PATCH: Remove sysdeps/x86_64/multiarch/strlen.S

["H.J. Lu" <hongjiu dot lu at intel dot com>]

>From our measurement, SSE4.2 strlen isn't faster than SSE2 strlen since
NULL char can be detected easily with SSE2 pcmpeqb. This patch removes it.

Thanks.


H.J.
---
2009-12-18  H.J. Lu  <hongjiu.lu@intel.com>

	* sysdeps/x86_64/multiarch/strlen.S: Removed.

diff --git a/sysdeps/x86_64/multiarch/strlen.S b/sysdeps/x86_64/multiarch/strlen.S
deleted file mode 100644
index 509f9c9..0000000
--- a/sysdeps/x86_64/multiarch/strlen.S
+++ /dev/null
@@ -1,95 +0,0 @@
-/* strlen(str) -- determine the length of the string STR.
-   Copyright (C) 2009 Free Software Foundation, Inc.
-   Contributed by Ulrich Drepper <drepper@redhat.com>.
-   This file is part of the GNU C Library.
-
-   The GNU C Library is free software; you can redistribute it and/or
-   modify it under the terms of the GNU Lesser General Public
-   License as published by the Free Software Foundation; either
-   version 2.1 of the License, or (at your option) any later version.
-
-   The GNU C Library is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   Lesser General Public License for more details.
-
-   You should have received a copy of the GNU Lesser General Public
-   License along with the GNU C Library; if not, write to the Free
-   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
-   02111-1307 USA.  */
-
-#include <sysdep.h>
-#include <init-arch.h>
-
-
-/* Define multiple versions only for the definition in libc and for
-   the DSO.  In static binaries we need strlen before the initialization
-   happened.  */
-#if defined SHARED && !defined NOT_IN_libc
-	.text
-ENTRY(strlen)
-	.type	strlen, @gnu_indirect_function
-	cmpl	$0, __cpu_features+KIND_OFFSET(%rip)
-	jne	1f
-	call	__init_cpu_features
-1:	leaq	__strlen_sse2(%rip), %rax
-	testl	$bit_SSE4_2, __cpu_features+CPUID_OFFSET+index_SSE4_2(%rip)
-	jz	2f
-	leaq	__strlen_sse42(%rip), %rax
-2:	ret
-END(strlen)
-
-
-	.section .text.sse4.2,"ax",@progbits
-	.align 	16
-	.type	__strlen_sse42, @function
-__strlen_sse42:
-	cfi_startproc
-	CALL_MCOUNT
-	pxor	%xmm2, %xmm2
-	movq	%rdi, %rcx
-	movq	%rdi, %r8
-	andq	$~15, %rdi
-	movdqa	%xmm2, %xmm1
-	pcmpeqb	(%rdi), %xmm2
-	orl	$0xffffffff, %esi
-	subq	%rdi, %rcx
-	shll	%cl, %esi
-	pmovmskb %xmm2, %edx
-	andl	%esi, %edx
-	jnz	1f
-
-2:	pcmpistri $0x08, 16(%rdi), %xmm1
-	leaq	16(%rdi), %rdi
-	jnz	2b
-
-	leaq	(%rdi,%rcx), %rax
-	subq	%r8, %rax
-	ret
-
-1:	subq	%r8, %rdi
-	bsfl	%edx, %eax
-	addq	%rdi, %rax
-	ret
-	cfi_endproc
-	.size	__strlen_sse42, .-__strlen_sse42
-
-
-# undef ENTRY
-# define ENTRY(name) \
-	.type __strlen_sse2, @function; \
-	.align 16; \
-	__strlen_sse2: cfi_startproc; \
-	CALL_MCOUNT
-# undef END
-# define END(name) \
-	cfi_endproc; .size __strlen_sse2, .-__strlen_sse2
-# undef libc_hidden_builtin_def
-/* It doesn't make sense to send libc-internal strlen calls through a PLT.
-   The speedup we get from using SSE4.2 instruction is likely eaten away
-   by the indirect call in the PLT.  */
-# define libc_hidden_builtin_def(name) \
-	.globl __GI_strlen; __GI_strlen = __strlen_sse2
-#endif
-
-#include "../strlen.S"