Mercurial > jdk8-mips64-public > hotspot / file revision
src/os_cpu/solaris_x86/vm/solaris_x86_32.s@9b8ce46870df
src/os_cpu/solaris_x86/vm/solaris_x86_32.s
Thu, 16 Feb 2012 17:12:49 -0800
- author
- kvn
- date
- Thu, 16 Feb 2012 17:12:49 -0800
- changeset 3577
- 9b8ce46870df
- parent 2384
- 0a8e0d4345b3
- child 5400
- 980532a806a5
- permissions
- -rw-r--r--
7145346: VerifyStackAtCalls is broken
Summary: Replace call_epilog() encoding with macroassembler use. Moved duplicated code to x86.ad. Fixed return_addr() definition.
Reviewed-by: never
1 //
2 // Copyright (c) 2004, 2010, Oracle and/or its affiliates. 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 Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20 // or visit www.oracle.com if you need additional information or have any
21 // questions.
22 //
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
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.
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
46 .section .text,"ax"
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
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
71 .align 16
72 .globl SpinPause
73 SpinPause:
74 rep
75 nop
76 movl $1, %eax
77 ret
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
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
100 .align 16
101 gs_thread:
102 movl %gs:0x0,%eax
103 ret
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 5: andl $3,%ecx / suffix byte count
158 jz 7f / no suffix
159 / copy suffix
160 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 5: andl $3,%ecx / suffix byte count
196 jz 7f / no suffix
197 / copy suffix
198 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
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 5: andl $3,%ecx
250 jz 7f
251 / copy suffix
252 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 5: andl $3,%ecx
286 jz 7f
287 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
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 subl $1,%ecx
322 jl 5f / zero count
323 movw (%esi),%dx
324 movw %dx,(%edi)
325 addl %eax,%esi / %eax == 2
326 addl %eax,%edi
327 1: movl %ecx,%eax / word count less prefix
328 sarl %ecx / dword count
329 jz 4f / no dwords to move
330 cmpl $32,%ecx
331 jbe 2f / <= 32 dwords
332 / copy aligned dwords
333 rep; smovl
334 jmp 4f
335 / copy aligned dwords
336 2: subl %esi,%edi
337 .align 16
338 3: movl (%esi),%edx
339 movl %edx,(%edi,%esi,1)
340 addl $4,%esi
341 subl $1,%ecx
342 jnz 3b
343 addl %esi,%edi
344 4: andl $1,%eax / suffix count
345 jz 5f / no suffix
346 / copy suffix
347 movw (%esi),%dx
348 movw %dx,(%edi)
349 5: popl %edi
350 popl %esi
351 ret
352 / copy from high to low
353 cs_CopyLeft:
354 std
355 leal -4(%edi,%ecx,2),%edi / to + count*2 - 4
356 movl %eax,%esi / from + count*2 - 2
357 movl %ecx,%eax
358 subl $2,%esi / from + count*2 - 4
359 1: sarl %ecx / dword count
360 jz 4f / no dwords to move
361 cmpl $32,%ecx
362 ja 3f / > 32 dwords
363 subl %esi,%edi
364 .align 16
365 2: movl (%esi),%edx
366 movl %edx,(%edi,%esi,1)
367 subl $4,%esi
368 subl $1,%ecx
369 jnz 2b
370 addl %esi,%edi
371 jmp 4f
372 3: rep; smovl
373 4: andl $1,%eax / suffix count
374 jz 5f / no suffix
