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
|
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2007 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
#pragma ident "%Z%%M% %I% %E% SMI"
#include <sys/asm_linkage.h>
#include <sys/regset.h>
#include <sys/privregs.h>
#if defined(__lint)
#include <sys/types.h>
#include <sys/archsystm.h>
#else
#include "assym.h"
#endif
/*
* Do block operations using Streaming SIMD extensions
*/
#if defined(DEBUG)
#if defined(__amd64)
#define ASSERT_KPREEMPT_DISABLED(t, r32, msg) \
movq %gs:CPU_THREAD, t; \
movsbl T_PREEMPT(t), r32; \
testl r32, r32; \
jne 5f; \
pushq %rbp; \
movq %rsp, %rbp; \
leaq msg(%rip), %rdi; \
xorl %eax, %eax; \
call panic; \
5:
#elif defined(__i386)
#define ASSERT_KPREEMPT_DISABLED(t, r32, msg) \
movl %gs:CPU_THREAD, t; \
movsbl T_PREEMPT(t), r32; \
testl r32, r32; \
jne 5f; \
pushl %ebp; \
movl %esp, %ebp; \
pushl $msg; \
call panic; \
5:
#endif /* __i386 */
#else /* DEBUG */
#define ASSERT_KPREEMPT_DISABLED(t, r32, msg)
#endif /* DEBUG */
#define BLOCKSHIFT 6
#define BLOCKSIZE 64 /* (1 << BLOCKSHIFT) */
#define BLOCKMASK 63 /* (BLOCKSIZE - 1) */
#if (1 << BLOCKSHIFT) != BLOCKSIZE || BLOCKMASK != (BLOCKSIZE - 1)
#error "mucked up constants"
#endif
#if defined(__lint)
/*ARGSUSED*/
void
hwblkclr(void *addr, size_t size)
{}
#else /* __lint */
#if defined(__amd64)
#define ADD addq
#define SUB subq
#else
#define ADD addl
#define SUB subl
#endif
#define SAVE_XMM0(r) \
SAVE_XMM_PROLOG(r, 1); \
movdqa %xmm0, (r)
#define ZERO_LOOP_INIT_XMM(dst) \
pxor %xmm0, %xmm0
#define ZERO_LOOP_BODY_XMM(dst, cnt) \
movntdq %xmm0, (dst); \
movntdq %xmm0, 0x10(dst); \
movntdq %xmm0, 0x20(dst); \
movntdq %xmm0, 0x30(dst); \
ADD $BLOCKSIZE, dst; \
SUB $1, cnt
#define ZERO_LOOP_FINI_XMM(dst) \
mfence
#define RSTOR_XMM0(r) \
movdqa 0x0(r), %xmm0; \
RSTOR_XMM_EPILOG(r, 1)
#if defined(__amd64)
/*
* %rdi dst
* %rsi size
* %rax saved %cr0 (#if DEBUG then %eax is t->t_preempt)
* %r8 pointer to %xmm register save area
*/
ENTRY(hwblkclr)
pushq %rbp
movq %rsp, %rbp
testl $BLOCKMASK, %edi /* address must be BLOCKSIZE aligned */
jne .dobzero
cmpq $BLOCKSIZE, %rsi /* size must be at least BLOCKSIZE */
jl .dobzero
testq $BLOCKMASK, %rsi /* .. and be a multiple of BLOCKSIZE */
jne .dobzero
shrq $BLOCKSHIFT, %rsi
ASSERT_KPREEMPT_DISABLED(%r11, %eax, .not_disabled)
movq %cr0, %rax
clts
testl $CR0_TS, %eax
jnz 1f
SAVE_XMM0(%r8)
1: ZERO_LOOP_INIT_XMM(%rdi)
9: ZERO_LOOP_BODY_XMM(%rdi, %rsi)
jnz 9b
ZERO_LOOP_FINI_XMM(%rdi)
testl $CR0_TS, %eax
jnz 2f
RSTOR_XMM0(%r8)
2: movq %rax, %cr0
leave
ret
.dobzero:
leave
jmp bzero
SET_SIZE(hwblkclr)
#elif defined(__i386)
/*
* %eax dst
* %ecx size in bytes, loop count
* %ebx saved %cr0 (#if DEBUG then t->t_preempt)
* %edi pointer to %xmm register save area
*/
ENTRY(hwblkclr)
movl 4(%esp), %eax
movl 8(%esp), %ecx
testl $BLOCKMASK, %eax /* address must be BLOCKSIZE aligned */
jne .dobzero
cmpl $BLOCKSIZE, %ecx /* size must be at least BLOCKSIZE */
jl .dobzero
testl $BLOCKMASK, %ecx /* .. and be a multiple of BLOCKSIZE */
jne .dobzero
shrl $BLOCKSHIFT, %ecx
