aboutsummaryrefslogtreecommitdiffstats
path: root/arch/sparc/kernel/perf_event.c
blob: fa5936e1c3b9e8d632c024d5dd6ea0ce8863eba1 (plain)
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
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
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
/* Performance event support for sparc64.
 *
 * Copyright (C) 2009 David S. Miller <davem@davemloft.net>
 *
 * This code is based almost entirely upon the x86 perf event
 * code, which is:
 *
 *  Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de>
 *  Copyright (C) 2008-2009 Red Hat, Inc., Ingo Molnar
 *  Copyright (C) 2009 Jaswinder Singh Rajput
 *  Copyright (C) 2009 Advanced Micro Devices, Inc., Robert Richter
 *  Copyright (C) 2008-2009 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com>
 */

#include <linux/perf_event.h>
#include <linux/kprobes.h>
#include <linux/kernel.h>
#include <linux/kdebug.h>
#include <linux/mutex.h>

#include <asm/cpudata.h>
#include <asm/atomic.h>
#include <asm/nmi.h>
#include <asm/pcr.h>

/* Sparc64 chips have two performance counters, 32-bits each, with
 * overflow interrupts generated on transition from 0xffffffff to 0.
 * The counters are accessed in one go using a 64-bit register.
 *
 * Both counters are controlled using a single control register.  The
 * only way to stop all sampling is to clear all of the context (user,
 * supervisor, hypervisor) sampling enable bits.  But these bits apply
 * to both counters, thus the two counters can't be enabled/disabled
 * individually.
 *
 * The control register has two event fields, one for each of the two
 * counters.  It's thus nearly impossible to have one counter going
 * while keeping the other one stopped.  Therefore it is possible to
 * get overflow interrupts for counters not currently "in use" and
 * that condition must be checked in the overflow interrupt handler.
 *
 * So we use a hack, in that we program inactive counters with the
 * "sw_count0" and "sw_count1" events.  These count how many times
 * the instruction "sethi %hi(0xfc000), %g0" is executed.  It's an
 * unusual way to encode a NOP and therefore will not trigger in
 * normal code.
 */

#define MAX_HWEVENTS			2
#define MAX_PERIOD			((1UL << 32) - 1)

#define PIC_UPPER_INDEX			0
#define PIC_LOWER_INDEX			1

struct cpu_hw_events {
	struct perf_event	*events[MAX_HWEVENTS];
	unsigned long		used_mask[BITS_TO_LONGS(MAX_HWEVENTS)];
	unsigned long		active_mask[BITS_TO_LONGS(MAX_HWEVENTS)];
	u64			pcr;
	int			enabled;
};
DEFINE_PER_CPU(struct cpu_hw_events, cpu_hw_events) = { .enabled = 1, };

struct perf_event_map {
	u16	encoding;
	u8	pic_mask;
#define PIC_NONE	0x00
#define PIC_UPPER	0x01
#define PIC_LOWER	0x02
};

static unsigned long perf_event_encode(const struct perf_event_map *pmap)
{
	return ((unsigned long) pmap->encoding << 16) | pmap->pic_mask;
}

static void perf_event_decode(unsigned long val, u16 *enc, u8 *msk)
{
	*msk = val & 0xff;
	*enc = val >> 16;
}

#define C(x) PERF_COUNT_HW_CACHE_##x

#define CACHE_OP_UNSUPPORTED	0xfffe
#define CACHE_OP_NONSENSE	0xffff

typedef struct perf_event_map cache_map_t
				[PERF_COUNT_HW_CACHE_MAX]
				[PERF_COUNT_HW_CACHE_OP_MAX]
				[PERF_COUNT_HW_CACHE_RESULT_MAX];

struct sparc_pmu {
	const struct perf_event_map	*(*event_map)(int);
	const cache_map_t		*cache_map;
	int				max_events;
	int				upper_shift;
	int				lower_shift;
	int				event_mask;
	int				hv_bit;
	int				irq_bit;
	int				upper_nop;
	int				lower_nop;
};

static const struct perf_event_map ultra3_perfmon_event_map[] = {
	[PERF_COUNT_HW_CPU_CYCLES] = { 0x0000, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_INSTRUCTIONS] = { 0x0001, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_CACHE_REFERENCES] = { 0x0009, PIC_LOWER },
	[PERF_COUNT_HW_CACHE_MISSES] = { 0x0009, PIC_UPPER },
};

