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
|
/*--------------------------------------------------------------------*/
/*--- The core dispatch loop, for jumping to a code address. ---*/
/*--- dispatch-ppc32-linux.S ---*/
/*--------------------------------------------------------------------*/
/*
This file is part of Valgrind, a dynamic binary instrumentation
framework.
Copyright (C) 2005-2017 Cerion Armour-Brown <cerion@open-works.co.uk>
This program is free software; you can redistribute it and/or
modify it under the terms of the GNU General Public License as
published by the Free Software Foundation; either version 2 of the
License, or (at your option) any later version.
This program 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
General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, see <http://www.gnu.org/licenses/>.
The GNU General Public License is contained in the file COPYING.
*/
#include "pub_core_basics_asm.h"
#if defined(VGP_ppc32_linux)
#include "pub_core_dispatch_asm.h"
#include "pub_core_transtab_asm.h"
#include "libvex_guest_offsets.h" /* for OFFSET_ppc32_CIA */
/*------------------------------------------------------------*/
/*--- ---*/
/*--- The dispatch loop. VG_(disp_run_translations) is ---*/
/*--- used to run all translations, ---*/
/*--- including no-redir ones. ---*/
/*--- ---*/
/*------------------------------------------------------------*/
/*----------------------------------------------------*/
/*--- Entry and preamble (set everything up) ---*/
/*----------------------------------------------------*/
/* signature:
void VG_(disp_run_translations)( UWord* two_words,
void* guest_state,
Addr host_addr );
*/
.text
.globl VG_(disp_run_translations)
.type VG_(disp_run_translations), @function
VG_(disp_run_translations):
/* r3 holds two_words */
/* r4 holds guest_state */
/* r5 holds host_addr */
/* ----- entry point to VG_(disp_run_translations) ----- */
/* For Linux/ppc32 we need the SysV ABI, which uses
LR->4(parent_sp), CR->anywhere.
(The AIX ABI, used on Darwin,
uses LR->8(prt_sp), CR->4(prt_sp))
*/
/* Save lr */
mflr 6
stw 6,4(1)
/* New stack frame */
stwu 1,-496(1) /* sp should maintain 16-byte alignment */
/* Save callee-saved registers... */
/* r3, r4, r5 are live here, so use r6 */
lis 6,VG_(machine_ppc32_has_FP)@ha
lwz 6,VG_(machine_ppc32_has_FP)@l(6)
cmplwi 6,0
beq LafterFP1
/* Floating-point reg save area : 144 bytes */
stfd 31,488(1)
stfd 30,480(1)
stfd 29,472(1)
stfd 28,464(1)
stfd 27,456(1)
stfd 26,448(1)
stfd 25,440(1)
stfd 24,432(1)
stfd 23,424(1)
stfd 22,416(1)
stfd 21,408(1)
stfd 20,400(1)
stfd 19,392(1)
stfd 18,384(1)
stfd 17,376(1)
stfd 16,368(1)
stfd 15,360(1)
stfd 14,352(1)
LafterFP1:
/* General reg save area : 76 bytes */
stw 31,348(1)
stw 30,344(1)
stw 29,340(1)
stw 28,336(1)
stw 27,332(1)
stw 26,328(1)
stw 25,324(1)
stw 24,320(1)
stw 23,316(1)
stw 22,312(1)
stw 21,308(1)
stw 20,304(1)
stw 19,300(1)
stw 18,296(1)
stw 17,292(1)
stw 16,288(1)
stw 15,284(1)
stw 14,280(1)
stw 13,276(1)
stw 3,272(1) /* save two_words for later */
/* It's necessary to save/restore VRSAVE in the AIX / Darwin ABI.
