michael@0: /*-
michael@0:  * Copyright (c) 1990, 1993
michael@0:  *	The Regents of the University of California.  All rights reserved.
michael@0:  *
michael@0:  * This code is derived from software contributed to Berkeley by
michael@0:  * Margo Seltzer.
michael@0:  *
michael@0:  * Redistribution and use in source and binary forms, with or without
michael@0:  * modification, are permitted provided that the following conditions
michael@0:  * are met:
michael@0:  * 1. Redistributions of source code must retain the above copyright
michael@0:  *    notice, this list of conditions and the following disclaimer.
michael@0:  * 2. Redistributions in binary form must reproduce the above copyright
michael@0:  *    notice, this list of conditions and the following disclaimer in the
michael@0:  *    documentation and/or other materials provided with the distribution.
michael@0:  * 3. ***REMOVED*** - see 
michael@0:  *    ftp://ftp.cs.berkeley.edu/pub/4bsd/README.Impt.License.Change
michael@0:  * 4. Neither the name of the University nor the names of its contributors
michael@0:  *    may be used to endorse or promote products derived from this software
michael@0:  *    without specific prior written permission.
michael@0:  *
michael@0:  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
michael@0:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
michael@0:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
michael@0:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
michael@0:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
michael@0:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
michael@0:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
michael@0:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
michael@0:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
michael@0:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
michael@0:  * SUCH DAMAGE.
michael@0:  */
michael@0: 
michael@0: #if defined(LIBC_SCCS) && !defined(lint)
michael@0: static char sccsid[] = "@(#)hash_func.c	8.2 (Berkeley) 2/21/94";
michael@0: #endif /* LIBC_SCCS and not lint */
michael@0: 
michael@0: #ifndef macintosh
michael@0: #include <sys/types.h>
michael@0: #endif
michael@0: #include "mcom_db.h"
michael@0: #include "hash.h"
michael@0: #include "page.h"
michael@0: /* #include "extern.h" */
michael@0: 
michael@0: #if 0
michael@0: static uint32 hash1 __P((const void *, size_t));
michael@0: static uint32 hash2 __P((const void *, size_t));
michael@0: static uint32 hash3 __P((const void *, size_t));
michael@0: #endif
michael@0: static uint32 hash4 __P((const void *, size_t));
michael@0: 
michael@0: /* Global default hash function */
michael@0: uint32 (*__default_hash) __P((const void *, size_t)) = hash4;
michael@0: 
michael@0: /*
michael@0:  * HASH FUNCTIONS
michael@0:  *
michael@0:  * Assume that we've already split the bucket to which this key hashes,
michael@0:  * calculate that bucket, and check that in fact we did already split it.
michael@0:  *
michael@0:  * This came from ejb's hsearch.
michael@0:  */
michael@0: 
michael@0: #define PRIME1		37
michael@0: #define PRIME2		1048583
michael@0: 
michael@0: #if 0
michael@0: static uint32
michael@0: hash1(const void *keyarg, register size_t len)
michael@0: {
michael@0: 	register const uint8 *key;
michael@0: 	register uint32 h;
michael@0: 
michael@0: 	/* Convert string to integer */
michael@0: 	for (key = (const uint8 *)keyarg, h = 0; len--;)
michael@0: 		h = h * PRIME1 ^ (*key++ - ' ');
michael@0: 	h %= PRIME2;
michael@0: 	return (h);
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * Phong's linear congruential hash
michael@0:  */
michael@0: #define dcharhash(h, c)	((h) = 0x63c63cd9*(h) + 0x9c39c33d + (c))
michael@0: 
michael@0: static uint32
michael@0: hash2(const void *keyarg, size_t len)
michael@0: {
michael@0: 	register const uint8 *e, *key;
michael@0: 	register uint32 h;
michael@0: 	register uint8 c;
michael@0: 
michael@0: 	key = (const uint8 *)keyarg;
michael@0: 	e = key + len;
michael@0: 	for (h = 0; key != e;) {
michael@0: 		c = *key++;
michael@0: 		if (!c && key > e)
michael@0: 			break;
michael@0: 		dcharhash(h, c);
michael@0: 	}
michael@0: 	return (h);
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * This is INCREDIBLY ugly, but fast.  We break the string up into 8 byte
michael@0:  * units.  On the first time through the loop we get the "leftover bytes"
michael@0:  * (strlen % 8).  On every other iteration, we perform 8 HASHC's so we handle
michael@0:  * all 8 bytes.  Essentially, this saves us 7 cmp & branch instructions.  If
michael@0:  * this routine is heavily used enough, it's worth the ugly coding.
michael@0:  *
michael@0:  * OZ's original sdbm hash
michael@0:  */
michael@0: static uint32
michael@0: hash3(const void *keyarg, register size_t len)
michael@0: {
michael@0: 	register const uint8 *key;
michael@0: 	register size_t loop;
michael@0: 	register uint32 h;
michael@0: 
michael@0: #define HASHC   h = *key++ + 65599 * h
michael@0: 
michael@0: 	h = 0;
michael@0: 	key = (const uint8 *)keyarg;
michael@0: 	if (len > 0) {
michael@0: 		loop = (len + 8 - 1) >> 3;
michael@0: 
michael@0: 		switch (len & (8 - 1)) {
michael@0: 		case 0:
michael@0: 			do {
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 7:
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 6:
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 5:
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 4:
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 3:
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 2:
michael@0: 				HASHC;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 1:
michael@0: 				HASHC;
michael@0: 			} while (--loop);
michael@0: 		}
michael@0: 	}
michael@0: 	return (h);
michael@0: }
michael@0: #endif /* 0 */
michael@0: 
michael@0: /* Hash function from Chris Torek. */
michael@0: static uint32
michael@0: hash4(const void *keyarg, register size_t len)
michael@0: {
michael@0: 	register const uint8 *key;
michael@0: 	register size_t loop;
michael@0: 	register uint32 h;
michael@0: 
michael@0: #define HASH4a   h = (h << 5) - h + *key++;
michael@0: #define HASH4b   h = (h << 5) + h + *key++;
michael@0: #define HASH4 HASH4b
michael@0: 
michael@0: 	h = 0;
michael@0: 	key = (const uint8 *)keyarg;
michael@0: 	if (len > 0) {
michael@0: 		loop = (len + 8 - 1) >> 3;
michael@0: 
michael@0: 		switch (len & (8 - 1)) {
michael@0: 		case 0:
michael@0: 			do {
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 7:
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 6:
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 5:
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 4:
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 3:
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 2:
michael@0: 				HASH4;
michael@0: 				/* FALLTHROUGH */
michael@0: 		case 1:
michael@0: 				HASH4;
michael@0: 			} while (--loop);
michael@0: 		}
michael@0: 	}
michael@0: 	return (h);
michael@0: }