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 #ifndef __minix 41 __RCSID("$NetBSD: bt_overflow.c,v 1.16 2008/09/11 12:58:00 joerg Exp $"); 42 #endif 43 44 #ifndef __minix 45 #include "namespace.h" 46 #endif 47 #include <sys/param.h> 48 49 #include <assert.h> 50 #include <stdio.h> 51 #include <stdlib.h> 52 #include <string.h> 53 54 #include <db.h> 55 #include "btree.h" 56 57 #define MAX(a, b) ((a) > (b) ? (a) : (b)) 58 #define MIN(a, b) ((a) < (b) ? (a) : (b)) 59 60 /* 61 * Big key/data code. 62 * 63 * Big key and data entries are stored on linked lists of pages. The initial 64 * reference is byte string stored with the key or data and is the page number 65 * and size. The actual record is stored in a chain of pages linked by the 66 * nextpg field of the PAGE header. 67 * 68 * The first page of the chain has a special property. If the record is used 69 * by an internal page, it cannot be deleted and the P_PRESERVE bit will be set 70 * in the header. 71 * 72 * XXX 73 * A single DBT is written to each chain, so a lot of space on the last page 74 * is wasted. This is a fairly major bug for some data sets. 75 */ 76 77 /* 78 * __OVFL_GET -- Get an overflow key/data item. 79 * 80 * Parameters: 81 * t: tree 82 * p: pointer to { pgno_t, uint32_t } 83 * buf: storage address 84 * bufsz: storage size 85 * 86 * Returns: 87 * RET_ERROR, RET_SUCCESS 88 */ 89 int 90 __ovfl_get(BTREE *t, void *p, size_t *ssz, void **buf, size_t *bufsz) 91 { 92 PAGE *h; 93 pgno_t pg; 94 uint32_t sz, nb, plen; 95 size_t temp; 96 97 memmove(&pg, p, sizeof(pgno_t)); 98 memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(uint32_t)); 99 *ssz = sz; 100 101 #ifdef DEBUG 102 if (pg == P_INVALID || sz == 0) 103 abort(); 104 #endif 105 /* Make the buffer bigger as necessary. */ 106 if (*bufsz < sz) { 107 *buf = (char *)(*buf == NULL ? malloc(sz) : realloc(*buf, sz)); 108 if (*buf == NULL) 109 return (RET_ERROR); 110 *bufsz = sz; 111 } 112 113 /* 114 * Step through the linked list of pages, copying the data on each one 115 * into the buffer. Never copy more than the data's length. 116 */ 117 temp = t->bt_psize - BTDATAOFF; 118 _DBFIT(temp, uint32_t); 119 plen = (uint32_t)temp; 120 for (p = *buf;; p = (char *)p + nb, pg = h->nextpg) { 121 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL) 122 return (RET_ERROR); 123 124 nb = MIN(sz, plen); 125 memmove(p, (char *)(void *)h + BTDATAOFF, nb); 126 mpool_put(t->bt_mp, h, 0); 127 128 if ((sz -= nb) == 0) 129 break; 130 } 131 return (RET_SUCCESS); 132 } 133 134 /* 135 * __OVFL_PUT -- Store an overflow key/data item. 136 * 137 * Parameters: 138 * t: tree 139 * data: DBT to store 140 * pgno: storage page number 141 * 142 * Returns: 143 * RET_ERROR, RET_SUCCESS 144 */ 145 int 146 __ovfl_put(BTREE *t, const DBT *dbt, pgno_t *pg) 147 { 148 PAGE *h, *last; 149 void *p; 150 pgno_t npg; 151 uint32_t sz, nb, plen; 152 size_t temp; 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 temp = t->bt_psize - BTDATAOFF; 159 _DBFIT(temp, uint32_t); 160 plen = (uint32_t)temp; 161 last = NULL; 162 p = dbt->data; 163 temp = dbt->size; 164 _DBFIT(temp, uint32_t); 165 sz = temp; 166 for (;; p = (char *)p + plen, last = h) { 167 if ((h = __bt_new(t, &npg)) == NULL) 168 return (RET_ERROR); 169 170 h->pgno = npg; 171 h->nextpg = h->prevpg = P_INVALID; 172 h->flags = P_OVERFLOW; 173 h->lower = h->upper = 0; 174 175 nb = MIN(sz, plen); 176 (void)memmove((char *)(void *)h + BTDATAOFF, p, (size_t)nb); 177 178 if (last) { 179 last->nextpg = h->pgno; 180 mpool_put(t->bt_mp, last, MPOOL_DIRTY); 181 } else 182 *pg = h->pgno; 183 184 if ((sz -= nb) == 0) { 185 mpool_put(t->bt_mp, h, MPOOL_DIRTY); 186 break; 187 } 188 } 189 return (RET_SUCCESS); 190 } 191 192 /* 193 * __OVFL_DELETE -- Delete an overflow chain. 194 * 195 * Parameters: 196 * t: tree 197 * p: pointer to { pgno_t, uint32_t } 198 * 199 * Returns: 200 * RET_ERROR, RET_SUCCESS 201 */ 202 int 203 __ovfl_delete(BTREE *t, void *p) 204 { 205 PAGE *h; 206 pgno_t pg; 207 uint32_t sz, plen; 208 size_t temp; 209 210 (void)memmove(&pg, p, sizeof(pgno_t)); 211 (void)memmove(&sz, (char *)p + sizeof(pgno_t), sizeof(uint32_t)); 212 213 #ifdef DEBUG 214 if (pg == P_INVALID || sz == 0) 215 abort(); 216 #endif 217 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL) 218 return (RET_ERROR); 219 220 /* Don't delete chains used by internal pages. */ 221 if (h->flags & P_PRESERVE) { 222 mpool_put(t->bt_mp, h, 0); 223 return (RET_SUCCESS); 224 } 225 226 /* Step through the chain, calling the free routine for each page. */ 227 temp = t->bt_psize - BTDATAOFF; 228 _DBFIT(temp, uint32_t); 229 plen = (uint32_t)temp; 230 for (;; sz -= plen) { 231 pg = h->nextpg; 232 __bt_free(t, h); 233 if (sz <= plen) 234 break; 235 if ((h = mpool_get(t->bt_mp, pg, 0)) == NULL) 236 return (RET_ERROR); 237 } 238 return (RET_SUCCESS); 239 } 240