xref: /freebsd/lib/libc/db/hash/hash_bigkey.c (revision 8657387683946d0c03e09fe77029edfe309eeb20)
1 /*-
2  * Copyright (c) 1990, 1993, 1994
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * This code is derived from software contributed to Berkeley by
6  * Margo Seltzer.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 #if defined(LIBC_SCCS) && !defined(lint)
34 static char sccsid[] = "@(#)hash_bigkey.c	8.3 (Berkeley) 5/31/94";
35 #endif /* LIBC_SCCS and not lint */
36 #include <sys/cdefs.h>
37 __FBSDID("$FreeBSD$");
38 
39 /*
40  * PACKAGE: hash
41  * DESCRIPTION:
42  *	Big key/data handling for the hashing package.
43  *
44  * ROUTINES:
45  * External
46  *	__big_keydata
47  *	__big_split
48  *	__big_insert
49  *	__big_return
50  *	__big_delete
51  *	__find_last_page
52  * Internal
53  *	collect_key
54  *	collect_data
55  */
56 
57 #include <sys/param.h>
58 
59 #include <errno.h>
60 #include <stdio.h>
61 #include <stdlib.h>
62 #include <string.h>
63 
64 #ifdef DEBUG
65 #include <assert.h>
66 #endif
67 
68 #include <db.h>
69 #include "hash.h"
70 #include "page.h"
71 #include "extern.h"
72 
73 static int collect_key(HTAB *, BUFHEAD *, int, DBT *, int);
74 static int collect_data(HTAB *, BUFHEAD *, int, int);
75 
76 /*
77  * Big_insert
78  *
79  * You need to do an insert and the key/data pair is too big
80  *
81  * Returns:
82  * 0 ==> OK
83  *-1 ==> ERROR
84  */
85 int
86 __big_insert(HTAB *hashp, BUFHEAD *bufp, const DBT *key, const DBT *val)
87 {
88 	u_int16_t *p;
89 	int key_size, n;
90 	unsigned int val_size;
91 	u_int16_t space, move_bytes, off;
92 	char *cp, *key_data, *val_data;
93 
94 	cp = bufp->page;		/* Character pointer of p. */
95 	p = (u_int16_t *)cp;
96 
97 	key_data = (char *)key->data;
98 	key_size = key->size;
99 	val_data = (char *)val->data;
100 	val_size = val->size;
101 
102 	/* First move the Key */
103 	for (space = FREESPACE(p) - BIGOVERHEAD; key_size;
104 	    space = FREESPACE(p) - BIGOVERHEAD) {
105 		move_bytes = MIN(space, key_size);
106 		off = OFFSET(p) - move_bytes;
107 		memmove(cp + off, key_data, move_bytes);
108 		key_size -= move_bytes;
109 		key_data += move_bytes;
110 		n = p[0];
111 		p[++n] = off;
112 		p[0] = ++n;
113 		FREESPACE(p) = off - PAGE_META(n);
114 		OFFSET(p) = off;
115 		p[n] = PARTIAL_KEY;
116 		bufp = __add_ovflpage(hashp, bufp);
117 		if (!bufp)
118 			return (-1);
119 		n = p[0];
120 		if (!key_size) {
121 			space = FREESPACE(p);
122 			if (space) {
123 				move_bytes = MIN(space, val_size);
124 				/*
125 				 * If the data would fit exactly in the
126 				 * remaining space, we must overflow it to the
127 				 * next page; otherwise the invariant that the
128 				 * data must end on a page with FREESPACE
129 				 * non-zero would fail.
130 				 */
131 				if (space == val_size && val_size == val->size)
132 					goto toolarge;
133 				off = OFFSET(p) - move_bytes;
134 				memmove(cp + off, val_data, move_bytes);
135 				val_data += move_bytes;
136 				val_size -= move_bytes;
137 				p[n] = off;
138 				p[n - 2] = FULL_KEY_DATA;
139 				FREESPACE(p) = FREESPACE(p) - move_bytes;
140 				OFFSET(p) = off;
141 			} else {
142 			toolarge:
143 				p[n - 2] = FULL_KEY;
144 			}
145 		}
146 		p = (u_int16_t *)bufp->page;
147 		cp = bufp->page;
148 		bufp->flags |= BUF_MOD;
149 	}
150 
151 	/* Now move the data */
152 	for (space = FREESPACE(p) - BIGOVERHEAD; val_size;
153 	    space = FREESPACE(p) - BIGOVERHEAD) {
154 		move_bytes = MIN(space, val_size);
155 		/*
156 		 * Here's the hack to make sure that if the data ends on the
157 		 * same page as the key ends, FREESPACE is at least one.
