0
|
1 //
|
|
2 // Copyright 2004-2007 Sun Microsystems, Inc. All Rights Reserved.
|
|
3 // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
|
|
4 //
|
|
5 // This code is free software; you can redistribute it and/or modify it
|
|
6 // under the terms of the GNU General Public License version 2 only, as
|
|
7 // published by the Free Software Foundation.
|
|
8 //
|
|
9 // This code is distributed in the hope that it will be useful, but WITHOUT
|
|
10 // ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
11 // FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
|
|
12 // version 2 for more details (a copy is included in the LICENSE file that
|
|
13 // accompanied this code).
|
|
14 //
|
|
15 // You should have received a copy of the GNU General Public License version
|
|
16 // 2 along with this work; if not, write to the Free Software Foundation,
|
|
17 // Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
|
|
18 //
|
|
19 // Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
|
|
20 // CA 95054 USA or visit www.sun.com if you need additional information or
|
|
21 // have any questions.
|
|
22 //
|
|
23
|
|
24 .globl fixcw
|
|
25 .globl sse_check
|
|
26 .globl sse_unavailable
|
|
27 .globl gs_load
|
|
28 .globl gs_thread
|
|
29 .globl _Atomic_cmpxchg_long_gcc
|
|
30
|
|
31 // NOTE WELL! The _Copy functions are called directly
|
|
32 // from server-compiler-generated code via CallLeafNoFP,
|
|
33 // which means that they *must* either not use floating
|
|
34 // point or use it in the same manner as does the server
|
|
35 // compiler.
|
|
36
|
|
37 .globl _Copy_conjoint_bytes
|
|
38 .globl _Copy_arrayof_conjoint_bytes
|
|
39 .globl _Copy_conjoint_jshorts_atomic
|
|
40 .globl _Copy_arrayof_conjoint_jshorts
|
|
41 .globl _Copy_conjoint_jints_atomic
|
|
42 .globl _Copy_arrayof_conjoint_jints
|
|
43 .globl _Copy_conjoint_jlongs_atomic
|
|
44 .globl _mmx_Copy_arrayof_conjoint_jshorts
|
|
45
|
|
46 .section .text,"ax"
|
|
47
|
|
48 / Support for void os::Solaris::init_thread_fpu_state() in os_solaris_i486.cpp
|
|
49 / Set fpu to 53 bit precision. This happens too early to use a stub.
|
|
50 .align 16
|
|
51 fixcw:
|
|
52 pushl $0x27f
|
|
53 fldcw 0(%esp)
|
|
54 popl %eax
|
|
55 ret
|
|
56
|
|
57 .align 16
|
|
58 .globl SafeFetch32
|
|
59 .globl SafeFetchN
|
|
60 .globl Fetch32PFI, Fetch32Resume
|
|
61 SafeFetch32:
|
|
62 SafeFetchN:
|
|
63 movl 0x8(%esp), %eax
|
|
64 movl 0x4(%esp), %ecx
|
|
65 Fetch32PFI:
|
|
66 movl (%ecx), %eax
|
|
67 Fetch32Resume:
|
|
68 ret
|
|
69
|
|
70
|
|
71 .align 16
|
|
72 .globl SpinPause
|
|
73 SpinPause:
|
|
74 rep
|
|
75 nop
|
|
76 movl $1, %eax
|
|
77 ret
|
|
78
|
|
79
|
|
80 / Test SSE availability, used by os_solaris_i486.cpp
|
|
81 .align 16
|
|
82 sse_check:
|
|
83 / Fault if SSE not available
|
|
84 xorps %xmm0,%xmm0
|
|
85 / No fault
|
|
86 movl $1,%eax
|
|
87 ret
|
|
88 / Signal handler continues here if SSE is not available
|
|
89 sse_unavailable:
|
|
90 xorl %eax,%eax
|
|
91 ret
|
|
92
|
|
93 / Fast thread accessors, used by threadLS_solaris_i486.cpp
|
|
94 .align 16
|
|
95 gs_load:
|
|
96 movl 4(%esp),%ecx
|
|
97 movl %gs:(%ecx),%eax
