xref: /freebsd/lib/libc/db/btree/bt_delete.c (revision a7623790fb345e6dc986dfd31df0ace115e6f2e4)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
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  * Mike Olson.
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 #if defined(LIBC_SCCS) && !defined(lint)
36 static char sccsid[] = "@(#)bt_delete.c	8.13 (Berkeley) 7/28/94";
37 #endif /* LIBC_SCCS and not lint */
38 #include <sys/cdefs.h>
39 __FBSDID("$FreeBSD$");
40 
41 #include <sys/types.h>
42 
43 #include <errno.h>
44 #include <stdio.h>
45 #include <string.h>
46 
47 #include <db.h>
48 #include "btree.h"
49 
50 static int __bt_bdelete(BTREE *, const DBT *);
51 static int __bt_curdel(BTREE *, const DBT *, PAGE *, u_int);
52 static int __bt_pdelete(BTREE *, PAGE *);
53 static int __bt_relink(BTREE *, PAGE *);
54 static int __bt_stkacq(BTREE *, PAGE **, CURSOR *);
55 
56 /*
57  * __bt_delete
58  *	Delete the item(s) referenced by a key.
59  *
60  * Return RET_SPECIAL if the key is not found.
61  */
62 int
63 __bt_delete(const DB *dbp, const DBT *key, u_int flags)
64 {
65 	BTREE *t;
66 	CURSOR *c;
67 	PAGE *h;
68 	int status;
69 
70 	t = dbp->internal;
71 
72 	/* Toss any page pinned across calls. */
73 	if (t->bt_pinned != NULL) {
74 		mpool_put(t->bt_mp, t->bt_pinned, 0);
75 		t->bt_pinned = NULL;
76 	}
77 
78 	/* Check for change to a read-only tree. */
79 	if (F_ISSET(t, B_RDONLY)) {
80 		errno = EPERM;
81 		return (RET_ERROR);
82 	}
83 
84 	switch (flags) {
85 	case 0:
86 		status = __bt_bdelete(t, key);
87 		break;
88 	case R_CURSOR:
89 		/*
90 		 * If flags is R_CURSOR, delete the cursor.  Must already
91 		 * have started a scan and not have already deleted it.
92 		 */
93 		c = &t->bt_cursor;
94 		if (F_ISSET(c, CURS_INIT)) {
95 			if (F_ISSET(c, CURS_ACQUIRE | CURS_AFTER | CURS_BEFORE))
96 				return (RET_SPECIAL);
97 			if ((h = mpool_get(t->bt_mp, c->pg.pgno, 0)) == NULL)
98 				return (RET_ERROR);
99 
100 			/*
101 			 * If the page is about to be emptied, we'll need to
102 			 * delete it, which means we have to acquire a stack.
103 			 */
104 			if (NEXTINDEX(h) == 1)
105 				if (__bt_stkacq(t, &h, &t->bt_cursor))
106 					return (RET_ERROR);
107 
108 			status = __bt_dleaf(t, NULL, h, c->pg.index);
109 
110 			if (NEXTINDEX(h) == 0 && status == RET_SUCCESS) {
111 				if (__bt_pdelete(t, h))
112 					return (RET_ERROR);
113 			} else
114 				mpool_put(t->bt_mp,
115 				    h, status == RET_SUCCESS ? MPOOL_DIRTY : 0);
116 			break;
117 		}
118 		/* FALLTHROUGH */
119 	default:
120 		errno = EINVAL;
121 		return (RET_ERROR);
122 	}
123 	if (status == RET_SUCCESS)
124 		F_SET(t, B_MODIFIED);
125 	return (status);
126 }
127 
128 /*
129  * __bt_stkacq --
130  *	Acquire a stack so we can delete a cursor entry.
