Home | History | Annotate | Line # | Download | only in btree
bt_delete.c revision 1.4
      1 /*-
      2  * Copyright (c) 1990, 1993
      3  *	The Regents of the University of California.  All rights reserved.
      4  *
      5  * This code is derived from software contributed to Berkeley by
      6  * Mike Olson.
      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. All advertising materials mentioning features or use of this software
     17  *    must display the following acknowledgement:
     18  *	This product includes software developed by the University of
     19  *	California, Berkeley and its contributors.
     20  * 4. Neither the name of the University nor the names of its contributors
     21  *    may be used to endorse or promote products derived from this software
     22  *    without specific prior written permission.
     23  *
     24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     34  * SUCH DAMAGE.
     35  */
     36 
     37 #if defined(LIBC_SCCS) && !defined(lint)
     38 /* from: static char sccsid[] = "@(#)bt_delete.c	8.2 (Berkeley) 9/7/93"; */
     39 static char *rcsid = "$Id: bt_delete.c,v 1.4 1993/09/09 02:41:22 cgd Exp $";
     40 #endif /* LIBC_SCCS and not lint */
     41 
     42 #include <sys/types.h>
     43 
     44 #include <errno.h>
     45 #include <stdio.h>
     46 #include <string.h>
     47 
     48 #include <db.h>
     49 #include "btree.h"
     50 
     51 static int bt_bdelete __P((BTREE *, const DBT *));
     52 
     53 /*
     54  * __BT_DELETE -- Delete the item(s) referenced by a key.
     55  *
     56  * Parameters:
     57  *	dbp:	pointer to access method
     58  *	key:	key to delete
     59  *	flags:	R_CURSOR if deleting what the cursor references
     60  *
     61  * Returns:
     62  *	RET_ERROR, RET_SUCCESS and RET_SPECIAL if the key not found.
     63  */
     64 int
     65 __bt_delete(dbp, key, flags)
     66 	const DB *dbp;
     67 	const DBT *key;
     68 	u_int flags;
     69 {
     70 	BTREE *t;
     71 	int status;
     72 
     73 	t = dbp->internal;
     74 
     75 	/* Toss any page pinned across calls. */
     76 	if (t->bt_pinned != NULL) {
     77 		mpool_put(t->bt_mp, t->bt_pinned, 0);
     78 		t->bt_pinned = NULL;
     79 	}
     80 
     81 	if (ISSET(t, B_RDONLY)) {
     82 		errno = EPERM;
     83 		return (RET_ERROR);
     84 	}
     85 
     86 	switch(flags) {
     87 	case 0:
     88 		status = bt_bdelete(t, key);
     89 		break;
     90 	case R_CURSOR:
     91 		/*
     92 		 * If flags is R_CURSOR, delete the cursor; must already have
     93 		 * started a scan and not have already deleted the record.  For
     94 		 * the delete cursor bit to have been set requires that the
     95 		 * scan be initialized, so no reason to check.
     96 		 */
     97 		if (!ISSET(t, B_SEQINIT))
     98                         goto einval;
     99 		status = ISSET(t, B_DELCRSR) ?
    100 		    RET_SPECIAL : __bt_crsrdel(t, &t->bt_bcursor);
    101 		break;
    102 	default:
    103 einval:		errno = EINVAL;
    104 		return (RET_ERROR);
    105 	}
    106 	if (status == RET_SUCCESS)
    107 		SET(t, B_MODIFIED);
    108 	return (status);
    109 }
    110 
    111 /*
    112  * BT_BDELETE -- Delete all key/data pairs matching the specified key.
    113  *
    114  * Parameters:
    115  *	tree:	tree
    116  *	key:	key to delete
    117  *
    118  * Returns:
    119  *	RET_ERROR, RET_SUCCESS and RET_SPECIAL if the key not found.