|
|
98 ret
|
|
99
|
|
100 .align 16
|
|
101 gs_thread:
|
|
102 movl %gs:0x0,%eax
|
|
103 ret
|
|
104
|
|
105 / Support for void Copy::conjoint_bytes(void* from,
|
|
106 / void* to,
|
|
107 / size_t count)
|
|
108 .align 16
|
|
109 _Copy_conjoint_bytes:
|
|
110 pushl %esi
|
|
111 movl 4+12(%esp),%ecx / count
|
|
112 pushl %edi
|
|
113 movl 8+ 4(%esp),%esi / from
|
|
114 movl 8+ 8(%esp),%edi / to
|
|
115 cmpl %esi,%edi
|
|
116 leal -1(%esi,%ecx),%eax / from + count - 1
|
|
117 jbe cb_CopyRight
|
|
118 cmpl %eax,%edi
|
|
119 jbe cb_CopyLeft
|
|
120 / copy from low to high
|
|
121 cb_CopyRight:
|
|
122 cmpl $3,%ecx
|
|
123 jbe 5f / <= 3 bytes
|
|
124 / align source address at dword address boundary
|
|
125 movl %ecx,%eax / original count
|
|
126 movl $4,%ecx
|
|
127 subl %esi,%ecx
|
|
128 andl $3,%ecx / prefix byte count
|
|
129 jz 1f / no prefix
|
|
130 subl %ecx,%eax / byte count less prefix
|
|
131 / copy prefix
|
|
132 subl %esi,%edi
|
|
133 0: movb (%esi),%dl
|
|
134 movb %dl,(%edi,%esi,1)
|
|
135 addl $1,%esi
|
|
136 subl $1,%ecx
|
|
137 jnz 0b
|
|
138 addl %esi,%edi
|
|
139 1: movl %eax,%ecx / byte count less prefix
|
|
140 shrl $2,%ecx / dword count
|
|
141 jz 4f / no dwords to move
|
|
142 cmpl $32,%ecx
|
|
143 jbe 2f / <= 32 dwords
|
|
144 / copy aligned dwords
|
|
145 rep; smovl
|
|
146 jmp 4f
|
|
147 / copy aligned dwords
|
|
148 2: subl %esi,%edi
|
|
149 .align 16
|
|
150 3: movl (%esi),%edx
|
|
151 movl %edx,(%edi,%esi,1)
|
|
152 addl $4,%esi
|
|
153 subl $1,%ecx
|
|
154 jnz 3b
|
|
155 addl %esi,%edi
|
|
156 4: movl %eax,%ecx / byte count less prefix
|
|
157 andl $3,%ecx / suffix byte count
|
|
158 jz 7f / no suffix
|
|
159 / copy suffix
|
|
160 5: xorl %eax,%eax
|
|
161 6: movb (%esi,%eax,1),%dl
|
|
162 movb %dl,(%edi,%eax,1)
|
|
163 addl $1,%eax
|
|
164 subl $1,%ecx
|
|
165 jnz 6b
|
|
166 7: popl %edi
|
|
167 popl %esi
|
|
168 ret
|
|
169 / copy from high to low
|
|
170 cb_CopyLeft:
|
|
171 std
|
|
172 leal -4(%edi,%ecx),%edi / to + count - 4
|
|
173 movl %eax,%esi / from + count - 1
|
|
174 movl %ecx,%eax
|
|
175 subl $3,%esi / from + count - 4
|
|
176 cmpl $3,%ecx
|
|
177 jbe 5f / <= 3 bytes
|
|
178 1: shrl $2,%ecx / dword count
|
|
179 jz 4f / no dwords to move
|
|
180 cmpl $32,%ecx
|
|
181 ja 3f / > 32 dwords
|
|
182 / copy dwords, aligned or not
|
|
183 subl %esi,%edi
|
|
184 .align 16
|
|
185 2: movl (%esi),%edx
|
|
186 movl %edx,(%edi,%esi,1)
|
|
187 subl $4,%esi
|
|
188 subl $1,%ecx
|
|
189 jnz 2b
|
|
190 addl %esi,%edi
|
|
191 jmp 4f
|
|
192 / copy dwords, aligned or not
|
|
193 3: rep; smovl
|
|
194 4: movl %eax,%ecx / byte count
|
|
195 andl $3,%ecx / suffix byte count
|
|
196 jz 7f / no suffix
|
|
197 / copy suffix
|
|
198 5: subl %esi,%edi
|
|
199 addl $3,%esi
|
|
200 6: movb (%esi),%dl
|
|
201 movb %dl,(%edi,%esi,1)
|
|
202 subl $1,%esi
|
|
203 subl $1,%ecx
|
|
204 jnz 6b
|
|
205 7: cld
|
|
206 popl %edi
|
|
207 popl %esi
|
|
208 ret
|
|
209
|
|
210 / Support for void Copy::arrayof_conjoint_bytes(void* from,
|
|
211 / void* to,
|
|
212 / size_t count)
|
|
213 /
|
|
214 / Same as _Copy_conjoint_bytes, except no source alignment check.
