1/*-
2 * Copyright (c) 2009, 2010 Xin LI <delphij@FreeBSD.org>
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 *
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24 * SUCH DAMAGE.
25 */
26
27#include <sys/cdefs.h>
28__FBSDID("$FreeBSD$");
29
30#include <sys/limits.h>
31#include <sys/types.h>
32#include <string.h>
33
34/*
35 * Portable strlen() for 32-bit and 64-bit systems.
36 *
37 * Rationale: it is generally much more efficient to do word length
38 * operations and avoid branches on modern computer systems, as
39 * compared to byte-length operations with a lot of branches.
40 *
41 * The expression:
42 *
43 *	((x - 0x01....01) & ~x & 0x80....80)
44 *
45 * would evaluate to a non-zero value iff any of the bytes in the
46 * original word is zero.
47 *
48 * On multi-issue processors, we can divide the above expression into:
49 *	a)  (x - 0x01....01)
50 *	b) (~x & 0x80....80)
51 *	c) a & b
52 *
53 * Where, a) and b) can be partially computed in parallel.
54 *
55 * The algorithm above is found on "Hacker's Delight" by
56 * Henry S. Warren, Jr.
57 */
58
59/* Magic numbers for the algorithm */
60#if LONG_BIT == 32
61static const unsigned long mask01 = 0x01010101;
62static const unsigned long mask80 = 0x80808080;
63#elif LONG_BIT == 64
64static const unsigned long mask01 = 0x0101010101010101;
65static const unsigned long mask80 = 0x8080808080808080;
66#else
67#error Unsupported word size
68#endif
69
70#define	LONGPTR_MASK (sizeof(long) - 1)
71
72/*
73 * Helper macro to return string length if we caught the zero
74 * byte.
75 */
76#define testbyte(x)				\
77	do {					\
78		if (p[x] == '\0')		\
79		    return (p - str + x);	\
80	} while (0)
81
82size_t
83strlen(const char *str)
84{
85	const char *p;
86	const unsigned long *lp;
87	long va, vb;
88
89	/*
90	 * Before trying the hard (unaligned byte-by-byte access) way
91	 * to figure out whether there is a nul character, try to see
92	 * if there is a nul character is within this accessible word
93	 * first.
94	 *
95	 * p and (p & ~LONGPTR_MASK) must be equally accessible since
96	 * they always fall in the same memory page, as long as page
97	 * boundaries is integral multiple of word size.
98	 */
99	lp = (const unsigned long *)((uintptr_t)str & ~LONGPTR_MASK);
100	va = (*lp - mask01);
101	vb = ((~*lp) & mask80);
102	lp++;
103	if (va & vb)
104		/* Check if we have \0 in the first part */
105		for (p = str; p < (const char *)lp; p++)
106			if (*p == '\0')
107				return (p - str);
108
109	/* Scan the rest of the string using word sized operation */
110	for (; ; lp++) {
111		va = (*lp - mask01);
112		vb = ((~*lp) & mask80);
113		if (va & vb) {
114			p = (const char *)(lp);
115			testbyte(0);
116			testbyte(1);
117			testbyte(2);
118			testbyte(3);
119#if (LONG_BIT >= 64)
120			testbyte(4);
121			testbyte(5);
122			testbyte(6);
123			testbyte(7);
124#endif
125		}
126	}
127
128	/* NOTREACHED */
129	return (0);
130}
131