375 / copy suffix
376 addl $2,%esi
377 addl $2,%edi
378 movw (%esi),%dx
379 movw %dx,(%edi)
380 5: cld
381 popl %edi
382 popl %esi
383 ret
385 / Support for void Copy::arrayof_conjoint_jshorts(void* from,
386 / void* to,
387 / size_t count)
388 .align 16
389 _Copy_arrayof_conjoint_jshorts:
390 pushl %esi
391 movl 4+12(%esp),%ecx / count
392 pushl %edi
393 movl 8+ 4(%esp),%esi / from
394 movl 8+ 8(%esp),%edi / to
395 cmpl %esi,%edi
396 leal -2(%esi,%ecx,2),%eax / from + count*2 - 2
397 jbe acs_CopyRight
398 cmpl %eax,%edi
399 jbe acs_CopyLeft
400 acs_CopyRight:
401 movl %ecx,%eax / word count
402 sarl %ecx / dword count
403 jz 4f / no dwords to move
404 cmpl $32,%ecx
405 jbe 2f / <= 32 dwords
406 / copy aligned dwords
407 rep; smovl
408 jmp 4f
409 / copy aligned dwords
410 .=.+5
411 2: subl %esi,%edi
412 .align 16
413 3: movl (%esi),%edx
414 movl %edx,(%edi,%esi,1)
415 addl $4,%esi
416 subl $1,%ecx
417 jnz 3b
418 addl %esi,%edi
419 4: andl $1,%eax / suffix count
420 jz 5f / no suffix
421 / copy suffix
422 movw (%esi),%dx
423 movw %dx,(%edi)
424 5: popl %edi
425 popl %esi
426 ret
427 acs_CopyLeft:
428 std
429 leal -4(%edi,%ecx,2),%edi / to + count*2 - 4
430 movl %eax,%esi / from + count*2 - 2
431 movl %ecx,%eax
432 subl $2,%esi / from + count*2 - 4
433 sarl %ecx / dword count
434 jz 4f / no dwords to move
435 cmpl $32,%ecx
436 ja 3f / > 32 dwords
437 subl %esi,%edi
438 .align 16
439 2: movl (%esi),%edx
440 movl %edx,(%edi,%esi,1)
441 subl $4,%esi
442 subl $1,%ecx
443 jnz 2b
444 addl %esi,%edi
445 jmp 4f
446 3: rep; smovl
447 4: andl $1,%eax / suffix count
448 jz 5f / no suffix
449 / copy suffix
450 addl $2,%esi
451 addl $2,%edi
452 movw (%esi),%dx
453 movw %dx,(%edi)
454 5: cld
455 popl %edi
456 popl %esi
457 ret
459 / Support for void Copy::conjoint_jints_atomic(void* from,
460 / void* to,
461 / size_t count)
462 / Equivalent to
463 / arrayof_conjoint_jints
464 .align 16
465 _Copy_conjoint_jints_atomic:
466 _Copy_arrayof_conjoint_jints:
467 pushl %esi
468 movl 4+12(%esp),%ecx / count
469 pushl %edi
470 movl 8+ 4(%esp),%esi / from
471 movl 8+ 8(%esp),%edi / to
472 cmpl %esi,%edi
473 leal -4(%esi,%ecx,4),%eax / from + count*4 - 4
474 jbe ci_CopyRight
475 cmpl %eax,%edi
476 jbe ci_CopyLeft
477 ci_CopyRight:
478 cmpl $32,%ecx
479 jbe 2f / <= 32 dwords
480 rep; smovl
481 popl %edi
482 popl %esi
483 ret
484 .=.+10
485 2: subl %esi,%edi
486 jmp 4f
487 .align 16
488 3: movl (%esi),%edx
489 movl %edx,(%edi,%esi,1)
490 addl $4,%esi
491 4: subl $1,%ecx
492 jge 3b
493 popl %edi
494 popl %esi
495 ret
496 ci_CopyLeft:
497 std
498 leal -4(%edi,%ecx,4),%edi / to + count*4 - 4
499 cmpl $32,%ecx
500 ja 4f / > 32 dwords
501 subl %eax,%edi / eax == from + count*4 - 4
502 jmp 3f
503 .align 16
504 2: movl (%eax),%edx
505 movl %edx,(%edi,%eax,1)
506 subl $4,%eax
507 3: subl $1,%ecx
508 jge 2b
509 cld
510 popl %edi
511 popl %esi
512 ret
513 4: movl %eax,%esi / from + count*4 - 4
514 rep; smovl
515 cld
516 popl %edi
517 popl %esi
518 ret
520 / Support for void Copy::conjoint_jlongs_atomic(jlong* from,
521 / jlong* to,
522 / size_t count)
523 /
524 / 32-bit
525 /
526 / count treated as signed
527 /
528 / if (from > to) {