static const struct perf_event_map *ultra3_event_map(int event_id)
{
	return &ultra3_perfmon_event_map[event_id];
}

static const cache_map_t ultra3_cache_map = {
[C(L1D)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x09, PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x09, PIC_UPPER, },
	},
	[C(OP_WRITE)] = {
		[C(RESULT_ACCESS)] = { 0x0a, PIC_LOWER },
		[C(RESULT_MISS)] = { 0x0a, PIC_UPPER },
	},
	[C(OP_PREFETCH)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(L1I)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x09, PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x09, PIC_UPPER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
		[ C(RESULT_MISS)   ] = { CACHE_OP_NONSENSE },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(LL)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x0c, PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x0c, PIC_UPPER, },
	},
	[C(OP_WRITE)] = {
		[C(RESULT_ACCESS)] = { 0x0c, PIC_LOWER },
		[C(RESULT_MISS)] = { 0x0c, PIC_UPPER },
	},
	[C(OP_PREFETCH)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(DTLB)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x12, PIC_UPPER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(ITLB)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x11, PIC_UPPER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(BPU)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
};

static const struct sparc_pmu ultra3_pmu = {
	.event_map	= ultra3_event_map,
	.cache_map	= &ultra3_cache_map,
	.max_events	= ARRAY_SIZE(ultra3_perfmon_event_map),
	.upper_shift	= 11,
	.lower_shift	= 4,
	.event_mask	= 0x3f,
	.upper_nop	= 0x1c,
	.lower_nop	= 0x14,
};

/* Niagara1 is very limited.  The upper PIC is hard-locked to count
 * only instructions, so it is free running which creates all kinds of
 * problems.  Some hardware designs make one wonder if the creator
 * even looked at how this stuff gets used by software.
 */
static const struct perf_event_map niagara1_perfmon_event_map[] = {
	[PERF_COUNT_HW_CPU_CYCLES] = { 0x00, PIC_UPPER },
	[PERF_COUNT_HW_INSTRUCTIONS] = { 0x00, PIC_UPPER },
	[PERF_COUNT_HW_CACHE_REFERENCES] = { 0, PIC_NONE },
	[PERF_COUNT_HW_CACHE_MISSES] = { 0x03, PIC_LOWER },
};

static const struct perf_event_map *niagara1_event_map(int event_id)
{
	return &niagara1_perfmon_event_map[event_id];
}

static const cache_map_t niagara1_cache_map = {
[C(L1D)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x03, PIC_LOWER, },
	},
	[C(OP_WRITE)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x03, PIC_LOWER, },
	},
	[C(OP_PREFETCH)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(L1I)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x00, PIC_UPPER },
		[C(RESULT_MISS)] = { 0x02, PIC_LOWER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
		[ C(RESULT_MISS)   ] = { CACHE_OP_NONSENSE },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(LL)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x07, PIC_LOWER, },
	},
	[C(OP_WRITE)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x07, PIC_LOWER, },
	},
	[C(OP_PREFETCH)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(DTLB)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x05, PIC_LOWER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(ITLB)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x04, PIC_LOWER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(BPU)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
};

static const struct sparc_pmu niagara1_pmu = {
	.event_map	= niagara1_event_map,
	.cache_map	= &niagara1_cache_map,
	.max_events	= ARRAY_SIZE(niagara1_perfmon_event_map),
	.upper_shift	= 0,
	.lower_shift	= 4,
	.event_mask	= 0x7,
	.upper_nop	= 0x0,
	.lower_nop	= 0x0,
};

static const struct perf_event_map niagara2_perfmon_event_map[] = {
	[PERF_COUNT_HW_CPU_CYCLES] = { 0x02ff, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_INSTRUCTIONS] = { 0x02ff, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_CACHE_REFERENCES] = { 0x0208, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_CACHE_MISSES] = { 0x0302, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_BRANCH_INSTRUCTIONS] = { 0x0201, PIC_UPPER | PIC_LOWER },
	[PERF_COUNT_HW_BRANCH_MISSES] = { 0x0202, PIC_UPPER | PIC_LOWER },
};

static const struct perf_event_map *niagara2_event_map(int event_id)
{
	return &niagara2_perfmon_event_map[event_id];
}