131  *
132  * Parameters:
133  *	  t:	tree
134  *	 hp:	pointer to current, pinned PAGE pointer
135  *	  c:	pointer to the cursor
136  *
137  * Returns:
138  *	0 on success, 1 on failure
139  */
140 static int
141 __bt_stkacq(BTREE *t, PAGE **hp, CURSOR *c)
142 {
143 	BINTERNAL *bi;
144 	EPG *e;
145 	EPGNO *parent;
146 	PAGE *h;
147 	indx_t idx;
148 	pgno_t pgno;
149 	recno_t nextpg, prevpg;
150 	int exact, level;
151 
152 	/*
153 	 * Find the first occurrence of the key in the tree.  Toss the
154 	 * currently locked page so we don't hit an already-locked page.
155 	 */
156 	h = *hp;
157 	mpool_put(t->bt_mp, h, 0);
158 	if ((e = __bt_search(t, &c->key, &exact)) == NULL)
159 		return (1);
160 	h = e->page;
161 
162 	/* See if we got it in one shot. */
163 	if (h->pgno == c->pg.pgno)
164 		goto ret;
165 
166 	/*
167 	 * Move right, looking for the page.  At each move we have to move
168 	 * up the stack until we don't have to move to the next page.  If
169 	 * we have to change pages at an internal level, we have to fix the
170 	 * stack back up.
171 	 */
172 	while (h->pgno != c->pg.pgno) {
173 		if ((nextpg = h->nextpg) == P_INVALID)
174 			break;
175 		mpool_put(t->bt_mp, h, 0);
176 
177 		/* Move up the stack. */
178 		for (level = 0; (parent = BT_POP(t)) != NULL; ++level) {
179 			/* Get the parent page. */
180 			if ((h = mpool_get(t->bt_mp, parent->pgno, 0)) == NULL)
181 				return (1);
182 
183 			/* Move to the next index. */
184 			if (parent->index != NEXTINDEX(h) - 1) {
185 				idx = parent->index + 1;
186 				BT_PUSH(t, h->pgno, idx);
187 				break;
188 			}
189 			mpool_put(t->bt_mp, h, 0);
190 		}
191 
192 		/* Restore the stack. */
193 		while (level--) {
194 			/* Push the next level down onto the stack. */
195 			bi = GETBINTERNAL(h, idx);
196 			pgno = bi->pgno;
197 			BT_PUSH(t, pgno, 0);
198 
199 			/* Lose the currently pinned page. */
200 			mpool_put(t->bt_mp, h, 0);
201 
202 			/* Get the next level down. */
203 			if ((h = mpool_get(t->bt_mp, pgno, 0)) == NULL)
204 				return (1);
205 			idx = 0;
206 		}
207 		mpool_put(t->bt_mp, h, 0);
208 		if ((h = mpool_get(t->bt_mp, nextpg, 0)) == NULL)
209 			return (1);
210 	}
211 
212 	if (h->pgno == c->pg.pgno)
213 		goto ret;
214 
215 	/* Reacquire the original stack. */
216 	mpool_put(t->bt_mp, h, 0);
217 	if ((e = __bt_search(t, &c->key, &exact)) == NULL)
218 		return (1);
219 	h = e->page;
220 
221 	/*
222 	 * Move left, looking for the page.  At each move we have to move
223 	 * up the stack until we don't have to change pages to move to the
224 	 * next page.  If we have to change pages at an internal level, we
225 	 * have to fix the stack back up.
