summaryrefslogtreecommitdiffstats
path: root/lib
diff options
context:
space:
mode:
authordelphij <delphij@FreeBSD.org>2009-01-25 23:08:47 +0000
committerdelphij <delphij@FreeBSD.org>2009-01-25 23:08:47 +0000
commit12e5c4fd4c1b731e55f00ff3926ba953ee29d39a (patch)
tree79333d07b8661f2894da7b8eb790e1819a93f964 /lib
parent34c3dd7ace1e27678447ebdb30dfb3fe11306aaf (diff)
downloadFreeBSD-src-12e5c4fd4c1b731e55f00ff3926ba953ee29d39a.zip
FreeBSD-src-12e5c4fd4c1b731e55f00ff3926ba953ee29d39a.tar.gz
Rewrite of MI strlen(3) in a way that can better utilize modern hardware by
reducing branches and doing word-sized operation. The idea is taken from J.T. Conklin's x86_64 optimized version of strlen(3) for NetBSD, and reimplemented in C by me. Discussed on: -arch@
Diffstat (limited to 'lib')
-rw-r--r--lib/libc/string/strlen.c93
1 files changed, 78 insertions, 15 deletions
diff --git a/lib/libc/string/strlen.c b/lib/libc/string/strlen.c
index db4fe26..b6e94b9 100644
--- a/lib/libc/string/strlen.c
+++ b/lib/libc/string/strlen.c
@@ -1,6 +1,6 @@
/*-
- * Copyright (c) 1990, 1993
- * The Regents of the University of California. All rights reserved.
+ * Copyright (c) 2009 Xin LI <delphij@FreeBSD.org>
+ * All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
@@ -10,14 +10,11 @@
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
- * 4. Neither the name of the University nor the names of its contributors
- * may be used to endorse or promote products derived from this software
- * without specific prior written permission.
*
- * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
@@ -27,21 +24,87 @@
* SUCH DAMAGE.
*/
-#if defined(LIBC_SCCS) && !defined(lint)
-static char sccsid[] = "@(#)strlen.c 8.1 (Berkeley) 6/4/93";
-#endif /* LIBC_SCCS and not lint */
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
+#include <sys/limits.h>
+#include <sys/types.h>
#include <string.h>
+/*
+ * Portable strlen() for 32-bit and 64-bit systems.
+ *
+ * Rationale: it is generally much more efficient to do word length
+ * operations and avoid branches on modern computer systems, as
+ * compared to byte-length operations with a lot of branches.
+ *
+ * The expression:
+ *
+ * ((x - 0x01....01) & ~x & 0x80....80)
+ *
+ * would evaluate to a non-zero value iff any of the bytes in the
+ * original word is zero. However, we can further reduce ~1/3 of
+ * time if we consider that strlen() usually operate on 7-bit ASCII
+ * by employing the following expression, which allows false positive
+ * when high bit of 1 and use the tail case to catch these case:
+ *
+ * ((x - 0x01....01) & 0x80....80)
+ *
+ * This is more than 5.2 times as compared to the raw implementation
+ * on Intel T7300 under EM64T mode for strings longer than word length.
+ */
+
+/* Magic numbers for the algorithm */
+#if LONG_BIT == 32
+static const unsigned long mask01 = 0x01010101;
+static const unsigned long mask80 = 0x80808080;
+#elif LONG_BIT == 64
+static const unsigned long mask01 = 0x0101010101010101;
+static const unsigned long mask80 = 0x8080808080808080;
+#else
+#error Unsupported word size
+#endif
+
+#define LONGPTR_MASK (sizeof(long) - 1)
+
+/*
+ * Helper macro to return string length if we caught the zero
+ * byte.
+ */
+#define testbyte(x) \
+ do { \
+ if (p[x] == '\0') \
+ return (p - str + x); \
+ } while (0)
+
size_t
-strlen(str)
- const char *str;
+strlen(const char *str)
{
- const char *s;
+ const char *p;
+ const unsigned long *lp;
+
+ /* Skip the first few bytes until we have an aligned p */
+ for (p = str; (uintptr_t)p & LONGPTR_MASK; p++)
+ if (*p == '\0')
+ return (p - str);
+
+ /* Scan the rest of the string using word sized operation */
+ for (lp = (const unsigned long *)p; ; lp++)
+ if ((*lp - mask01) & mask80) {
+ p = (const char *)(lp);
+ testbyte(0);
+ testbyte(1);
+ testbyte(2);
+ testbyte(3);
+#if (LONG_BIT >= 64)
+ testbyte(4);
+ testbyte(5);
+ testbyte(6);
+ testbyte(7);
+#endif
+ }
- for (s = str; *s; ++s);
- return(s - str);
+ /* NOTREACHED */
+ return 0;
}
OpenPOWER on IntegriCloud