529 / while (--count >= 0) {
530 / *to++ = *from++;
531 / }
532 / } else {
533 / while (--count >= 0) {
534 / to[count] = from[count];
535 / }
536 / }
537 .align 16
538 _Copy_conjoint_jlongs_atomic:
539 movl 4+8(%esp),%ecx / count
540 movl 4+0(%esp),%eax / from
541 movl 4+4(%esp),%edx / to
542 cmpl %eax,%edx
543 jae cla_CopyLeft
544 cla_CopyRight:
545 subl %eax,%edx
546 jmp 2f
547 .align 16
548 1: fildll (%eax)
549 fistpll (%edx,%eax,1)
550 addl $8,%eax
551 2: subl $1,%ecx
552 jge 1b
553 ret
554 .align 16
555 3: fildll (%eax,%ecx,8)
556 fistpll (%edx,%ecx,8)
557 cla_CopyLeft:
558 subl $1,%ecx
559 jge 3b
560 ret
562 / Support for void Copy::arrayof_conjoint_jshorts(void* from,
563 / void* to,
564 / size_t count)
565 .align 16
566 _mmx_Copy_arrayof_conjoint_jshorts:
567 pushl %esi
568 movl 4+12(%esp),%ecx
569 pushl %edi
570 movl 8+ 4(%esp),%esi
571 movl 8+ 8(%esp),%edi
572 cmpl %esi,%edi
573 leal -2(%esi,%ecx,2),%eax
574 jbe mmx_acs_CopyRight
575 cmpl %eax,%edi
576 jbe mmx_acs_CopyLeft
577 mmx_acs_CopyRight:
578 movl %ecx,%eax
579 sarl %ecx
580 je 5f
581 cmpl $33,%ecx
582 jae 3f
583 1: subl %esi,%edi
584 .align 16
585 2: movl (%esi),%edx
586 movl %edx,(%edi,%esi,1)
587 addl $4,%esi
588 subl $1,%ecx
589 jnz 2b
590 addl %esi,%edi
591 jmp 5f
592 3: smovl / align to 8 bytes, we know we are 4 byte aligned to start
593 subl $1,%ecx
594 4: .align 16
595 movq 0(%esi),%mm0
596 addl $64,%edi
597 movq 8(%esi),%mm1
598 subl $16,%ecx
599 movq 16(%esi),%mm2
600 movq %mm0,-64(%edi)
601 movq 24(%esi),%mm0
602 movq %mm1,-56(%edi)
603 movq 32(%esi),%mm1
604 movq %mm2,-48(%edi)
605 movq 40(%esi),%mm2
606 movq %mm0,-40(%edi)
607 movq 48(%esi),%mm0
608 movq %mm1,-32(%edi)
609 movq 56(%esi),%mm1
610 movq %mm2,-24(%edi)
611 movq %mm0,-16(%edi)
612 addl $64,%esi
613 movq %mm1,-8(%edi)
614 cmpl $16,%ecx
615 jge 4b
616 emms
617 testl %ecx,%ecx
618 ja 1b
619 5: andl $1,%eax
620 je 7f
621 6: movw (%esi),%dx
622 movw %dx,(%edi)
623 7: popl %edi
624 popl %esi
625 ret
626 mmx_acs_CopyLeft:
627 std
628 leal -4(%edi,%ecx,2),%edi
629 movl %eax,%esi
630 movl %ecx,%eax
631 subl $2,%esi
632 sarl %ecx
633 je 4f
634 cmpl $32,%ecx
635 ja 3f
636 subl %esi,%edi
637 .align 16
638 2: movl (%esi),%edx
639 movl %edx,(%edi,%esi,1)
640 subl $4,%esi
641 subl $1,%ecx
642 jnz 2b
643 addl %esi,%edi
644 jmp 4f
645 3: rep; smovl
646 4: andl $1,%eax
647 je 6f
648 addl $2,%esi
649 addl $2,%edi
650 5: movw (%esi),%dx
651 movw %dx,(%edi)
652 6: cld
653 popl %edi
654 popl %esi
655 ret
658 / Support for jlong Atomic::cmpxchg(jlong exchange_value,
659 / volatile jlong* dest,
660 / jlong compare_value,
661 / bool is_MP)
662 / Used only for Solaris/gcc builds
663 .align 16
664 _Atomic_cmpxchg_long_gcc:
665 / 8(%esp) : return PC
666 pushl %ebx / 4(%esp) : old %ebx
667 pushl %edi / 0(%esp) : old %edi
668 movl 12(%esp), %ebx / 12(%esp) : exchange_value (low)
669 movl 16(%esp), %ecx / 16(%esp) : exchange_value (high)
670 movl 24(%esp), %eax / 24(%esp) : compare_value (low)
671 movl 28(%esp), %edx / 28(%esp) : compare_value (high)
672 movl 20(%esp), %edi / 20(%esp) : dest
673 cmpl $0, 32(%esp) / 32(%esp) : is_MP
674 je 1f
675 lock
676 1: cmpxchg8b (%edi)
677 popl %edi
678 popl %ebx
679 ret
