aboutsummaryrefslogblamecommitdiffstats
path: root/arch/blackfin/mach-bf561/atomic.S
blob: 52d6f73fccedd152fb215702bebf28556ab3487e (plain) (tree)
1
2
3
4
5
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
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
  

                                                
  
                                     















                                









                                         













                                                               
                            




                                   
                                                                       
















                                                                    
                            









































































































































































































































































































































































































































































































































































































































                                                                                 

                                                                       





























































































































































































































                                                                                

                                                                       




                                  
/*
 * Copyright 2007-2008 Analog Devices Inc.
 *              Philippe Gerum <rpm@xenomai.org>
 *
 * Licensed under the GPL-2 or later.
 */

#include <linux/linkage.h>
#include <asm/blackfin.h>
#include <asm/cache.h>
#include <asm/asm-offsets.h>
#include <asm/rwlock.h>
#include <asm/cplb.h>

.text

.macro coreslot_loadaddr reg:req
	\reg\().l = _corelock;
	\reg\().h = _corelock;
.endm

.macro safe_testset addr:req, scratch:req
#if ANOMALY_05000477
	cli \scratch;
	testset (\addr);
	sti \scratch;
#else
	testset (\addr);
#endif
.endm

/*
 * r0 = address of atomic data to flush and invalidate (32bit).
 *
 * Clear interrupts and return the old mask.
 * We assume that no atomic data can span cachelines.
 *
 * Clobbers: r2:0, p0
 */
ENTRY(_get_core_lock)
	r1 = -L1_CACHE_BYTES;
	r1 = r0 & r1;
	cli r0;
	coreslot_loadaddr p0;
.Lretry_corelock:
	safe_testset p0, r2;
	if cc jump .Ldone_corelock;
	SSYNC(r2);
	jump .Lretry_corelock
.Ldone_corelock:
	p0 = r1;
	/* flush core internal write buffer before invalidate dcache */
	CSYNC(r2);
	flushinv[p0];
	SSYNC(r2);
	rts;
ENDPROC(_get_core_lock)

/*
 * r0 = address of atomic data in uncacheable memory region (32bit).
 *
 * Clear interrupts and return the old mask.
 *
 * Clobbers: r0, p0
 */
ENTRY(_get_core_lock_noflush)
	cli r0;
	coreslot_loadaddr p0;
.Lretry_corelock_noflush:
	safe_testset p0, r2;
	if cc jump .Ldone_corelock_noflush;
	SSYNC(r2);
	jump .Lretry_corelock_noflush
.Ldone_corelock_noflush:
	rts;
ENDPROC(_get_core_lock_noflush)

/*
 * r0 = interrupt mask to restore.
 * r1 = address of atomic data to flush and invalidate (32bit).
 *
 * Interrupts are masked on entry (see _get_core_lock).
 * Clobbers: r2:0, p0
 */
ENTRY(_put_core_lock)
	/* Write-through cache assumed, so no flush needed here. */
	coreslot_loadaddr p0;
	r1 = 0;
	[p0] = r1;
	SSYNC(r2);
	sti r0;
	rts;
ENDPROC(_put_core_lock)

#ifdef __ARCH_SYNC_CORE_DCACHE

ENTRY(___raw_smp_mark_barrier_asm)
	[--sp] = rets;
	[--sp] = ( r7:5 );
	[--sp] = r0;
	[--sp] = p1;
	[--sp] = p0;
	call _get_core_lock_noflush;

	/*
	 * Calculate current core mask
	 */
	GET_CPUID(p1, r7);
	r6 = 1;
	r6 <<= r7;

	/*
	 * Set bit of other cores in barrier mask. Don't change current core bit.
	 */
	p1.l = _barrier_mask;
	p1.h = _barrier_mask;
	r7 = [p1];
	r5 = r7 & r6;
	r7 = ~r6;
	cc = r5 == 0;
	if cc jump 1f;
	r7 = r7 | r6;
1:
	[p1] = r7;
	SSYNC(r2);

	call _put_core_lock;
	p0 = [sp++];
	p1 = [sp++];
	r0 = [sp++];
	( r7:5 ) = [sp++];
	rets = [sp++];
	rts;
ENDPROC(___raw_smp_mark_barrier_asm)

ENTRY(___raw_smp_check_barrier_asm)
	[--sp] = rets;
	[--sp] = ( r7:5 );
	[--sp] = r0;
	[--sp] = p1;
	[--sp] = p0;
	call _get_core_lock_noflush;

	/*
	 * Calculate current core mask
	 */
	GET_CPUID(p1, r7);
	r6 = 1;
	r6 <<= r7;

