about summary refs log tree commit diff
path: root/sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S
diff options
context:
space:
mode:
authorLiubov Dmitrieva <liubov.dmitrieva@gmail.com>2011-10-12 11:42:04 -0400
committerUlrich Drepper <drepper@gmail.com>2011-10-12 11:42:04 -0400
commit951fbcec70e65c49705fcdbf4630bee5ce2a5691 (patch)
tree33fa5a0729ad3f67f693290d450b42a1a431f5ea /sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S
parent0ac5ae2335292908f39031b1ea9fe8edce433c0f (diff)
downloadglibc-951fbcec70e65c49705fcdbf4630bee5ce2a5691.tar.gz
glibc-951fbcec70e65c49705fcdbf4630bee5ce2a5691.tar.xz
glibc-951fbcec70e65c49705fcdbf4630bee5ce2a5691.zip
Optimized memchr, memrchr, rawmemchr for x86-32
Diffstat (limited to 'sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S')
-rw-r--r--sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S418
1 files changed, 418 insertions, 0 deletions
diff --git a/sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S b/sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S
new file mode 100644
index 0000000000..1c2a867ca2
--- /dev/null
+++ b/sysdeps/i386/i686/multiarch/memrchr-sse2-bsf.S
@@ -0,0 +1,418 @@
+/* Optimized memrchr with sse2
+   Copyright (C) 2011 Free Software Foundation, Inc.
+   Contributed by Intel Corporation.
+   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.  */
+
+#ifndef  NOT_IN_libc
+
+# include <sysdep.h>
+
+# define CFI_PUSH(REG)	\
+	cfi_adjust_cfa_offset (4);	\
+	cfi_rel_offset (REG, 0)
+
+# define CFI_POP(REG)	\
+	cfi_adjust_cfa_offset (-4);	\
+	cfi_restore (REG)
+
+# define PUSH(REG) pushl REG; CFI_PUSH (REG)
+# define POP(REG) popl REG; CFI_POP (REG)
+
+# define PARMS  4
+# define STR1  PARMS
+# define STR2  STR1+4
+# define LEN   STR2+4
+
+# define MEMCHR __memrchr_sse2_bsf
+
+	.text
+ENTRY (MEMCHR)
+	mov	STR1(%esp), %ecx
+	movd	STR2(%esp), %xmm1
+	mov	LEN(%esp), %edx
+
+	sub	$16, %edx
+	jbe	L(length_less16)
+
+	punpcklbw %xmm1, %xmm1
+	add	%edx, %ecx
+	punpcklbw %xmm1, %xmm1
+
+	movdqu	(%ecx), %xmm0
+	pshufd	$0, %xmm1, %xmm1
+	pcmpeqb	%xmm1, %xmm0
+
+/* Check if there is a match.  */
+	pmovmskb %xmm0, %eax
+	test	%eax, %eax
+	jnz	L(matches0)
+
+	sub	$64, %ecx
+	mov	%ecx, %eax
+	and	$15, %eax
+	jz	L(loop_prolog)
+
+	add	$16, %ecx
+	add	$16, %edx
+	sub	%eax, %ecx
+	sub	%eax, %edx
+
+	.p2align 4
+/* Loop start on aligned string.  */
+L(loop_prolog):
+	sub	$64, %edx
+	jbe	L(exit_loop)
+
+	movdqa	48(%ecx), %xmm0
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %eax
+	test	%eax, %eax
+	jnz	L(matches48)
+
