1 /*- 2 * Copyright (c) 1990, 1993, 1994 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Margo Seltzer. 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. Neither the name of the University nor the names of its contributors 17 * may be used to endorse or promote products derived from this software 18 * without specific prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 30 * SUCH DAMAGE. 31 * 32 * @(#)hash.h 8.3 (Berkeley) 5/31/94 33 * $FreeBSD: src/lib/libc/db/hash/hash.h,v 1.6 2002/03/21 22:46:26 obrien Exp $ 34 */ 35 36 #include <sys/param.h> 37 #define __need_size_t 38 #include <stddef.h> 39 #include <stdint.h> 40 41 /* Check that newlib understands the byte order of its target system. */ 42 #ifndef _BYTE_ORDER 43 #error _BYTE_ORDER not defined by sys/param.h 44 #endif 45 46 /* Define DB endianness constants based on target endianness. */ 47 #define DB_LITTLE_ENDIAN 1234 48 #define DB_BIG_ENDIAN 4321 49 #if (_BYTE_ORDER == _LITTLE_ENDIAN) 50 #define DB_BYTE_ORDER DB_LITTLE_ENDIAN 51 #else 52 #define DB_BYTE_ORDER DB_BIG_ENDIAN 53 #endif 54 55 /* Operations */ 56 typedef enum { 57 HASH_GET, HASH_PUT, HASH_PUTNEW, HASH_DELETE, HASH_FIRST, HASH_NEXT 58 } HASH_ACTION; 59 60 /* Buffer Management structures */ 61 typedef struct _bufhead BUFHEAD; 62 63 struct _bufhead { 64 BUFHEAD *prev; /* LRU links */ 65 BUFHEAD *next; /* LRU links */ 66 BUFHEAD *ovfl; /* Overflow page buffer header */ 67 __uint32_t addr; /* Address of this page */ 68 char *page; /* Actual page data */ 69 char flags; 70 #define BUF_MOD 0x0001 71 #define BUF_DISK 0x0002 72 #define BUF_BUCKET 0x0004 73 #define BUF_PIN 0x0008 74 }; 75 76 #define IS_BUCKET(X) ((X) & BUF_BUCKET) 77 78 typedef BUFHEAD **SEGMENT; 79 80 #ifdef __GNUC__ 81 #pragma GCC diagnostic ignored "-Wpragmas" 82 #pragma GCC diagnostic ignored "-Wunknown-warning-option" 83 /* 'bsize' is used directly with malloc/realloc which confuses -fanalyzer */ 84 #pragma GCC diagnostic ignored "-Wanalyzer-allocation-size" 85 #endif 86 87 /* Hash Table Information */ 88 typedef struct hashhdr { /* Disk resident portion */ 89 int32_t magic; /* Magic NO for hash tables */ 90 int32_t version; /* Version ID */ 91 __uint32_t lorder; /* Byte Order */ 92 int32_t bsize; /* Bucket/Page Size */ 93 int32_t bshift; /* Bucket shift */ 94 int32_t dsize; /* Directory Size */ 95 int32_t ssize; /* Segment Size */ 96 int32_t sshift; /* Segment shift */ 97 int32_t ovfl_point; /* Where overflow pages are being 98 * allocated */ 99 int32_t last_freed; /* Last overflow page freed */ 100 int32_t max_bucket; /* ID of Maximum bucket in use */ 101 int32_t high_mask; /* Mask to modulo into entire table */ 102 int32_t low_mask; /* Mask to modulo into lower half of 103 * table */ 104 int32_t ffactor; /* Fill factor */ 105 int32_t nkeys; /* Number of keys in hash table */ 106 int32_t hdrpages; /* Size of table header */ 107 uint32_t h_charkey; /* value of hash(CHARKEY) */ 108 #define NCACHED 32 /* number of bit maps and spare 109 * points */ 110 int32_t spares[NCACHED];/* spare pages for overflow */ 111 __uint16_t bitmaps[NCACHED]; /* address of overflow page 112 * bitmaps */ 113 } HASHHDR; 114 115 typedef struct htab { /* Memory resident data structure */ 116 HASHHDR hdr; /* Header */ 117 int nsegs; /* Number of allocated segments */ 118 int exsegs; /* Number of extra allocated 119 * segments */ 120 __uint32_t /* Hash function */ 121 (*hash)(const void *, size_t); 122 int flags; /* Flag values */ 123 int fp; /* File pointer */ 124 char *tmp_buf; /* Temporary Buffer for BIG data */ 125 char *tmp_key; /* Temporary Buffer for BIG keys */ 126 BUFHEAD *cpage; /* Current page */ 127 int32_t cbucket; /* Current bucket */ 128 int cndx; /* Index of next item on cpage */ 129 int error; /* Error Number -- for DBM 130 * compatibility */ 131 int new_file; /* Indicates if fd is backing store 132 * or no */ 133 int save_file; /* Indicates whether we need to flush 134 * file at 135 * exit */ 136 __uint32_t *mapp[NCACHED]; /* Pointers to page maps */ 137 int nmaps; /* Initial number of bitmaps */ 138 int nbufs; /* Number of buffers left to 139 * allocate */ 140 BUFHEAD bufhead; /* Header of buffer lru list */ 141 SEGMENT *dir; /* Hash Bucket directory */ 142 } HTAB; 143 144 /* 145 * Constants 146 */ 147 #if INT_MAX == 32767 148 #define MAX_BSIZE 4096 149 #else 150 #define MAX_BSIZE 65536 /* 2^16 */ 151 #endif 152 #define MIN_BUFFERS 6 153 #define MINHDRSIZE 512 154 #if INT_MAX == 32767 155 #define DEF_BUFSIZE 4096 156 #else 157 #define DEF_BUFSIZE 65536 /* 64 K */ 158 #endif 159 #define DEF_BUCKET_SIZE 4096 160 #define DEF_BUCKET_SHIFT 12 /* log2(BUCKET) */ 161 #define DEF_SEGSIZE 256 162 #define DEF_SEGSIZE_SHIFT 8 /* log2(SEGSIZE) */ 163 #define DEF_DIRSIZE 256 164 #define DEF_FFACTOR 65536 165 #define MIN_FFACTOR 4 166 #define SPLTMAX 8 167 #define CHARKEY "%$sniglet^&" 168 #define NUMKEY 1038583 169 #define BYTE_SHIFT 3 170 #define INT_TO_BYTE 2 171 #define INT_BYTE_SHIFT 5 172 #define ALL_SET ((__uint32_t)0xFFFFFFFF) 173 #define ALL_CLEAR 0 174 175 #define PTROF(X) ((BUFHEAD *)((ptrdiff_t)(X)&~0x3)) 176 #define ISMOD(X) ((__uint32_t)(ptrdiff_t)(X)&0x1) 177 #define DOMOD(X) ((X) = (char *)((ptrdiff_t)(X)|0x1)) 178 #define ISDISK(X) ((__uint32_t)(ptrdiff_t)(X)&0x2) 179 #define DODISK(X) ((X) = (char *)((ptrdiff_t)(X)|0x2)) 180 181 #define BITS_PER_MAP 32 182 183 /* Given the address of the beginning of a big map, clear/set the nth bit */ 184 #define CLRBIT(A, N) ((A)[(N)/BITS_PER_MAP] &= ~(1<<((N)%BITS_PER_MAP))) 185 #define SETBIT(A, N) ((A)[(N)/BITS_PER_MAP] |= (1<<((N)%BITS_PER_MAP))) 186 #define ISSET(A, N) ((A)[(N)/BITS_PER_MAP] & (1<<((N)%BITS_PER_MAP))) 187 188 /* Overflow management */ 189 /* 190 * Overflow page numbers are allocated per split point. At each doubling of 191 * the table, we can allocate extra pages. So, an overflow page number has 192 * the top 5 bits indicate which split point and the lower 11 bits indicate 193 * which page at that split point is indicated (pages within split points are 194 * numberered starting with 1). 195 */ 196 197 #define SPLITSHIFT 11 198 #define SPLITMASK 0x7FF 199 #define SPLITNUM(N) (((__uint32_t)(N)) >> SPLITSHIFT) 200 #define OPAGENUM(N) ((N) & SPLITMASK) 201 #define OADDR_OF(S,O) ((__uint32_t)((__uint32_t)(S) << SPLITSHIFT) + (O)) 202 203 #define BUCKET_TO_PAGE(B) \ 204 (B) + hashp->HDRPAGES + ((B) ? hashp->SPARES[__log2((B)+1)-1] : 0) 205 #define OADDR_TO_PAGE(B) \ 206 BUCKET_TO_PAGE ( (1 << SPLITNUM((B))) -1 ) + OPAGENUM((B)); 207 208 /* 209 * page.h contains a detailed description of the page format. 