	/*
	 * Clear current core bit in barrier mask if it is set.
	 */
	p1.l = _barrier_mask;
	p1.h = _barrier_mask;
	r7 = [p1];
	r5 = r7 & r6;
	cc = r5 == 0;
	if cc jump 1f;
	r6 = ~r6;
	r7 = r7 & r6;
	[p1] = r7;
	SSYNC(r2);

	call _put_core_lock;

	/*
	 * Invalidate the entire D-cache of current core.
	 */
	sp += -12;
	call _resync_core_dcache
	sp += 12;
	jump 2f;
1:
	call _put_core_lock;
2:
	p0 = [sp++];
	p1 = [sp++];
	r0 = [sp++];
	( r7:5 ) = [sp++];
	rets = [sp++];
	rts;
ENDPROC(___raw_smp_check_barrier_asm)

/*
 * r0 = irqflags
 * r1 = address of atomic data
 *
 * Clobbers: r2:0, p1:0
 */
_start_lock_coherent:

	[--sp] = rets;
	[--sp] = ( r7:6 );
	r7 = r0;
	p1 = r1;

	/*
	 * Determine whether the atomic data was previously
	 * owned by another CPU (=r6).
	 */
	GET_CPUID(p0, r2);
	r1 = 1;
	r1 <<= r2;
	r2 = ~r1;

	r1 = [p1];
	r1 >>= 28;   /* CPU fingerprints are stored in the high nibble. */
	r6 = r1 & r2;
	r1 = [p1];
	r1 <<= 4;
	r1 >>= 4;
	[p1] = r1;

	/*
	 * Release the core lock now, but keep IRQs disabled while we are
	 * performing the remaining housekeeping chores for the current CPU.
	 */
	coreslot_loadaddr p0;
	r1 = 0;
	[p0] = r1;

	/*
	 * If another CPU has owned the same atomic section before us,
	 * then our D-cached copy of the shared data protected by the
	 * current spin/write_lock may be obsolete.
	 */
	cc = r6 == 0;
	if cc jump .Lcache_synced

	/*
	 * Invalidate the entire D-cache of the current core.
	 */
	sp += -12;
	call _resync_core_dcache
	sp += 12;

.Lcache_synced:
	SSYNC(r2);
	sti r7;
	( r7:6 ) = [sp++];
	rets = [sp++];
	rts

/*
 * r0 = irqflags
 * r1 = address of atomic data
 *
 * Clobbers: r2:0, p1:0
 */
_end_lock_coherent:

	p1 = r1;
	GET_CPUID(p0, r2);
	r2 += 28;
	r1 = 1;
	r1 <<= r2;
	r2 = [p1];
	r2 = r1 | r2;
	[p1] = r2;
	r1 = p1;
	jump _put_core_lock;

#endif /* __ARCH_SYNC_CORE_DCACHE */

/*
 * r0 = &spinlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_spin_is_locked_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
	r3 = [p1];
	cc = bittst( r3, 0 );
	r3 = cc;
	r1 = p1;
	call _put_core_lock;
	rets = [sp++];
	r0 = r3;
	rts;
ENDPROC(___raw_spin_is_locked_asm)

/*
 * r0 = &spinlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_spin_lock_asm)
	p1 = r0;
	[--sp] = rets;
.Lretry_spinlock:
	call _get_core_lock;
	r1 = p1;
	r2 = [p1];
	cc = bittst( r2, 0 );
	if cc jump .Lbusy_spinlock
#ifdef __ARCH_SYNC_CORE_DCACHE
	r3 = p1;
	bitset ( r2, 0 ); /* Raise the lock bit. */
	[p1] = r2;
	call _start_lock_coherent
#else
	r2 = 1;
	[p1] = r2;
	call _put_core_lock;
#endif
	rets = [sp++];
	rts;

.Lbusy_spinlock:
	/* We don't touch the atomic area if busy, so that flush
	   will behave like nop in _put_core_lock. */
	call _put_core_lock;
	SSYNC(r2);
	r0 = p1;
	jump .Lretry_spinlock
ENDPROC(___raw_spin_lock_asm)

/*
 * r0 = &spinlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_spin_trylock_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
	r1 = p1;
	r3 = [p1];
	cc = bittst( r3, 0 );
	if cc jump .Lfailed_trylock
#ifdef __ARCH_SYNC_CORE_DCACHE
	bitset ( r3, 0 ); /* Raise the lock bit. */
	[p1] = r3;
	call _start_lock_coherent
#else
	r2 = 1;
	[p1] = r2;
	call _put_core_lock;
#endif
	r0 = 1;
	rets = [sp++];
	rts;
.Lfailed_trylock:
	call _put_core_lock;
	r0 = 0;
	rets = [sp++];
	rts;
ENDPROC(___raw_spin_trylock_asm)

