xref: /netbsd-src/lib/libc/db/hash/hash_func.c (revision 8b0f9554ff8762542c4defc4f70e1eb76fb508fa)
1 /*	$NetBSD: hash_func.c,v 1.11 2007/02/03 23:46:09 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. 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 #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 1.11 2007/02/03 23:46:09 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 #if 0
56 static u_int32_t hash1(const void *, size_t) __attribute__((__unused__));
57 static u_int32_t hash2(const void *, size_t) __attribute__((__unused__));
58 static u_int32_t hash3(const void *, size_t) __attribute__((__unused__));
59 #endif
60 static u_int32_t hash4(const void *, size_t) __attribute__((__unused__));
61 
62 /* Global default hash function */
63 u_int32_t (*__default_hash)(const void *, size_t) = hash4;
64 #if 0
65 /*
66  * HASH FUNCTIONS
67  *
68  * Assume that we've already split the bucket to which this key hashes,
69  * calculate that bucket, and check that in fact we did already split it.
70  *
71  * This came from ejb's hsearch.
72  */
73 
74 #define PRIME1		37
75 #define PRIME2		1048583
76 
77 static u_int32_t
78 hash1(const void *keyarg, size_t len)
79 {
80 	const u_char *key;
81 	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(const void *keyarg, size_t len)
97 {
98 	const u_char *e, *key;
99 	u_int32_t h;
100 	u_char c;
101 
102 	key = keyarg;
103 	e = key + len;
104 	for (h = 0; key != e;) {
105 		c = *key++;
106 		if (!c && key > e)
107 			break;
108 		dcharhash(h, c);
109 	}
110 	return (h);
111 }
112 
113 /*
114  * This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
115  * units.  On the first time through the loop we get the "leftover bytes"
116  * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
117  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
118  * this routine is heavily used enough, it's worth the ugly coding.
119  *
120  * OZ's original sdbm hash
121  */
122 static u_int32_t
123 hash3(const void *keyarg, size_t len)
124 {
125 	const u_char *key;
126 	size_t loop;
127 	u_int32_t h;
128 
129 #define HASHC   h = *key++ + 65599 * h
130 
131 	h = 0;
132 	key = keyarg;
133 	if (len > 0) {
134 		loop = (len + 8 - 1) >> 3;
135 
136 		switch (len & (8 - 1)) {
137 		case 0:
138 			do {
139 				HASHC;
140 				/* FALLTHROUGH */
141 		case 7:
142 				HASHC;
143 				/* FALLTHROUGH */
144 		case 6:
145 				HASHC;
146 				/* FALLTHROUGH */
147 		case 5:
148 				HASHC;
149 				/* FALLTHROUGH */
150 		case 4:
151 				HASHC;
152 				/* FALLTHROUGH */
153 		case 3:
154 				HASHC;
155 				/* FALLTHROUGH */
156 		case 2:
157 				HASHC;
158 				/* FALLTHROUGH */
159 		case 1:
160 				HASHC;
161 			} while (--loop);
162 		}
163 	}
164 	return (h);
165 }
166 #endif
167 
168 /* Hash function from Chris Torek. */
169 static u_int32_t
170 hash4(const void *keyarg, size_t len)
171 {
172 	const u_char *key;
173 	size_t loop;
174 	u_int32_t h;
175 
176 #define HASH4a   h = (h << 5) - h + *key++;
177 #define HASH4b   h = (h << 5) + h + *key++;
178 #define HASH4 HASH4b
179 
180 	h = 0;
181 	key = keyarg;
182 	if (len > 0) {
183 		loop = (len + 8 - 1) >> 3;
184 
185 		switch (len & (8 - 1)) {
186 		case 0:
187 			do {
188 				HASH4;
189 				/* FALLTHROUGH */
190 		case 7:
191 				HASH4;
192 				/* FALLTHROUGH */
193 		case 6:
194 				HASH4;
195 				/* FALLTHROUGH */
196 		case 5:
197 				HASH4;
198 				/* FALLTHROUGH */
199 		case 4:
200 				HASH4;
201 				/* FALLTHROUGH */
202 		case 3:
203 				HASH4;
204 				/* FALLTHROUGH */
205 		case 2:
206 				HASH4;
207 				/* FALLTHROUGH */
208 		case 1:
209 				HASH4;
210 			} while (--loop);
211 		}
212 	}
213 	return (h);
214 }
215