|
|
215 .align 16
|
|
216 _Copy_arrayof_conjoint_bytes:
|
|
217 pushl %esi
|
|
218 movl 4+12(%esp),%ecx / count
|
|
219 pushl %edi
|
|
220 movl 8+ 4(%esp),%esi / from
|
|
221 movl 8+ 8(%esp),%edi / to
|
|
222 cmpl %esi,%edi
|
|
223 leal -1(%esi,%ecx),%eax / from + count - 1
|
|
224 jbe acb_CopyRight
|
|
225 cmpl %eax,%edi
|
|
226 jbe acb_CopyLeft
|
|
227 / copy from low to high
|
|
228 acb_CopyRight:
|
|
229 cmpl $3,%ecx
|
|
230 jbe 5f
|
|
231 1: movl %ecx,%eax
|
|
232 shrl $2,%ecx
|
|
233 jz 4f
|
|
234 cmpl $32,%ecx
|
|
235 ja 3f
|
|
236 / copy aligned dwords
|
|
237 subl %esi,%edi
|
|
238 .align 16
|
|
239 2: movl (%esi),%edx
|
|
240 movl %edx,(%edi,%esi,1)
|
|
241 addl $4,%esi
|
|
242 subl $1,%ecx
|
|
243 jnz 2b
|
|
244 addl %esi,%edi
|
|
245 jmp 4f
|
|
246 / copy aligned dwords
|
|
247 3: rep; smovl
|
|
248 4: movl %eax,%ecx
|
|
249 andl $3,%ecx
|
|
250 jz 7f
|
|
251 / copy suffix
|
|
252 5: xorl %eax,%eax
|
|
253 6: movb (%esi,%eax,1),%dl
|
|
254 movb %dl,(%edi,%eax,1)
|
|
255 addl $1,%eax
|
|
256 subl $1,%ecx
|
|
257 jnz 6b
|
|
258 7: popl %edi
|
|
259 popl %esi
|
|
260 ret
|
|
261 acb_CopyLeft:
|
|
262 std
|
|
263 leal -4(%edi,%ecx),%edi / to + count - 4
|
|
264 movl %eax,%esi / from + count - 1
|
|
265 movl %ecx,%eax
|
|
266 subl $3,%esi / from + count - 4
|
|
267 cmpl $3,%ecx
|
|
268 jbe 5f
|
|
269 1: shrl $2,%ecx
|
|
270 jz 4f
|
|
271 cmpl $32,%ecx
|
|
272 jbe 2f / <= 32 dwords
|
|
273 rep; smovl
|
|
274 jmp 4f
|
|
275 .=.+8
|
|
276 2: subl %esi,%edi
|
|
277 .align 16
|
|
278 3: movl (%esi),%edx
|
|
279 movl %edx,(%edi,%esi,1)
|
|
280 subl $4,%esi
|
|
281 subl $1,%ecx
|
|
282 jnz 3b
|
|
283 addl %esi,%edi
|
|
284 4: movl %eax,%ecx
|
|
285 andl $3,%ecx
|
|
286 jz 7f
|
|
287 5: subl %esi,%edi
|
|
288 addl $3,%esi
|
|
289 6: movb (%esi),%dl
|
|
290 movb %dl,(%edi,%esi,1)
|
|
291 subl $1,%esi
|
|
292 subl $1,%ecx
|
|
293 jnz 6b
|
|
294 7: cld
|
|
295 popl %edi
|
|
296 popl %esi
|
|
297 ret
|
|
298
|
|
299 / Support for void Copy::conjoint_jshorts_atomic(void* from,
|
|
300 / void* to,
|
|
301 / size_t count)
|
|
302 .align 16
|
|
303 _Copy_conjoint_jshorts_atomic:
|
|
304 pushl %esi
|
|
305 movl 4+12(%esp),%ecx / count
|
|
306 pushl %edi
|
|
307 movl 8+ 4(%esp),%esi / from
|
|
308 movl 8+ 8(%esp),%edi / to
|
|
309 cmpl %esi,%edi
|
|
310 leal -2(%esi,%ecx,2),%eax / from + count*2 - 2
