1/* $NetBSD: bt_overflow.c,v 1.16 2008/09/11 12:58:00 joerg 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 1.16 2008/09/11 12:58:00 joerg 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 */ 82int 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 *buf = (*buf == NULL ? malloc(sz) : realloc(*buf, sz)); 101 if (*buf == NULL) 102 return (RET_ERROR); 103 *bufsz = sz; 104 } 105 106 /* 107 * Step through the linked list of pages, copying the data on each one 108 * into the buffer. Never copy more than the data's length. 109 */ 110 temp = t->bt_psize - BTDATAOFF; 111 _DBFIT(temp, uint32_t); 112 plen = (uint32_t)temp; 113 for (p = *buf;; p = (char *)p + nb, pg = h->nextpg) { 114 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL) 115 return (RET_ERROR); 116 117 nb = MIN(sz, plen); 118 memmove(p, (char *)(void *)h + BTDATAOFF, nb); 119 mpool_put(t->bt_mp, h, 0); 120 121 if ((sz -= nb) == 0) 122 break; 123 } 124 return (RET_SUCCESS); 125} 126 127/* 128 * __OVFL_PUT -- Store an overflow key/data item. 129 * 130 * Parameters: 131 * t: tree 132 * data: DBT to store 133 * pgno: storage page number 134 * 135 * Returns: 136 * RET_ERROR, RET_SUCCESS 137 */ 138int 139__ovfl_put(BTREE *t, const DBT *dbt, pgno_t *pg) 140{ 141 PAGE *h, *last; 142 void *p; 143 pgno_t npg; 144 uint32_t sz, nb, plen; 145 size_t temp; 146 147 /* 148 * Allocate pages and copy the key/data record into them. Store the 149 * number of the first page in the chain. 150 */ 151 temp = t->bt_psize - BTDATAOFF; 152 _DBFIT(temp, uint32_t); 153 plen = (uint32_t)temp; 154 last = NULL; 155 p = dbt->data; 156 temp = dbt->size; 157 _DBFIT(temp, uint32_t); 158 sz = temp; 159 for (;; p = (char *)p + plen, last = h) { 160 if ((h = __bt_new(t, &npg)) == NULL) 161 return (RET_ERROR); 162 163 h->pgno = npg; 164 h->nextpg = h->prevpg = P_INVALID; 165 h->flags = P_OVERFLOW; 166 h->lower = h->upper = 0; 167 168 nb = MIN(sz, plen); 169 (void)memmove((char *)(void *)h + BTDATAOFF, p, (size_t)nb); 170 171 if (last) { 172 last->nextpg = h->pgno; 173 mpool_put(t->bt_mp, last, MPOOL_DIRTY); 174 } else 175 *pg = h->pgno; 176 177 if ((sz -= nb) == 0) { 178 mpool_put(t->bt_mp, h, MPOOL_DIRTY); 179 break; 180 } 181 } 182 return (RET_SUCCESS); 183} 184 185/* 186 * __OVFL_DELETE -- Delete an overflow chain. 187 * 188 * Parameters: 189 * t: tree 190 * p: pointer to { pgno_t, uint32_t } 191 * 192 * Returns: 193 * RET_ERROR, RET_SUCCESS 194 */ 195int 196__ovfl_delete(BTREE *t, void *p) 197{ 198 PAGE *h; 199 pgno_t pg; 200 uint32_t sz, plen; 201 size_t temp; 202 203 (void)memmove(&pg, p, sizeof(pg)); 204 (void)memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(uint32_t)); 205 206#ifdef DEBUG 207 if (pg == P_INVALID || sz == 0) 208 abort(); 209#endif 210 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL) 211 return (RET_ERROR); 212 213 /* Don't delete chains used by internal pages. */ 214 if (h->flags & P_PRESERVE) { 215 mpool_put(t->bt_mp, h, 0); 216 return (RET_SUCCESS); 217 } 218 219 /* Step through the chain, calling the free routine for each page. */ 220 temp = t->bt_psize - BTDATAOFF; 221 _DBFIT(temp, uint32_t); 222 plen = (uint32_t)temp; 223 for (;; 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