movl 0xc(%esp), %edx
pushl %ebx
pushl %esi
ASSERT_KPREEMPT_DISABLED(%esi, %ebx, .not_disabled)
popl %esi
movl %cr0, %ebx
clts
testl $CR0_TS, %ebx
jnz 1f
pushl %edi
SAVE_XMM0(%edi)
1: ZERO_LOOP_INIT_XMM(%eax)
9: ZERO_LOOP_BODY_XMM(%eax, %ecx)
jnz 9b
ZERO_LOOP_FINI_XMM(%eax)
testl $CR0_TS, %ebx
jnz 2f
RSTOR_XMM0(%edi)
popl %edi
2: movl %ebx, %cr0
popl %ebx
ret
.dobzero:
jmp bzero
SET_SIZE(hwblkclr)
#endif /* __i386 */
#endif /* __lint */
#if defined(__lint)
/*ARGSUSED*/
void
hwblkpagecopy(const void *src, void *dst)
{}
#else /* __lint */
#define PREFETCH_START(src) \
prefetchnta 0x0(src); \
prefetchnta 0x40(src)
#define SAVE_XMMS(r) \
SAVE_XMM_PROLOG(r, 8); \
movdqa %xmm0, (r); \
movdqa %xmm1, 0x10(r); \
movdqa %xmm2, 0x20(r); \
movdqa %xmm3, 0x30(r); \
movdqa %xmm4, 0x40(r); \
movdqa %xmm5, 0x50(r); \
movdqa %xmm6, 0x60(r); \
movdqa %xmm7, 0x70(r)
#define COPY_LOOP_INIT_XMM(src) \
prefetchnta 0x80(src); \
prefetchnta 0xc0(src); \
movdqa 0x0(src), %xmm0; \
movdqa 0x10(src), %xmm1; \
movdqa 0x20(src), %xmm2; \
movdqa 0x30(src), %xmm3; \
movdqa 0x40(src), %xmm4; \
movdqa 0x50(src), %xmm5; \
movdqa 0x60(src), %xmm6; \
movdqa 0x70(src), %xmm7; \
ADD $0x80, src
#define COPY_LOOP_BODY_XMM(src, dst, cnt) \
prefetchnta 0x80(src); \
prefetchnta 0xc0(src); \
prefetchnta 0x100(src); \
prefetchnta 0x140(src); \
movntdq %xmm0, (dst); \
movntdq %xmm1, 0x10(dst); \
movntdq %xmm2, 0x20(dst); \
movntdq %xmm3, 0x30(dst); \
movdqa 0x0(src), %xmm0; \
movdqa 0x10(src), %xmm1; \
movntdq %xmm4, 0x40(dst); \
movntdq %xmm5, 0x50(dst); \
movdqa 0x20(src), %xmm2; \
movdqa 0x30(src), %xmm3; \
movntdq %xmm6, 0x60(dst); \
movntdq %xmm7, 0x70(dst); \
movdqa 0x40(src), %xmm4; \
movdqa 0x50(src), %xmm5; \
ADD $0x80, dst; \
movdqa 0x60(src), %xmm6; \
movdqa 0x70(src), %xmm7; \
ADD $0x80, src; \
subl $1, cnt
#define COPY_LOOP_FINI_XMM(dst) \
movntdq %xmm0, 0x0(dst); \
movntdq %xmm1, 0x10(dst); \
movntdq %xmm2, 0x20(dst); \
movntdq %xmm3, 0x30(dst); \
movntdq %xmm4, 0x40(dst); \
movntdq %xmm5, 0x50(dst); \
movntdq %xmm6, 0x60(dst); \
movntdq %xmm7, 0x70(dst)
#define RSTOR_XMMS(r) \
movdqa 0x0(r), %xmm0; \
movdqa 0x10(r), %xmm1; \
movdqa 0x20(r), %xmm2; \
movdqa 0x30(r), %xmm3; \
movdqa 0x40(r), %xmm4; \
movdqa 0x50(r), %xmm5; \
movdqa 0x60(r), %xmm6; \
movdqa 0x70(r), %xmm7; \
RSTOR_XMM_EPILOG(r, 8)
#if defined(__amd64)
/*
* %rdi src
* %rsi dst
* %rdx #if DEBUG then curthread
* %ecx loop count
* %rax saved %cr0 (#if DEBUG then %eax is t->t_prempt)
* %r8 pointer to %xmm register save area
*/
ENTRY(hwblkpagecopy)
pushq %rbp
movq %rsp, %rbp
PREFETCH_START(%rdi)
/*
* PAGESIZE is 4096, each loop moves 128 bytes, but the initial
* load and final store save us on loop count
*/
movl $_CONST(32 - 1), %ecx
ASSERT_KPREEMPT_DISABLED(%rdx, %eax, .not_disabled)
movq %cr0, %rax
clts
testl $CR0_TS, %eax
jnz 3f
SAVE_XMMS(%r8)
3: COPY_LOOP_INIT_XMM(%rdi)
4: COPY_LOOP_BODY_XMM(%rdi, %rsi, %ecx)
jnz 4b
COPY_LOOP_FINI_XMM(%rsi)
testl $CR0_TS, %eax
jnz 5f
RSTOR_XMMS(%r8)
5: movq %rax, %cr0
mfence
leave
ret
SET_SIZE(hwblkpagecopy)
#elif defined(__i386)
/*
* %eax src
* %edx dst