/*
 * r0 = &spinlock->lock
 *
 * Clobbers: r2:0, p1:0
 */
ENTRY(___raw_spin_unlock_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
	r2 = [p1];
	bitclr ( r2, 0 );
	[p1] = r2;
	r1 = p1;
#ifdef __ARCH_SYNC_CORE_DCACHE
	call _end_lock_coherent
#else
	call _put_core_lock;
#endif
	rets = [sp++];
	rts;
ENDPROC(___raw_spin_unlock_asm)

/*
 * r0 = &rwlock->lock
 *
 * Clobbers: r2:0, p1:0
 */
ENTRY(___raw_read_lock_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
.Lrdlock_try:
	r1 = [p1];
	r1 += -1;
	[p1] = r1;
	cc = r1 < 0;
	if cc jump .Lrdlock_failed
	r1 = p1;
#ifdef __ARCH_SYNC_CORE_DCACHE
	call _start_lock_coherent
#else
	call _put_core_lock;
#endif
	rets = [sp++];
	rts;

.Lrdlock_failed:
	r1 += 1;
	[p1] = r1;
.Lrdlock_wait:
	r1 = p1;
	call _put_core_lock;
	SSYNC(r2);
	r0 = p1;
	call _get_core_lock;
	r1 = [p1];
	cc = r1 < 2;
	if cc jump .Lrdlock_wait;
	jump .Lrdlock_try
ENDPROC(___raw_read_lock_asm)

/*
 * r0 = &rwlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_read_trylock_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
	r1 = [p1];
	cc = r1 <= 0;
	if cc jump .Lfailed_tryrdlock;
	r1 += -1;
	[p1] = r1;
	r1 = p1;
#ifdef __ARCH_SYNC_CORE_DCACHE
	call _start_lock_coherent
#else
	call _put_core_lock;
#endif
	rets = [sp++];
	r0 = 1;
	rts;
.Lfailed_tryrdlock:
	r1 = p1;
	call _put_core_lock;
	rets = [sp++];
	r0 = 0;
	rts;
ENDPROC(___raw_read_trylock_asm)

/*
 * r0 = &rwlock->lock
 *
 * Note: Processing controlled by a reader lock should not have
 * any side-effect on cache issues with the other core, so we
 * just release the core lock and exit (no _end_lock_coherent).
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_read_unlock_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
	r1 = [p1];
	r1 += 1;
	[p1] = r1;
	r1 = p1;
	call _put_core_lock;
	rets = [sp++];
	rts;
ENDPROC(___raw_read_unlock_asm)

/*
 * r0 = &rwlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_write_lock_asm)
	p1 = r0;
	r3.l = lo(RW_LOCK_BIAS);
	r3.h = hi(RW_LOCK_BIAS);
	[--sp] = rets;
	call _get_core_lock;
.Lwrlock_try:
	r1 = [p1];
	r1 = r1 - r3;
#ifdef __ARCH_SYNC_CORE_DCACHE
	r2 = r1;
	r2 <<= 4;
	r2 >>= 4;
	cc = r2 == 0;
#else
	cc = r1 == 0;
#endif
	if !cc jump .Lwrlock_wait
	[p1] = r1;
	r1 = p1;
#ifdef __ARCH_SYNC_CORE_DCACHE
	call _start_lock_coherent
#else
	call _put_core_lock;
#endif
	rets = [sp++];
	rts;

.Lwrlock_wait:
	r1 = p1;
	call _put_core_lock;
	SSYNC(r2);
	r0 = p1;
	call _get_core_lock;
	r1 = [p1];
#ifdef __ARCH_SYNC_CORE_DCACHE
	r1 <<= 4;
	r1 >>= 4;
#endif
	cc = r1 == r3;
	if !cc jump .Lwrlock_wait;
	jump .Lwrlock_try
ENDPROC(___raw_write_lock_asm)

/*
 * r0 = &rwlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_write_trylock_asm)
	p1 = r0;
	[--sp] = rets;
	call _get_core_lock;
	r1 = [p1];
	r2.l = lo(RW_LOCK_BIAS);
	r2.h = hi(RW_LOCK_BIAS);
	cc = r1 == r2;
	if !cc jump .Lfailed_trywrlock;
#ifdef __ARCH_SYNC_CORE_DCACHE
	r1 >>= 28;
	r1 <<= 28;
#else
	r1 = 0;
#endif
	[p1] = r1;
	r1 = p1;
#ifdef __ARCH_SYNC_CORE_DCACHE
	call _start_lock_coherent
#else
	call _put_core_lock;
#endif
	rets = [sp++];
	r0 = 1;
	rts;