226 	 */
227 	while (h->pgno != c->pg.pgno) {
228 		if ((prevpg = h->prevpg) == P_INVALID)
229 			break;
230 		mpool_put(t->bt_mp, h, 0);
231 
232 		/* Move up the stack. */
233 		for (level = 0; (parent = BT_POP(t)) != NULL; ++level) {
234 			/* Get the parent page. */
235 			if ((h = mpool_get(t->bt_mp, parent->pgno, 0)) == NULL)
236 				return (1);
237 
238 			/* Move to the next index. */
239 			if (parent->index != 0) {
240 				idx = parent->index - 1;
241 				BT_PUSH(t, h->pgno, idx);
242 				break;
243 			}
244 			mpool_put(t->bt_mp, h, 0);
245 		}
246 
247 		/* Restore the stack. */
248 		while (level--) {
249 			/* Push the next level down onto the stack. */
250 			bi = GETBINTERNAL(h, idx);
251 			pgno = bi->pgno;
252 
253 			/* Lose the currently pinned page. */
254 			mpool_put(t->bt_mp, h, 0);
255 
256 			/* Get the next level down. */
257 			if ((h = mpool_get(t->bt_mp, pgno, 0)) == NULL)
258 				return (1);
259 
260 			idx = NEXTINDEX(h) - 1;
261 			BT_PUSH(t, pgno, idx);
262 		}
263 		mpool_put(t->bt_mp, h, 0);
264 		if ((h = mpool_get(t->bt_mp, prevpg, 0)) == NULL)
265 			return (1);
266 	}
267 
268 
269 ret:	mpool_put(t->bt_mp, h, 0);
270 	return ((*hp = mpool_get(t->bt_mp, c->pg.pgno, 0)) == NULL);
271 }
272 
273 /*
274  * __bt_bdelete --
275  *	Delete all key/data pairs matching the specified key.
276  *
277  * Parameters:
278  *	  t:	tree
279  *	key:	key to delete
280  *
281  * Returns:
282  *	RET_ERROR, RET_SUCCESS and RET_SPECIAL if the key not found.
283  */
284 static int
285 __bt_bdelete(BTREE *t, const DBT *key)
286 {
287 	EPG *e;
288 	PAGE *h;
289 	int deleted, exact, redo;
290 
291 	deleted = 0;
292 
293 	/* Find any matching record; __bt_search pins the page. */
294 loop:	if ((e = __bt_search(t, key, &exact)) == NULL)
295 		return (deleted ? RET_SUCCESS : RET_ERROR);
296 	if (!exact) {
297 		mpool_put(t->bt_mp, e->page, 0);
298 		return (deleted ? RET_SUCCESS : RET_SPECIAL);
299 	}
300 
301 	/*
302 	 * Delete forward, then delete backward, from the found key.  If
303 	 * there are duplicates and we reach either side of the page, do
304 	 * the key search again, so that we get them all.
305 	 */
306 	redo = 0;
307 	h = e->page;
308 	do {
309 		if (__bt_dleaf(t, key, h, e->index)) {
310 			mpool_put(t->bt_mp, h, 0);
311 			return (RET_ERROR);
312 		}
313 		if (F_ISSET(t, B_NODUPS)) {
314 			if (NEXTINDEX(h) == 0) {
315 				if (__bt_pdelete(t, h))
316 					return (RET_ERROR);
317 			} else
318 				mpool_put(t->bt_mp, h, MPOOL_DIRTY);
319 			return (RET_SUCCESS);
320 		}
321 		deleted = 1;
322 	} while (e->index < NEXTINDEX(h) && __bt_cmp(t, key, e) == 0);
323 
324 	/* Check for right-hand edge of the page. */
325 	if (e->index == NEXTINDEX(h))
326 		redo = 1;
327 
328 	/* Delete from the key to the beginning of the page. */
329 	while (e->index-- > 0) {
330 		if (__bt_cmp(t, key, e) != 0)
331 			break;
332 		if (__bt_dleaf(t, key, h, e->index) == RET_ERROR) {
333 			mpool_put(t->bt_mp, h, 0);
334 			return (RET_ERROR);
335 		}
336 		if (e->index == 0)
337 			redo = 1;
338 	}
339 
340 	/* Check for an empty page. */
341 	if (NEXTINDEX(h) == 0) {
342 		if (__bt_pdelete(t, h))
343 			return (RET_ERROR);
344 		goto loop;
345 	}
346 
347 	/* Put the page. */
348 	mpool_put(t->bt_mp, h, MPOOL_DIRTY);
349 
350 	if (redo)
351 		goto loop;
352 	return (RET_SUCCESS);
353 }
354 
355 /*
356  * __bt_pdelete --
357  *	Delete a single page from the tree.