* %ecx loop count
* %ebx saved %cr0 (#if DEBUG then t->t_prempt)
* %edi pointer to %xmm register save area
* %esi #if DEBUG temporary thread pointer
*/
ENTRY(hwblkpagecopy)
movl 4(%esp), %eax
movl 8(%esp), %edx
PREFETCH_START(%eax)
pushl %ebx
/*
* PAGESIZE is 4096, each loop moves 128 bytes, but the initial
* load and final store save us one loop count
*/
movl $_CONST(32 - 1), %ecx
pushl %esi
ASSERT_KPREEMPT_DISABLED(%esi, %ebx, .not_disabled)
popl %esi
movl %cr0, %ebx
clts
testl $CR0_TS, %ebx
jnz 3f
pushl %edi
SAVE_XMMS(%edi)
3: COPY_LOOP_INIT_XMM(%eax)
4: COPY_LOOP_BODY_XMM(%eax, %edx, %ecx)
jnz 4b
COPY_LOOP_FINI_XMM(%edx)
testl $CR0_TS, %ebx
jnz 5f
RSTOR_XMMS(%edi)
popl %edi
5: movl %ebx, %cr0
popl %ebx
mfence
ret
SET_SIZE(hwblkpagecopy)
#endif /* __i386 */
#endif /* __lint */
#if defined(__lint)
/*
* Version of hwblkclr which doesn't use XMM registers.
* Note that it requires aligned dst and len.
*
* XXPV This needs to be performance tuned at some point.
* Is 4 the best number of iterations to unroll?
*/
/*ARGSUSED*/
void
block_zero_no_xmm(void *dst, int len)
{}
#else /* __lint */
#if defined(__amd64)
ENTRY(block_zero_no_xmm)
pushq %rbp
movq %rsp, %rbp
xorl %eax, %eax
addq %rsi, %rdi
negq %rsi
1:
movnti %rax, (%rdi, %rsi)
movnti %rax, 8(%rdi, %rsi)
movnti %rax, 16(%rdi, %rsi)
movnti %rax, 24(%rdi, %rsi)
addq $32, %rsi
jnz 1b
mfence
leave
ret
SET_SIZE(block_zero_no_xmm)
#elif defined(__i386)
ENTRY(block_zero_no_xmm)
pushl %ebp
movl %esp, %ebp
xorl %eax, %eax
movl 8(%ebp), %edx
movl 12(%ebp), %ecx
addl %ecx, %edx
negl %ecx
1:
movnti %eax, (%edx, %ecx)
movnti %eax, 4(%edx, %ecx)
movnti %eax, 8(%edx, %ecx)
movnti %eax, 12(%edx, %ecx)
addl $16, %ecx
jnz 1b
mfence
leave
ret
SET_SIZE(block_zero_no_xmm)
#endif /* __i386 */
#endif /* __lint */
#if defined(__lint)
/*
* Version of page copy which doesn't use XMM registers.
*
* XXPV This needs to be performance tuned at some point.
* Is 4 the right number of iterations to unroll?
* Is the load/store order optimal? Should it use prefetch?
*/
/*ARGSUSED*/
void
page_copy_no_xmm(void *dst, void *src)
{}
#else /* __lint */
#if defined(__amd64)
ENTRY(page_copy_no_xmm)
movq $MMU_STD_PAGESIZE, %rcx
addq %rcx, %rdi
addq %rcx, %rsi
negq %rcx
1:
movq (%rsi, %rcx), %rax
movnti %rax, (%rdi, %rcx)
movq 8(%rsi, %rcx), %rax
movnti %rax, 8(%rdi, %rcx)
movq 16(%rsi, %rcx), %rax
movnti %rax, 16(%rdi, %rcx)
movq 24(%rsi, %rcx), %rax
movnti %rax, 24(%rdi, %rcx)
addq $32, %rcx
jnz 1b
mfence
ret
SET_SIZE(page_copy_no_xmm)
#elif defined(__i386)
ENTRY(page_copy_no_xmm)
pushl %esi
movl $MMU_STD_PAGESIZE, %ecx
movl 8(%esp), %edx
movl 12(%esp), %esi
addl %ecx, %edx
addl %ecx, %esi
negl %ecx
1:
movl (%esi, %ecx), %eax
movnti %eax, (%edx, %ecx)
movl 4(%esi, %ecx), %eax
movnti %eax, 4(%edx, %ecx)
movl 8(%esi, %ecx), %eax
movnti %eax, 8(%edx, %ecx)
movl 12(%esi, %ecx), %eax
movnti %eax, 12(%edx, %ecx)
addl $16, %ecx
jnz 1b
mfence
popl %esi
ret
SET_SIZE(page_copy_no_xmm)
#endif /* __i386 */
#endif /* __lint */
#if defined(DEBUG) && !defined(__lint)
.text
.not_disabled:
.string "sseblk: preemption not disabled!"
#endif
|