hash_func.c revision 1.1 1 1.1 cgd /*-
2 1.1 cgd * Copyright (c) 1990, 1993
3 1.1 cgd * The Regents of the University of California. All rights reserved.
4 1.1 cgd *
5 1.1 cgd * This code is derived from software contributed to Berkeley by
6 1.1 cgd * Margo Seltzer.
7 1.1 cgd *
8 1.1 cgd * Redistribution and use in source and binary forms, with or without
9 1.1 cgd * modification, are permitted provided that the following conditions
10 1.1 cgd * are met:
11 1.1 cgd * 1. Redistributions of source code must retain the above copyright
12 1.1 cgd * notice, this list of conditions and the following disclaimer.
13 1.1 cgd * 2. Redistributions in binary form must reproduce the above copyright
14 1.1 cgd * notice, this list of conditions and the following disclaimer in the
15 1.1 cgd * documentation and/or other materials provided with the distribution.
16 1.1 cgd * 3. All advertising materials mentioning features or use of this software
17 1.1 cgd * must display the following acknowledgement:
18 1.1 cgd * This product includes software developed by the University of
19 1.1 cgd * California, Berkeley and its contributors.
20 1.1 cgd * 4. Neither the name of the University nor the names of its contributors
21 1.1 cgd * may be used to endorse or promote products derived from this software
22 1.1 cgd * without specific prior written permission.
23 1.1 cgd *
24 1.1 cgd * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 1.1 cgd * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 1.1 cgd * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 1.1 cgd * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 1.1 cgd * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 1.1 cgd * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 1.1 cgd * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 1.1 cgd * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 1.1 cgd * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 1.1 cgd * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 1.1 cgd * SUCH DAMAGE.
35 1.1 cgd */
36 1.1 cgd
37 1.1 cgd #if defined(LIBC_SCCS) && !defined(lint)
38 1.1 cgd static char sccsid[] = "@(#)hash_func.c 8.1 (Berkeley) 6/4/93";
39 1.1 cgd #endif /* LIBC_SCCS and not lint */
40 1.1 cgd
41 1.1 cgd #include <sys/types.h>
42 1.1 cgd
43 1.1 cgd #include <db.h>
44 1.1 cgd #include "hash.h"
45 1.1 cgd #include "page.h"
46 1.1 cgd #include "extern.h"
47 1.1 cgd
48 1.1 cgd static int hash1 __P((u_char *, int));
49 1.1 cgd static int hash2 __P((u_char *, int));
50 1.1 cgd static int hash3 __P((u_char *, int));
51 1.1 cgd static int hash4 __P((u_char *, int));
52 1.1 cgd
53 1.1 cgd /* Global default hash function */
54 1.1 cgd int (*__default_hash) __P((u_char *, int)) = hash4;
55 1.1 cgd
56 1.1 cgd /******************************* HASH FUNCTIONS **************************/
57 1.1 cgd /*
58 1.1 cgd * Assume that we've already split the bucket to which this key hashes,
59 1.1 cgd * calculate that bucket, and check that in fact we did already split it.
60 1.1 cgd *
61 1.1 cgd * This came from ejb's hsearch.
62 1.1 cgd */
63 1.1 cgd
64 1.1 cgd #define PRIME1 37
65 1.1 cgd #define PRIME2 1048583
66 1.1 cgd
67 1.1 cgd static int
68 1.1 cgd hash1(key, len)
69 1.1 cgd register u_char *key;
70 1.1 cgd register int len;
71 1.1 cgd {
72 1.1 cgd register int h;
73 1.1 cgd
74 1.1 cgd h = 0;
75 1.1 cgd /* Convert string to integer */
76 1.1 cgd while (len--)
77 1.1 cgd h = h * PRIME1 ^ (*key++ - ' ');
78 1.1 cgd h %= PRIME2;
79 1.1 cgd return (h);
80 1.1 cgd }
81 1.1 cgd
82 1.1 cgd /*
83 1.1 cgd * Phong's linear congruential hash
84 1.1 cgd */
85 1.1 cgd #define dcharhash(h, c) ((h) = 0x63c63cd9*(h) + 0x9c39c33d + (c))
86 1.1 cgd
87 1.1 cgd static int
88 1.1 cgd hash2(key, len)
89 1.1 cgd register u_char *key;
90 1.1 cgd int len;
91 1.1 cgd {
92 1.1 cgd register u_char *e, c;
93 1.1 cgd register int h;
94 1.1 cgd
95 1.1 cgd e = key + len;
96 1.1 cgd for (h = 0; key != e;) {
97 1.1 cgd c = *key++;
98 1.1 cgd if (!c && key > e)
99 1.1 cgd break;
100 1.1 cgd dcharhash(h, c);
101 1.1 cgd }
102 1.1 cgd return (h);
103 1.1 cgd }
104 1.1 cgd
105 1.1 cgd /*
106 1.1 cgd * This is INCREDIBLY ugly, but fast. We break the string up into 8 byte
107 1.1 cgd * units. On the first time through the loop we get the "leftover bytes"
108 1.1 cgd * (strlen % 8). On every other iteration, we perform 8 HASHC's so we handle
109 1.1 cgd * all 8 bytes. Essentially, this saves us 7 cmp & branch instructions. If
110 1.1 cgd * this routine is heavily used enough, it's worth the ugly coding.
