summaryrefslogtreecommitdiff
path: root/sysdeps/sparc/sparc64/sparcv9v2/memcpy.S
blob: b261f461a4ccfbbd800f272cbedcee4ea0066e51 (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
/* Copy SIZE bytes from SRC to DEST.  For SUN4V Niagara-2.
   Copyright (C) 2007, 2008 Free Software Foundation, Inc.
   This file is part of the GNU C Library.
   Contributed by David S. Miller (davem@davemloft.net)

   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.  */

#include <sysdep.h>

#define ASI_BLK_INIT_QUAD_LDD_P	0xe2
#define ASI_BLK_P		0xf0
#define ASI_P			0x80
#define ASI_PNF			0x82

#define FPRS_FEF		0x04

#define VISEntryHalf			\
	rd	%fprs, %o5;		\
	wr	%g0, FPRS_FEF, %fprs

#define VISExitHalf			\
	and	%o5, FPRS_FEF, %o5;	\
	wr	%o5, 0x0, %fprs

#define STORE_ASI		ASI_BLK_INIT_QUAD_LDD_P

#define LOAD(type,addr,dest)	type [addr], dest
#define LOAD_BLK(addr,dest)	ldda [addr] ASI_BLK_P, dest
#define STORE(type,src,addr)	type src, [addr]
#define STORE_BLK(src,addr)	stda src, [addr] ASI_BLK_P
#define STORE_INIT(src,addr)	stxa src, [addr] STORE_ASI

#ifndef XCC
#define USE_BPR
#define XCC xcc
#endif

#define FREG_FROB(x0, x1, x2, x3, x4, x5, x6, x7, x8) \
	faligndata	%x0, %x1, %f0; \
	faligndata	%x1, %x2, %f2; \
	faligndata	%x2, %x3, %f4; \
	faligndata	%x3, %x4, %f6; \
	faligndata	%x4, %x5, %f8; \
	faligndata	%x5, %x6, %f10; \
	faligndata	%x6, %x7, %f12; \
	faligndata	%x7, %x8, %f14;