static const cache_map_t niagara2_cache_map = {
[C(L1D)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x0208, PIC_UPPER | PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x0302, PIC_UPPER | PIC_LOWER, },
	},
	[C(OP_WRITE)] = {
		[C(RESULT_ACCESS)] = { 0x0210, PIC_UPPER | PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x0302, PIC_UPPER | PIC_LOWER, },
	},
	[C(OP_PREFETCH)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(L1I)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x02ff, PIC_UPPER | PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x0301, PIC_UPPER | PIC_LOWER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_NONSENSE },
		[ C(RESULT_MISS)   ] = { CACHE_OP_NONSENSE },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(LL)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { 0x0208, PIC_UPPER | PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x0330, PIC_UPPER | PIC_LOWER, },
	},
	[C(OP_WRITE)] = {
		[C(RESULT_ACCESS)] = { 0x0210, PIC_UPPER | PIC_LOWER, },
		[C(RESULT_MISS)] = { 0x0320, PIC_UPPER | PIC_LOWER, },
	},
	[C(OP_PREFETCH)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(DTLB)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0x0b08, PIC_UPPER | PIC_LOWER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(ITLB)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { 0xb04, PIC_UPPER | PIC_LOWER, },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
[C(BPU)] = {
	[C(OP_READ)] = {
		[C(RESULT_ACCESS)] = { CACHE_OP_UNSUPPORTED },
		[C(RESULT_MISS)] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_WRITE) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
	[ C(OP_PREFETCH) ] = {
		[ C(RESULT_ACCESS) ] = { CACHE_OP_UNSUPPORTED },
		[ C(RESULT_MISS)   ] = { CACHE_OP_UNSUPPORTED },
	},
},
};

static const struct sparc_pmu niagara2_pmu = {
	.event_map	= niagara2_event_map,
	.cache_map	= &niagara2_cache_map,
	.max_events	= ARRAY_SIZE(niagara2_perfmon_event_map),
	.upper_shift	= 19,
	.lower_shift	= 6,
	.event_mask	= 0xfff,
	.hv_bit		= 0x8,
	.irq_bit	= 0x30,
	.upper_nop	= 0x220,
	.lower_nop	= 0x220,
};

static const struct sparc_pmu *sparc_pmu __read_mostly;

static u64 event_encoding(u64 event_id, int idx)
{
	if (idx == PIC_UPPER_INDEX)
		event_id <<= sparc_pmu->upper_shift;
	else
		event_id <<= sparc_pmu->lower_shift;
	return event_id;
}

static u64 mask_for_index(int idx)
{
	return event_encoding(sparc_pmu->event_mask, idx);
}

static u64 nop_for_index(int idx)
{
	return event_encoding(idx == PIC_UPPER_INDEX ?
			      sparc_pmu->upper_nop :
			      sparc_pmu->lower_nop, idx);
}

static inline void sparc_pmu_enable_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc, int idx)
{
	u64 val, mask = mask_for_index(idx);

	val = cpuc->pcr;
	val &= ~mask;
	val |= hwc->config;
	cpuc->pcr = val;

	pcr_ops->write(cpuc->pcr);
}

static inline void sparc_pmu_disable_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc, int idx)
{
	u64 mask = mask_for_index(idx);
	u64 nop = nop_for_index(idx);
	u64 val;

	val = cpuc->pcr;
	val &= ~mask;
	val |= nop;
	cpuc->pcr = val;

	pcr_ops->write(cpuc->pcr);
}

void hw_perf_enable(void)
{
	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
	u64 val;
	int i;

	if (cpuc->enabled)
		return;

	cpuc->enabled = 1;
	barrier();

	val = cpuc->pcr;

	for (i = 0; i < MAX_HWEVENTS; i++) {
		struct perf_event *cp = cpuc->events[i];
		struct hw_perf_event *hwc;

		if (!cp)
			continue;
		hwc = &cp->hw;
		val |= hwc->config_base;
	}

	cpuc->pcr = val;

	pcr_ops->write(cpuc->pcr);
}

void hw_perf_disable(void)
{
	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
	u64 val;

	if (!cpuc->enabled)
		return;

	cpuc->enabled = 0;

	val = cpuc->pcr;
	val &= ~(PCR_UTRACE | PCR_STRACE |
		 sparc_pmu->hv_bit | sparc_pmu->irq_bit);
	cpuc->pcr = val;

