1 /* $NetBSD: bt_overflow.c,v 1.22 2016/09/24 21:31:25 christos 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 #if HAVE_NBTOOL_CONFIG_H
36 #include "nbtool_config.h"
37 #endif
38
39 #include <sys/cdefs.h>
40 __RCSID("$NetBSD: bt_overflow.c,v 1.22 2016/09/24 21:31:25 christos Exp $");
41
42 #include "namespace.h"
43 #include <sys/param.h>
44
45 #include <assert.h>
46 #include <stdio.h>
47 #include <stdlib.h>
48 #include <string.h>
49
50 #include <db.h>
51 #include "btree.h"
52
53 /*
54 * Big key/data code.
55 *
56 * Big key and data entries are stored on linked lists of pages. The initial
57 * reference is byte string stored with the key or data and is the page number
58 * and size. The actual record is stored in a chain of pages linked by the
59 * nextpg field of the PAGE header.
60 *
61 * The first page of the chain has a special property. If the record is used
62 * by an internal page, it cannot be deleted and the P_PRESERVE bit will be set
63 * in the header.
64 *
65 * XXX
66 * A single DBT is written to each chain, so a lot of space on the last page
67 * is wasted. This is a fairly major bug for some data sets.
68 */
69
70 /*
71 * __OVFL_GET -- Get an overflow key/data item.
72 *
73 * Parameters:
74 * t: tree
75 * p: pointer to { pgno_t, uint32_t }
76 * buf: storage address
77 * bufsz: storage size
78 *
79 * Returns:
80 * RET_ERROR, RET_SUCCESS
81 */
82 int
__ovfl_get(BTREE * t,void * p,size_t * ssz,void ** buf,size_t * bufsz)83 __ovfl_get(BTREE *t, void *p, size_t *ssz, void **buf, size_t *bufsz)
84 {
85 PAGE *h;
86 pgno_t pg;
87 uint32_t sz, nb, plen;
88 size_t temp;
89
90 memmove(&pg, p, sizeof(pg));
91 memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(uint32_t));
92 *ssz = sz;
93
94 #ifdef DEBUG
95 if (pg == P_INVALID || sz == 0)
96 abort();
97 #endif
98 /* Make the buffer bigger as necessary. */
99 if (*bufsz < sz) {
100 void *nbuf = realloc(*buf, sz);
101 if (nbuf == NULL)
102 return (RET_ERROR);
103 *buf = nbuf;
104 *bufsz = sz;
105 }
106
107 /*
108 * Step through the linked list of pages, copying the data on each one
109 * into the buffer. Never copy more than the data's length.
110 */
111 temp = t->bt_psize - BTDATAOFF;
112 _DBFIT(temp, uint32_t);
113 plen = (uint32_t)temp;
114 for (p = *buf;; p = (char *)p + nb, pg = h->nextpg) {
115 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
116 return (RET_ERROR);
117
118 nb = MIN(sz, plen);
119 memmove(p, (char *)(void *)h + BTDATAOFF, nb);
120 mpool_put(t->bt_mp, h, 0);
121
122 if ((sz -= nb) == 0)
123 break;
124 }
125 return (RET_SUCCESS);
126 }
127
128 /*
129 * __OVFL_PUT -- Store an overflow key/data item.
130 *
131 * Parameters:
132 * t: tree
133 * data: DBT to store
134 * pgno: storage page number
135 *
136 * Returns:
137 * RET_ERROR, RET_SUCCESS
138 */
139 int
__ovfl_put(BTREE * t,const DBT * dbt,pgno_t * pg)140 __ovfl_put(BTREE *t, const DBT *dbt, pgno_t *pg)
141 {
142 PAGE *h, *last;
143 void *p;
144 pgno_t npg;
145 uint32_t sz, nb, plen;
146 size_t temp;
147
148 /*
149 * Allocate pages and copy the key/data record into them. Store the
150 * number of the first page in the chain.
151 */
152 temp = t->bt_psize - BTDATAOFF;
153 _DBFIT(temp, uint32_t);
154 plen = (uint32_t)temp;
155 last = NULL;
156 p = dbt->data;
157 temp = dbt->size;
158 _DBFIT(temp, uint32_t);
159 sz = (uint32_t)temp;
160 for (;; 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 (void)memmove((char *)(void *)h + BTDATAOFF, p, (size_t)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, uint32_t }
192 *
193 * Returns:
194 * RET_ERROR, RET_SUCCESS
195 */
196 int
__ovfl_delete(BTREE * t,void * p)197 __ovfl_delete(BTREE *t, void *p)
198 {
199 PAGE *h;
200 pgno_t pg;
201 uint32_t sz, plen;
202 size_t temp;
203
204 (void)memmove(&pg, p, sizeof(pg));
205 (void)memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(uint32_t));
206
207 #ifdef DEBUG
208 if (pg == P_INVALID || sz == 0)
209 abort();
210 #endif
211 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
212 return (RET_ERROR);
213
214 /* Don't delete chains used by internal pages. */
215 if (h->flags & P_PRESERVE) {
216 mpool_put(t->bt_mp, h, 0);
217 return (RET_SUCCESS);
218 }
219
220 /* Step through the chain, calling the free routine for each page. */
221 temp = t->bt_psize - BTDATAOFF;
222 _DBFIT(temp, uint32_t);
223 plen = (uint32_t)temp;
224 for (;; sz -= plen) {
225 pg = h->nextpg;
226 __bt_free(t, h);
227 if (sz <= plen)
228 break;
229 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL)
230 return (RET_ERROR);
231 }
232 return (RET_SUCCESS);
233 }
234