summaryrefslogtreecommitdiff
path: root/libc/string/arm/_memcpy.S
blob: 1dc0887a474c210a431713dbb9ba3ae33c20bbf8 (plain)
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
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
/*-
 * 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.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *        This product includes software developed by the NetBSD
 *        Foundation, Inc. and its contributors.
 * 4. Neither the name of The NetBSD Foundation nor the names of its
 *    contributors may be used to endorse or promote products derived
 *    from this software without specific prior written permission.
 *
 * 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.
 *
 * Adapted for uClibc from NetBSD _memcpy.S,v 1.6 2003/10/09
 * by Erik Andersen <andersen@codepoet.org>
 */

#include <features.h>
#include <endian.h>
#include <bits/arm_asm.h>

#if !defined(THUMB1_ONLY)
/*
 * 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()
 * memmove() and bcopy().
 *
 * 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 ;-)
 */

.text
.global _memcpy
.hidden _memcpy
.type _memcpy,%function
.align 4

/* XXX: The Thumb-2 conditionals can be removed if/when we require an
   assembler that supports unified syntax.  */
.macro copy regs
#if defined(__thumb2__)
	ittt	ge
	ldmiage	r1!, \regs
	stmiage	r0!, \regs
#else
	ldmgeia	r1!, \regs
	stmgeia	r0!, \regs
#endif
.endm

.macro copydb regs
#if defined(__thumb2__)
	ittt	ge
	ldmdbge	r1!, \regs
	stmdbge	r0!, \regs
#else
	ldmgedb	r1!, \regs
	stmgedb	r0!, \regs
#endif
.endm

_memcpy:
	/* Determine copy direction */
	cmp	r1, r0
	bcc	.Lmemcpy_backwards

	IT(t, eq)			/* Quick abort for src=dst */
#if defined(__USE_BX__)
        bxeq    lr
#else
        moveq   pc, lr
#endif
	stmdb	sp!, {r0, lr}		/* memcpy() returns dest addr */
	subs	r2, r2, #4
	blt	.Lmemcpy_fl4		/* less than 4 bytes */
	ands	r12, r0, #3
	bne	.Lmemcpy_fdestul	/* oh unaligned destination addr */
	ands	r12, r1, #3
	bne	.Lmemcpy_fsrcul		/* oh unaligned source addr */

.Lmemcpy_ft8:
	/* We have aligned source and destination */
	subs	r2, r2, #8
	blt	.Lmemcpy_fl12		/* less than 12 bytes (4 from above) */
	subs	r2, r2, #0x14         
	blt	.Lmemcpy_fl32		/* less than 32 bytes (12 from above) */
	str	r4, [sp, #-4]!		/* borrow r4 */

	/* blat 32 bytes at a time */
	/* XXX for really big copies perhaps we should use more registers */
.Lmemcpy_floop32:	
	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_floop32

	cmn	r2, #0x10
	/* blat a remaining 16 bytes */
	copy	"{r3, r4, r12, lr}"
	subge	r2, r2, #0x10         
	ldr	r4, [sp], #4		/* restore r4 */

.Lmemcpy_fl32:
	adds	r2, r2, #0x14         

	/* blat 12 bytes at a time */
.Lmemcpy_floop12:
	copy	"{r3, r12, lr}"
#if defined(__thumb2__)
	subsge	r2, r2, #0x0c         
#else
	subges	r2, r2, #0x0c         
#endif
	bge	.Lmemcpy_floop12

.Lmemcpy_fl12:
	adds	r2, r2, #8
	blt	.Lmemcpy_fl4

	subs	r2, r2, #4
	IT(tt, lt)
	ldrlt	r3, [r1], #4
	strlt	r3, [r0], #4
	copy	"{r3, r12}"
	subge	r2, r2, #4

.Lmemcpy_fl4:
	/* less than 4 bytes to go */
	adds	r2, r2, #4
#if defined(__thumb2__)
	it	eq
	popeq	{r0, pc}		/* done */
#elif defined(__ARM_ARCH_4T__)
	ldmeqia	sp!, {r0, r3}		/* done */
	bxeq	r3
#else
	ldmeqia	sp!, {r0, pc}		/* done */
#endif

