xref: /original-bsd/lib/libc/db/hash/hash_func.c (revision e59fb703)
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	5.2 (Berkeley) 09/04/91";
13 #endif /* LIBC_SCCS and not lint */
14 
15 #include <sys/types.h>
16 #include <db.h>
17 #include "hash.h"
18 #include "page.h"
19 #include "extern.h"
20 
21 static int hash1 __P((u_char *, int));
22 static int hash2 __P((u_char *, int));
23 static int hash3 __P((u_char *, int));
24 static int hash4 __P((u_char *, int));
25 
26 /* Global default hash function */
27 int (*default_hash) __P((u_char *, int)) = hash4;
28 
29 /******************************* HASH FUNCTIONS **************************/
30 /*
31  * Assume that we've already split the bucket to which this key hashes,
32  * calculate that bucket, and check that in fact we did already split it.
33  *
34  * This came from ejb's hsearch.
35  */
36 
37 #define PRIME1		37
38 #define PRIME2		1048583
39 
40 static int
41 hash1(key, len)
42 	register u_char *key;
43 	register int len;
44 {
45 	register int h;
46 
47 	h = 0;
48 	/* Convert string to integer */
49 	while (len--)
50 		h = h * PRIME1 ^ (*key++ - ' ');
51 	h %= PRIME2;
52 	return (h);
53 }
54 
55 /*
56  * Phong's linear congruential hash
57  */
58 #define dcharhash(h, c)	((h) = 0x63c63cd9*(h) + 0x9c39c33d + (c))
59 
60 static int
61 hash2(key, len)
62 	register u_char *key;
63 	int len;
64 {
65 	register u_char *e, c;
66 	register int h;
67 
68 	e = key + len;
69 	for (h = 0; key != e;) {
70 		c = *key++;
71 		if (!c && key > e)
72 			break;
73 		dcharhash(h, c);
74 	}
75 	return (h);
76 }
77 
78 /*
79  * This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
80  * units.  On the first time through the loop we get the "leftover bytes"
81  * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
82  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
83  * this routine is heavily used enough, it's worth the ugly coding.
84  *
85  * OZ's original sdbm hash
86  */
87 static int
88 hash3(key, len)
89 	register u_char *key;
90 	register int len;
91 {
92 	register int n, loop;
93 
94 #define HASHC   n = *key++ + 65599 * n
95 
96 	n = 0;
97 	if (len > 0) {
98 		loop = (len + 8 - 1) >> 3;
99 
100 		switch (len & (8 - 1)) {
101 		case 0:
102 			do {	/* All fall throughs */
103 				HASHC;
104 		case 7:
105 				HASHC;
106 		case 6:
107 				HASHC;
108 		case 5:
109 				HASHC;
110 		case 4:
111 				HASHC;
112 		case 3:
113 				HASHC;
114 		case 2:
115 				HASHC;
116 		case 1:
117 				HASHC;
118 			} while (--loop);
119 		}
120 
121 	}
122 	return (n);
123 }
124 
125 /* Hash function from Chris Torek. */
126 static int
127 hash4(key, len)
128 	register u_char *key;
129 	register int len;
130 {
131 	register int h, loop;
132 
133 #define HASH4a   h = (h << 5) - h + *key++;
134 #define HASH4b   h = (h << 5) + h + *key++;
135 #define HASH4 HASH4b
136 
137 	h = 0;
138 	if (len > 0) {
139 		loop = (len + 8 - 1) >> 3;
140 
141 		switch (len & (8 - 1)) {
142 		case 0:
143 			do {	/* All fall throughs */
144 				HASH4;
145 		case 7:
146 				HASH4;
147 		case 6:
148 				HASH4;
149 		case 5:
150 				HASH4;
151 		case 4:
152 				HASH4;
153 		case 3:
154 				HASH4;
155 		case 2:
156 				HASH4;
157 		case 1:
158 				HASH4;
159 			} while (--loop);
160 		}
161 
162 	}
163 	return (h);
164 }
165