]> git.ipfire.org Git - thirdparty/glibc.git/blame - sysdeps/x86_64/multiarch/strlen.S
Unroll x86-64 strlen
[thirdparty/glibc.git] / sysdeps / x86_64 / multiarch / strlen.S
CommitLineData
3ab2d57a 1/* strlen(str) -- determine the length of the string STR.
623aac7f 2 Copyright (C) 2009, 2010 Free Software Foundation, Inc.
3ab2d57a
UD
3 Contributed by Ulrich Drepper <drepper@redhat.com>.
4 This file is part of the GNU C Library.
5
6 The GNU C Library is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the License, or (at your option) any later version.
10
11 The GNU C Library is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 Lesser General Public License for more details.
15
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, write to the Free
18 Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
19 02111-1307 USA. */
20
21#include <sysdep.h>
51ddd2c0 22#include <init-arch.h>
3ab2d57a
UD
23
24
25/* Define multiple versions only for the definition in libc and for
26 the DSO. In static binaries we need strlen before the initialization
27 happened. */
28#if defined SHARED && !defined NOT_IN_libc
29 .text
30ENTRY(strlen)
31 .type strlen, @gnu_indirect_function
32 cmpl $0, __cpu_features+KIND_OFFSET(%rip)
33 jne 1f
34 call __init_cpu_features
351: leaq __strlen_sse2(%rip), %rax
51ddd2c0 36 testl $bit_SSE4_2, __cpu_features+CPUID_OFFSET+index_SSE4_2(%rip)
3ab2d57a
UD
37 jz 2f
38 leaq __strlen_sse42(%rip), %rax
5a7af22f 39 ret
623aac7f
L
402: testl $bit_Slow_BSF, __cpu_features+FEATURE_OFFSET+index_Slow_BSF(%rip)
41 jz 3f
42 leaq __strlen_no_bsf(%rip), %rax
433: ret
44END(strlen)
3ab2d57a
UD
45
46# undef ENTRY
47# define ENTRY(name) \
6f9eea15 48 .type __strlen_sse2, @function; \
d6485c98 49 .align 16; \
6f9eea15
UD
50 __strlen_sse2: cfi_startproc; \
51 CALL_MCOUNT
3ab2d57a
UD
52# undef END
53# define END(name) \
6f9eea15 54 cfi_endproc; .size __strlen_sse2, .-__strlen_sse2
3ab2d57a
UD
55# undef libc_hidden_builtin_def
56/* It doesn't make sense to send libc-internal strlen calls through a PLT.
57 The speedup we get from using SSE4.2 instruction is likely eaten away
58 by the indirect call in the PLT. */
59# define libc_hidden_builtin_def(name) \
60 .globl __GI_strlen; __GI_strlen = __strlen_sse2
61#endif
62
63#include "../strlen.S"