The Linux kernel might not actually use VRSAVE for its intended
purpose, but it should be harmless to preserve anyway. */
/* r3, r4, r5 are live here, so use r6 */
lis 6,VG_(machine_ppc32_has_VMX)@ha
lwz 6,VG_(machine_ppc32_has_VMX)@l(6)
cmplwi 6,0
beq LafterVMX1
#ifdef HAS_ALTIVEC
/* VRSAVE save word : 32 bytes */
mfspr 6,256 /* vrsave reg is spr number 256 */
stw 6,244(1)
/* Alignment padding : 4 bytes */
/* Vector reg save area (quadword aligned) : 192 bytes */
li 6,224
stvx 31,6,1
li 6,208
stvx 30,6,1
li 6,192
stvx 29,6,1
li 6,176
stvx 28,6,1
li 6,160
stvx 27,6,1
li 6,144
stvx 26,6,1
li 6,128
stvx 25,6,1
li 6,112
stvx 24,6,1
li 6,96
stvx 23,6,1
li 6,80
stvx 22,6,1
li 6,64
stvx 21,6,1
li 6,48
stvx 20,6,1
#endif
LafterVMX1:
/* Save cr */
mfcr 6
stw 6,44(1)
/* Local variable space... */
/* 32(sp) used later to check FPSCR[RM] */
/* r3 holds two_words */
/* r4 holds guest_state */
/* r5 holds host_addr */
/* 24(sp) used later to stop ctr reg being clobbered */
/* 20(sp) used later to load fpscr with zero */
/* 8:16(sp) free */
/* Linkage Area (reserved)
4(sp) : LR
0(sp) : back-chain
*/
/* set host FPU control word to the default mode expected
by VEX-generated code. See comments in libvex.h for
more info. */
lis 6,VG_(machine_ppc32_has_FP)@ha
lwz 6,VG_(machine_ppc32_has_FP)@l(6)
cmplwi 6,0
beq LafterFP2
/* get zero into f3 (tedious) */
/* note: fsub 3,3,3 is not a reliable way to do this,
since if f3 holds a NaN or similar then we don't necessarily
wind up with zero. */
li 6,0
stw 6,20(1)
lfs 3,20(1)
mtfsf 0xFF,3 /* fpscr = f3 */
LafterFP2:
/* set host AltiVec control word to the default mode expected
by VEX-generated code. */
lis 6,VG_(machine_ppc32_has_VMX)@ha
lwz 6,VG_(machine_ppc32_has_VMX)@l(6)
cmplwi 6,0
beq LafterVMX2
#ifdef HAS_ALTIVEC
vspltisw 3,0x0 /* generate zero */
mtvscr 3 /* sets VSCR[NJ]=0 */
#endif
LafterVMX2:
/* make a stack frame for the code we are calling */
stwu 1,-16(1)
/* Set up the guest state ptr */
mr 31,4 /* r31 (generated code gsp) = r4 */
/* and jump into the code cache. Chained translations in
the code cache run, until for whatever reason, they can't
continue. When that happens, the translation in question
will jump (or call) to one of the continuation points
VG_(cp_...) below. */
mtctr 5
bctr
/*NOTREACHED*/
/*----------------------------------------------------*/
/*--- Postamble and exit. ---*/
/*----------------------------------------------------*/
postamble:
/* At this point, r6 and r7 contain two
words to be returned to the caller. r6
holds a TRC value, and r7 optionally may
hold another word (for CHAIN_ME exits, the
address of the place to patch.) */
/* We're leaving. Check that nobody messed with
VSCR or FPSCR in ways we don't expect. */
/* Using r10 - value used again further on, so don't trash! */
lis 10,VG_(machine_ppc32_has_FP)@ha
lwz 10,VG_(machine_ppc32_has_FP)@l(10)
/* Using r11 - value used again further on, so don't trash! */
lis 11,VG_(machine_ppc32_has_VMX)@ha
lwz 11,VG_(machine_ppc32_has_VMX)@l(11)
cmplwi 10,0 /* Do we have FP ? */
beq LafterFP8
/* Set fpscr back to a known state, since vex-generated code
may have messed with fpscr[rm]. */
li 5,0
addi 1,1,-16
stw 5,0(1)
lfs 3,0(1)
addi 1,1,16
mtfsf 0xFF,3 /* fpscr = f3 */
LafterFP8:
cmplwi 11,0 /* Do we have altivec? */
beq LafterVMX8
#ifdef HAS_ALTIVEC