#define FREG_MOVE_1(x0) \
	fmovd		%x0, %f0;
#define FREG_MOVE_2(x0, x1) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2;
#define FREG_MOVE_3(x0, x1, x2) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2; \
	fmovd		%x2, %f4;
#define FREG_MOVE_4(x0, x1, x2, x3) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2; \
	fmovd		%x2, %f4; \
	fmovd		%x3, %f6;
#define FREG_MOVE_5(x0, x1, x2, x3, x4) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2; \
	fmovd		%x2, %f4; \
	fmovd		%x3, %f6; \
	fmovd		%x4, %f8;
#define FREG_MOVE_6(x0, x1, x2, x3, x4, x5) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2; \
	fmovd		%x2, %f4; \
	fmovd		%x3, %f6; \
	fmovd		%x4, %f8; \
	fmovd		%x5, %f10;
#define FREG_MOVE_7(x0, x1, x2, x3, x4, x5, x6) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2; \
	fmovd		%x2, %f4; \
	fmovd		%x3, %f6; \
	fmovd		%x4, %f8; \
	fmovd		%x5, %f10; \
	fmovd		%x6, %f12;
#define FREG_MOVE_8(x0, x1, x2, x3, x4, x5, x6, x7) \
	fmovd		%x0, %f0; \
	fmovd		%x1, %f2; \
	fmovd		%x2, %f4; \
	fmovd		%x3, %f6; \
	fmovd		%x4, %f8; \
	fmovd		%x5, %f10; \
	fmovd		%x6, %f12; \
	fmovd		%x7, %f14;
#define FREG_LOAD_1(base, x0) \
	LOAD(ldd, base + 0x00, %x0)
#define FREG_LOAD_2(base, x0, x1) \
	LOAD(ldd, base + 0x00, %x0); \
	LOAD(ldd, base + 0x08, %x1);
#define FREG_LOAD_3(base, x0, x1, x2) \
	LOAD(ldd, base + 0x00, %x0); \
	LOAD(ldd, base + 0x08, %x1); \
	LOAD(ldd, base + 0x10, %x2);
#define FREG_LOAD_4(base, x0, x1, x2, x3) \
	LOAD(ldd, base + 0x00, %x0); \
	LOAD(ldd, base + 0x08, %x1); \
	LOAD(ldd, base + 0x10, %x2); \
	LOAD(ldd, base + 0x18, %x3);
#define FREG_LOAD_5(base, x0, x1, x2, x3, x4) \
	LOAD(ldd, base + 0x00, %x0); \
	LOAD(ldd, base + 0x08, %x1); \
	LOAD(ldd, base + 0x10, %x2); \
	LOAD(ldd, base + 0x18, %x3); \
	LOAD(ldd, base + 0x20, %x4);
#define FREG_LOAD_6(base, x0, x1, x2, x3, x4, x5) \
	LOAD(ldd, base + 0x00, %x0); \
	LOAD(ldd, base + 0x08, %x1); \
	LOAD(ldd, base + 0x10, %x2); \
	LOAD(ldd, base + 0x18, %x3); \
	LOAD(ldd, base + 0x20, %x4); \
	LOAD(ldd, base + 0x28, %x5);
#define FREG_LOAD_7(base, x0, x1, x2, x3, x4, x5, x6) \
	LOAD(ldd, base + 0x00, %x0); \
	LOAD(ldd, base + 0x08, %x1); \
	LOAD(ldd, base + 0x10, %x2); \
	LOAD(ldd, base + 0x18, %x3); \
	LOAD(ldd, base + 0x20, %x4); \
	LOAD(ldd, base + 0x28, %x5); \
	LOAD(ldd, base + 0x30, %x6);

	.register	%g2,#scratch
	.register	%g3,#scratch
	.register	%g6,#scratch

	.text
	.align		32

ENTRY(bcopy)
	sub		%o1, %o0, %o4
	mov		%o0, %g4
	cmp		%o4, %o2
	mov		%o1, %o0
	bgeu,pt		%XCC, 100f
	 mov		%g4, %o1
#ifndef USE_BPR
	srl		%o2, 0, %o2
#endif
	brnz,pn		%o2, 220f
	 add		%o0, %o2, %o0
	retl
	 nop
END(bcopy)

	.align		32
ENTRY(memcpy)
#ifndef USE_BPR
	srl		%o2, 0, %o2
#endif
100:	/* %o0=dst, %o1=src, %o2=len */
	mov		%o0, %g5
	cmp		%o2, 0
	be,pn		%XCC, 85f
218:	 or		%o0, %o1, %o3
	cmp		%o2, 16
	blu,a,pn	%XCC, 80f
	 or		%o3, %o2, %o3

	/* 2 blocks (128 bytes) is the minimum we can do the block
	 * copy with.  We need to ensure that we'll iterate at least
	 * once in the block copy loop.  At worst we'll need to align
	 * the destination to a 64-byte boundary which can chew up
	 * to (64 - 1) bytes from the length before we perform the
	 * block copy loop.
	 *
	 * However, the cut-off point, performance wise, is around
	 * 4 64-byte blocks.
	 */
	cmp		%o2, (4 * 64)
	blu,pt		%XCC, 75f
	 andcc		%o3, 0x7, %g0

	/* %o0:	dst
	 * %o1:	src
	 * %o2:	len  (known to be >= 128)
	 *
	 * The block copy loops can use %o4, %g2, %g3 as
	 * temporaries while copying the data.  %o5 must
	 * be preserved between VISEntryHalf and VISExitHalf
	 */

	LOAD(prefetch, %o1 + 0x000, #one_read)
	LOAD(prefetch, %o1 + 0x040, #one_read)
	LOAD(prefetch, %o1 + 0x080, #one_read)