.Lfailed_trywrlock:
	r1 = p1;
	call _put_core_lock;
	rets = [sp++];
	r0 = 0;
	rts;
ENDPROC(___raw_write_trylock_asm)

/*
 * r0 = &rwlock->lock
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_write_unlock_asm)
	p1 = r0;
	r3.l = lo(RW_LOCK_BIAS);
	r3.h = hi(RW_LOCK_BIAS);
	[--sp] = rets;
	call _get_core_lock;
	r1 = [p1];
	r1 = r1 + r3;
	[p1] = r1;
	r1 = p1;
#ifdef __ARCH_SYNC_CORE_DCACHE
	call _end_lock_coherent
#else
	call _put_core_lock;
#endif
	rets = [sp++];
	rts;
ENDPROC(___raw_write_unlock_asm)

/*
 * r0 = ptr
 * r1 = value
 *
 * Add a signed value to a 32bit word and return the new value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_atomic_update_asm)
	p1 = r0;
	r3 = r1;
	[--sp] = rets;
	call _get_core_lock;
	r2 = [p1];
	r3 = r3 + r2;
	[p1] = r3;
	r1 = p1;
	call _put_core_lock;
	r0 = r3;
	rets = [sp++];
	rts;
ENDPROC(___raw_atomic_update_asm)

/*
 * r0 = ptr
 * r1 = mask
 *
 * Clear the mask bits from a 32bit word and return the old 32bit value
 * atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_atomic_clear_asm)
	p1 = r0;
	r3 = ~r1;
	[--sp] = rets;
	call _get_core_lock;
	r2 = [p1];
	r3 = r2 & r3;
	[p1] = r3;
	r3 = r2;
	r1 = p1;
	call _put_core_lock;
	r0 = r3;
	rets = [sp++];
	rts;
ENDPROC(___raw_atomic_clear_asm)

/*
 * r0 = ptr
 * r1 = mask
 *
 * Set the mask bits into a 32bit word and return the old 32bit value
 * atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_atomic_set_asm)
	p1 = r0;
	r3 = r1;
	[--sp] = rets;
	call _get_core_lock;
	r2 = [p1];
	r3 = r2 | r3;
	[p1] = r3;
	r3 = r2;
	r1 = p1;
	call _put_core_lock;
	r0 = r3;
	rets = [sp++];
	rts;
ENDPROC(___raw_atomic_set_asm)

/*
 * r0 = ptr
 * r1 = mask
 *
 * XOR the mask bits with a 32bit word and return the old 32bit value
 * atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_atomic_xor_asm)
	p1 = r0;
	r3 = r1;
	[--sp] = rets;
	call _get_core_lock;
	r2 = [p1];
	r3 = r2 ^ r3;
	[p1] = r3;
	r3 = r2;
	r1 = p1;
	call _put_core_lock;
	r0 = r3;
	rets = [sp++];
	rts;
ENDPROC(___raw_atomic_xor_asm)

/*
 * r0 = ptr
 * r1 = mask
 *
 * Perform a logical AND between the mask bits and a 32bit word, and
 * return the masked value. We need this on this architecture in
 * order to invalidate the local cache before testing.
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_atomic_test_asm)
	p1 = r0;
	r3 = r1;
	r1 = -L1_CACHE_BYTES;
	r1 = r0 & r1;
	p0 = r1;
	/* flush core internal write buffer before invalidate dcache */
	CSYNC(r2);
	flushinv[p0];
	SSYNC(r2);
	r0 = [p1];
	r0 = r0 & r3;
	rts;
ENDPROC(___raw_atomic_test_asm)

/*
 * r0 = ptr
 * r1 = value
 *
 * Swap *ptr with value and return the old 32bit value atomically.
 * Clobbers: r3:0, p1:0
 */
#define	__do_xchg(src, dst) 		\
	p1 = r0;			\
	r3 = r1;			\
	[--sp] = rets;			\
	call _get_core_lock;		\
	r2 = src;			\
	dst = r3;			\
	r3 = r2;			\
	r1 = p1;			\
	call _put_core_lock;		\
	r0 = r3;			\
	rets = [sp++];			\
	rts;

ENTRY(___raw_xchg_1_asm)
	__do_xchg(b[p1] (z), b[p1])
ENDPROC(___raw_xchg_1_asm)

ENTRY(___raw_xchg_2_asm)
	__do_xchg(w[p1] (z), w[p1])
ENDPROC(___raw_xchg_2_asm)