/* Expect VSCR[NJ] to be 0, call invariant_violation if
VSCR[NJ] == 1 . */
/* first generate 4x 0x00010000 */
vspltisw 4,0x1 /* 4x 0x00000001 */
vspltisw 5,0x0 /* zero */
vsldoi 6,4,5,0x2 /* <<2*8 => 4x 0x00010000 */
/* retrieve VSCR and mask wanted bits */
mfvscr 7
vand 7,7,6 /* gives NJ flag */
vspltw 7,7,0x3 /* flags-word to all lanes */
vcmpequw. 8,6,7 /* CR[24] = 1 if v6 == v7 */
bt 24,invariant_violation /* branch if all_equal */
#endif
LafterVMX8:
/* otherwise we're OK */
b remove_frame
invariant_violation:
li 6,VG_TRC_INVARIANT_FAILED
li 7,0
/* fall through */
remove_frame:
/* Restore FP regs */
/* r10 already holds VG_(machine_ppc32_has_FP) value */
cmplwi 10,0
beq LafterFP9
/* Floating-point regs */
lfd 31,488(1)
lfd 30,480(1)
lfd 29,472(1)
lfd 28,464(1)
lfd 27,456(1)
lfd 26,448(1)
lfd 25,440(1)
lfd 24,432(1)
lfd 23,424(1)
lfd 22,416(1)
lfd 21,408(1)
lfd 20,400(1)
lfd 19,392(1)
lfd 18,384(1)
lfd 17,376(1)
lfd 16,368(1)
lfd 15,360(1)
lfd 14,352(1)
LafterFP9:
/* r11 already holds VG_(machine_ppc32_has_VMX) value */
cmplwi 11,0
beq LafterVMX9
/* Restore Altivec regs */
#ifdef HAS_ALTIVEC
/* VRSAVE */
lwz 4,244(1)
mfspr 4,256 /* VRSAVE reg is spr number 256 */
/* Vector regs */
li 4,224
lvx 31,4,1
li 4,208
lvx 30,4,1
li 4,192
lvx 29,4,1
li 4,176
lvx 28,4,1
li 4,160
lvx 27,4,1
li 4,144
lvx 26,4,1
li 4,128
lvx 25,4,1
li 4,112
lvx 24,4,1
li 4,96
lvx 23,4,1
li 4,80
lvx 22,4,1
li 4,64
lvx 21,4,1
li 4,48
lvx 20,4,1
#endif
LafterVMX9:
/* restore int regs, including importantly r3 (two_words) */
addi 1,1,16
lwz 31,348(1)
lwz 30,344(1)
lwz 29,340(1)
lwz 28,336(1)
lwz 27,332(1)
lwz 26,328(1)
lwz 25,324(1)
lwz 24,320(1)
lwz 23,316(1)
lwz 22,312(1)
lwz 21,308(1)
lwz 20,304(1)
lwz 19,300(1)
lwz 18,296(1)
lwz 17,292(1)
lwz 16,288(1)
lwz 15,284(1)
lwz 14,280(1)
lwz 13,276(1)
lwz 3,272(1)
/* Stash return values */
stw 6,0(3)
stw 7,4(3)
/* restore lr & sp, and leave */
lwz 0,500(1) /* stack_size + 4 */
mtlr 0
addi 1,1,496 /* stack_size */
blr
/*----------------------------------------------------*/
/*--- Continuation points ---*/
/*----------------------------------------------------*/
/* ------ Chain me to slow entry point ------ */
.global VG_(disp_cp_chain_me_to_slowEP)
VG_(disp_cp_chain_me_to_slowEP):
/* We got called. The return address indicates
where the patching needs to happen. Collect
the return address and, exit back to C land,
handing the caller the pair (Chain_me_S, RA) */
li 6, VG_TRC_CHAIN_ME_TO_SLOW_EP
mflr 7
/* 8 = imm32-fixed2 r30, disp_cp_chain_me_to_slowEP
4 = mtctr r30
4 = btctr
*/
subi 7,7,8+4+4
b postamble
/* ------ Chain me to fast entry point ------ */
.global VG_(disp_cp_chain_me_to_fastEP)
VG_(disp_cp_chain_me_to_fastEP):
/* We got called. The return address indicates
where the patching needs to happen. Collect
the return address and, exit back to C land,
handing the caller the pair (Chain_me_S, RA) */
li 6, VG_TRC_CHAIN_ME_TO_FAST_EP
mflr 7
/* 8 = imm32-fixed2 r30, disp_cp_chain_me_to_fastEP
4 = mtctr r30
4 = btctr
*/
subi 7,7,8+4+4
b postamble
/* ------ Indirect but boring jump ------ */
.global VG_(disp_cp_xindir)
VG_(disp_cp_xindir):
/* Where are we going? */
lwz 20, OFFSET_ppc32_CIA(31)
/* stats only */
lis 24, VG_(stats__n_xIndirs_32)@ha
addi 24, 24, VG_(stats__n_xIndirs_32)@l
lwz 25, 0(24)
addi 25, 25, 1
stw 25, 0(24)
// LIVE: r31 (guest state ptr), r20 (guest address to go to).