358  *
359  * Parameters:
360  *	t:	tree
361  *	h:	leaf page
362  *
363  * Returns:
364  *	RET_SUCCESS, RET_ERROR.
365  *
366  * Side-effects:
367  *	mpool_put's the page
368  */
369 static int
370 __bt_pdelete(BTREE *t, PAGE *h)
371 {
372 	BINTERNAL *bi;
373 	PAGE *pg;
374 	EPGNO *parent;
375 	indx_t cnt, idx, *ip, offset;
376 	u_int32_t nksize;
377 	char *from;
378 
379 	/*
380 	 * Walk the parent page stack -- a LIFO stack of the pages that were
381 	 * traversed when we searched for the page where the delete occurred.
382 	 * Each stack entry is a page number and a page index offset.  The
383 	 * offset is for the page traversed on the search.  We've just deleted
384 	 * a page, so we have to delete the key from the parent page.
385 	 *
386 	 * If the delete from the parent page makes it empty, this process may
387 	 * continue all the way up the tree.  We stop if we reach the root page
388 	 * (which is never deleted, it's just not worth the effort) or if the
389 	 * delete does not empty the page.
390 	 */
391 	while ((parent = BT_POP(t)) != NULL) {
392 		/* Get the parent page. */
393 		if ((pg = mpool_get(t->bt_mp, parent->pgno, 0)) == NULL)
394 			return (RET_ERROR);
395 
396 		idx = parent->index;
397 		bi = GETBINTERNAL(pg, idx);
398 
399 		/* Free any overflow pages. */
400 		if (bi->flags & P_BIGKEY &&
401 		    __ovfl_delete(t, bi->bytes) == RET_ERROR) {
402 			mpool_put(t->bt_mp, pg, 0);
403 			return (RET_ERROR);
404 		}
405 
406 		/*
407 		 * Free the parent if it has only the one key and it's not the
408 		 * root page. If it's the rootpage, turn it back into an empty
409 		 * leaf page.
410 		 */
411 		if (NEXTINDEX(pg) == 1) {
412 			if (pg->pgno == P_ROOT) {
413 				pg->lower = BTDATAOFF;
414 				pg->upper = t->bt_psize;
415 				pg->flags = P_BLEAF;
416 			} else {
417 				if (__bt_relink(t, pg) || __bt_free(t, pg))
418 					return (RET_ERROR);
419 				continue;
420 			}
421 		} else {
422 			/* Pack remaining key items at the end of the page. */
423 			nksize = NBINTERNAL(bi->ksize);
424 			from = (char *)pg + pg->upper;
425 			memmove(from + nksize, from, (char *)bi - from);
426 			pg->upper += nksize;
427 
428 			/* Adjust indices' offsets, shift the indices down. */
429 			offset = pg->linp[idx];
430 			for (cnt = idx, ip = &pg->linp[0]; cnt--; ++ip)
431 				if (ip[0] < offset)
432 					ip[0] += nksize;
433 			for (cnt = NEXTINDEX(pg) - idx; --cnt; ++ip)
434 				ip[0] = ip[1] < offset ? ip[1] + nksize : ip[1];
435 			pg->lower -= sizeof(indx_t);
436 		}
437 
438 		mpool_put(t->bt_mp, pg, MPOOL_DIRTY);
439 		break;
440 	}
441 
442 	/* Free the leaf page, as long as it wasn't the root. */
443 	if (h->pgno == P_ROOT) {
444 		mpool_put(t->bt_mp, h, MPOOL_DIRTY);
445 		return (RET_SUCCESS);
446 	}
447 	return (__bt_relink(t, h) || __bt_free(t, h));
448 }
449 
450 /*
451  * __bt_dleaf --
452  *	Delete a single record from a leaf page.