|
|
311 jbe cs_CopyRight
|
|
312 cmpl %eax,%edi
|
|
313 jbe cs_CopyLeft
|
|
314 / copy from low to high
|
|
315 cs_CopyRight:
|
|
316 / align source address at dword address boundary
|
|
317 movl %esi,%eax / original from
|
|
318 andl $3,%eax / either 0 or 2
|
|
319 jz 1f / no prefix
|
|
320 / copy prefix
|
|
321 movw (%esi),%dx
|
|
322 movw %dx,(%edi)
|
|
323 addl %eax,%esi / %eax == 2
|
|
324 addl %eax,%edi
|
|
325 subl $1,%ecx
|
|
326 1: movl %ecx,%eax / word count less prefix
|
|
327 sarl %ecx / dword count
|
|
328 jz 4f / no dwords to move
|
|
329 cmpl $32,%ecx
|
|
330 jbe 2f / <= 32 dwords
|
|
331 / copy aligned dwords
|
|
332 rep; smovl
|
|
333 jmp 4f
|
|
334 / copy aligned dwords
|
|
335 2: subl %esi,%edi
|
|
336 .align 16
|
|
337 3: movl (%esi),%edx
|
|
338 movl %edx,(%edi,%esi,1)
|
|
339 addl $4,%esi
|
|
340 subl $1,%ecx
|
|
341 jnz 3b
|
|
342 addl %esi,%edi
|
|
343 4: andl $1,%eax / suffix count
|
|
344 jz 5f / no suffix
|
|
345 / copy suffix
|
|
346 movw (%esi),%dx
|
|
347 movw %dx,(%edi)
|
|
348 5: popl %edi
|
|
349 popl %esi
|
|
350 ret
|
|
351 / copy from high to low
|
|
352 cs_CopyLeft:
|
|
353 std
|
|
354 leal -4(%edi,%ecx,2),%edi / to + count*2 - 4
|
|
355 movl %eax,%esi / from + count*2 - 2
|
|
356 movl %ecx,%eax
|
|
357 subl $2,%esi / from + count*2 - 4
|
|
358 1: sarl %ecx / dword count
|
|
359 jz 4f / no dwords to move
|
|
360 cmpl $32,%ecx
|
|
361 ja 3f / > 32 dwords
|
|
362 subl %esi,%edi
|
|
363 .align 16
|
|
364 2: movl (%esi),%edx
|
|
365 movl %edx,(%edi,%esi,1)
|
|
366 subl $4,%esi
|
|
367 subl $1,%ecx
|
|
368 jnz 2b
|
|
369 addl %esi,%edi
|
|
370 jmp 4f
|
|
371 3: rep; smovl
|
|
372 4: andl $1,%eax / suffix count
|
|
373 jz 5f / no suffix
|
|
374 / copy suffix
|
|
375 addl $2,%esi
|
|
376 addl $2,%edi
|
|
377 movw (%esi),%dx
|
|
378 movw %dx,(%edi)
|
|
379 5: cld
|
|
380 popl %edi
|
|
381 popl %esi
|
|
382 ret
|
|
383
|
|
384 / Support for void Copy::arrayof_conjoint_jshorts(void* from,
|
|
385 / void* to,
|
|
386 / size_t count)
|
|
387 .align 16
|
|
388 _Copy_arrayof_conjoint_jshorts:
|
|
389 pushl %esi
|
|
390 movl 4+12(%esp),%ecx / count
|
|
391 pushl %edi
|
|
392 movl 8+ 4(%esp),%esi / from
|
|
393 movl 8+ 8(%esp),%edi / to
|
|
394 cmpl %esi,%edi
|
|
395 leal -2(%esi,%ecx,2),%eax / from + count*2 - 2
|
|
396 jbe acs_CopyRight