	/* Align destination on 64-byte boundary.  */
	andcc		%o0, (64 - 1), %o4
	be,pt		%XCC, 2f
	 sub		%o4, 64, %o4
	sub		%g0, %o4, %o4	! bytes to align dst
	sub		%o2, %o4, %o2
1:	subcc		%o4, 1, %o4
	LOAD(ldub, %o1, %g1)
	STORE(stb, %g1, %o0)
	add		%o1, 1, %o1
	bne,pt		%XCC, 1b
	add		%o0, 1, %o0

2:
	/* Clobbers o5/g1/g2/g3/g7/icc/xcc.  We must preserve
	 * o5 from here until we hit VISExitHalf.
	 */
	VISEntryHalf

	alignaddr	%o1, %g0, %g0

	add		%o1, (64 - 1), %o4
	andn		%o4, (64 - 1), %o4
	andn		%o2, (64 - 1), %g1
	sub		%o2, %g1, %o2

	and		%o1, (64 - 1), %g2
	add		%o1, %g1, %o1
	sub		%o0, %o4, %g3
	brz,pt		%g2, 190f
	 cmp		%g2, 32
	blu,a		5f
	 cmp		%g2, 16
	cmp		%g2, 48
	blu,a		4f
	 cmp		%g2, 40
	cmp		%g2, 56
	blu		170f
	 nop
	ba,a,pt		%xcc, 180f

4:	/* 32 <= low bits < 48 */
	blu		150f
	 nop
	ba,a,pt		%xcc, 160f
5:	/* 0 < low bits < 32 */
	blu,a		6f
	 cmp		%g2, 8
	cmp		%g2, 24
	blu		130f
	 nop
	ba,a,pt		%xcc, 140f
6:	/* 0 < low bits < 16 */
	bgeu		120f
	 nop
	/* fall through for 0 < low bits < 8 */
110:	sub		%o4, 64, %g2
	LOAD_BLK(%g2, %f0)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f4, f6, f8, f10, f12, f14, f16)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_8(f16, f18, f20, f22, f24, f26, f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

120:	sub		%o4, 56, %g2
	FREG_LOAD_7(%g2, f0, f2, f4, f6, f8, f10, f12)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f4, f6, f8, f10, f12, f16, f18)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_7(f18, f20, f22, f24, f26, f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

130:	sub		%o4, 48, %g2
	FREG_LOAD_6(%g2, f0, f2, f4, f6, f8, f10)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f4, f6, f8, f10, f16, f18, f20)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_6(f20, f22, f24, f26, f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

140:	sub		%o4, 40, %g2
	FREG_LOAD_5(%g2, f0, f2, f4, f6, f8)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f4, f6, f8, f16, f18, f20, f22)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_5(f22, f24, f26, f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

150:	sub		%o4, 32, %g2
	FREG_LOAD_4(%g2, f0, f2, f4, f6)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f4, f6, f16, f18, f20, f22, f24)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_4(f24, f26, f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

160:	sub		%o4, 24, %g2
	FREG_LOAD_3(%g2, f0, f2, f4)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f4, f16, f18, f20, f22, f24, f26)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_3(f26, f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

170:	sub		%o4, 16, %g2
	FREG_LOAD_2(%g2, f0, f2)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f2, f16, f18, f20, f22, f24, f26, f28)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_2(f28, f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

180:	sub		%o4, 8, %g2
	FREG_LOAD_1(%g2, f0)
1:	STORE_INIT(%g0, %o4 + %g3)
	LOAD_BLK(%o4, %f16)
	FREG_FROB(f0, f16, f18, f20, f22, f24, f26, f28, f30)
	STORE_BLK(%f0, %o4 + %g3)
	FREG_MOVE_1(f30)
	subcc		%g1, 64, %g1
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)
	ba,pt		%xcc, 195f
	 nop