    120  */
    121 static int
    122 bt_bdelete(t, key)
    123 	BTREE *t;
    124 	const DBT *key;
    125 {
    126 	EPG *e, save;
    127 	PAGE *h;
    128 	pgno_t cpgno, pg;
    129 	indx_t cindex;
    130 	int deleted, dirty1, dirty2, exact;
    131 
    132 	/* Find any matching record; __bt_search pins the page. */
    133 	if ((e = __bt_search(t, key, &exact)) == NULL)
    134 		return (RET_ERROR);
    135 	if (!exact) {
    136 		mpool_put(t->bt_mp, e->page, 0);
    137 		return (RET_SPECIAL);
    138 	}
    139 
    140 	/*
    141 	 * Delete forward, then delete backward, from the found key.  The
    142 	 * ordering is so that the deletions don't mess up the page refs.
    143 	 * The first loop deletes the key from the original page, the second
    144 	 * unpins the original page.  In the first loop, dirty1 is set if
    145 	 * the original page is modified, and dirty2 is set if any subsequent
    146 	 * pages are modified.  In the second loop, dirty1 starts off set if
    147 	 * the original page has been modified, and is set if any subsequent
    148 	 * pages are modified.
    149 	 *
    150 	 * If find the key referenced by the cursor, don't delete it, just
    151 	 * flag it for future deletion.  The cursor page number is P_INVALID
    152 	 * unless the sequential scan is initialized, so no reason to check.
    153 	 * A special case is when the already deleted cursor record was the
    154 	 * only record found.  If so, then the delete opertion fails as no
    155 	 * records were deleted.
    156 	 *
    157 	 * Cycle in place in the current page until the current record doesn't
    158 	 * match the key or the page is empty.  If the latter, walk forward,
    159 	 * skipping empty pages and repeating until a record doesn't match
    160 	 * the key or the end of the tree is reached.
    161 	 */
    162 	cpgno = t->bt_bcursor.pgno;
    163 	cindex = t->bt_bcursor.index;
    164 	save = *e;
    165 	dirty1 = 0;
    166 	for (h = e->page, deleted = 0;;) {
    167 		dirty2 = 0;
    168 		do {
    169 			if (h->pgno == cpgno && e->index == cindex) {
    170 				if (!ISSET(t, B_DELCRSR)) {
    171 					SET(t, B_DELCRSR);
    172 					deleted = 1;
    173 				}
    174 				++e->index;
    175 			} else {
    176 				if (__bt_dleaf(t, h, e->index)) {
    177 					if (h->pgno != save.page->pgno)
    178 						mpool_put(t->bt_mp, h, dirty2);
    179 					mpool_put(t->bt_mp, save.page, dirty1);
    180 					return (RET_ERROR);
    181 				}
    182 				if (h->pgno == save.page->pgno)
    183 					dirty1 = MPOOL_DIRTY;
    184 				else
    185 					dirty2 = MPOOL_DIRTY;
    186 				deleted = 1;
    187 			}
    188 		} while (e->index < NEXTINDEX(h) && __bt_cmp(t, key, e) == 0);
    189 
    190 		/*
    191 		 * Quit if didn't find a match, no next page, or first key on
    192 		 * the next page doesn't match.  Don't unpin the original page
    193 		 * unless an error occurs.
    194 		 */
    195 		if (e->index < NEXTINDEX(h))
    196 			break;
    197 		for (;;) {
    198 			if ((pg = h->nextpg) == P_INVALID)
    199 				goto done1;
    200 			if (h->pgno != save.page->pgno)
    201 				mpool_put(t->bt_mp, h, dirty2);
    202 			if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL) {
    203 				mpool_put(t->bt_mp, save.page, dirty1);
    204 				return (RET_ERROR);
    205 			}
    206 			if (NEXTINDEX(h) != 0) {
    207 				e->page = h;
    208 				e->index = 0;
    209 				break;
    210 			}
    211 		}
    212 
    213 		if (__bt_cmp(t, key, e) != 0)
    214 			break;
    215 	}
    216 
    217 	/*
    218 	 * Reach here with the original page and the last page referenced
    219 	 * pinned (they may be the same).  Release it if not the original.
    220 	 */
    221 done1:	if (h->pgno != save.page->pgno)
    222 		mpool_put(t->bt_mp, h, dirty2);
    223 
    224 	/*
    225 	 * Walk backwards from the record previous to the record returned by
    226 	 * __bt_search, skipping empty pages, until a record doesn't match
    227 	 * the key or reach the beginning of the tree.