453  *
454  * Parameters:
455  *	t:	tree
456  *    key:	referenced key
457  *	h:	page
458  *	idx:	index on page to delete
459  *
460  * Returns:
461  *	RET_SUCCESS, RET_ERROR.
462  */
463 int
464 __bt_dleaf(BTREE *t, const DBT *key, PAGE *h, u_int idx)
465 {
466 	BLEAF *bl;
467 	indx_t cnt, *ip, offset;
468 	u_int32_t nbytes;
469 	void *to;
470 	char *from;
471 
472 	/* If this record is referenced by the cursor, delete the cursor. */
473 	if (F_ISSET(&t->bt_cursor, CURS_INIT) &&
474 	    !F_ISSET(&t->bt_cursor, CURS_ACQUIRE) &&
475 	    t->bt_cursor.pg.pgno == h->pgno && t->bt_cursor.pg.index == idx &&
476 	    __bt_curdel(t, key, h, idx))
477 		return (RET_ERROR);
478 
479 	/* If the entry uses overflow pages, make them available for reuse. */
480 	to = bl = GETBLEAF(h, idx);
481 	if (bl->flags & P_BIGKEY && __ovfl_delete(t, bl->bytes) == RET_ERROR)
482 		return (RET_ERROR);
483 	if (bl->flags & P_BIGDATA &&
484 	    __ovfl_delete(t, bl->bytes + bl->ksize) == RET_ERROR)
485 		return (RET_ERROR);
486 
487 	/* Pack the remaining key/data items at the end of the page. */
488 	nbytes = NBLEAF(bl);
489 	from = (char *)h + h->upper;
490 	memmove(from + nbytes, from, (char *)to - from);
491 	h->upper += nbytes;
492 
493 	/* Adjust the indices' offsets, shift the indices down. */
494 	offset = h->linp[idx];
495 	for (cnt = idx, ip = &h->linp[0]; cnt--; ++ip)
496 		if (ip[0] < offset)
497 			ip[0] += nbytes;
498 	for (cnt = NEXTINDEX(h) - idx; --cnt; ++ip)
499 		ip[0] = ip[1] < offset ? ip[1] + nbytes : ip[1];
500 	h->lower -= sizeof(indx_t);
501 
502 	/* If the cursor is on this page, adjust it as necessary. */
503 	if (F_ISSET(&t->bt_cursor, CURS_INIT) &&
504 	    !F_ISSET(&t->bt_cursor, CURS_ACQUIRE) &&
505 	    t->bt_cursor.pg.pgno == h->pgno && t->bt_cursor.pg.index > idx)
506 		--t->bt_cursor.pg.index;
507 
508 	return (RET_SUCCESS);
509 }
510 
511 /*
512  * __bt_curdel --
513  *	Delete the cursor.
514  *
515  * Parameters:
516  *	t:	tree
517  *    key:	referenced key (or NULL)
518  *	h:	page
519  *    idx:	index on page to delete
520  *
521  * Returns:
522  *	RET_SUCCESS, RET_ERROR.
523  */
524 static int
525 __bt_curdel(BTREE *t, const DBT *key, PAGE *h, u_int idx)
526 {
527 	CURSOR *c;
528 	EPG e;
529 	PAGE *pg;
530 	int curcopy, status;
531 
532 	/*
533 	 * If there are duplicates, move forward or backward to one.
534 	 * Otherwise, copy the key into the cursor area.
535 	 */
536 	c = &t->bt_cursor;
537 	F_CLR(c, CURS_AFTER | CURS_BEFORE | CURS_ACQUIRE);
538 
539 	curcopy = 0;
540 	if (!F_ISSET(t, B_NODUPS)) {
541 		/*
542 		 * We're going to have to do comparisons.  If we weren't
543 		 * provided a copy of the key, i.e. the user is deleting
544 		 * the current cursor position, get one.