190:
1:	STORE_INIT(%g0, %o4 + %g3)
	subcc		%g1, 64, %g1
	LOAD_BLK(%o4, %f0)
	STORE_BLK(%f0, %o4 + %g3)
	add		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 LOAD(prefetch, %o4 + 64, #one_read)

195:
	add		%o4, %g3, %o0
	membar		#Sync

	VISExitHalf

	/* %o2 contains any final bytes still needed to be copied
	 * over. If anything is left, we copy it one byte at a time.
	 */
	brz,pt		%o2, 85f
	 sub		%o0, %o1, %o3
	ba,a,pt		%XCC, 90f

	.align		64
75: /* 16 < len <= 64 */
	bne,pn		%XCC, 75f
	 sub		%o0, %o1, %o3

72:
	andn		%o2, 0xf, %o4
	and		%o2, 0xf, %o2
1:	subcc		%o4, 0x10, %o4
	LOAD(ldx, %o1, %o5)
	add		%o1, 0x08, %o1
	LOAD(ldx, %o1, %g1)
	sub		%o1, 0x08, %o1
	STORE(stx, %o5, %o1 + %o3)
	add		%o1, 0x8, %o1
	STORE(stx, %g1, %o1 + %o3)
	bgu,pt		%XCC, 1b
	 add		%o1, 0x8, %o1
73:	andcc		%o2, 0x8, %g0
	be,pt		%XCC, 1f
	 nop
	sub		%o2, 0x8, %o2
	LOAD(ldx, %o1, %o5)
	STORE(stx, %o5, %o1 + %o3)
	add		%o1, 0x8, %o1
1:	andcc		%o2, 0x4, %g0
	be,pt		%XCC, 1f
	 nop
	sub		%o2, 0x4, %o2
	LOAD(lduw, %o1, %o5)
	STORE(stw, %o5, %o1 + %o3)
	add		%o1, 0x4, %o1
1:	cmp		%o2, 0
	be,pt		%XCC, 85f
	 nop
	ba,pt		%xcc, 90f
	 nop

75:
	andcc		%o0, 0x7, %g1
	sub		%g1, 0x8, %g1
	be,pn		%icc, 2f
	 sub		%g0, %g1, %g1
	sub		%o2, %g1, %o2

1:	subcc		%g1, 1, %g1
	LOAD(ldub, %o1, %o5)
	STORE(stb, %o5, %o1 + %o3)
	bgu,pt		%icc, 1b
	 add		%o1, 1, %o1

2:	add		%o1, %o3, %o0
	andcc		%o1, 0x7, %g1
	bne,pt		%icc, 8f
	 sll		%g1, 3, %g1

	cmp		%o2, 16
	bgeu,pt		%icc, 72b
	 nop
	ba,a,pt		%xcc, 73b

8:	mov		64, %o3
	andn		%o1, 0x7, %o1
	LOAD(ldx, %o1, %g2)
	sub		%o3, %g1, %o3
	andn		%o2, 0x7, %o4
	sllx		%g2, %g1, %g2
1:	add		%o1, 0x8, %o1
	LOAD(ldx, %o1, %g3)
	subcc		%o4, 0x8, %o4
	srlx		%g3, %o3, %o5
	or		%o5, %g2, %o5
	STORE(stx, %o5, %o0)
	add		%o0, 0x8, %o0
	bgu,pt		%icc, 1b
	 sllx		%g3, %g1, %g2

	srl		%g1, 3, %g1
	andcc		%o2, 0x7, %o2
	be,pn		%icc, 85f
	 add		%o1, %g1, %o1
	ba,pt		%xcc, 90f
	 sub		%o0, %o1, %o3

	.align		64
80: /* 0 < len <= 16 */
	andcc		%o3, 0x3, %g0
	bne,pn		%XCC, 90f
	 sub		%o0, %o1, %o3