	/* copy the crud byte at a time */
	cmp	r2, #2
	ldrb	r3, [r1], #1
	strb	r3, [r0], #1
#if defined(__thumb2__)
	itt	ge
	ldrbge	r3, [r1], #1
	strbge	r3, [r0], #1
	itt	gt
	ldrbgt	r3, [r1], #1
	strbgt	r3, [r0], #1
#else
	ldrgeb	r3, [r1], #1
	strgeb	r3, [r0], #1
	ldrgtb	r3, [r1], #1
	strgtb	r3, [r0], #1
#endif
#if defined(__ARM_ARCH_4T__)
	ldmia	sp!, {r0, r3}
	bx	r3
#else
	ldmia	sp!, {r0, pc}
#endif

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

	/* align destination with byte copies */
	ldrb	r3, [r1], #1
	strb	r3, [r0], #1
#if defined(__thumb2__)
	itt	ge
	ldrbge	r3, [r1], #1
	strbge	r3, [r0], #1
	itt	gt
	ldrbgt	r3, [r1], #1
	strbgt	r3, [r0], #1
#else
	ldrgeb	r3, [r1], #1
	strgeb	r3, [r0], #1
	ldrgtb	r3, [r1], #1
	strgtb	r3, [r0], #1
#endif
	subs	r2, r2, r12
	blt	.Lmemcpy_fl4		/* less the 4 bytes */

	ands	r12, r1, #3
	beq	.Lmemcpy_ft8		/* we have an aligned source */

	/* erg - unaligned source */
	/* This is where it gets nasty ... */
.Lmemcpy_fsrcul:
	bic	r1, r1, #3
	ldr	lr, [r1], #4
	cmp	r12, #2
	bgt	.Lmemcpy_fsrcul3
	beq	.Lmemcpy_fsrcul2
	cmp	r2, #0x0c            
	blt	.Lmemcpy_fsrcul1loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5}

.Lmemcpy_fsrcul1loop16:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r3, lr, lsl #8
	ldmia	r1!, {r4, r5, r12, lr}
	orr	r3, r3, r4, lsr #24
	mov	r4, r4, lsl #8
	orr	r4, r4, r5, lsr #24
	mov	r5, r5, lsl #8
	orr	r5, r5, r12, lsr #24
	mov	r12, r12, lsl #8
	orr	r12, r12, lr, lsr #24
#else
	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
#endif
	stmia	r0!, {r3-r5, r12}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_fsrcul1loop16
	ldmia	sp!, {r4, r5}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_fsrcul1l4

.Lmemcpy_fsrcul1loop4:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r12, lr, lsl #8
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsr #24
#else
	mov	r12, lr, lsr #8
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsl #24
#endif
	str	r12, [r0], #4
	subs	r2, r2, #4
	bge	.Lmemcpy_fsrcul1loop4

.Lmemcpy_fsrcul1l4:
	sub	r1, r1, #3
	b	.Lmemcpy_fl4

.Lmemcpy_fsrcul2:
	cmp	r2, #0x0c            
	blt	.Lmemcpy_fsrcul2loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5}

.Lmemcpy_fsrcul2loop16:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r3, lr, lsl #16
	ldmia	r1!, {r4, r5, r12, lr}
	orr	r3, r3, r4, lsr #16
	mov	r4, r4, lsl #16
	orr	r4, r4, r5, lsr #16
	mov	r5, r5, lsl #16
	orr	r5, r5, r12, lsr #16
	mov	r12, r12, lsl #16
	orr	r12, r12, lr, lsr #16
#else
	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
#endif
	stmia	r0!, {r3-r5, r12}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_fsrcul2loop16
	ldmia	sp!, {r4, r5}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_fsrcul2l4

.Lmemcpy_fsrcul2loop4:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r12, lr, lsl #16
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsr #16
#else
	mov	r12, lr, lsr #16
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsl #16
#endif
	str	r12, [r0], #4
	subs	r2, r2, #4
	bge	.Lmemcpy_fsrcul2loop4

.Lmemcpy_fsrcul2l4:
	sub	r1, r1, #2
	b	.Lmemcpy_fl4

.Lmemcpy_fsrcul3:
	cmp	r2, #0x0c            
	blt	.Lmemcpy_fsrcul3loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5}