158 		 */
159 		if (space == val_size && val_size == val->size)
160 			move_bytes--;
161 		off = OFFSET(p) - move_bytes;
162 		memmove(cp + off, val_data, move_bytes);
163 		val_size -= move_bytes;
164 		val_data += move_bytes;
165 		n = p[0];
166 		p[++n] = off;
167 		p[0] = ++n;
168 		FREESPACE(p) = off - PAGE_META(n);
169 		OFFSET(p) = off;
170 		if (val_size) {
171 			p[n] = FULL_KEY;
172 			bufp = __add_ovflpage(hashp, bufp);
173 			if (!bufp)
174 				return (-1);
175 			cp = bufp->page;
176 			p = (u_int16_t *)cp;
177 		} else
178 			p[n] = FULL_KEY_DATA;
179 		bufp->flags |= BUF_MOD;
180 	}
181 	return (0);
182 }
183 
184 /*
185  * Called when bufp's page  contains a partial key (index should be 1)
186  *
187  * All pages in the big key/data pair except bufp are freed.  We cannot
188  * free bufp because the page pointing to it is lost and we can't get rid
189  * of its pointer.
190  *
191  * Returns:
192  * 0 => OK
193  *-1 => ERROR
194  */
195 int
196 __big_delete(HTAB *hashp, BUFHEAD *bufp)
197 {
198 	BUFHEAD *last_bfp, *rbufp;
199 	u_int16_t *bp, pageno;
200 	int key_done, n;
201 
202 	rbufp = bufp;
203 	last_bfp = NULL;
204 	bp = (u_int16_t *)bufp->page;
205 	pageno = 0;
206 	key_done = 0;
207 
208 	while (!key_done || (bp[2] != FULL_KEY_DATA)) {
209 		if (bp[2] == FULL_KEY || bp[2] == FULL_KEY_DATA)
210 			key_done = 1;
211 
212 		/*
213 		 * If there is freespace left on a FULL_KEY_DATA page, then
214 		 * the data is short and fits entirely on this page, and this
215 		 * is the last page.
216 		 */
217 		if (bp[2] == FULL_KEY_DATA && FREESPACE(bp))
218 			break;
219 		pageno = bp[bp[0] - 1];
220 		rbufp->flags |= BUF_MOD;
221 		rbufp = __get_buf(hashp, pageno, rbufp, 0);
222 		if (last_bfp)
223 			__free_ovflpage(hashp, last_bfp);
224 		last_bfp = rbufp;
225 		if (!rbufp)
226 			return (-1);		/* Error. */
227 		bp = (u_int16_t *)rbufp->page;
228 	}
229 
230 	/*
231 	 * If we get here then rbufp points to the last page of the big
232 	 * key/data pair.  Bufp points to the first one -- it should now be
233 	 * empty pointing to the next page after this pair.  Can't free it
234 	 * because we don't have the page pointing to it.
235 	 */
236 
237 	/* This is information from the last page of the pair. */
238 	n = bp[0];
239 	pageno = bp[n - 1];
240 
241 	/* Now, bp is the first page of the pair. */
242 	bp = (u_int16_t *)bufp->page;
243 	if (n > 2) {
244 		/* There is an overflow page. */
245 		bp[1] = pageno;
246 		bp[2] = OVFLPAGE;
247 		bufp->ovfl = rbufp->ovfl;
248 	} else
249 		/* This is the last page. */
250 		bufp->ovfl = NULL;
251 	n -= 2;
252 	bp[0] = n;
253 	FREESPACE(bp) = hashp->BSIZE - PAGE_META(n);
254 	OFFSET(bp) = hashp->BSIZE;
255 
256 	bufp->flags |= BUF_MOD;
257 	if (rbufp)
258 		__free_ovflpage(hashp, rbufp);
259 	if (last_bfp && last_bfp != rbufp)
260 		__free_ovflpage(hashp, last_bfp);
261 
262 	hashp->NKEYS--;
263 	return (0);
264 }
265 /*
266  * Returns:
267  *  0 = key not found
268  * -1 = get next overflow page
269  * -2 means key not found and this is big key/data
270  * -3 error
271  */
272 int
273 __find_bigpair(HTAB *hashp, BUFHEAD *bufp, int ndx, char *key, int size)
274 {
275 	u_int16_t *bp;
276 	char *p;
277 	int ksize;
278 	u_int16_t bytes;
279 	char *kkey;
280 
281 	bp = (u_int16_t *)bufp->page;
282 	p = bufp->page;
283 	ksize = size;
284 	kkey = key;
285 
286 	for (bytes = hashp->BSIZE - bp[ndx];
287 	    bytes <= size && bp[ndx + 1] == PARTIAL_KEY;
288 	    bytes = hashp->BSIZE - bp[ndx]) {
289 		if (memcmp(p + bp[ndx], kkey, bytes))
290 			return (-2);
291 		kkey += bytes;
292 		ksize -= bytes;
293 		bufp = __get_buf(hashp, bp[ndx + 2], bufp, 0);
294 		if (!bufp)
295 			return (-3);
296 		p = bufp->page;
297 		bp = (u_int16_t *)p;
298 		ndx = 1;
299 	}
300 
301 	if (bytes != ksize || memcmp(p + bp[ndx], kkey, bytes)) {
302 #ifdef HASH_STATISTICS
303 		++hash_collisions;
304 #endif
305 		return (-2);
306 	} else
307 		return (ndx);
308 }
309 
310 /*
311  * Given the buffer pointer of the first overflow page of a big pair,
312  * find the end of the big pair
313  *
314  * This will set bpp to the buffer header of the last page of the big pair.