1:
	subcc		%o2, 4, %o2
	LOAD(lduw, %o1, %g1)
	STORE(stw, %g1, %o1 + %o3)
	bgu,pt		%XCC, 1b
	 add		%o1, 4, %o1

85:	retl
	 mov		%g5, %o0

	.align		32
90:
	subcc		%o2, 1, %o2
	LOAD(ldub, %o1, %g1)
	STORE(stb, %g1, %o1 + %o3)
	bgu,pt		%XCC, 90b
	 add		%o1, 1, %o1
	retl
	 mov		%g5, %o0

END(memcpy)

#define RMOVE_BIGCHUNK(src, dst, offset, t0, t1, t2, t3)	\
	ldx		[%src - offset - 0x20], %t0; 		\
	ldx		[%src - offset - 0x18], %t1; 		\
	ldx		[%src - offset - 0x10], %t2; 		\
	ldx		[%src - offset - 0x08], %t3; 		\
	stw		%t0, [%dst - offset - 0x1c]; 		\
	srlx		%t0, 32, %t0;				\
	stw		%t0, [%dst - offset - 0x20]; 		\
	stw		%t1, [%dst - offset - 0x14]; 		\
	srlx		%t1, 32, %t1;				\
	stw		%t1, [%dst - offset - 0x18]; 		\
	stw		%t2, [%dst - offset - 0x0c]; 		\
	srlx		%t2, 32, %t2;				\
	stw		%t2, [%dst - offset - 0x10]; 		\
	stw		%t3, [%dst - offset - 0x04];		\
	srlx		%t3, 32, %t3;				\
	stw		%t3, [%dst - offset - 0x08];

#define RMOVE_BIGALIGNCHUNK(src, dst, offset, t0, t1, t2, t3)	\
	ldx		[%src - offset - 0x20], %t0; 		\
	ldx		[%src - offset - 0x18], %t1; 		\
	ldx		[%src - offset - 0x10], %t2; 		\
	ldx		[%src - offset - 0x08], %t3; 		\
	stx		%t0, [%dst - offset - 0x20]; 		\
	stx		%t1, [%dst - offset - 0x18]; 		\
	stx		%t2, [%dst - offset - 0x10]; 		\
	stx		%t3, [%dst - offset - 0x08];		\
	ldx		[%src - offset - 0x40], %t0; 		\
	ldx		[%src - offset - 0x38], %t1; 		\
	ldx		[%src - offset - 0x30], %t2; 		\
	ldx		[%src - offset - 0x28], %t3; 		\
	stx		%t0, [%dst - offset - 0x40]; 		\
	stx		%t1, [%dst - offset - 0x38]; 		\
	stx		%t2, [%dst - offset - 0x30]; 		\
	stx		%t3, [%dst - offset - 0x28];

#define RMOVE_LASTCHUNK(src, dst, offset, t0, t1, t2, t3)	\
	ldx		[%src + offset + 0x00], %t0;		\
	ldx		[%src + offset + 0x08], %t1; 		\
	stw		%t0, [%dst + offset + 0x04]; 		\
	srlx		%t0, 32, %t2;				\
	stw		%t2, [%dst + offset + 0x00]; 		\
	stw		%t1, [%dst + offset + 0x0c]; 		\
	srlx		%t1, 32, %t3;				\
	stw		%t3, [%dst + offset + 0x08];

#define RMOVE_LASTALIGNCHUNK(src, dst, offset, t0, t1)		\
	ldx		[%src + offset + 0x00], %t0; 		\
	ldx		[%src + offset + 0x08], %t1; 		\
	stx		%t0, [%dst + offset + 0x00]; 		\
	stx		%t1, [%dst + offset + 0x08];

