xref: /freebsd-13.1/lib/libc/amd64/string/memset.S (revision ca9e7ac2)
1/*-
2 * Copyright (c) 2018 The FreeBSD Foundation
3 *
4 * This software was developed by Mateusz Guzik <[email protected]>
5 * under sponsorship from the FreeBSD Foundation.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 *    notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 *    notice, this list of conditions and the following disclaimer in the
14 *    documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 * SUCH DAMAGE.
27 *
28 * $FreeBSD$
29 */
30
31#include <machine/asm.h>
32__FBSDID("$FreeBSD$");
33
34#define ALIGN_TEXT      .p2align 4,0x90 /* 16-byte alignment, nop filled */
35
36.macro MEMSET erms
37	movq	%rdi,%rax
38	movq	%rdx,%rcx
39	movzbq	%sil,%r8
40	movabs	$0x0101010101010101,%r10
41	imulq	%r8,%r10
42
43	cmpq	$32,%rcx
44	jbe	101632f
45
46	cmpq	$256,%rcx
47	ja	1256f
48
49	ALIGN_TEXT
50103200:
51	movq	%r10,(%rdi)
52	movq	%r10,8(%rdi)
53	movq	%r10,16(%rdi)
54	movq	%r10,24(%rdi)
55	leaq	32(%rdi),%rdi
56	subq	$32,%rcx
57	cmpq	$32,%rcx
58	ja	103200b
59	cmpb	$16,%cl
60	ja	201632f
61	movq	%r10,-16(%rdi,%rcx)
62	movq	%r10,-8(%rdi,%rcx)
63	ret
64	ALIGN_TEXT
65101632:
66	cmpb	$16,%cl
67	jl	100816f
68201632:
69	movq	%r10,(%rdi)
70	movq	%r10,8(%rdi)
71	movq	%r10,-16(%rdi,%rcx)
72	movq	%r10,-8(%rdi,%rcx)
73	ret
74	ALIGN_TEXT
75100816:
76	cmpb	$8,%cl
77	jl	100408f
78	movq	%r10,(%rdi)
79	movq	%r10,-8(%rdi,%rcx)
80	ret
81	ALIGN_TEXT
82100408:
83	cmpb	$4,%cl
84	jl	100204f
85	movl	%r10d,(%rdi)
86	movl	%r10d,-4(%rdi,%rcx)
87	ret
88	ALIGN_TEXT
89100204:
90	cmpb	$2,%cl
91	jl	100001f
92	movw	%r10w,(%rdi)
93	movw	%r10w,-2(%rdi,%rcx)
94	ret
95	ALIGN_TEXT
96100001:
97	cmpb	$0,%cl
98	je	100000f
99	movb	%r10b,(%rdi)
100100000:
101	ret
102	ALIGN_TEXT
1031256:
104	movq	%rdi,%r9
105	movq	%r10,%rax
106	testl	$15,%edi
107	jnz	3f
1081:
109.if \erms == 1
110	rep
111	stosb
112	movq	%r9,%rax
113.else
114	movq	%rcx,%rdx
115	shrq	$3,%rcx
116	rep
117	stosq
118	movq	%r9,%rax
119	andl	$7,%edx
120	jnz	2f
121	ret
1222:
123	movq	%r10,-8(%rdi,%rdx)
124.endif
125	ret
126	ALIGN_TEXT
1273:
128	movq	%r10,(%rdi)
129	movq	%r10,8(%rdi)
130	movq	%rdi,%r8
131	andq	$15,%r8
132	leaq	-16(%rcx,%r8),%rcx
133	neg	%r8
134	leaq	16(%rdi,%r8),%rdi
135	jmp	1b
136.endm
137
138
139ENTRY(memset)
140	MEMSET erms=0
141END(memset)
142
143	.section .note.GNU-stack,"",%progbits
144