315  * It will return the pageno of the overflow page following the last page
316  * of the pair; 0 if there isn't any (i.e. big pair is the last key in the
317  * bucket)
318  */
319 u_int16_t
320 __find_last_page(HTAB *hashp, BUFHEAD **bpp)
321 {
322 	BUFHEAD *bufp;
323 	u_int16_t *bp, pageno;
324 	int n;
325 
326 	bufp = *bpp;
327 	bp = (u_int16_t *)bufp->page;
328 	for (;;) {
329 		n = bp[0];
330 
331 		/*
332 		 * This is the last page if: the tag is FULL_KEY_DATA and
333 		 * either only 2 entries OVFLPAGE marker is explicit there
334 		 * is freespace on the page.
335 		 */
336 		if (bp[2] == FULL_KEY_DATA &&
337 		    ((n == 2) || (bp[n] == OVFLPAGE) || (FREESPACE(bp))))
338 			break;
339 
340 		pageno = bp[n - 1];
341 		bufp = __get_buf(hashp, pageno, bufp, 0);
342 		if (!bufp)
343 			return (0);	/* Need to indicate an error! */
344 		bp = (u_int16_t *)bufp->page;
345 	}
346 
347 	*bpp = bufp;
348 	if (bp[0] > 2)
349 		return (bp[3]);
350 	else
351 		return (0);
352 }
353 
354 /*
355  * Return the data for the key/data pair that begins on this page at this
356  * index (index should always be 1).
357  */
358 int
359 __big_return(HTAB *hashp, BUFHEAD *bufp, int ndx, DBT *val, int set_current)
360 {
361 	BUFHEAD *save_p;
362 	u_int16_t *bp, len, off, save_addr;
363 	char *tp;
364 
365 	bp = (u_int16_t *)bufp->page;
366 	while (bp[ndx + 1] == PARTIAL_KEY) {
367 		bufp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
368 		if (!bufp)
369 			return (-1);
370 		bp = (u_int16_t *)bufp->page;
371 		ndx = 1;
372 	}
373 
374 	if (bp[ndx + 1] == FULL_KEY) {
375 		bufp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
376 		if (!bufp)
377 			return (-1);
378 		bp = (u_int16_t *)bufp->page;
379 		save_p = bufp;
380 		save_addr = save_p->addr;
381 		off = bp[1];
382 		len = 0;
383 	} else
384 		if (!FREESPACE(bp)) {
385 			/*
386 			 * This is a hack.  We can't distinguish between
387 			 * FULL_KEY_DATA that contains complete data or
388 			 * incomplete data, so we require that if the data
389 			 * is complete, there is at least 1 byte of free
390 			 * space left.