+	movdqa	32(%ecx), %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	pmovmskb %xmm2, %eax
+	test	%eax, %eax
+	jnz	L(matches32)
+
+	movdqa	16(%ecx), %xmm3
+	pcmpeqb	%xmm1, %xmm3
+	pmovmskb %xmm3, %eax
+	test	%eax, %eax
+	jnz	L(matches16)
+
+	movdqa	(%ecx), %xmm4
+	pcmpeqb	%xmm1, %xmm4
+	pmovmskb %xmm4, %eax
+	test	%eax, %eax
+	jnz	L(matches0)
+
+	sub	$64, %ecx
+	sub	$64, %edx
+	jbe	L(exit_loop)
+
+	movdqa	48(%ecx), %xmm0
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %eax
+	test	%eax, %eax
+	jnz	L(matches48)
+
+	movdqa	32(%ecx), %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	pmovmskb %xmm2, %eax
+	test	%eax, %eax
+	jnz	L(matches32)
+
+	movdqa	16(%ecx), %xmm3
+	pcmpeqb	%xmm1, %xmm3
+	pmovmskb %xmm3, %eax
+	test	%eax, %eax
+	jnz	L(matches16)
+
+	movdqa	(%ecx), %xmm3
+	pcmpeqb	%xmm1, %xmm3
+	pmovmskb %xmm3, %eax
+	test	%eax, %eax
+	jnz	L(matches0)
+
+	mov	%ecx, %eax
+	and	$63, %eax
+	test	%eax, %eax
+	jz	L(align64_loop)    
+
+	add	$64, %ecx
+	add	$64, %edx
+	sub	%eax, %ecx
+	sub	%eax, %edx
+
+	.p2align 4
+L(align64_loop):
+	sub	$64, %ecx
+	sub	$64, %edx
+	jbe	L(exit_loop)
+
+	movdqa	(%ecx), %xmm0
+	movdqa	16(%ecx), %xmm2
+	movdqa	32(%ecx), %xmm3
+	movdqa	48(%ecx), %xmm4
+
+	pcmpeqb	%xmm1, %xmm0
+	pcmpeqb	%xmm1, %xmm2
+	pcmpeqb	%xmm1, %xmm3
+	pcmpeqb	%xmm1, %xmm4
+
+	pmaxub	%xmm3, %xmm0
+	pmaxub	%xmm4, %xmm2
+	pmaxub	%xmm0, %xmm2
+	pmovmskb %xmm2, %eax
+
+	test	%eax, %eax
+	jz	L(align64_loop)
+
+	pmovmskb %xmm4, %eax
+	test	%eax, %eax
+	jnz	L(matches48)
+
+	pmovmskb %xmm3, %eax
+	test	%eax, %eax
+	jnz	L(matches32)
+
+	movdqa	16(%ecx), %xmm2
+
+	pcmpeqb	%xmm1, %xmm2
+	pcmpeqb	(%ecx), %xmm1
+
+	pmovmskb %xmm2, %eax
+	test	%eax, %eax
+	jnz	L(matches16)
+
+	pmovmskb %xmm1, %eax
+	bsr	%eax, %eax
+
+	add	%ecx, %eax
+	ret
+
+	.p2align 4
+L(exit_loop):
+	add	$64, %edx
+	cmp	$32, %edx
+	jbe	L(exit_loop_32)
+
+	movdqa	48(%ecx), %xmm0
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %eax
+	test	%eax, %eax
+	jnz	L(matches48)
+
+	movdqa	32(%ecx), %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	pmovmskb %xmm2, %eax
+	test	%eax, %eax
+	jnz	L(matches32)
+
+	movdqa	16(%ecx), %xmm3
+	pcmpeqb	%xmm1, %xmm3
+	pmovmskb %xmm3, %eax
+	test	%eax, %eax
+	jnz	L(matches16_1)
+	cmp	$48, %edx
+	jbe	L(return_null)
+
+	pcmpeqb	(%ecx), %xmm1
+	pmovmskb %xmm1, %eax
+	test	%eax, %eax
+	jnz	L(matches0_1)
+	xor	%eax, %eax
+	ret
+
+	.p2align 4
+L(exit_loop_32):
+	movdqa	48(%ecx), %xmm0
+	pcmpeqb	%xmm1, %xmm0
+	pmovmskb %xmm0, %eax
+	test	%eax, %eax
+	jnz	L(matches48_1)
+	cmp	$16, %edx
+	jbe	L(return_null)
+
+	pcmpeqb	32(%ecx), %xmm1