// We use 6 temporaries:
// r26 (to point at the relevant FastCacheSet),
// r21, r22, r23 (scratch, for swapping entries within a set)
// r24, r25 (other scratch)
/* Try a fast lookup in the translation cache. This is pretty much
a handcoded version of VG_(lookupInFastCache). */
// Compute r26 = VG_TT_FAST_HASH(guest)
srwi 26, 20, 2 // g2 = guest >> 2
srwi 25, 20, (VG_TT_FAST_BITS + 2) // (g2 >> VG_TT_FAST_BITS)
xor 26, 26, 25 // (g2 >> VG_TT_FAST_BITS) ^ g2
andi. 26, 26, VG_TT_FAST_MASK // setNo
// Compute r6 = &VG_(tt_fast)[r6]
lis 25, VG_(tt_fast)@ha
addi 25, 25, VG_(tt_fast)@l
slwi 26, 26, VG_FAST_CACHE_SET_BITS
add 26, 26, 25
// LIVE: r31 (guest state ptr), r20 (guest addr), r26 (cache set)
// try way 0
lwz 24, FCS_g0(26) // .guest0
lwz 25, FCS_h0(26) // .host0
cmpw 24, 20 // cmp against .guest0
bne 1f
// hit at way 0
// goto .host0
mtctr 25
bctr
/*NOTREACHED*/
1: // try way 1
lwz 24, FCS_g1(26)
cmpw 24, 20 // cmp against .guest1
bne 2f
// hit at way 1; swap upwards
lwz 21, FCS_g0(26) // 21 = old .guest0
lwz 22, FCS_h0(26) // 22 = old .host0
lwz 23, FCS_h1(26) // 23 = old .host1
stw 20, FCS_g0(26) // new .guest0 = guest
stw 23, FCS_h0(26) // new .host0 = old .host1
stw 21, FCS_g1(26) // new .guest1 = old .guest0
stw 22, FCS_h1(26) // new .host1 = old .host0
// stats only
lis 24, VG_(stats__n_xIndir_hits1_32)@ha
addi 24, 24, VG_(stats__n_xIndir_hits1_32)@l
lwz 25, 0(24)
addi 25, 25, 1
stw 25, 0(24)
// goto old .host1 a.k.a. new .host0
mtctr 23
bctr
/*NOTREACHED*/
2: // try way 2
lwz 24, FCS_g2(26)
cmpw 24, 20 // cmp against .guest2
bne 3f
// hit at way 2; swap upwards
lwz 21, FCS_g1(26)
lwz 22, FCS_h1(26)
lwz 23, FCS_h2(26)
stw 20, FCS_g1(26)
stw 23, FCS_h1(26)
stw 21, FCS_g2(26)
stw 22, FCS_h2(26)
// stats only
lis 24, VG_(stats__n_xIndir_hits2_32)@ha
addi 24, 24, VG_(stats__n_xIndir_hits2_32)@l
lwz 25, 0(24)
addi 25, 25, 1
stw 25, 0(24)
// goto old .host2 a.k.a. new .host1
mtctr 23
bctr
/*NOTREACHED*/
3: // try way 3
lwz 24, FCS_g3(26)
cmpw 24, 20 // cmp against .guest3
bne 4f
// hit at way 3; swap upwards
lwz 21, FCS_g2(26)
lwz 22, FCS_h2(26)
lwz 23, FCS_h3(26)
stw 20, FCS_g2(26)
stw 23, FCS_h2(26)
stw 21, FCS_g3(26)
stw 22, FCS_h3(26)
// stats only
lis 24, VG_(stats__n_xIndir_hits3_32)@ha
addi 24, 24, VG_(stats__n_xIndir_hits3_32)@l
lwz 25, 0(24)
addi 25, 25, 1
stw 25, 0(24)
// goto old .host3 a.k.a. new .host2
mtctr 23
bctr
/*NOTREACHED*/
4: // fast lookup failed:
/* stats only */
lis 24, VG_(stats__n_xIndir_misses_32)@ha
addi 24, 24, VG_(stats__n_xIndir_misses_32)@l
lwz 25, 0(24)
addi 25, 25, 1
stw 25, 0(24)
li 6, VG_TRC_INNER_FASTMISS
li 7, 0
b postamble
/*NOTREACHED*/
/* ------ Assisted jump ------ */
.global VG_(disp_cp_xassisted)
VG_(disp_cp_xassisted):
/* r31 contains the TRC */
mr 6,31
li 7,0
b postamble
/* ------ Event check failed ------ */
.global VG_(disp_cp_evcheck_fail)
VG_(disp_cp_evcheck_fail):
li 6,VG_TRC_INNER_COUNTERZERO
li 7,0
b postamble
.size VG_(disp_run_translations), .-VG_(disp_run_translations)
#endif // defined(VGP_ppc32_linux)
/* Let the linker know we don't need an executable stack */
MARK_STACK_NO_EXEC
/*--------------------------------------------------------------------*/
/*--- end ---*/
/*--------------------------------------------------------------------*/
|