391 			 */
392 			off = bp[bp[0]];
393 			len = bp[1] - off;
394 			save_p = bufp;
395 			save_addr = bufp->addr;
396 			bufp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
397 			if (!bufp)
398 				return (-1);
399 			bp = (u_int16_t *)bufp->page;
400 		} else {
401 			/* The data is all on one page. */
402 			tp = (char *)bp;
403 			off = bp[bp[0]];
404 			val->data = (u_char *)tp + off;
405 			val->size = bp[1] - off;
406 			if (set_current) {
407 				if (bp[0] == 2) {	/* No more buckets in
408 							 * chain */
409 					hashp->cpage = NULL;
410 					hashp->cbucket++;
411 					hashp->cndx = 1;
412 				} else {
413 					hashp->cpage = __get_buf(hashp,
414 					    bp[bp[0] - 1], bufp, 0);
415 					if (!hashp->cpage)
416 						return (-1);
417 					hashp->cndx = 1;
418 					if (!((u_int16_t *)
419 					    hashp->cpage->page)[0]) {
420 						hashp->cbucket++;
421 						hashp->cpage = NULL;
422 					}
423 				}
424 			}
425 			return (0);
426 		}
427 
428 	val->size = (size_t)collect_data(hashp, bufp, (int)len, set_current);
429 	if (val->size == (size_t)-1)
430 		return (-1);
431 	if (save_p->addr != save_addr) {
432 		/* We are pretty short on buffers. */
433 		errno = EINVAL;			/* OUT OF BUFFERS */
434 		return (-1);
435 	}
436 	memmove(hashp->tmp_buf, (save_p->page) + off, len);
437 	val->data = (u_char *)hashp->tmp_buf;
438 	return (0);
439 }
440 /*
441  * Count how big the total datasize is by recursing through the pages.  Then
442  * allocate a buffer and copy the data as you recurse up.
443  */
444 static int
445 collect_data(HTAB *hashp, BUFHEAD *bufp, int len, int set)
446 {
447 	u_int16_t *bp;
448 	char *p;
449 	BUFHEAD *xbp;
450 	u_int16_t save_addr;
451 	int mylen, totlen;
452 
453 	p = bufp->page;
454 	bp = (u_int16_t *)p;
455 	mylen = hashp->BSIZE - bp[1];
456 	save_addr = bufp->addr;
457 
458 	if (bp[2] == FULL_KEY_DATA) {		/* End of Data */
459 		totlen = len + mylen;
460 		if (hashp->tmp_buf)
461 			free(hashp->tmp_buf);
462 		if ((hashp->tmp_buf = (char *)malloc(totlen)) == NULL)
463 			return (-1);
464 		if (set) {
465 			hashp->cndx = 1;
466 			if (bp[0] == 2) {	/* No more buckets in chain */
467 				hashp->cpage = NULL;
468 				hashp->cbucket++;
469 			} else {
470 				hashp->cpage =
471 				    __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
472 				if (!hashp->cpage)
473 					return (-1);
474 				else if (!((u_int16_t *)hashp->cpage->page)[0]) {
475 					hashp->cbucket++;
476 					hashp->cpage = NULL;
477 				}
478 			}
479 		}
480 	} else {
481 		xbp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
482 		if (!xbp || ((totlen =
483 		    collect_data(hashp, xbp, len + mylen, set)) < 1))
484 			return (-1);
485 	}
486 	if (bufp->addr != save_addr) {
487 		errno = EINVAL;			/* Out of buffers. */
488 		return (-1);
489 	}
490 	memmove(&hashp->tmp_buf[len], (bufp->page) + bp[1], mylen);
491 	return (totlen);
492 }
493 
494 /*
495  * Fill in the key and data for this big pair.
496  */
497 int
498 __big_keydata(HTAB *hashp, BUFHEAD *bufp, DBT *key, DBT *val, int set)
499 {
500 	key->size = (size_t)collect_key(hashp, bufp, 0, val, set);
501 	if (key->size == (size_t)-1)
502 		return (-1);
503 	key->data = (u_char *)hashp->tmp_key;
504 	return (0);
505 }
506 
507 /*
508  * Count how big the total key size is by recursing through the pages.  Then
509  * collect the data, allocate a buffer and copy the key as you recurse up.
