about summary refs log tree commit diff
path: root/sysdeps/sparc/sparc64/multiarch/memcpy-niagara1.S
blob: a3b69f9ef1934c09a8f7283432eb7990fccbb46f (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
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
/* Copy SIZE bytes from SRC to DEST.  For SUN4V Niagara.
   Copyright (C) 2006-2016 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, see
   <http://www.gnu.org/licenses/>.  */

#include <sysdep.h>

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

#define LOAD(type,addr,dest)	type##a [addr] ASI_P, dest
#define LOAD_TWIN(addr_reg,dest0,dest1)	\
	ldda [addr_reg] ASI_BLK_INIT_QUAD_LDD_P, dest0

#define STORE(type,src,addr)	type src, [addr]
#define STORE_INIT(src,addr)	stxa src, [addr] %asi

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

#if IS_IN (libc)

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

	.text

ENTRY(__mempcpy_niagara1)
	ba,pt		%XCC, 101f
	 add		%o0, %o2, %g5
END(__mempcpy_niagara1)

	.align		32
ENTRY(__memcpy_niagara1)
100:	/* %o0=dst, %o1=src, %o2=len */
	mov		%o0, %g5
101:
# ifndef USE_BPR
	srl		%o2, 0, %o2
# endif
	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.
	 */
	cmp		%o2, (2 * 64)
	blu,pt		%XCC, 70f
	 andcc		%o3, 0x7, %g0

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

	LOAD(prefetch, %o1, #one_read)
	wr		%g0, ASI_BLK_INIT_QUAD_LDD_P, %asi

	/* 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

	/* If the source is on a 16-byte boundary we can do
	 * the direct block copy loop.  If it is 8-byte aligned
	 * we can do the 16-byte loads offset by -8 bytes and the
	 * init stores offset by one register.
	 *
	 * If the source is not even 8-byte aligned, we need to do
	 * shifting and masking (basically integer faligndata).
	 *
	 * The careful bit with init stores is that if we store
	 * to any part of the cache line we have to store the whole
	 * cacheline else we can end up with corrupt L2 cache line
	 * contents.  Since the loop works on 64-bytes of 64-byte
	 * aligned store data at a time, this is easy to ensure.
	 */
2:
	andcc		%o1, (16 - 1), %o4
	andn		%o2, (64 - 1), %g1	! block copy loop iterator
	sub		%o2, %g1, %o2		! final sub-block copy bytes
	be,pt		%XCC, 50f
	 cmp		%o4, 8
	be,a,pt		%XCC, 10f
	 sub		%o1, 0x8, %o1

	/* Neither 8-byte nor 16-byte aligned, shift and mask.  */
	mov		%g1, %o4
	and		%o1, 0x7, %g1
	sll		%g1, 3, %g1
	mov		64, %o3
	andn		%o1, 0x7, %o1
	LOAD(ldx, %o1, %g2)
	sub		%o3, %g1, %o3
	sllx		%g2, %g1, %g2

#define SWIVEL_ONE_DWORD(SRC, TMP1, TMP2, PRE_VAL, PRE_SHIFT, POST_SHIFT, DST)\
	LOAD(ldx, SRC, TMP1); \
	srlx		TMP1, PRE_SHIFT, TMP2; \
	or		TMP2, PRE_VAL, TMP2; \
	STORE_INIT(TMP2, DST); \
	sllx		TMP1, POST_SHIFT, PRE_VAL;

1:	add		%o1, 0x8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x00)
	add		%o1, 0x8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x08)
	add		%o1, 0x8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x10)
	add		%o1, 0x8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x18)
	add		%o1, 32, %o1
	LOAD(prefetch, %o1, #one_read)
	sub		%o1, 32 - 8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x20)
	add		%o1, 8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x28)
	add		%o1, 8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x30)
	add		%o1, 8, %o1
	SWIVEL_ONE_DWORD(%o1, %g3, %o5, %g2, %o3, %g1, %o0 + 0x38)
	subcc		%o4, 64, %o4
	bne,pt		%XCC, 1b
	 add		%o0, 64, %o0

#undef SWIVEL_ONE_DWORD

	srl		%g1, 3, %g1
	ba,pt		%XCC, 60f
	 add		%o1, %g1, %o1

10:	/* Destination is 64-byte aligned, source was only 8-byte
	 * aligned but it has been subtracted by 8 and we perform
	 * one twin load ahead, then add 8 back into source when
	 * we finish the loop.
	 */
	LOAD_TWIN(%o1, %o4, %o5)
1:	add		%o1, 16, %o1
	LOAD_TWIN(%o1, %g2, %g3)
	add		%o1, 16 + 32, %o1
	LOAD(prefetch, %o1, #one_read)
	sub		%o1, 32, %o1
	STORE_INIT(%o5, %o0 + 0x00)		! initializes cache line
	STORE_INIT(%g2, %o0 + 0x08)
	LOAD_TWIN(%o1, %o4, %o5)
	add		%o1, 16, %o1
	STORE_INIT(%g3, %o0 + 0x10)
	STORE_INIT(%o4, %o0 + 0x18)
	LOAD_TWIN(%o1, %g2, %g3)
	add		%o1, 16, %o1
	STORE_INIT(%o5, %o0 + 0x20)
	STORE_INIT(%g2, %o0 + 0x28)
	LOAD_TWIN(%o1, %o4, %o5)
	STORE_INIT(%g3, %o0 + 0x30)
	STORE_INIT(%o4, %o0 + 0x38)
	subcc		%g1, 64, %g1
	bne,pt		%XCC, 1b
	 add		%o0, 64, %o0

	ba,pt		%XCC, 60f
	 add		%o1, 0x8, %o1

50:	/* Destination is 64-byte aligned, and source is 16-byte
	 * aligned.
	 */
1:	LOAD_TWIN(%o1, %o4, %o5)
	add	%o1, 16, %o1
	LOAD_TWIN(%o1, %g2, %g3)
	add	%o1, 16 + 32, %o1
	LOAD(prefetch, %o1, #one_read)
	sub	%o1, 32, %o1
	STORE_INIT(%o4, %o0 + 0x00)		! initializes cache line
	STORE_INIT(%o5, %o0 + 0x08)
	LOAD_TWIN(%o1, %o4, %o5)
	add	%o1, 16, %o1
	STORE_INIT(%g2, %o0 + 0x10)
	STORE_INIT(%g3, %o0 + 0x18)
	LOAD_TWIN(%o1, %g2, %g3)
	add	%o1, 16, %o1
	STORE_INIT(%o4, %o0 + 0x20)
	STORE_INIT(%o5, %o0 + 0x28)
	STORE_INIT(%g2, %o0 + 0x30)
	STORE_INIT(%g3, %o0 + 0x38)
	subcc	%g1, 64, %g1
	bne,pt	%XCC, 1b
	 add	%o0, 64, %o0
	/* fall through */

60:
	/* %o2 contains any final bytes still needed to be copied
	 * over. If anything is left, we copy it one byte at a time.
	 */
	wr		%g0, ASI_PNF, %asi
	brz,pt		%o2, 85f
	 sub		%o0, %o1, %o3
	ba,a,pt		%XCC, 90f

	.align		64
70: /* 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_niagara1)

#endif