2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2009, 2010 Xin LI <delphij@FreeBSD.org>
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
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.
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
28 #include <sys/cdefs.h>
29 __FBSDID("$FreeBSD$");
31 #include <sys/libkern.h>
32 #include <sys/limits.h>
35 * Portable strlen() for 32-bit and 64-bit systems.
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.
43 * ((x - 0x01....01) & ~x & 0x80....80)
45 * would evaluate to a non-zero value iff any of the bytes in the
46 * original word is zero.
48 * On multi-issue processors, we can divide the above expression into:
50 * b) (~x & 0x80....80)
53 * Where, a) and b) can be partially computed in parallel.
55 * The algorithm above is found on "Hacker's Delight" by
56 * Henry S. Warren, Jr.
59 /* Magic numbers for the algorithm */
61 static const unsigned long mask01 = 0x01010101;
62 static const unsigned long mask80 = 0x80808080;
64 static const unsigned long mask01 = 0x0101010101010101;
65 static const unsigned long mask80 = 0x8080808080808080;
67 #error Unsupported word size
70 #define LONGPTR_MASK (sizeof(long) - 1)
73 * Helper macro to return string length if we caught the zero
79 return (p - str + x); \
83 (strlen)(const char *str)
86 const unsigned long *lp;
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
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.
99 lp = (const unsigned long *)((uintptr_t)str & ~LONGPTR_MASK);
101 vb = ((~*lp) & mask80);
104 /* Check if we have \0 in the first part */
105 for (p = str; p < (const char *)lp; p++)
109 /* Scan the rest of the string using word sized operation */
112 vb = ((~*lp) & mask80);
114 p = (const char *)(lp);