210 * 211 * Normally, keys and data are accessed from offset tables in the top of 212 * each page which point to the beginning of the key and data. There are 213 * four flag values which may be stored in these offset tables which indicate 214 * the following: 215 * 216 * 217 * OVFLPAGE Rather than a key data pair, this pair contains 218 * the address of an overflow page. The format of 219 * the pair is: 220 * OVERFLOW_PAGE_NUMBER OVFLPAGE 221 * 222 * PARTIAL_KEY This must be the first key/data pair on a page 223 * and implies that page contains only a partial key. 224 * That is, the key is too big to fit on a single page 225 * so it starts on this page and continues on the next. 226 * The format of the page is: 227 * KEY_OFF PARTIAL_KEY OVFL_PAGENO OVFLPAGE 228 * 229 * KEY_OFF -- offset of the beginning of the key 230 * PARTIAL_KEY -- 1 231 * OVFL_PAGENO - page number of the next overflow page 232 * OVFLPAGE -- 0 233 * 234 * FULL_KEY This must be the first key/data pair on the page. It 235 * is used in two cases. 236 * 237 * Case 1: 238 * There is a complete key on the page but no data 239 * (because it wouldn't fit). The next page contains 240 * the data. 241 * 242 * Page format it: 243 * KEY_OFF FULL_KEY OVFL_PAGENO OVFL_PAGE 244 * 245 * KEY_OFF -- offset of the beginning of the key 246 * FULL_KEY -- 2 247 * OVFL_PAGENO - page number of the next overflow page 248 * OVFLPAGE -- 0 249 * 250 * Case 2: 251 * This page contains no key, but part of a large 252 * data field, which is continued on the next page. 253 * 254 * Page format it: 255 * DATA_OFF FULL_KEY OVFL_PAGENO OVFL_PAGE 256 * 257 * KEY_OFF -- offset of the beginning of the data on 258 * this page 259 * FULL_KEY -- 2 260 * OVFL_PAGENO - page number of the next overflow page 261 * OVFLPAGE -- 0 262 * 263 * FULL_KEY_DATA 264 * This must be the first key/data pair on the page. 265 * There are two cases: 266 * 267 * Case 1: 268 * This page contains a key and the beginning of the 269 * data field, but the data field is continued on the 270 * next page. 271 * 272 * Page format is: 273 * KEY_OFF FULL_KEY_DATA OVFL_PAGENO DATA_OFF 274 * 275 * KEY_OFF -- offset of the beginning of the key 276 * FULL_KEY_DATA -- 3 277 * OVFL_PAGENO - page number of the next overflow page 278 * DATA_OFF -- offset of the beginning of the data 279 * 280 * Case 2: 281 * This page contains the last page of a big data pair. 282 * There is no key, only the tail end of the data 283 * on this page. 284 * 285 * Page format is: 286 * DATA_OFF FULL_KEY_DATA <OVFL_PAGENO> <OVFLPAGE> 287 * 288 * DATA_OFF -- offset of the beginning of the data on 289 * this page 290 * FULL_KEY_DATA -- 3 291 * OVFL_PAGENO - page number of the next overflow page 292 * OVFLPAGE -- 0 293 * 294 * OVFL_PAGENO and OVFLPAGE are optional (they are 295 * not present if there is no next page). 296 */ 297 298 #define OVFLPAGE 0 299 #define PARTIAL_KEY 1 300 #define FULL_KEY 2 301 #define FULL_KEY_DATA 3 302 #define REAL_KEY 4 303 304 /* Short hands for accessing structure */ 305 #define BSIZE hdr.bsize 306 #define BSHIFT hdr.bshift 307 #define DSIZE hdr.dsize 308 #define SGSIZE hdr.ssize 309 #define SSHIFT hdr.sshift 310 #define LORDER hdr.lorder 311 #define OVFL_POINT hdr.ovfl_point 312 #define LAST_FREED hdr.last_freed 313 #define MAX_BUCKET hdr.max_bucket 314 #define FFACTOR hdr.ffactor 315 #define HIGH_MASK hdr.high_mask 316 #define LOW_MASK hdr.low_mask 317 #define NKEYS hdr.nkeys 318 #define HDRPAGES hdr.hdrpages 319 #define SPARES hdr.spares 320 #define BITMAPS hdr.bitmaps 321 #define HASH_VERSION hdr.version 322 #define MAGIC hdr.magic 323 #define NEXT_FREE hdr.next_free 324 #define H_CHARKEY hdr.h_charkey 325