+	pmovmskb %xmm1, %eax
+	test	%eax, %eax
+	jnz	L(matches32_1)
+	xor	%eax, %eax
+	ret
+
+	.p2align 4
+L(matches0):
+	bsr	%eax, %eax
+	add	%ecx, %eax
+	ret
+
+	.p2align 4
+L(matches16):
+	bsr	%eax, %eax
+	lea	16(%eax, %ecx), %eax
+	ret
+
+	.p2align 4
+L(matches32):
+	bsr	%eax, %eax
+	lea	32(%eax, %ecx), %eax
+	ret
+
+	.p2align 4
+L(matches48):
+	bsr	%eax, %eax
+	lea	48(%eax, %ecx), %eax
+	ret
+
+	.p2align 4
+L(matches0_1):
+	bsr	%eax, %eax
+	sub	$64, %edx
+	add	%eax, %edx
+	jl	L(return_null)
+	add	%ecx, %eax
+	ret
+
+	.p2align 4
+L(matches16_1):
+	bsr	%eax, %eax
+	sub	$48, %edx
+	add	%eax, %edx
+	jl	L(return_null)
+	lea	16(%ecx, %eax), %eax
+	ret
+
+	.p2align 4
+L(matches32_1):
+	bsr	%eax, %eax
+	sub	$32, %edx
+	add	%eax, %edx
+	jl	L(return_null)
+	lea	32(%ecx, %eax), %eax
+	ret
+
+	.p2align 4
+L(matches48_1):
+	bsr	%eax, %eax
+	sub	$16, %edx
+	add	%eax, %edx
+	jl	L(return_null)
+	lea	48(%ecx, %eax), %eax
+	ret
+
+	.p2align 4
+L(return_null):
+	xor	%eax, %eax
+	ret
+
+	.p2align 4
+L(length_less16_offset0):
+	mov	%dl, %cl
+	pcmpeqb	(%eax), %xmm1
+
+	mov	$1, %edx
+	sal	%cl, %edx
+	sub	$1, %edx
+	mov	%edx, %ecx
+
+	pmovmskb %xmm1, %edx
+
+	and	%ecx, %edx
+	test	%edx, %edx
+	jz	L(return_null)
+
+	bsr	%edx, %ecx
+	add	%ecx, %eax
+	ret
+
+	.p2align 4
+L(length_less16):
+	punpcklbw %xmm1, %xmm1
+	mov	%ecx, %eax
+	punpcklbw %xmm1, %xmm1
+	add	$16, %edx
+	jz	L(return_null)
+
+	pshufd	$0, %xmm1, %xmm1
+	and	$15, %ecx
+	jz	L(length_less16_offset0)
+	
+	PUSH	(%edi)
+	mov	%cl, %dh
+	add	%dl, %dh
+	and	$-16, %eax
+
+	sub	$16, %dh
+	ja	L(length_less16_part2)
+
+	pcmpeqb	(%eax), %xmm1
+	pmovmskb %xmm1, %edi
+
+	sar	%cl, %edi
+	add	%ecx, %eax
+	mov	%dl, %cl
+
+	mov	$1, %edx
+	sal	%cl, %edx
+	sub	$1, %edx
+
+	and	%edx, %edi
+	test	%edi, %edi
+	jz	L(ret_null)
+
+	bsr	%edi, %edi
+	add	%edi, %eax
+	POP	(%edi)
+	ret
+
+	CFI_PUSH     (%edi)
+
+	.p2align 4
+L(length_less16_part2):
+	movdqa	16(%eax), %xmm2
+	pcmpeqb	%xmm1, %xmm2
+	pmovmskb %xmm2, %edi
+
+	mov	%cl, %ch
+
+	mov	%dh, %cl
+	mov	$1, %edx
+	sal	%cl, %edx
+	sub	$1, %edx
+
+	and	%edx, %edi
+
+	test	%edi, %edi
+	jnz	L(length_less16_part2_return)
+
+	pcmpeqb	(%eax), %xmm1
+	pmovmskb %xmm1, %edi
+
+	mov	%ch, %cl
+	sar	%cl, %edi
+	test	%edi, %edi
+	jz	L(ret_null)
+
+	bsr	%edi, %edi
+	add	%edi, %eax
+	xor	%ch, %ch
+	add	%ecx, %eax
+	POP	(%edi)
+	ret
+
+	CFI_PUSH     (%edi)
+
+	.p2align 4
+L(length_less16_part2_return):
+	bsr	%edi, %edi
+	lea	16(%eax, %edi), %eax
+	POP	(%edi)
+	ret
+
+	CFI_PUSH     (%edi)
+
+	.p2align 4
+L(ret_null):
+	xor	%eax, %eax
+	POP	(%edi)
+	ret
+
+END (MEMCHR)
+#endif