.Lmemcpy_fsrcul3loop16:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r3, lr, lsl #24
	ldmia	r1!, {r4, r5, r12, lr}
	orr	r3, r3, r4, lsr #8
	mov	r4, r4, lsl #24
	orr	r4, r4, r5, lsr #8
	mov	r5, r5, lsl #24
	orr	r5, r5, r12, lsr #8
	mov	r12, r12, lsl #24
	orr	r12, r12, lr, lsr #8
#else
	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
#endif
	stmia	r0!, {r3-r5, r12}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_fsrcul3loop16
	ldmia	sp!, {r4, r5}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_fsrcul3l4

.Lmemcpy_fsrcul3loop4:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r12, lr, lsl #24
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsr #8
#else
	mov	r12, lr, lsr #24
	ldr	lr, [r1], #4
	orr	r12, r12, lr, lsl #8
#endif
	str	r12, [r0], #4
	subs	r2, r2, #4
	bge	.Lmemcpy_fsrcul3loop4

.Lmemcpy_fsrcul3l4:
	sub	r1, r1, #1
	b	.Lmemcpy_fl4

.Lmemcpy_backwards:
	add	r1, r1, r2
	add	r0, r0, r2
	subs	r2, r2, #4
	blt	.Lmemcpy_bl4		/* less than 4 bytes */
	ands	r12, r0, #3
	bne	.Lmemcpy_bdestul	/* oh unaligned destination addr */
	ands	r12, r1, #3
	bne	.Lmemcpy_bsrcul		/* oh unaligned source addr */

.Lmemcpy_bt8:
	/* We have aligned source and destination */
	subs	r2, r2, #8
	blt	.Lmemcpy_bl12		/* less than 12 bytes (4 from above) */
	stmdb	sp!, {r4, lr}
	subs	r2, r2, #0x14		/* less than 32 bytes (12 from above) */
	blt	.Lmemcpy_bl32

	/* blat 32 bytes at a time */
	/* XXX for really big copies perhaps we should use more registers */
.Lmemcpy_bloop32:
	ldmdb	r1!, {r3, r4, r12, lr}
	stmdb	r0!, {r3, r4, r12, lr}
	ldmdb	r1!, {r3, r4, r12, lr}
	stmdb	r0!, {r3, r4, r12, lr}
	subs	r2, r2, #0x20         
	bge	.Lmemcpy_bloop32

.Lmemcpy_bl32:
	cmn	r2, #0x10            
	/* blat a remaining 16 bytes */
	copydb	"{r3, r4, r12, lr}"
	subge	r2, r2, #0x10         
	adds	r2, r2, #0x14         
	/* blat a remaining 12 bytes */
	copydb	"{r3, r12, lr}"
	subge	r2, r2, #0x0c         
	ldmia	sp!, {r4, lr}

.Lmemcpy_bl12:
	adds	r2, r2, #8
	blt	.Lmemcpy_bl4
	subs	r2, r2, #4
	IT(tt, lt)
	ldrlt	r3, [r1, #-4]!
	strlt	r3, [r0, #-4]!
	copydb	"{r3, r12}"
	subge	r2, r2, #4

.Lmemcpy_bl4:
	/* less than 4 bytes to go */
	adds	r2, r2, #4
	IT(t, eq)
#if defined(__USE_BX__)
        bxeq    lr
#else
	moveq	pc, lr			/* done */
#endif
	/* copy the crud byte at a time */
	cmp	r2, #2
	ldrb	r3, [r1, #-1]!
	strb	r3, [r0, #-1]!
#ifdef __thumb2__
	itt	ge
	ldrbge	r3, [r1, #-1]!
	strbge	r3, [r0, #-1]!
	itt	gt
	ldrbgt	r3, [r1, #-1]!
	strbgt	r3, [r0, #-1]!
#else
	ldrgeb	r3, [r1, #-1]!
	strgeb	r3, [r0, #-1]!
	ldrgtb	r3, [r1, #-1]!
	strgtb	r3, [r0, #-1]!
#endif
#if defined(__USE_BX__)
        bx      lr
#else
	mov	pc, lr
#endif
	/* erg - unaligned destination */
.Lmemcpy_bdestul:
	cmp	r12, #2

