int16_t addr;
824 	u_int32_t *freep;
825 	int bit_address, free_page, free_bit;
826 	u_int16_t ndx;
827 
828 	addr = obufp->addr;
829 #ifdef DEBUG1
830 	(void)fprintf(stderr, "Freeing %d\n", addr);
831 #endif
832 	ndx = (((u_int16_t)addr) >> SPLITSHIFT);
833 	bit_address =
834 	    (ndx ? hashp->SPARES[ndx - 1] : 0) + (addr & SPLITMASK) - 1;
835 	 if (bit_address < hashp->LAST_FREED)
836 		hashp->LAST_FREED = bit_address;
837 	free_page = (bit_address >> (hashp->BSHIFT + BYTE_SHIFT));
838 	free_bit = bit_address & ((hashp->BSIZE << BYTE_SHIFT) - 1);
839 
840 	if (!(freep = hashp->mapp[free_page]))
841 		freep = fetch_bitmap(hashp, free_page);
842 #ifdef DEBUG
843 	/*
844 	 * This had better never happen.  It means we tried to read a bitmap
845 	 * that has already had overflow pages allocated off it, and we
846 	 * failed to read it from the file.
847 	 */
848 	if (!freep)
849 		assert(0);
850 #endif
851 	CLRBIT(freep, free_bit);
852 #ifdef DEBUG2
853 	(void)fprintf(stderr, "FREE_OVFLPAGE: ADDR: %d BIT: %d PAGE %d\n",
854 	    obufp->addr, free_bit, free_page);
855 #endif
856 	__reclaim_buf(hashp, obufp);
857 }
858 
859 /*
860  * Returns:
861  *	 0 success
862  *	-1 failure
863  */
864 static int
865 open_temp(hashp)
866 	HTAB *hashp;
867 {
868 	sigset_t set, oset;
869 	static char namestr[] = "_hashXXXXXX";
870 
871 	/* Block signals; make sure file goes away at process exit. */
872 	(void)sigfillset(&set);
873 	(void)sigprocmask(SIG_BLOCK, &set, &oset);
874 	if ((hashp->fp = mkstemp(namestr)) != -1) {
875 		(void)unlink(namestr);
876 		(void)fcntl(hashp->fp, F_SETFD, 1);
877 	}
878 	(void)sigprocmask(SIG_SETMASK, &oset, (sigset_t *)NULL);
879 	return (hashp->fp != -1 ? 0 : -1);
880 }
881 
882 /*
883  * We have to know that the key will fit, but the last entry on the page is
884  * an overflow pair, so we need to shift things.
885  */
886 static void
887 squeeze_key(sp, key, val)
888 	u_int16_t *sp;
889 	const DBT *key, *val;
890 {
891 	register char *p;
892 	u_int16_t free_space, n, off, pageno;
893 
894 	p = (char *)sp;
895 	n = sp[0];
896 	free_space = FREESPACE(sp);
897 	off = OFFSET(sp);
898 
899 	pageno = sp[n - 1];
900 	off -= key->size;
901 	sp[n - 1] = off;
902 	memmove(p + off, key->data, key->size);
903 	off -= val->size;
904 	sp[n] = off;
905 	memmove(p + off, val->data, val->size);
906 	sp[0] = n + 2;
907 	sp[n + 1] = pageno;
908 	sp[n + 2] = OVFLPAGE;
909 	FREESPACE(sp) = free_space - PAIRSIZE(key, val);
910 	OFFSET(sp) = off;
911 }
912 
913 static u_int32_t *
914 fetch_bitmap(hashp, ndx)
915 	HTAB *hashp;
916 	int ndx;
917 {
918 	if (ndx >= hashp->nmaps)
919 		return (NULL);
920 	if ((hashp->mapp[ndx] = (u_int32_t *)malloc(hashp->BSIZE)) == NULL)
921 		return (NULL);
922 	if (__get_page(hashp,
923 	    (char *)hashp->mapp[ndx], hashp->BITMAPS[ndx], 0, 1, 1)) {
924 		free(hashp->mapp[ndx]);
925 		return (NULL);
926 	}
927 	return (hashp->mapp[ndx]);
928 }
929 
930 #ifdef DEBUG4
931 int
932 print_chain(addr)
933 	int addr;
934 {
935 	BUFHEAD *bufp;
936 	short *bp, oaddr;
937 
938 	(void)fprintf(stderr, "%d ", addr);
939 	bufp = __get_buf(hashp, addr, NULL, 0);
940 	bp = (short *)bufp->page;
941 	while (bp[0] && ((bp[bp[0]] == OVFLPAGE) ||
942 		((bp[0] > 2) && bp[2] < REAL_KEY))) {
943 		oaddr = bp[bp[0] - 1];
944 		(void)fprintf(stderr, "%d ", (int)oaddr);
945 		bufp = __get_buf(hashp, (int)oaddr, bufp, 0);
946 		bp = (short *)bufp->page;
947 	}
948 	(void)fprintf(stderr, "\n");
949 }
950 #endif
951