aboutsummaryrefslogtreecommitdiff
path: root/lib/libc/arm/string/memcpy_arm.S
blob: 050ea8fb315fb36047a94afc358dcd2724a0c7ba (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
/*	$NetBSD: memcpy_arm.S,v 1.1 2003/10/14 07:51:45 scw Exp $	*/

/*-
 * Copyright (c) 1997 The NetBSD Foundation, Inc.
 * All rights reserved.
 *
 * This code is derived from software contributed to The NetBSD Foundation
 * by Neil A. Carson and Mark Brinicombe
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 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.
 *
 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. 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 FOUNDATION 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) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 * POSSIBILITY OF SUCH DAMAGE.
 */

#include <machine/asm.h>
__FBSDID("$FreeBSD$");

.syntax	unified

/*
 * This is one fun bit of code ...
 * Some easy listening music is suggested while trying to understand this
 * code e.g. Iron Maiden
 *
 * For anyone attempting to understand it :
 *
 * The core code is implemented here with simple stubs for memcpy().
 *
 * All local labels are prefixed with Lmemcpy_
 * Following the prefix a label starting f is used in the forward copy code
 * while a label using b is used in the backwards copy code
 * The source and destination addresses determine whether a forward or
 * backward copy is performed.
 * Separate bits of code are used to deal with the following situations
 * for both the forward and backwards copy.
 * unaligned source address
 * unaligned destination address
 * Separate copy routines are used to produce an optimised result for each
 * of these cases.
 * The copy code will use LDM/STM instructions to copy up to 32 bytes at
 * a time where possible.
 *
 * Note: r12 (aka ip) can be trashed during the function along with
 * r0-r3 although r0-r2 have defined uses i.e. src, dest, len through out.
 * Additional registers are preserved prior to use i.e. r4, r5 & lr
 *
 * Apologies for the state of the comments ;-)
 */
/* LINTSTUB: Func: void *memcpy(void *dst, const void *src, size_t len) */
ENTRY(memcpy)
	/* save leaf functions having to store this away */
	stmdb	sp!, {r0, lr}		/* memcpy() returns dest addr */

	subs	r2, r2, #4
	blt	.Lmemcpy_l4		/* less than 4 bytes */
	ands	r12, r0, #3
	bne	.Lmemcpy_destul		/* oh unaligned destination addr */
	ands	r12, r1, #3
	bne	.Lmemcpy_srcul		/* oh unaligned source addr */

.Lmemcpy_t8:
	/* We have aligned source and destination */
	subs	r2, r2, #8
	blt	.Lmemcpy_l12		/* less than 12 bytes (4 from above) */
	subs	r2, r2, #0x14         
	blt	.Lmemcpy_l32		/* less than 32 bytes (12 from above) */
	stmdb	sp!, {r4}		/* borrow r4 */

	/* blat 32 bytes at a time */
	/* XXX for really big copies perhaps we should use more registers */
.Lmemcpy_loop32:	
	ldmia	r1!, {r3, r4, r12, lr}
	stmia	r0!, {r3, r4, r12, lr}
	ldmia	r1!, {r3, r4, r12, lr}
	stmia	r0!, {r3, r4, r12, lr}
	subs	r2, r2, #0x20         
	bge	.Lmemcpy_loop32

	cmn	r2, #0x10
	ldmiage	r1!, {r3, r4, r12, lr}	/* blat a remaining 16 bytes */
	stmiage	r0!, {r3, r4, r12, lr}
	subge	r2, r2, #0x10         
	ldmia	sp!, {r4}		/* return r4 */

.Lmemcpy_l32:
	adds	r2, r2, #0x14         

	/* blat 12 bytes at a time */
.Lmemcpy_loop12:
	ldmiage	r1!, {r3, r12, lr}
	stmiage	r0!, {r3, r12, lr}
	subsge	r2, r2, #0x0c         
	bge	.Lmemcpy_loop12

.Lmemcpy_l12:
	adds	r2, r2, #8
	blt	.Lmemcpy_l4

	subs	r2, r2, #4
	ldrlt	r3, [r1], #4
	strlt	r3, [r0], #4
	ldmiage	r1!, {r3, r12}
	stmiage	r0!, {r3, r12}
	subge	r2, r2, #4

