xref: /netbsd-src/lib/libc/db/hash/hash_func.c (revision 2a399c6883d870daece976daec6ffa7bb7f934ce)
1 /*	$NetBSD: hash_func.c,v 1.7 1997/07/13 18:52:05 christos 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 #include <sys/cdefs.h>
40 #if defined(LIBC_SCCS) && !defined(lint)
41 #if 0
42 static char sccsid[] = "@(#)hash_func.c	8.2 (Berkeley) 2/21/94";
43 #else
44 __RCSID("$NetBSD: hash_func.c,v 1.7 1997/07/13 18:52:05 christos Exp $");
45 #endif
46 #endif /* LIBC_SCCS and not lint */
47 
48 #include <sys/types.h>
49 
50 #include <db.h>
51 #include "hash.h"
52 #include "page.h"
53 #include "extern.h"
54 
55 static u_int32_t hash1 __P((const void *, size_t)) __attribute__((__unused__));
56 static u_int32_t hash2 __P((const void *, size_t)) __attribute__((__unused__));
57 static u_int32_t hash3 __P((const void *, size_t)) __attribute__((__unused__));
58 static u_int32_t hash4 __P((const void *, size_t)) __attribute__((__unused__));
59 
60 /* Global default hash function */
61 u_int32_t (*__default_hash) __P((const void *, size_t)) = hash4;
62 
63 /*
64  * HASH FUNCTIONS
65  *
66  * Assume that we've already split the bucket to which this key hashes,
67  * calculate that bucket, and check that in fact we did already split it.
68  *
69  * This came from ejb's hsearch.
70  */
71 
72 #define PRIME1		37
73 #define PRIME2		1048583
74 
75 static u_int32_t
76 hash1(keyarg, len)
77 	const void *keyarg;
78 	register size_t len;
79 {
80 	register const u_char *key;
81 	register u_int32_t h;
82 
83 	/* Convert string to integer */
84 	for (key = keyarg, h = 0; len--;)
85 		h = h * PRIME1 ^ (*key++ - ' ');
86 	h %= PRIME2;
87 	return (h);
88 }
89 
90 /*
91  * Phong's linear congruential hash
92  */
93 #define dcharhash(h, c)	((h) = 0x63c63cd9*(h) + 0x9c39c33d + (c))
94 
95 static u_int32_t
96 hash2(keyarg, len)
97 	const void *keyarg;
98 	size_t len;
99 {
100 	register const u_char *e, *key;
101 	register u_int32_t h;
102 	register u_char c;
103 
104 	key = keyarg;
105 	e = key + len;
106 	for (h = 0; key != e;) {
107 		c = *key++;
108 		if (!c && key > e)
109 			break;
110 		dcharhash(h, c);
111 	}
112 	return (h);
113 }
114 
115 /*
116  * This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
117  * units.  On the first time through the loop we get the "leftover bytes"
118  * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
119  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
120  * this routine is heavily used enough, it's worth the ugly coding.
121  *
122  * OZ's original sdbm hash
123  */
124 static u_int32_t
125 hash3(keyarg, len)
126 	const void *keyarg;
127 	register size_t len;
128 {
129 	register const u_char *key;
130 	register size_t loop;
131 	register u_int32_t h;
132 
133 #define HASHC   h = *key++ + 65599 * h
134 
135 	h = 0;
136 	key = keyarg;
137 	if (len > 0) {
138 		loop = (len + 8 - 1) >> 3;
139 
140 		switch (len & (8 - 1)) {
141 		case 0:
142 			do {
143 				HASHC;
144 				/* FALLTHROUGH */
145 		case 7:
146 				HASHC;
147 				/* FALLTHROUGH */
148 		case 6:
149 				HASHC;
150 				/* FALLTHROUGH */
151 		case 5:
152 				HASHC;
153 				/* FALLTHROUGH */
154 		case 4:
155 				HASHC;
156 				/* FALLTHROUGH */
157 		case 3:
158 				HASHC;
159 				/* FALLTHROUGH */
160 		case 2:
161 				HASHC;
162 				/* FALLTHROUGH */
163 		case 1:
164 				HASHC;
165 			} while (--loop);
166 		}
167 	}
168 	return (h);
169 }
170 
171 /* Hash function from Chris Torek. */
172 static u_int32_t
173 hash4(keyarg, len)
174 	const void *keyarg;
175 	register size_t len;
176 {
177 	register const u_char *key;
178 	register size_t loop;
179 	register u_int32_t h;
180 
181 #define HASH4a   h = (h << 5) - h + *key++;
182 #define HASH4b   h = (h << 5) + h + *key++;
183 #define HASH4 HASH4b
184 
185 	h = 0;
186 	key = keyarg;
187 	if (len > 0) {
188 		loop = (len + 8 - 1) >> 3;
189 
190 		switch (len & (8 - 1)) {
191 		case 0:
192 			do {
193 				HASH4;
194 				/* FALLTHROUGH */
195 		case 7:
196 				HASH4;
197 				/* FALLTHROUGH */
198 		case 6:
199 				HASH4;
200 				/* FALLTHROUGH */
201 		case 5:
202 				HASH4;
203 				/* FALLTHROUGH */
204 		case 4:
205 				HASH4;
206 				/* FALLTHROUGH */
207 		case 3:
208 				HASH4;
209 				/* FALLTHROUGH */
210 		case 2:
211 				HASH4;
212 				/* FALLTHROUGH */
213 		case 1:
214 				HASH4;
215 			} while (--loop);
216 		}
217 	}
218 	return (h);
219 }
220