xref: /openbmc/u-boot/arch/arm/lib/memcpy.S (revision d77fa2ff)
1/*
2 *  linux/arch/arm/lib/memcpy.S
3 *
4 *  Author:	Nicolas Pitre
5 *  Created:	Sep 28, 2005
6 *  Copyright:	MontaVista Software, Inc.
7 *
8 *  This program is free software; you can redistribute it and/or modify
9 *  it under the terms of the GNU General Public License version 2 as
10 *  published by the Free Software Foundation.
11 */
12
13#include <linux/linkage.h>
14#include <asm/assembler.h>
15
16#define LDR1W_SHIFT	0
17#define STR1W_SHIFT	0
18
19	.macro ldr1w ptr reg abort
20	W(ldr) \reg, [\ptr], #4
21	.endm
22
23	.macro ldr4w ptr reg1 reg2 reg3 reg4 abort
24	ldmia \ptr!, {\reg1, \reg2, \reg3, \reg4}
25	.endm
26
27	.macro ldr8w ptr reg1 reg2 reg3 reg4 reg5 reg6 reg7 reg8 abort
28	ldmia \ptr!, {\reg1, \reg2, \reg3, \reg4, \reg5, \reg6, \reg7, \reg8}
29	.endm
30
31	.macro ldr1b ptr reg cond=al abort
32	ldrb\cond\() \reg, [\ptr], #1
33	.endm
34
35	.macro str1w ptr reg abort
36	W(str) \reg, [\ptr], #4
37	.endm
38
39	.macro str8w ptr reg1 reg2 reg3 reg4 reg5 reg6 reg7 reg8 abort
40	stmia \ptr!, {\reg1, \reg2, \reg3, \reg4, \reg5, \reg6, \reg7, \reg8}
41	.endm
42
43	.macro str1b ptr reg cond=al abort
44	strb\cond\() \reg, [\ptr], #1
45	.endm
46
47	.macro enter reg1 reg2
48	stmdb sp!, {r0, \reg1, \reg2}
49	.endm
50
51	.macro exit reg1 reg2
52	ldmfd sp!, {r0, \reg1, \reg2}
53	.endm
54
55	.text
56
57/* Prototype: void *memcpy(void *dest, const void *src, size_t n); */
58	.syntax unified
59#if defined(CONFIG_SYS_THUMB_BUILD) && !defined(MEMCPY_NO_THUMB_BUILD)
60	.thumb
61	.thumb_func
62#endif
63ENTRY(memcpy)
64		cmp	r0, r1
65		moveq	pc, lr
66
67		enter	r4, lr
68
69		subs	r2, r2, #4
70		blt	8f
71		ands	ip, r0, #3
72	PLD(	pld	[r1, #0]		)
73		bne	9f
74		ands	ip, r1, #3
75		bne	10f
76
771:		subs	r2, r2, #(28)
78		stmfd	sp!, {r5 - r8}
79		blt	5f
80
81	CALGN(	ands	ip, r0, #31		)
82	CALGN(	rsb	r3, ip, #32		)
83	CALGN(	sbcsne	r4, r3, r2		)  @ C is always set here
84	CALGN(	bcs	2f			)
85	CALGN(	adr	r4, 6f			)
86	CALGN(	subs	r2, r2, r3		)  @ C gets set
87	CALGN(	add	pc, r4, ip		)
88
89	PLD(	pld	[r1, #0]		)
902:	PLD(	subs	r2, r2, #96		)
91	PLD(	pld	[r1, #28]		)
92	PLD(	blt	4f			)
93	PLD(	pld	[r1, #60]		)
94	PLD(	pld	[r1, #92]		)
95
963:	PLD(	pld	[r1, #124]		)
974:		ldr8w	r1, r3, r4, r5, r6, r7, r8, ip, lr, abort=20f
98		subs	r2, r2, #32
99		str8w	r0, r3, r4, r5, r6, r7, r8, ip, lr, abort=20f
100		bge	3b
101	PLD(	cmn	r2, #96			)
102	PLD(	bge	4b			)
103
1045:		ands	ip, r2, #28
105		rsb	ip, ip, #32
106#if LDR1W_SHIFT > 0
107		lsl	ip, ip, #LDR1W_SHIFT
108#endif
109		addne	pc, pc, ip		@ C is always clear here
110		b	7f
1116:
112		.rept	(1 << LDR1W_SHIFT)
113		W(nop)
114		.endr
115		ldr1w	r1, r3, abort=20f
116		ldr1w	r1, r4, abort=20f
117		ldr1w	r1, r5, abort=20f
118		ldr1w	r1, r6, abort=20f
119		ldr1w	r1, r7, abort=20f
120		ldr1w	r1, r8, abort=20f
121		ldr1w	r1, lr, abort=20f
122
123#if LDR1W_SHIFT < STR1W_SHIFT
124		lsl	ip, ip, #STR1W_SHIFT - LDR1W_SHIFT
125#elif LDR1W_SHIFT > STR1W_SHIFT
126		lsr	ip, ip, #LDR1W_SHIFT - STR1W_SHIFT
127#endif
128		add	pc, pc, ip
129		nop
130		.rept	(1 << STR1W_SHIFT)