	/* align destination with byte copies */
	ldrb	r3, [r1, #-1]!
	strb	r3, [r0, #-1]!
#ifdef __thumb2__
	itt	ge
	ldrbge	r3, [r1, #-1]!
	strbge	r3, [r0, #-1]!
	itt	gt
	ldrbgt	r3, [r1, #-1]!
	strbgt	r3, [r0, #-1]!
#else
	ldrgeb	r3, [r1, #-1]!
	strgeb	r3, [r0, #-1]!
	ldrgtb	r3, [r1, #-1]!
	strgtb	r3, [r0, #-1]!
#endif
	subs	r2, r2, r12
	blt	.Lmemcpy_bl4		/* less than 4 bytes to go */
	ands	r12, r1, #3
	beq	.Lmemcpy_bt8		/* we have an aligned source */

	/* erg - unaligned source */
	/* This is where it gets nasty ... */
.Lmemcpy_bsrcul:
	bic	r1, r1, #3
	ldr	r3, [r1, #0]
	cmp	r12, #2
	blt	.Lmemcpy_bsrcul1
	beq	.Lmemcpy_bsrcul2
	cmp	r2, #0x0c            
	blt	.Lmemcpy_bsrcul3loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5, lr}

.Lmemcpy_bsrcul3loop16:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	lr, r3, lsr #8
	ldmdb	r1!, {r3-r5, r12}
	orr	lr, lr, r12, lsl #24
	mov	r12, r12, lsr #8
	orr	r12, r12, r5, lsl #24
	mov	r5, r5, lsr #8
	orr	r5, r5, r4, lsl #24
	mov	r4, r4, lsr #8
	orr	r4, r4, r3, lsl #24
#else
	mov	lr, r3, lsl #8
	ldmdb	r1!, {r3-r5, r12}
	orr	lr, lr, r12, lsr #24
	mov	r12, r12, lsl #8
	orr	r12, r12, r5, lsr #24
	mov	r5, r5, lsl #8
	orr	r5, r5, r4, lsr #24
	mov	r4, r4, lsl #8
	orr	r4, r4, r3, lsr #24
#endif
	stmdb	r0!, {r4, r5, r12, lr}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_bsrcul3loop16
	ldmia	sp!, {r4, r5, lr}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_bsrcul3l4

.Lmemcpy_bsrcul3loop4:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r12, r3, lsr #8
	ldr	r3, [r1, #-4]!
	orr	r12, r12, r3, lsl #24
#else
	mov	r12, r3, lsl #8
	ldr	r3, [r1, #-4]!
	orr	r12, r12, r3, lsr #24
#endif
	str	r12, [r0, #-4]!
	subs	r2, r2, #4
	bge	.Lmemcpy_bsrcul3loop4

.Lmemcpy_bsrcul3l4:
	add	r1, r1, #3
	b	.Lmemcpy_bl4

.Lmemcpy_bsrcul2:
	cmp	r2, #0x0c            
	blt	.Lmemcpy_bsrcul2loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5, lr}

.Lmemcpy_bsrcul2loop16:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	lr, r3, lsr #16
	ldmdb	r1!, {r3-r5, r12}
	orr	lr, lr, r12, lsl #16
	mov	r12, r12, lsr #16
	orr	r12, r12, r5, lsl #16
	mov	r5, r5, lsr #16
	orr	r5, r5, r4, lsl #16
	mov	r4, r4, lsr #16
	orr	r4, r4, r3, lsl #16
#else
	mov	lr, r3, lsl #16
	ldmdb	r1!, {r3-r5, r12}
	orr	lr, lr, r12, lsr #16
	mov	r12, r12, lsl #16
	orr	r12, r12, r5, lsr #16
	mov	r5, r5, lsl #16
	orr	r5, r5, r4, lsr #16
	mov	r4, r4, lsl #16
	orr	r4, r4, r3, lsr #16
#endif
	stmdb	r0!, {r4, r5, r12, lr}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_bsrcul2loop16
	ldmia	sp!, {r4, r5, lr}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_bsrcul2l4

.Lmemcpy_bsrcul2loop4:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r12, r3, lsr #16
	ldr	r3, [r1, #-4]!
	orr	r12, r12, r3, lsl #16
#else
	mov	r12, r3, lsl #16
	ldr	r3, [r1, #-4]!
	orr	r12, r12, r3, lsr #16
#endif
	str	r12, [r0, #-4]!
	subs	r2, r2, #4
	bge	.Lmemcpy_bsrcul2loop4