|
|
397 cmpl %eax,%edi
|
|
398 jbe acs_CopyLeft
|
|
399 acs_CopyRight:
|
|
400 movl %ecx,%eax / word count
|
|
401 sarl %ecx / dword count
|
|
402 jz 4f / no dwords to move
|
|
403 cmpl $32,%ecx
|
|
404 jbe 2f / <= 32 dwords
|
|
405 / copy aligned dwords
|
|
406 rep; smovl
|
|
407 jmp 4f
|
|
408 / copy aligned dwords
|
|
409 .=.+5
|
|
410 2: subl %esi,%edi
|
|
411 .align 16
|
|
412 3: movl (%esi),%edx
|
|
413 movl %edx,(%edi,%esi,1)
|
|
414 addl $4,%esi
|
|
415 subl $1,%ecx
|
|
416 jnz 3b
|
|
417 addl %esi,%edi
|
|
418 4: andl $1,%eax / suffix count
|
|
419 jz 5f / no suffix
|
|
420 / copy suffix
|
|
421 movw (%esi),%dx
|
|
422 movw %dx,(%edi)
|
|
423 5: popl %edi
|
|
424 popl %esi
|
|
425 ret
|
|
426 acs_CopyLeft:
|
|
427 std
|
|
428 leal -4(%edi,%ecx,2),%edi / to + count*2 - 4
|
|
429 movl %eax,%esi / from + count*2 - 2
|
|
430 movl %ecx,%eax
|
|
431 subl $2,%esi / from + count*2 - 4
|
|
432 sarl %ecx / dword count
|
|
433 jz 4f / no dwords to move
|
|
434 cmpl $32,%ecx
|
|
435 ja 3f / > 32 dwords
|
|
436 subl %esi,%edi
|
|
437 .align 16
|
|
438 2: movl (%esi),%edx
|
|
439 movl %edx,(%edi,%esi,1)
|
|
440 subl $4,%esi
|
|
441 subl $1,%ecx
|
|
442 jnz 2b
|
|
443 addl %esi,%edi
|
|
444 jmp 4f
|
|
445 3: rep; smovl
|
|
446 4: andl $1,%eax / suffix count
|
|
447 jz 5f / no suffix
|
|
448 / copy suffix
|
|
449 addl $2,%esi
|
|
450 addl $2,%edi
|
|
451 movw (%esi),%dx
|
|
452 movw %dx,(%edi)
|
|
453 5: cld
|
|
454 popl %edi
|
|
455 popl %esi
|
|
456 ret
|
|
457
|
|
458 / Support for void Copy::conjoint_jints_atomic(void* from,
|
|
459 / void* to,
|
|
460 / size_t count)
|
|
461 / Equivalent to
|
|
462 / arrayof_conjoint_jints
|
|
463 .align 16
|
|
464 _Copy_conjoint_jints_atomic:
|
|
465 _Copy_arrayof_conjoint_jints:
|
|
466 pushl %esi
|
|
467 movl 4+12(%esp),%ecx / count
|
|
468 pushl %edi
|
|
469 movl 8+ 4(%esp),%esi / from
|
|
470 movl 8+ 8(%esp),%edi / to
|
|
471 cmpl %esi,%edi
|
|
472 leal -4(%esi,%ecx,4),%eax / from + count*4 - 4
|
|
473 jbe ci_CopyRight
|
|
474 cmpl %eax,%edi
|
|
475 jbe ci_CopyLeft
|
|
476 ci_CopyRight:
|
|
477 cmpl $32,%ecx
|
|
478 jbe 2f / <= 32 dwords
|
|
479 rep; smovl
|
|
480 popl %edi
|
|
481 popl %esi
|
|
482 ret
|
|
483 .=.+10
|
|
484 2: subl %esi,%edi
|
|
485 .align 16
|
|
486 3: movl (%esi),%edx
|
|
487 movl %edx,(%edi,%esi,1)
|
|
488 addl $4,%esi