.Lmemcpy_l4:
	/* less than 4 bytes to go */
	adds	r2, r2, #4
#ifdef __APCS_26_
	ldmiaeq sp!, {r0, pc}^		/* done */
#else
	ldmiaeq	sp!, {r0, pc}		/* done */
#endif
	/* copy the crud byte at a time */
	cmp	r2, #2
	ldrb	r3, [r1], #1
	strb	r3, [r0], #1
	ldrbge	r3, [r1], #1
	strbge	r3, [r0], #1
	ldrbgt	r3, [r1], #1
	strbgt	r3, [r0], #1
	ldmia	sp!, {r0, pc}

	/* erg - unaligned destination */
.Lmemcpy_destul:
	rsb	r12, r12, #4
	cmp	r12, #2

	/* align destination with byte copies */
	ldrb	r3, [r1], #1
	strb	r3, [r0], #1
	ldrbge	r3, [r1], #1
	strbge	r3, [r0], #1
	ldrbgt	r3, [r1], #1
	strbgt	r3, [r0], #1
	subs	r2, r2, r12
	blt	.Lmemcpy_l4		/* less the 4 bytes */

	ands	r12, r1, #3
	beq	.Lmemcpy_t8		/* we have an aligned source */

	/* erg - unaligned source */
	/* This is where it gets nasty ... */
.Lmemcpy_srcul:
	bic	r1, r1, #3
	ldr	lr, [r1], #4
	cmp	r12, #2
	bgt	.Lmemcpy_srcul3
	beq	.Lmemcpy_srcul2
	cmp	r2, #0x0c            
	blt	.Lmemcpy_srcul1loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5}

.Lmemcpy_srcul1loop16:
	mov	r3, lr, lsr #8
	ldmia	r1!, {r4, r5, r12, lr}
	orr	r3, r3, r4, lsl #24
	mov	r4, r4, lsr #8
	orr	r4, r4, r5, lsl #24
	mov	r5, r5, lsr #8
	orr	r5, r5, r12, lsl #24
	mov	r12, r12, lsr #8
	orr	r12, r12, lr, lsl #24
	stmia	r0!, {r3-r5, r12}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_srcul1loop16
	ldmia	sp!, {r4, r5}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_srcul1l4

.Lmemcpy_srcul1loop4:
	mov	r12, lr, lsr #8
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsl #24
	str	r12, [r0], #4
	subs	r2, r2, #4
	bge	.Lmemcpy_srcul1loop4

.Lmemcpy_srcul1l4:
	sub	r1, r1, #3
	b	.Lmemcpy_l4

.Lmemcpy_srcul2:
	cmp	r2, #0x0c            
	blt	.Lmemcpy_srcul2loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5}

.Lmemcpy_srcul2loop16:
	mov	r3, lr, lsr #16
	ldmia	r1!, {r4, r5, r12, lr}
	orr	r3, r3, r4, lsl #16
	mov	r4, r4, lsr #16
	orr	r4, r4, r5, lsl #16
	mov	r5, r5, lsr #16
	orr	r5, r5, r12, lsl #16
	mov	r12, r12, lsr #16
	orr	r12, r12, lr, lsl #16
	stmia	r0!, {r3-r5, r12}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_srcul2loop16
	ldmia	sp!, {r4, r5}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_srcul2l4

.Lmemcpy_srcul2loop4:
	mov	r12, lr, lsr #16
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsl #16
	str	r12, [r0], #4
	subs	r2, r2, #4
	bge	.Lmemcpy_srcul2loop4

.Lmemcpy_srcul2l4:
	sub	r1, r1, #2
	b	.Lmemcpy_l4

.Lmemcpy_srcul3:
	cmp	r2, #0x0c            
	blt	.Lmemcpy_srcul3loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5}

.Lmemcpy_srcul3loop16:
	mov	r3, lr, lsr #24
	ldmia	r1!, {r4, r5, r12, lr}
	orr	r3, r3, r4, lsl #8
	mov	r4, r4, lsr #24
	orr	r4, r4, r5, lsl #8
	mov	r5, r5, lsr #24
	orr	r5, r5, r12, lsl #8
	mov	r12, r12, lsr #24
	orr	r12, r12, lr, lsl #8
	stmia	r0!, {r3-r5, r12}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_srcul3loop16
	ldmia	sp!, {r4, r5}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_srcul3l4

.Lmemcpy_srcul3loop4:
	mov	r12, lr, lsr #24
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsl #8
	str	r12, [r0], #4
	subs	r2, r2, #4
	bge	.Lmemcpy_srcul3loop4

.Lmemcpy_srcul3l4:
	sub	r1, r1, #1
	b	.Lmemcpy_l4
END(memcpy)

	.section .note.GNU-stack,"",%progbits