ENTRY(___raw_xchg_4_asm)
	__do_xchg([p1], [p1])
ENDPROC(___raw_xchg_4_asm)

/*
 * r0 = ptr
 * r1 = new
 * r2 = old
 *
 * Swap *ptr with new if *ptr == old and return the previous *ptr
 * value atomically.
 *
 * Clobbers: r3:0, p1:0
 */
#define	__do_cmpxchg(src, dst) 		\
	[--sp] = rets;			\
	[--sp] = r4;			\
	p1 = r0;			\
	r3 = r1;			\
	r4 = r2;			\
	call _get_core_lock;		\
	r2 = src;			\
	cc = r2 == r4;			\
	if !cc jump 1f;			\
	dst = r3;			\
     1: r3 = r2;			\
	r1 = p1;			\
	call _put_core_lock;		\
	r0 = r3;			\
	r4 = [sp++];			\
	rets = [sp++];			\
	rts;

ENTRY(___raw_cmpxchg_1_asm)
	__do_cmpxchg(b[p1] (z), b[p1])
ENDPROC(___raw_cmpxchg_1_asm)

ENTRY(___raw_cmpxchg_2_asm)
	__do_cmpxchg(w[p1] (z), w[p1])
ENDPROC(___raw_cmpxchg_2_asm)

ENTRY(___raw_cmpxchg_4_asm)
	__do_cmpxchg([p1], [p1])
ENDPROC(___raw_cmpxchg_4_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Set a bit in a 32bit word and return the old 32bit value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_set_asm)
	r2 = r1;
	r1 = 1;
	r1 <<= r2;
	jump ___raw_atomic_set_asm
ENDPROC(___raw_bit_set_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Clear a bit in a 32bit word and return the old 32bit value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_clear_asm)
	r2 = r1;
	r1 = 1;
	r1 <<= r2;
	jump ___raw_atomic_clear_asm
ENDPROC(___raw_bit_clear_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Toggle a bit in a 32bit word and return the old 32bit value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_toggle_asm)
	r2 = r1;
	r1 = 1;
	r1 <<= r2;
	jump ___raw_atomic_xor_asm
ENDPROC(___raw_bit_toggle_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Test-and-set a bit in a 32bit word and return the old bit value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_test_set_asm)
	[--sp] = rets;
	[--sp] = r1;
	call ___raw_bit_set_asm
	r1 = [sp++];
	r2 = 1;
	r2 <<= r1;
	r0 = r0 & r2;
	cc = r0 == 0;
	if cc jump 1f
	r0 = 1;
1:
	rets = [sp++];
	rts;
ENDPROC(___raw_bit_test_set_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Test-and-clear a bit in a 32bit word and return the old bit value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_test_clear_asm)
	[--sp] = rets;
	[--sp] = r1;
	call ___raw_bit_clear_asm
	r1 = [sp++];
	r2 = 1;
	r2 <<= r1;
	r0 = r0 & r2;
	cc = r0 == 0;
	if cc jump 1f
	r0 = 1;
1:
	rets = [sp++];
	rts;
ENDPROC(___raw_bit_test_clear_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Test-and-toggle a bit in a 32bit word,
 * and return the old bit value atomically.
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_test_toggle_asm)
	[--sp] = rets;
	[--sp] = r1;
	call ___raw_bit_toggle_asm
	r1 = [sp++];
	r2 = 1;
	r2 <<= r1;
	r0 = r0 & r2;
	cc = r0 == 0;
	if cc jump 1f
	r0 = 1;
1:
	rets = [sp++];
	rts;
ENDPROC(___raw_bit_test_toggle_asm)

/*
 * r0 = ptr
 * r1 = bitnr
 *
 * Test a bit in a 32bit word and return its value.
 * We need this on this architecture in order to invalidate
 * the local cache before testing.
 *
 * Clobbers: r3:0, p1:0
 */
ENTRY(___raw_bit_test_asm)
	r2 = r1;
	r1 = 1;
	r1 <<= r2;
	jump ___raw_atomic_test_asm
ENDPROC(___raw_bit_test_asm)

/*
 * r0 = ptr
 *
 * Fetch and return an uncached 32bit value.
 *
 * Clobbers: r2:0, p1:0
 */
ENTRY(___raw_uncached_fetch_asm)
	p1 = r0;
	r1 = -L1_CACHE_BYTES;
	r1 = r0 & r1;
	p0 = r1;
	/* flush core internal write buffer before invalidate dcache */
	CSYNC(r2);
	flushinv[p0];
	SSYNC(r2);
	r0 = [p1];
	rts;
ENDPROC(___raw_uncached_fetch_asm)