|
|
489 subl $1,%ecx
|
|
490 jnz 3b
|
|
491 popl %edi
|
|
492 popl %esi
|
|
493 ret
|
|
494 ci_CopyLeft:
|
|
495 std
|
|
496 leal -4(%edi,%ecx,4),%edi / to + count*4 - 4
|
|
497 cmpl $32,%ecx
|
|
498 ja 3f / > 32 dwords
|
|
499 subl %eax,%edi / eax == from + count*4 - 4
|
|
500 .align 16
|
|
501 2: movl (%eax),%edx
|
|
502 movl %edx,(%edi,%eax,1)
|
|
503 subl $4,%eax
|
|
504 subl $1,%ecx
|
|
505 jnz 2b
|
|
506 cld
|
|
507 popl %edi
|
|
508 popl %esi
|
|
509 ret
|
|
510 3: movl %eax,%esi / from + count*4 - 4
|
|
511 rep; smovl
|
|
512 cld
|
|
513 popl %edi
|
|
514 popl %esi
|
|
515 ret
|
|
516
|
|
517 / Support for void Copy::conjoint_jlongs_atomic(jlong* from,
|
|
518 / jlong* to,
|
|
519 / size_t count)
|
|
520 /
|
|
521 / 32-bit
|
|
522 /
|
|
523 / count treated as signed
|
|
524 /
|
|
525 / if (from > to) {
|
|
526 / while (--count >= 0) {
|
|
527 / *to++ = *from++;
|
|
528 / }
|
|
529 / } else {
|
|
530 / while (--count >= 0) {
|
|
531 / to[count] = from[count];
|
|
532 / }
|
|
533 / }
|
|
534 .align 16
|
|
535 _Copy_conjoint_jlongs_atomic:
|
|
536 movl 4+8(%esp),%ecx / count
|
|
537 movl 4+0(%esp),%eax / from
|
|
538 movl 4+4(%esp),%edx / to
|
|
539 cmpl %eax,%edx
|
|
540 jae cla_CopyLeft
|
|
541 cla_CopyRight:
|
|
542 subl %eax,%edx
|
|
543 jmp 2f
|
|
544 .align 16
|
|
545 1: fildll (%eax)
|
|
546 fistpll (%edx,%eax,1)
|
|
547 addl $8,%eax
|
|
548 2: subl $1,%ecx
|
|
549 jge 1b
|
|
550 ret
|
|
551 .align 16
|
|
552 3: fildll (%eax,%ecx,8)
|
|
553 fistpll (%edx,%ecx,8)
|
|
554 cla_CopyLeft:
|
|
555 subl $1,%ecx
|
|
556 jge 3b
|
|
557 ret
|
|
558
|
|
559 / Support for void Copy::arrayof_conjoint_jshorts(void* from,
|
|
560 / void* to,
|
|
561 / size_t count)
|
|
562 .align 16
|
|
563 _mmx_Copy_arrayof_conjoint_jshorts:
|
|
564 pushl %esi
|
|
565 movl 4+12(%esp),%ecx
|
|
566 pushl %edi
|
|
567 movl 8+ 4(%esp),%esi
|
|
568 movl 8+ 8(%esp),%edi
|
|
569 cmpl %esi,%edi
|
|
570 leal -2(%esi,%ecx,2),%eax
|
|
571 jbe mmx_acs_CopyRight
|
|
572 cmpl %eax,%edi
|
|
573 jbe mmx_acs_CopyLeft
|
|
574 mmx_acs_CopyRight:
|
|
575 movl %ecx,%eax
|
|
576 sarl %ecx
|
|
577 je 5f
|
|
578 cmpl $33,%ecx
|
|
579 jae 3f
|
|
580 1: subl %esi,%edi
|
|
581 .align 16
|
|
582 2: movl (%esi),%edx
|
|
583 movl %edx,(%edi,%esi,1)
|
|
584 addl $4,%esi
|
|
585 subl $1,%ecx
|
|
586 jnz 2b
|
|