131		W(nop)
132		.endr
133		str1w	r0, r3, abort=20f
134		str1w	r0, r4, abort=20f
135		str1w	r0, r5, abort=20f
136		str1w	r0, r6, abort=20f
137		str1w	r0, r7, abort=20f
138		str1w	r0, r8, abort=20f
139		str1w	r0, lr, abort=20f
140
141	CALGN(	bcs	2b			)
142
1437:		ldmfd	sp!, {r5 - r8}
144
1458:		movs	r2, r2, lsl #31
146		ldr1b	r1, r3, ne, abort=21f
147		ldr1b	r1, r4, cs, abort=21f
148		ldr1b	r1, ip, cs, abort=21f
149		str1b	r0, r3, ne, abort=21f
150		str1b	r0, r4, cs, abort=21f
151		str1b	r0, ip, cs, abort=21f
152
153		exit	r4, pc
154
1559:		rsb	ip, ip, #4
156		cmp	ip, #2
157		ldr1b	r1, r3, gt, abort=21f
158		ldr1b	r1, r4, ge, abort=21f
159		ldr1b	r1, lr, abort=21f
160		str1b	r0, r3, gt, abort=21f
161		str1b	r0, r4, ge, abort=21f
162		subs	r2, r2, ip
163		str1b	r0, lr, abort=21f
164		blt	8b
165		ands	ip, r1, #3
166		beq	1b
167
16810:		bic	r1, r1, #3
169		cmp	ip, #2
170		ldr1w	r1, lr, abort=21f
171		beq	17f
172		bgt	18f
173
174
175		.macro	forward_copy_shift pull push
176
177		subs	r2, r2, #28
178		blt	14f
179
180	CALGN(	ands	ip, r0, #31		)
181	CALGN(	rsb	ip, ip, #32		)
182	CALGN(	sbcsne	r4, ip, r2		)  @ C is always set here
183	CALGN(	subcc	r2, r2, ip		)
184	CALGN(	bcc	15f			)
185
18611:		stmfd	sp!, {r5 - r9}
187
188	PLD(	pld	[r1, #0]		)
189	PLD(	subs	r2, r2, #96		)
190	PLD(	pld	[r1, #28]		)
191	PLD(	blt	13f			)
192	PLD(	pld	[r1, #60]		)
193	PLD(	pld	[r1, #92]		)
194
19512:	PLD(	pld	[r1, #124]		)
19613:		ldr4w	r1, r4, r5, r6, r7, abort=19f
197		mov	r3, lr, lspull #\pull
198		subs	r2, r2, #32
199		ldr4w	r1, r8, r9, ip, lr, abort=19f
200		orr	r3, r3, r4, lspush #\push
201		mov	r4, r4, lspull #\pull
202		orr	r4, r4, r5, lspush #\push
203		mov	r5, r5, lspull #\pull
204		orr	r5, r5, r6, lspush #\push
205		mov	r6, r6, lspull #\pull
206		orr	r6, r6, r7, lspush #\push
207		mov	r7, r7, lspull #\pull
208		orr	r7, r7, r8, lspush #\push
209		mov	r8, r8, lspull #\pull
210		orr	r8, r8, r9, lspush #\push
211		mov	r9, r9, lspull #\pull
212		orr	r9, r9, ip, lspush #\push
213		mov	ip, ip, lspull #\pull
214		orr	ip, ip, lr, lspush #\push
215		str8w	r0, r3, r4, r5, r6, r7, r8, r9, ip, , abort=19f
216		bge	12b
217	PLD(	cmn	r2, #96			)
218	PLD(	bge	13b			)
219
220		ldmfd	sp!, {r5 - r9}
221
22214:		ands	ip, r2, #28
223		beq	16f
224
22515:		mov	r3, lr, lspull #\pull
226		ldr1w	r1, lr, abort=21f
227		subs	ip, ip, #4
228		orr	r3, r3, lr, lspush #\push
229		str1w	r0, r3, abort=21f
230		bgt	15b
231	CALGN(	cmp	r2, #0			)
232	CALGN(	bge	11b			)
233
23416:		sub	r1, r1, #(\push / 8)
235		b	8b
236
237		.endm
238
239
240		forward_copy_shift	pull=8	push=24
241
24217:		forward_copy_shift	pull=16	push=16
243
24418:		forward_copy_shift	pull=24	push=8
245
246
247/*
248 * Abort preamble and completion macros.
249 * If a fixup handler is required then those macros must surround it.
250 * It is assumed that the fixup code will handle the private part of
251 * the exit macro.
252 */
253
254	.macro	copy_abort_preamble
25519:	ldmfd	sp!, {r5 - r9}
256	b	21f
25720:	ldmfd	sp!, {r5 - r8}
25821:
259	.endm
260
261	.macro	copy_abort_end
262	ldmfd	sp!, {r4, pc}
263	.endm
264
265ENDPROC(memcpy)
266