.Lmemcpy_bsrcul2l4:
	add	r1, r1, #2
	b	.Lmemcpy_bl4

.Lmemcpy_bsrcul1:
	cmp	r2, #0x0c            
	blt	.Lmemcpy_bsrcul1loop4
	sub	r2, r2, #0x0c         
	stmdb	sp!, {r4, r5, lr}

.Lmemcpy_bsrcul1loop32:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	lr, r3, lsr #24
	ldmdb	r1!, {r3-r5, r12}
	orr	lr, lr, r12, lsl #8
	mov	r12, r12, lsr #24
	orr	r12, r12, r5, lsl #8
	mov	r5, r5, lsr #24
	orr	r5, r5, r4, lsl #8
	mov	r4, r4, lsr #24
	orr	r4, r4, r3, lsl #8
#else
	mov	lr, r3, lsl #24
	ldmdb	r1!, {r3-r5, r12}
	orr	lr, lr, r12, lsr #8
	mov	r12, r12, lsl #24
	orr	r12, r12, r5, lsr #8
	mov	r5, r5, lsl #24
	orr	r5, r5, r4, lsr #8
	mov	r4, r4, lsl #24
	orr	r4, r4, r3, lsr #8
#endif
	stmdb	r0!, {r4, r5, r12, lr}
	subs	r2, r2, #0x10         
	bge	.Lmemcpy_bsrcul1loop32
	ldmia	sp!, {r4, r5, lr}
	adds	r2, r2, #0x0c         
	blt	.Lmemcpy_bsrcul1l4

.Lmemcpy_bsrcul1loop4:
#if __BYTE_ORDER == __BIG_ENDIAN
	mov	r12, r3, lsr #24
	ldr	r3, [r1, #-4]!
	orr	r12, r12, r3, lsl #8
#else
	mov	r12, r3, lsl #24
	ldr	r3, [r1, #-4]!
	orr	r12, r12, r3, lsr #8
#endif
	str	r12, [r0, #-4]!
	subs	r2, r2, #4
	bge	.Lmemcpy_bsrcul1loop4

.Lmemcpy_bsrcul1l4:
	add	r1, r1, #1
	b	.Lmemcpy_bl4

#else /* THUMB1_ONLY */

/* This is a fairly dumb implementation for when we can't use the 32-bit code
   above.  */
.text
.global _memcpy
.hidden _memcpy
.type _memcpy,%function
.align 4
.thumb
_memcpy:
	push	{r0, r4}
	cmp	r2, #0
	beq	.Lmemcpy_exit
	@ See if we have overlapping regions, and need to reverse the
	@ direction of the copy
	cmp	r0, r1
	bls	.Lmemcpy_forwards
	add	r4, r1, r2
	cmp	r0, r4
	bcc	.Lmemcpy_backwards
.Lmemcpy_forwards:
	/* Forwards.  */
	mov	r3, r0
	eor	r3, r1
	mov	r4, #3
	tst	r3, r4
	bne	.Lmemcpy_funaligned
	cmp	r2, #8
	bcc	.Lmemcpy_funaligned
1:	@ copy up to the first word boundary.
	tst	r0, r4
	beq	1f
	ldrb	r3, [r1]
	add	r1, r1, #1
	strb	r3, [r0]
	add	r0, r0, #1
	sub	r2, r2, #1
	b	1b
1:	@ Copy aligned words
	ldr	r3, [r1]
	add	r1, r1, #4
	str	r3, [r0]
	add	r0, r0, #4
	sub	r2, r2, #4
	cmp	r2, #4
	bcs	1b
	cmp	r2, #0
	beq	.Lmemcpy_exit
.Lmemcpy_funaligned:
1:
	ldrb	r3, [r1]
	add	r1, r1, #1
	strb	r3, [r0]
	add	r0, r0, #1
	sub	r2, r2, #1
	bne	1b
.Lmemcpy_exit:
	pop	{r0, r4}
	bx	lr

.Lmemcpy_backwards:
	add	r0, r0, r2
	add	r1, r1, r2
1:
	sub	r0, r0, #1
	sub	r1, r1, #1
	ldrb	r3, [r1]
	strb	r3, [r0]
	sub	r2, r2, #1
	bne	1b
	b	.Lmemcpy_exit
#endif