510  */
511 static int
512 collect_key(HTAB *hashp, BUFHEAD *bufp, int len, DBT *val, int set)
513 {
514 	BUFHEAD *xbp;
515 	char *p;
516 	int mylen, totlen;
517 	u_int16_t *bp, save_addr;
518 
519 	p = bufp->page;
520 	bp = (u_int16_t *)p;
521 	mylen = hashp->BSIZE - bp[1];
522 
523 	save_addr = bufp->addr;
524 	totlen = len + mylen;
525 	if (bp[2] == FULL_KEY || bp[2] == FULL_KEY_DATA) {    /* End of Key. */
526 		if (hashp->tmp_key != NULL)
527 			free(hashp->tmp_key);
528 		if ((hashp->tmp_key = (char *)malloc(totlen)) == NULL)
529 			return (-1);
530 		if (__big_return(hashp, bufp, 1, val, set))
531 			return (-1);
532 	} else {
533 		xbp = __get_buf(hashp, bp[bp[0] - 1], bufp, 0);
534 		if (!xbp || ((totlen =
535 		    collect_key(hashp, xbp, totlen, val, set)) < 1))
536 			return (-1);
537 	}
538 	if (bufp->addr != save_addr) {
539 		errno = EINVAL;		/* MIS -- OUT OF BUFFERS */
540 		return (-1);
541 	}
542 	memmove(&hashp->tmp_key[len], (bufp->page) + bp[1], mylen);
543 	return (totlen);
544 }
545 
546 /*
547  * Returns:
548  *  0 => OK
549  * -1 => error
550  */
551 int
552 __big_split(HTAB *hashp,
553     BUFHEAD *op,	/* Pointer to where to put keys that go in old bucket */
554     BUFHEAD *np,	/* Pointer to new bucket page */
555     BUFHEAD *big_keyp,	/* Pointer to first page containing the big key/data */
556     int addr,		/* Address of big_keyp */
557     u_int32_t obucket,	/* Old Bucket */
558     SPLIT_RETURN *ret)
559 {
560 	BUFHEAD *bp, *tmpp;
561 	DBT key, val;
562 	u_int32_t change;
563 	u_int16_t free_space, n, off, *tp;
564 
565 	bp = big_keyp;
566 
567 	/* Now figure out where the big key/data goes */
568 	if (__big_keydata(hashp, big_keyp, &key, &val, 0))
569 		return (-1);
570 	change = (__call_hash(hashp, key.data, key.size) != obucket);
571 
572 	if ( (ret->next_addr = __find_last_page(hashp, &big_keyp)) ) {
573 		if (!(ret->nextp =
574 		    __get_buf(hashp, ret->next_addr, big_keyp, 0)))
575 			return (-1);
576 	} else
577 		ret->nextp = NULL;
578 
579 	/* Now make one of np/op point to the big key/data pair */
580 #ifdef DEBUG
581 	assert(np->ovfl == NULL);
582 #endif
583 	if (change)
584 		tmpp = np;
585 	else
586 		tmpp = op;
587 
588 	tmpp->flags |= BUF_MOD;
589 #ifdef DEBUG1
590 	(void)fprintf(stderr,
591 	    "BIG_SPLIT: %d->ovfl was %d is now %d\n", tmpp->addr,
592 	    (tmpp->ovfl ? tmpp->ovfl->addr : 0), (bp ? bp->addr : 0));
593 #endif
594 	tmpp->ovfl = bp;	/* one of op/np point to big_keyp */
595 	tp = (u_int16_t *)tmpp->page;
596 #ifdef DEBUG
597 	assert(FREESPACE(tp) >= OVFLSIZE);
598 #endif
599 	n = tp[0];
600 	off = OFFSET(tp);
601 	free_space = FREESPACE(tp);
602 	tp[++n] = (u_int16_t)addr;
603 	tp[++n] = OVFLPAGE;
604 	tp[0] = n;
605 	OFFSET(tp) = off;
606 	FREESPACE(tp) = free_space - OVFLSIZE;
607 
608 	/*
609 	 * Finally, set the new and old return values. BIG_KEYP contains a
610 	 * pointer to the last page of the big key_data pair. Make sure that
611 	 * big_keyp has no following page (2 elements) or create an empty
612 	 * following page.
613 	 */
614 
615 	ret->newp = np;
616 	ret->oldp = op;
617 
618 	tp = (u_int16_t *)big_keyp->page;
619 	big_keyp->flags |= BUF_MOD;
620 	if (tp[0] > 2) {
621 		/*
622 		 * There may be either one or two offsets on this page.  If
623 		 * there is one, then the overflow page is linked on normally
624 		 * and tp[4] is OVFLPAGE.  If there are two, tp[4] contains
625 		 * the second offset and needs to get stuffed in after the
626 		 * next overflow page is added.
627 		 */
628 		n = tp[4];
629 		free_space = FREESPACE(tp);
630 		off = OFFSET(tp);
631 		tp[0] -= 2;
632 		FREESPACE(tp) = free_space + OVFLSIZE;
633 		OFFSET(tp) = off;
634 		tmpp = __add_ovflpage(hashp, big_keyp);
635 		if (!tmpp)
636 			return (-1);
637 		tp[4] = n;
638 	} else
639 		tmpp = big_keyp;
640 
641 	if (change)
642 		ret->newp = tmpp;
643 	else
644 		ret->oldp = tmpp;
645 	return (0);
646 }
647