	.align		32
228:	andcc		%o2, 1, %g0
	be,pt		%icc, 2f+4
1:	 ldub		[%o1 - 1], %o5
	sub		%o1, 1, %o1
	sub		%o0, 1, %o0
	subcc		%o2, 1, %o2
	be,pn		%XCC, 229f
	 stb		%o5, [%o0]
2:	ldub		[%o1 - 1], %o5
	sub		%o0, 2, %o0
	ldub		[%o1 - 2], %g5
	sub		%o1, 2, %o1
	subcc		%o2, 2, %o2
	stb		%o5, [%o0 + 1]
	bne,pt		%XCC, 2b
	 stb		%g5, [%o0]
229:	retl
	 mov		%g4, %o0
out:	retl
	 mov		%g5, %o0

	.align		32
ENTRY(memmove)
	mov		%o0, %g5
#ifndef USE_BPR
	srl		%o2, 0, %o2
#endif
	brz,pn		%o2, out
	 sub		%o0, %o1, %o4
	cmp		%o4, %o2
	bgeu,pt		%XCC, 218b
	 mov		%o0, %g4
	add		%o0, %o2, %o0
220:	add		%o1, %o2, %o1
	cmp		%o2, 15
	bleu,pn		%XCC, 228b
	 andcc		%o0, 7, %g2
	sub		%o0, %o1, %g5
	andcc		%g5, 3, %o5
	bne,pn		%XCC, 232f
	 andcc		%o1, 3, %g0
	be,a,pt		%XCC, 236f
	 andcc		%o1, 4, %g0
	andcc		%o1, 1, %g0
	be,pn		%XCC, 4f
	 andcc		%o1, 2, %g0
	ldub		[%o1 - 1], %g2
	sub		%o1, 1, %o1
	sub		%o0, 1, %o0
	sub		%o2, 1, %o2
	be,pn		%XCC, 5f
	 stb		%g2, [%o0]
4:	lduh		[%o1 - 2], %g2
	sub		%o1, 2, %o1
	sub		%o0, 2, %o0
	sub		%o2, 2, %o2
	sth		%g2, [%o0]
5:	andcc		%o1, 4, %g0
236:	be,a,pn		%XCC, 2f
	 andcc		%o2, -128, %g6
	lduw		[%o1 - 4], %g5
	sub		%o1, 4, %o1
	sub		%o0, 4, %o0
	sub		%o2, 4, %o2
	stw		%g5, [%o0]
	andcc		%o2, -128, %g6
2:	be,pn		%XCC, 235f
	 andcc		%o0, 4, %g0
	be,pn		%XCC, 282f + 4
5:	RMOVE_BIGCHUNK(o1, o0, 0x00, g1, g3, g5, o5)
	RMOVE_BIGCHUNK(o1, o0, 0x20, g1, g3, g5, o5)
	RMOVE_BIGCHUNK(o1, o0, 0x40, g1, g3, g5, o5)
	RMOVE_BIGCHUNK(o1, o0, 0x60, g1, g3, g5, o5)
	subcc		%g6, 128, %g6
	sub		%o1, 128, %o1
	bne,pt		%XCC, 5b
	 sub		%o0, 128, %o0
235:	andcc		%o2, 0x70, %g6
41:	be,pn		%XCC, 280f
	 andcc		%o2, 8, %g0