    228 	 */
    229 	*e = save;
    230 	for (;;) {
    231 		if (e->index)
    232 			--e->index;
    233 		for (h = e->page; e->index; --e->index) {
    234 			if (__bt_cmp(t, key, e) != 0)
    235 				goto done2;
    236 			if (h->pgno == cpgno && e->index == cindex) {
    237 				if (!ISSET(t, B_DELCRSR)) {
    238 					SET(t, B_DELCRSR);
    239 					deleted = 1;
    240 				}
    241 			} else {
    242 				if (__bt_dleaf(t, h, e->index) == RET_ERROR) {
    243 					mpool_put(t->bt_mp, h, dirty1);
    244 					return (RET_ERROR);
    245 				}
    246 				if (h->pgno == save.page->pgno)
    247 					dirty1 = MPOOL_DIRTY;
    248 				deleted = 1;
    249 			}
    250 		}
    251 
    252 		if ((pg = h->prevpg) == P_INVALID)
    253 			goto done2;
    254 		mpool_put(t->bt_mp, h, dirty1);
    255 		dirty1 = 0;
    256 		if ((e->page = mpool_get(t->bt_mp, pg, 0)) == NULL)
    257 			return (RET_ERROR);
    258 		e->index = NEXTINDEX(e->page);
    259 	}
    260 
    261 	/*
    262 	 * Reach here with the last page that was looked at pinned.  Release
    263 	 * it.
    264 	 */
    265 done2:	mpool_put(t->bt_mp, h, dirty1);
    266 	return (deleted ? RET_SUCCESS : RET_SPECIAL);
    267 }
    268 
    269 /*
    270  * __BT_DLEAF -- Delete a single record from a leaf page.
    271  *
    272  * Parameters:
    273  *	t:	tree
    274  *	index:	index on current page to delete
    275  *
    276  * Returns:
    277  *	RET_SUCCESS, RET_ERROR.
    278  */
    279 int
    280 __bt_dleaf(t, h, index)
    281 	BTREE *t;
    282 	PAGE *h;
    283 	int index;
    284 {
    285 	register BLEAF *bl;
    286 	register indx_t *ip, offset;
    287 	register size_t nbytes;
    288 	register int cnt;
    289 	char *from;
    290 	void *to;
    291 
    292 	/*
    293 	 * Delete a record from a btree leaf page.  Internal records are never
    294 	 * deleted from internal pages, regardless of the records that caused
    295 	 * them to be added being deleted.  Pages made empty by deletion are
    296 	 * not reclaimed.  They are, however, made available for reuse.
    297 	 *
    298 	 * Pack the remaining entries at the end of the page, shift the indices
    299 	 * down, overwriting the deleted record and its index.  If the record
    300 	 * uses overflow pages, make them available for reuse.
    301 	 */
    302 	to = bl = GETBLEAF(h, index);
    303 	if (bl->flags & P_BIGKEY && __ovfl_delete(t, bl->bytes) == RET_ERROR)
    304 		return (RET_ERROR);
    305 	if (bl->flags & P_BIGDATA &&
    306 	    __ovfl_delete(t, bl->bytes + bl->ksize) == RET_ERROR)
    307 		return (RET_ERROR);
    308 	nbytes = NBLEAF(bl);
    309 
    310 	/*
    311 	 * Compress the key/data pairs.  Compress and adjust the [BR]LEAF
    312 	 * offsets.  Reset the headers.
    313 	 */
    314 	from = (char *)h + h->upper;
    315 	memmove(from + nbytes, from, (char *)to - from);
    316 	h->upper += nbytes;
    317 
    318 	offset = h->linp[index];
    319 	for (cnt = index, ip = &h->linp[0]; cnt--; ++ip)
    320 		if (ip[0] < offset)
    321 			ip[0] += nbytes;
    322 	for (cnt = NEXTINDEX(h) - index; --cnt; ++ip)
    323 		ip[0] = ip[1] < offset ? ip[1] + nbytes : ip[1];
    324 	h->lower -= sizeof(indx_t);
    325 	return (RET_SUCCESS);
    326 }
    327