111 1.1 cgd *
112 1.1 cgd * OZ's original sdbm hash
113 1.1 cgd */
114 1.1 cgd static int
115 1.1 cgd hash3(key, len)
116 1.1 cgd register u_char *key;
117 1.1 cgd register int len;
118 1.1 cgd {
119 1.1 cgd register int n, loop;
120 1.1 cgd
121 1.1 cgd #define HASHC n = *key++ + 65599 * n
122 1.1 cgd
123 1.1 cgd n = 0;
124 1.1 cgd if (len > 0) {
125 1.1 cgd loop = (len + 8 - 1) >> 3;
126 1.1 cgd
127 1.1 cgd switch (len & (8 - 1)) {
128 1.1 cgd case 0:
129 1.1 cgd do { /* All fall throughs */
130 1.1 cgd HASHC;
131 1.1 cgd case 7:
132 1.1 cgd HASHC;
133 1.1 cgd case 6:
134 1.1 cgd HASHC;
135 1.1 cgd case 5:
136 1.1 cgd HASHC;
137 1.1 cgd case 4:
138 1.1 cgd HASHC;
139 1.1 cgd case 3:
140 1.1 cgd HASHC;
141 1.1 cgd case 2:
142 1.1 cgd HASHC;
143 1.1 cgd case 1:
144 1.1 cgd HASHC;
145 1.1 cgd } while (--loop);
146 1.1 cgd }
147 1.1 cgd
148 1.1 cgd }
149 1.1 cgd return (n);
150 1.1 cgd }
151 1.1 cgd
152 1.1 cgd /* Hash function from Chris Torek. */
153 1.1 cgd static int
154 1.1 cgd hash4(key, len)
155 1.1 cgd register u_char *key;
156 1.1 cgd register int len;
157 1.1 cgd {
158 1.1 cgd register int h, loop;
159 1.1 cgd
160 1.1 cgd #define HASH4a h = (h << 5) - h + *key++;
161 1.1 cgd #define HASH4b h = (h << 5) + h + *key++;
162 1.1 cgd #define HASH4 HASH4b
163 1.1 cgd
164 1.1 cgd h = 0;
165 1.1 cgd if (len > 0) {
166 1.1 cgd loop = (len + 8 - 1) >> 3;
167 1.1 cgd
168 1.1 cgd switch (len & (8 - 1)) {
169 1.1 cgd case 0:
170 1.1 cgd do { /* All fall throughs */
171 1.1 cgd HASH4;
172 1.1 cgd case 7:
173 1.1 cgd HASH4;
174 1.1 cgd case 6:
175 1.1 cgd HASH4;
176 1.1 cgd case 5:
177 1.1 cgd HASH4;
178 1.1 cgd case 4:
179 1.1 cgd HASH4;
180 1.1 cgd case 3:
181 1.1 cgd HASH4;
182 1.1 cgd case 2:
183 1.1 cgd HASH4;
184 1.1 cgd case 1:
185 1.1 cgd HASH4;
186 1.1 cgd } while (--loop);
187 1.1 cgd }
188 1.1 cgd
189 1.1 cgd }
190 1.1 cgd return (h);
191 1.1 cgd }
192