279:	rd		%pc, %o5
	sll		%g6, 1, %g5
	sub		%o1, %g6, %o1
	sub		%o5, %g5, %o5
	jmpl		%o5 + %lo(280f - 279b), %g0
	 sub		%o0, %g6, %o0
	RMOVE_LASTCHUNK(o1, o0, 0x60, g2, g3, g5, o5)
	RMOVE_LASTCHUNK(o1, o0, 0x50, g2, g3, g5, o5)
	RMOVE_LASTCHUNK(o1, o0, 0x40, g2, g3, g5, o5)
	RMOVE_LASTCHUNK(o1, o0, 0x30, g2, g3, g5, o5)
	RMOVE_LASTCHUNK(o1, o0, 0x20, g2, g3, g5, o5)
	RMOVE_LASTCHUNK(o1, o0, 0x10, g2, g3, g5, o5)
	RMOVE_LASTCHUNK(o1, o0, 0x00, g2, g3, g5, o5)
280:	be,pt		%XCC, 281f
	 andcc		%o2, 4, %g0
	ldx		[%o1 - 8], %g2
	sub		%o0, 8, %o0
	stw		%g2, [%o0 + 4]
	sub		%o1, 8, %o1
	srlx		%g2, 32, %g2
	stw		%g2, [%o0]
281:	be,pt		%XCC, 1f
	 andcc		%o2, 2, %g0
	lduw		[%o1 - 4], %g2
	sub		%o1, 4, %o1
	stw		%g2, [%o0 - 4]
	sub		%o0, 4, %o0
1:	be,pt		%XCC, 1f
	 andcc		%o2, 1, %g0
	lduh		[%o1 - 2], %g2
	sub		%o1, 2, %o1
	sth		%g2, [%o0 - 2]
	sub		%o0, 2, %o0
1:	be,pt		%XCC, 211f
	 nop
	ldub		[%o1 - 1], %g2
	stb		%g2, [%o0 - 1]
211:	retl
	 mov		%g4, %o0

282:	RMOVE_BIGALIGNCHUNK(o1, o0, 0x00, g1, g3, g5, o5)
	RMOVE_BIGALIGNCHUNK(o1, o0, 0x40, g1, g3, g5, o5)
	subcc		%g6, 128, %g6
	sub		%o1, 128, %o1
	bne,pt		%XCC, 282b
	 sub		%o0, 128, %o0
	andcc		%o2, 0x70, %g6
	be,pn		%XCC, 284f
	 andcc		%o2, 8, %g0

283:	rd		%pc, %o5
	sub		%o1, %g6, %o1
	sub		%o5, %g6, %o5
	jmpl		%o5 + %lo(284f - 283b), %g0
	 sub		%o0, %g6, %o0
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x60, g2, g3)
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x50, g2, g3)
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x40, g2, g3)
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x30, g2, g3)
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x20, g2, g3)
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x10, g2, g3)
	RMOVE_LASTALIGNCHUNK(o1, o0, 0x00, g2, g3)
284:	be,pt		%XCC, 285f
	 andcc		%o2, 4, %g0
	ldx		[%o1 - 8], %g2
	sub		%o0, 8, %o0
	sub		%o1, 8, %o1
	stx		%g2, [%o0]
285:	be,pt		%XCC, 1f
	 andcc		%o2, 2, %g0
	lduw		[%o1 - 4], %g2
	sub		%o0, 4, %o0
	sub		%o1, 4, %o1
	stw		%g2, [%o0]
1:	be,pt		%XCC, 1f
	 andcc		%o2, 1, %g0
	lduh		[%o1 - 2], %g2
	sub		%o0, 2, %o0
	sub		%o1, 2, %o1
	sth		%g2, [%o0]
1:	be,pt		%XCC, 1f
	 nop
	ldub		[%o1 - 1], %g2
	stb		%g2, [%o0 - 1]
1:	retl
	 mov		%g4, %o0

232:	ldub		[%o1 - 1], %g5
	sub		%o1, 1, %o1
	sub		%o0, 1, %o0
	subcc		%o2, 1, %o2
	bne,pt		%XCC, 232b
	 stb		%g5, [%o0]
234:	retl
	 mov		%g4, %o0
END(memmove)

#ifdef USE_BPR
weak_alias (memcpy, __align_cpy_1)
weak_alias (memcpy, __align_cpy_2)
weak_alias (memcpy, __align_cpy_4)
weak_alias (memcpy, __align_cpy_8)
weak_alias (memcpy, __align_cpy_16)
#endif
libc_hidden_builtin_def (memcpy)
libc_hidden_builtin_def (memmove)