Home | History | Annotate | Line # | Download | only in hash
hash_func.c revision 1.5
      1 /*	$NetBSD: hash_func.c,v 1.5 1995/02/27 13:22:27 cgd Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 1990, 1993
      5  *	The Regents of the University of California.  All rights reserved.
      6  *
      7  * This code is derived from software contributed to Berkeley by
      8  * Margo Seltzer.
      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. All advertising materials mentioning features or use of this software
     19  *    must display the following acknowledgement:
     20  *	This product includes software developed by the University of
     21  *	California, Berkeley and its contributors.
     22  * 4. Neither the name of the University nor the names of its contributors
     23  *    may be used to endorse or promote products derived from this software
     24  *    without specific prior written permission.
     25  *
     26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     36  * SUCH DAMAGE.
     37  */
     38 
     39 #if defined(LIBC_SCCS) && !defined(lint)
     40 #if 0
     41 static char sccsid[] = "@(#)hash_func.c	8.2 (Berkeley) 2/21/94";
     42 #else
     43 static char rcsid[] = "$NetBSD: hash_func.c,v 1.5 1995/02/27 13:22:27 cgd Exp $";
     44 #endif
     45 #endif /* LIBC_SCCS and not lint */
     46 
     47 #include <sys/types.h>
     48 
     49 #include <db.h>
     50 #include "hash.h"
     51 #include "page.h"
     52 #include "extern.h"
     53 
     54 static u_int32_t hash1 __P((const void *, size_t));
     55 static u_int32_t hash2 __P((const void *, size_t));
     56 static u_int32_t hash3 __P((const void *, size_t));
     57 static u_int32_t hash4 __P((const void *, size_t));
     58 
     59 /* Global default hash function */
     60 u_int32_t (*__default_hash) __P((const void *, size_t)) = hash4;
     61 
     62 /*
     63  * HASH FUNCTIONS
     64  *
     65  * Assume that we've already split the bucket to which this key hashes,
     66  * calculate that bucket, and check that in fact we did already split it.
     67  *
     68  * This came from ejb's hsearch.
     69  */
     70 
     71 #define PRIME1		37
     72 #define PRIME2		1048583
     73 
     74 static u_int32_t
     75 hash1(keyarg, len)
     76 	const void *keyarg;
     77 	register size_t len;
     78 {
     79 	register const u_char *key;
     80 	register u_int32_t h;
     81 
     82 	/* Convert string to integer */
     83 	for (key = keyarg, h = 0; len--;)
     84 		h = h * PRIME1 ^ (*key++ - ' ');
     85 	h %= PRIME2;
     86 	return (h);
     87 }
     88 
     89 /*
     90  * Phong's linear congruential hash
     91  */
     92 #define dcharhash(h, c)	((h) = 0x63c63cd9*(h) + 0x9c39c33d + (c))
     93 
     94 static u_int32_t
     95 hash2(keyarg, len)
     96 	const void *keyarg;
     97 	size_t len;
     98 {
     99 	register const u_char *e, *key;
    100 	register u_int32_t h;
    101 	register u_char c;
    102 
    103 	key = keyarg;
    104 	e = key + len;
    105 	for (h = 0; key != e;) {
    106 		c = *key++;
    107 		if (!c && key > e)
    108 			break;
    109 		dcharhash(h, c);
    110 	}
    111 	return (h);
    112 }
    113 
    114 /*
    115  * This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
    116  * units.  On the first time through the loop we get the "leftover bytes"
    117  * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
    118  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
    119  * this routine is heavily used enough, it's worth the ugly coding.
    120  *
    121  * OZ's original sdbm hash
    122  */
    123 static u_int32_t
    124 hash3(keyarg, len)
    125 	const void *keyarg;
    126 	register size_t len;
    127 {
    128 	register const u_char *key;
    129 	register size_t loop;
    130 	register u_int32_t h;
    131 
    132 #define HASHC   h = *key++ + 65599 * h
    133 
    134 	h = 0;
    135 	key = keyarg;
    136 	if (len > 0) {
    137 		loop = (len + 8 - 1) >> 3;
    138 
    139 		switch (len & (8 - 1)) {
    140 		case 0:
    141 			do {
    142 				HASHC;
    143 				/* FALLTHROUGH */
    144 		case 7:
    145 				HASHC;
    146 				/* FALLTHROUGH */
    147 		case 6:
    148 				HASHC;
    149 				/* FALLTHROUGH */
    150 		case 5:
    151 				HASHC;
    152 				/* FALLTHROUGH */
    153 		case 4:
    154 				HASHC;
    155 				/* FALLTHROUGH */
    156 		case 3:
    157 				HASHC;
    158 				/* FALLTHROUGH */
    159 		case 2:
    160 				HASHC;
    161 				/* FALLTHROUGH */
    162 		case 1:
    163 				HASHC;
    164 			} while (--loop);
    165 		}
    166 	}
    167 	return (h);
    168 }
    169 
    170 /* Hash function from Chris Torek. */
    171 static u_int32_t
    172 hash4(keyarg, len)
    173 	const void *keyarg;
    174 	register size_t len;
    175 {
    176 	register const u_char *key;
    177 	register size_t loop;
    178 	register u_int32_t h;
    179 
    180 #define HASH4a   h = (h << 5) - h + *key++;
    181 #define HASH4b   h = (h << 5) + h + *key++;
    182 #define HASH4 HASH4b
    183 
    184 	h = 0;
    185 	key = keyarg;
    186 	if (len > 0) {
    187 		loop = (len + 8 - 1) >> 3;
    188 
    189 		switch (len & (8 - 1)) {
    190 		case 0:
    191 			do {
    192 				HASH4;
    193 				/* FALLTHROUGH */
    194 		case 7:
    195 				HASH4;
    196 				/* FALLTHROUGH */
    197 		case 6:
    198 				HASH4;
    199 				/* FALLTHROUGH */
    200 		case 5:
    201 				HASH4;
    202 				/* FALLTHROUGH */
    203 		case 4:
    204 				HASH4;
    205 				/* FALLTHROUGH */
    206 		case 3:
    207 				HASH4;
    208 				/* FALLTHROUGH */
    209 		case 2:
    210 				HASH4;
    211 				/* FALLTHROUGH */
    212 		case 1:
    213 				HASH4;
    214 			} while (--loop);
    215 		}
    216 	}
    217 	return (h);
    218 }
    219