xref: /csrg-svn/lib/libc/db/hash/hash_func.c (revision 57586)
1 /*-
2  * Copyright (c) 1990 The Regents of the University of California.
3  * All rights reserved.
4  *
5  * This code is derived from software contributed to Berkeley by
6  * Margo Seltzer.
7  *
8  * %sccs.include.redist.c%
9  */
10 
11 #if defined(LIBC_SCCS) && !defined(lint)
12 static char sccsid[] = "@(#)hash_func.c	5.3 (Berkeley) 01/17/93";
13 #endif /* LIBC_SCCS and not lint */
14 
15 #include <sys/types.h>
16 
17 #include <db.h>
18 
19 #include "hash.h"
20 #include "page.h"
21 #include "extern.h"
22 
23 static int hash1 __P((u_char *, int));
24 static int hash2 __P((u_char *, int));
25 static int hash3 __P((u_char *, int));
26 static int hash4 __P((u_char *, int));
27 
28 /* Global default hash function */
29 int (*__default_hash) __P((u_char *, int)) = hash4;
30 
31 /******************************* HASH FUNCTIONS **************************/
32 /*
33  * Assume that we've already split the bucket to which this key hashes,
34  * calculate that bucket, and check that in fact we did already split it.
35  *
36  * This came from ejb's hsearch.
37  */
38 
39 #define PRIME1		37
40 #define PRIME2		1048583
41 
42 static int
43 hash1(key, len)
44 	register u_char *key;
45 	register int len;
46 {
47 	register int h;
48 
49 	h = 0;
50 	/* Convert string to integer */
51 	while (len--)
52 		h = h * PRIME1 ^ (*key++ - ' ');
53 	h %= PRIME2;
54 	return (h);
55 }
56 
57 /*
58  * Phong's linear congruential hash
59  */
60 #define dcharhash(h, c)	((h) = 0x63c63cd9*(h) + 0x9c39c33d + (c))
61 
62 static int
63 hash2(key, len)
64 	register u_char *key;
65 	int len;
66 {
67 	register u_char *e, c;
68 	register int h;
69 
70 	e = key + len;
71 	for (h = 0; key != e;) {
72 		c = *key++;
73 		if (!c && key > e)
74 			break;
75 		dcharhash(h, c);
76 	}
77 	return (h);
78 }
79 
80 /*
81  * This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
82  * units.  On the first time through the loop we get the "leftover bytes"
83  * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
84  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
85  * this routine is heavily used enough, it's worth the ugly coding.
86  *
87  * OZ's original sdbm hash
88  */
89 static int
90 hash3(key, len)
91 	register u_char *key;
92 	register int len;
93 {
94 	register int n, loop;
95 
96 #define HASHC   n = *key++ + 65599 * n
97 
98 	n = 0;
99 	if (len > 0) {
100 		loop = (len + 8 - 1) >> 3;
101 
102 		switch (len & (8 - 1)) {
103 		case 0:
104 			do {	/* All fall throughs */
105 				HASHC;
106 		case 7:
107 				HASHC;
108 		case 6:
109 				HASHC;
110 		case 5:
111 				HASHC;
112 		case 4:
113 				HASHC;
114 		case 3:
115 				HASHC;
116 		case 2:
117 				HASHC;
118 		case 1:
119 				HASHC;
120 			} while (--loop);
121 		}
122 
123 	}
124 	return (n);
125 }
126 
127 /* Hash function from Chris Torek. */
128 static int
129 hash4(key, len)
130 	register u_char *key;
131 	register int len;
132 {
133 	register int h, loop;
134 
135 #define HASH4a   h = (h << 5) - h + *key++;
136 #define HASH4b   h = (h << 5) + h + *key++;
137 #define HASH4 HASH4b
138 
139 	h = 0;
140 	if (len > 0) {
141 		loop = (len + 8 - 1) >> 3;
142 
143 		switch (len & (8 - 1)) {
144 		case 0:
145 			do {	/* All fall throughs */
146 				HASH4;
147 		case 7:
148 				HASH4;
149 		case 6:
150 				HASH4;
151 		case 5:
152 				HASH4;
153 		case 4:
154 				HASH4;
155 		case 3:
156 				HASH4;
157 		case 2:
158 				HASH4;
159 		case 1:
160 				HASH4;
161 			} while (--loop);
162 		}
163 
164 	}
165 	return (h);
166 }
167