bt_overflow.c revision 1.12 1 /* $NetBSD: bt_overflow.c,v 1.12 2003/08/07 16:42:41 agc 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 * 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 #include <sys/cdefs.h>
36 #if defined(LIBC_SCCS) && !defined(lint)
37 #if 0
38 static char sccsid[] = "@(#)bt_overflow.c 8.5 (Berkeley) 7/16/94";
39 #else
40 __RCSID("$NetBSD: bt_overflow.c,v 1.12 2003/08/07 16:42:41 agc Exp $");
41 #endif
42 #endif /* LIBC_SCCS and not lint */
43
44 #include "namespace.h"
45 #include <sys/param.h>
46
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50
51 #include <db.h>
52 #include "btree.h"
53
54 /*
55 * Big key/data code.
56 *
57 * Big key and data entries are stored on linked lists of pages. The initial
58 * reference is byte string stored with the key or data and is the page number
59 * and size. The actual record is stored in a chain of pages linked by the
60 * nextpg field of the PAGE header.
61 *
62 * The first page of the chain has a special property. If the record is used
63 * by an internal page, it cannot be deleted and the P_PRESERVE bit will be set
64 * in the header.
65 *
66 * XXX
67 * A single DBT is written to each chain, so a lot of space on the last page
68 * is wasted. This is a fairly major bug for some data sets.
69 */
70
71 /*
72 * __OVFL_GET -- Get an overflow key/data item.
73 *
74 * Parameters:
75 * t: tree
76 * p: pointer to { pgno_t, u_int32_t }
77 * buf: storage address
78 * bufsz: storage size
79 *
80 * Returns:
81 * RET_ERROR, RET_SUCCESS
82 */
83 int
84 __ovfl_get(t, p, ssz, buf, bufsz)
85 BTREE *t;
86 void *p;
87 size_t *ssz;
88 void **buf;
89 size_t *bufsz;
90 {
91 PAGE *h;
92 pgno_t pg;
93 size_t nb, plen;
94 u_int32_t sz;
95
96 memmove(&pg, p, sizeof(pgno_t));
97 memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(u_int32_t));
98 *ssz = sz;
99
100 #ifdef DEBUG
101 if (pg == P_INVALID || sz == 0)
102 abort();
103 #endif
104 /* Make the buffer bigger as necessary. */
105 if (*bufsz < sz) {
106 *buf = (char *)(*buf == NULL ? malloc(sz) : realloc(*buf, sz));
107 if (*buf == NULL)
108 return (RET_ERROR);
109 *bufsz = sz;
110 }
111
112 /*
113 * Step through the linked list of pages, copying the data on each one
114 * into the buffer. Never copy more than the data's length.
115 */
116 plen = t->bt_psize - BTDATAOFF;
117 for (p = *buf;; p = (char *)p + nb, pg = h->nextpg) {
118 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
119 return (RET_ERROR);
120
121 nb = MIN(sz, plen);
122 memmove(p, (char *)(void *)h + BTDATAOFF, nb);
123 mpool_put(t->bt_mp, h, 0);
124
125 if ((sz -= nb) == 0)
126 break;
127 }
128 return (RET_SUCCESS);
129 }
130
131 /*
132 * __OVFL_PUT -- Store an overflow key/data item.
133 *
134 * Parameters:
135 * t: tree
136 * data: DBT to store
137 * pgno: storage page number
138 *
139 * Returns:
140 * RET_ERROR, RET_SUCCESS
141 */
142 int
143 __ovfl_put(t, dbt, pg)
144 BTREE *t;
145 const DBT *dbt;
146 pgno_t *pg;
147 {
148 PAGE *h, *last;
149 void *p;
150 pgno_t npg;
151 size_t nb, plen;
152 u_int32_t sz;
153
154 /*
155 * Allocate pages and copy the key/data record into them. Store the
156 * number of the first page in the chain.
157 */
158 plen = t->bt_psize - BTDATAOFF;
159 for (last = NULL, p = dbt->data, sz = dbt->size;;
160 p = (char *)p + plen, last = h) {
161 if ((h = __bt_new(t, &npg)) == NULL)
162 return (RET_ERROR);
163
164 h->pgno = npg;
165 h->nextpg = h->prevpg = P_INVALID;
166 h->flags = P_OVERFLOW;
167 h->lower = h->upper = 0;
168
169 nb = MIN(sz, plen);
170 memmove((char *)(void *)h + BTDATAOFF, p, nb);
171
172 if (last) {
173 last->nextpg = h->pgno;
174 mpool_put(t->bt_mp, last, MPOOL_DIRTY);
175 } else
176 *pg = h->pgno;
177
178 if ((sz -= nb) == 0) {
179 mpool_put(t->bt_mp, h, MPOOL_DIRTY);
180 break;
181 }
182 }
183 return (RET_SUCCESS);
184 }
185
186 /*
187 * __OVFL_DELETE -- Delete an overflow chain.
188 *
189 * Parameters:
190 * t: tree
191 * p: pointer to { pgno_t, u_int32_t }
192 *
193 * Returns:
194 * RET_ERROR, RET_SUCCESS
195 */
196 int
197 __ovfl_delete(t, p)
198 BTREE *t;
199 void *p;
200 {
201 PAGE *h;
202 pgno_t pg;
203 size_t plen;
204 u_int32_t sz;
205
206 memmove(&pg, p, sizeof(pgno_t));
207 memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(u_int32_t));
208
209 #ifdef DEBUG
210 if (pg == P_INVALID || sz == 0)
211 abort();
212 #endif
213 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
214 return (RET_ERROR);
215
216 /* Don't delete chains used by internal pages. */
217 if (h->flags & P_PRESERVE) {
218 mpool_put(t->bt_mp, h, 0);
219 return (RET_SUCCESS);
220 }
221
222 /* Step through the chain, calling the free routine for each page. */
223 for (plen = t->bt_psize - BTDATAOFF;; sz -= plen) {
224 pg = h->nextpg;
225 __bt_free(t, h);
226 if (sz <= plen)
227 break;
228 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
229 return (RET_ERROR);
230 }
231 return (RET_SUCCESS);
232 }
233