xref: /freebsd/lib/libc/string/strlen.c (revision bdd1243df58e60e85101c09001d9812a789b6bc4)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
3  *
4  * Copyright (c) 2009, 2010 Xin LI <delphij@FreeBSD.org>
5  *
6  * Redistribution and use in source and binary forms, with or without
7  * modification, are permitted provided that the following conditions
8  * are met:
9  * 1. Redistributions of source code must retain the above copyright
10  *    notice, this list of conditions and the following disclaimer.
11  * 2. Redistributions in binary form must reproduce the above copyright
12  *    notice, this list of conditions and the following disclaimer in the
13  *    documentation and/or other materials provided with the distribution.
14  *
15  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
16  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
19  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25  * SUCH DAMAGE.
26  */
27 
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
30 
31 #include <sys/limits.h>
32 #include <sys/types.h>
33 #include <string.h>
34 
35 /*
36  * Portable strlen() for 32-bit and 64-bit systems.
37  *
38  * The expression:
39  *
40  *	((x - 0x01....01) & ~x & 0x80....80)
41  *
42  * would evaluate to a non-zero value iff any of the bytes in the
43  * original word is zero.
44  *
45  * The algorithm above is found on "Hacker's Delight" by
46  * Henry S. Warren, Jr.
47  *
48  * Note: this leaves performance on the table and each architecture
49  * would be best served with a tailor made routine instead, even if
50  * using the same trick.
51  */
52 
53 /* Magic numbers for the algorithm */
54 #if LONG_BIT == 32
55 static const unsigned long mask01 = 0x01010101;
56 static const unsigned long mask80 = 0x80808080;
57 #elif LONG_BIT == 64
58 static const unsigned long mask01 = 0x0101010101010101;
59 static const unsigned long mask80 = 0x8080808080808080;
60 #else
61 #error Unsupported word size
62 #endif
63 
64 #define	LONGPTR_MASK (sizeof(long) - 1)
65 
66 /*
67  * Helper macro to return string length if we caught the zero
68  * byte.
69  */
70 #define testbyte(x)				\
71 	do {					\
72 		if (p[x] == '\0')		\
73 		    return (p - str + x);	\
74 	} while (0)
75 
76 size_t
77 strlen(const char *str)
78 {
79 	const char *p;
80 	const unsigned long *lp;
81 	long va, vb;
82 
83 	/*
84 	 * Before trying the hard (unaligned byte-by-byte access) way
85 	 * to figure out whether there is a nul character, try to see
86 	 * if there is a nul character is within this accessible word
87 	 * first.
88 	 *
89 	 * p and (p & ~LONGPTR_MASK) must be equally accessible since
90 	 * they always fall in the same memory page, as long as page
91 	 * boundaries is integral multiple of word size.
92 	 */
93 	lp = (const unsigned long *)((uintptr_t)str & ~LONGPTR_MASK);
94 	va = (*lp - mask01);
95 	vb = ((~*lp) & mask80);
96 	lp++;
97 	if (va & vb)
98 		/* Check if we have \0 in the first part */
99 		for (p = str; p < (const char *)lp; p++)
100 			if (*p == '\0')
101 				return (p - str);
102 
103 	/* Scan the rest of the string using word sized operation */
104 	for (; ; lp++) {
105 		va = (*lp - mask01);
106 		vb = ((~*lp) & mask80);
107 		if (va & vb) {
108 			p = (const char *)(lp);
109 			testbyte(0);
110 			testbyte(1);
111 			testbyte(2);
112 			testbyte(3);
113 #if (LONG_BIT >= 64)
114 			testbyte(4);
115 			testbyte(5);
116 			testbyte(6);
117 			testbyte(7);
118 #endif
119 		}
120 	}
121 
122 	/* NOTREACHED */
123 	return (0);
124 }
125