545 		 */
546 		if (key == NULL) {
547 			e.page = h;
548 			e.index = idx;
549 			if ((status = __bt_ret(t, &e,
550 			    &c->key, &c->key, NULL, NULL, 1)) != RET_SUCCESS)
551 				return (status);
552 			curcopy = 1;
553 			key = &c->key;
554 		}
555 		/* Check previous key, if not at the beginning of the page. */
556 		if (idx > 0) {
557 			e.page = h;
558 			e.index = idx - 1;
559 			if (__bt_cmp(t, key, &e) == 0) {
560 				F_SET(c, CURS_BEFORE);
561 				goto dup2;
562 			}
563 		}
564 		/* Check next key, if not at the end of the page. */
565 		if (idx < NEXTINDEX(h) - 1) {
566 			e.page = h;
567 			e.index = idx + 1;
568 			if (__bt_cmp(t, key, &e) == 0) {
569 				F_SET(c, CURS_AFTER);
570 				goto dup2;
571 			}
572 		}
573 		/* Check previous key if at the beginning of the page. */
574 		if (idx == 0 && h->prevpg != P_INVALID) {
575 			if ((pg = mpool_get(t->bt_mp, h->prevpg, 0)) == NULL)
576 				return (RET_ERROR);
577 			e.page = pg;
578 			e.index = NEXTINDEX(pg) - 1;
579 			if (__bt_cmp(t, key, &e) == 0) {
580 				F_SET(c, CURS_BEFORE);
581 				goto dup1;
582 			}
583 			mpool_put(t->bt_mp, pg, 0);
584 		}
585 		/* Check next key if at the end of the page. */
586 		if (idx == NEXTINDEX(h) - 1 && h->nextpg != P_INVALID) {
587 			if ((pg = mpool_get(t->bt_mp, h->nextpg, 0)) == NULL)
588 				return (RET_ERROR);
589 			e.page = pg;
590 			e.index = 0;
591 			if (__bt_cmp(t, key, &e) == 0) {
592 				F_SET(c, CURS_AFTER);
593 dup1:				mpool_put(t->bt_mp, pg, 0);
594 dup2:				c->pg.pgno = e.page->pgno;
595 				c->pg.index = e.index;
596 				return (RET_SUCCESS);
597 			}
598 			mpool_put(t->bt_mp, pg, 0);
599 		}
600 	}
601 	e.page = h;
602 	e.index = idx;
603 	if (curcopy || (status =
604 	    __bt_ret(t, &e, &c->key, &c->key, NULL, NULL, 1)) == RET_SUCCESS) {
605 		F_SET(c, CURS_ACQUIRE);
606 		return (RET_SUCCESS);
607 	}
608 	return (status);
609 }
610 
611 /*
612  * __bt_relink --
613  *	Link around a deleted page.
614  *
615  * Parameters:
616  *	t:	tree
617  *	h:	page to be deleted
618  */
619 static int
620 __bt_relink(BTREE *t, PAGE *h)
621 {
622 	PAGE *pg;
623 
624 	if (h->nextpg != P_INVALID) {
625 		if ((pg = mpool_get(t->bt_mp, h->nextpg, 0)) == NULL)
626 			return (RET_ERROR);
627 		pg->prevpg = h->prevpg;
628 		mpool_put(t->bt_mp, pg, MPOOL_DIRTY);
629 	}
630 	if (h->prevpg != P_INVALID) {
631 		if ((pg = mpool_get(t->bt_mp, h->prevpg, 0)) == NULL)
632 			return (RET_ERROR);
633 		pg->nextpg = h->nextpg;
634 		mpool_put(t->bt_mp, pg, MPOOL_DIRTY);
635 	}
636 	return (0);
637 }
638