587 addl %esi,%edi
|
|
588 jmp 5f
|
|
589 3: smovl / align to 8 bytes, we know we are 4 byte aligned to start
|
|
590 subl $1,%ecx
|
|
591 4: .align 16
|
|
592 movq 0(%esi),%mm0
|
|
593 addl $64,%edi
|
|
594 movq 8(%esi),%mm1
|
|
595 subl $16,%ecx
|
|
596 movq 16(%esi),%mm2
|
|
597 movq %mm0,-64(%edi)
|
|
598 movq 24(%esi),%mm0
|
|
599 movq %mm1,-56(%edi)
|
|
600 movq 32(%esi),%mm1
|
|
601 movq %mm2,-48(%edi)
|
|
602 movq 40(%esi),%mm2
|
|
603 movq %mm0,-40(%edi)
|
|
604 movq 48(%esi),%mm0
|
|
605 movq %mm1,-32(%edi)
|
|
606 movq 56(%esi),%mm1
|
|
607 movq %mm2,-24(%edi)
|
|
608 movq %mm0,-16(%edi)
|
|
609 addl $64,%esi
|
|
610 movq %mm1,-8(%edi)
|
|
611 cmpl $16,%ecx
|
|
612 jge 4b
|
|
613 emms
|
|
614 testl %ecx,%ecx
|
|
615 ja 1b
|
|
616 5: andl $1,%eax
|
|
617 je 7f
|
|
618 6: movw (%esi),%dx
|
|
619 movw %dx,(%edi)
|
|
620 7: popl %edi
|
|
621 popl %esi
|
|
622 ret
|
|
623 mmx_acs_CopyLeft:
|
|
624 std
|
|
625 leal -4(%edi,%ecx,2),%edi
|
|
626 movl %eax,%esi
|
|
627 movl %ecx,%eax
|
|
628 subl $2,%esi
|
|
629 sarl %ecx
|
|
630 je 4f
|
|
631 cmpl $32,%ecx
|
|
632 ja 3f
|
|
633 subl %esi,%edi
|
|
634 .align 16
|
|
635 2: movl (%esi),%edx
|
|
636 movl %edx,(%edi,%esi,1)
|
|
637 subl $4,%esi
|
|
638 subl $1,%ecx
|
|
639 jnz 2b
|
|
640 addl %esi,%edi
|
|
641 jmp 4f
|
|
642 3: rep; smovl
|
|
643 4: andl $1,%eax
|
|
644 je 6f
|
|
645 addl $2,%esi
|
|
646 addl $2,%edi
|
|
647 5: movw (%esi),%dx
|
|
648 movw %dx,(%edi)
|
|
649 6: cld
|
|
650 popl %edi
|
|
651 popl %esi
|
|
652 ret
|
|
653
|
|
654
|
|
655 / Support for jlong Atomic::cmpxchg(jlong exchange_value,
|
|
656 / volatile jlong* dest,
|
|
657 / jlong compare_value,
|
|
658 / bool is_MP)
|
|
659 / Used only for Solaris/gcc builds
|
|
660 .align 16
|
|
661 _Atomic_cmpxchg_long_gcc:
|
|
662 / 8(%esp) : return PC
|
|
663 pushl %ebx / 4(%esp) : old %ebx
|
|
664 pushl %edi / 0(%esp) : old %edi
|
|
665 movl 12(%esp), %ebx / 12(%esp) : exchange_value (low)
|
|
666 movl 16(%esp), %ecx / 16(%esp) : exchange_value (high)
|
|
667 movl 24(%esp), %eax / 24(%esp) : compare_value (low)
|
|
668 movl 28(%esp), %edx / 28(%esp) : compare_value (high)
|
|
669 movl 20(%esp), %edi / 20(%esp) : dest
|
|
670 cmpl $0, 32(%esp) / 32(%esp) : is_MP
|
|
671 je 1f
|
|
672 lock
|
|
673 1: cmpxchg8b (%edi)
|
|
674 popl %edi
|
|
675 popl %ebx
|
|
676 ret
|