	pcr_ops->write(cpuc->pcr);
}

static u32 read_pmc(int idx)
{
	u64 val;

	read_pic(val);
	if (idx == PIC_UPPER_INDEX)
		val >>= 32;

	return val & 0xffffffff;
}

static void write_pmc(int idx, u64 val)
{
	u64 shift, mask, pic;

	shift = 0;
	if (idx == PIC_UPPER_INDEX)
		shift = 32;

	mask = ((u64) 0xffffffff) << shift;
	val <<= shift;

	read_pic(pic);
	pic &= ~mask;
	pic |= val;
	write_pic(pic);
}

static int sparc_perf_event_set_period(struct perf_event *event,
				       struct hw_perf_event *hwc, int idx)
{
	s64 left = atomic64_read(&hwc->period_left);
	s64 period = hwc->sample_period;
	int ret = 0;

	if (unlikely(left <= -period)) {
		left = period;
		atomic64_set(&hwc->period_left, left);
		hwc->last_period = period;
		ret = 1;
	}

	if (unlikely(left <= 0)) {
		left += period;
		atomic64_set(&hwc->period_left, left);
		hwc->last_period = period;
		ret = 1;
	}
	if (left > MAX_PERIOD)
		left = MAX_PERIOD;

	atomic64_set(&hwc->prev_count, (u64)-left);

	write_pmc(idx, (u64)(-left) & 0xffffffff);

	perf_event_update_userpage(event);

	return ret;
}

static int sparc_pmu_enable(struct perf_event *event)
{
	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
	struct hw_perf_event *hwc = &event->hw;
	int idx = hwc->idx;

	if (test_and_set_bit(idx, cpuc->used_mask))
		return -EAGAIN;

	sparc_pmu_disable_event(cpuc, hwc, idx);

	cpuc->events[idx] = event;
	set_bit(idx, cpuc->active_mask);

	sparc_perf_event_set_period(event, hwc, idx);
	sparc_pmu_enable_event(cpuc, hwc, idx);
	perf_event_update_userpage(event);
	return 0;
}

static u64 sparc_perf_event_update(struct perf_event *event,
				   struct hw_perf_event *hwc, int idx)
{
	int shift = 64 - 32;
	u64 prev_raw_count, new_raw_count;
	s64 delta;

again:
	prev_raw_count = atomic64_read(&hwc->prev_count);
	new_raw_count = read_pmc(idx);

	if (atomic64_cmpxchg(&hwc->prev_count, prev_raw_count,
			     new_raw_count) != prev_raw_count)
		goto again;

	delta = (new_raw_count << shift) - (prev_raw_count << shift);
	delta >>= shift;

	atomic64_add(delta, &event->count);
	atomic64_sub(delta, &hwc->period_left);

	return new_raw_count;
}

static void sparc_pmu_disable(struct perf_event *event)
{
	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
	struct hw_perf_event *hwc = &event->hw;
	int idx = hwc->idx;

	clear_bit(idx, cpuc->active_mask);
	sparc_pmu_disable_event(cpuc, hwc, idx);

	barrier();

	sparc_perf_event_update(event, hwc, idx);
	cpuc->events[idx] = NULL;
	clear_bit(idx, cpuc->used_mask);

	perf_event_update_userpage(event);
}

static void sparc_pmu_read(struct perf_event *event)
{
	struct hw_perf_event *hwc = &event->hw;

	sparc_perf_event_update(event, hwc, hwc->idx);
}

static void sparc_pmu_unthrottle(struct perf_event *event)
{
	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);
	struct hw_perf_event *hwc = &event->hw;

	sparc_pmu_enable_event(cpuc, hwc, hwc->idx);
}

static atomic_t active_events = ATOMIC_INIT(0);
static DEFINE_MUTEX(pmc_grab_mutex);

static void perf_stop_nmi_watchdog(void *unused)
{
	struct cpu_hw_events *cpuc = &__get_cpu_var(cpu_hw_events);

	stop_nmi_watchdog(NULL);
	cpuc->pcr = pcr_ops->read();
}

void perf_event_grab_pmc(void)
{
	if (atomic_inc_not_zero(&active_events))
		return;

	mutex_lock(&pmc_grab_mutex);
	if (atomic_read(&active_events) == 0) {
		if (atomic_read(&nmi_active) > 0) {
			on_each_cpu(perf_stop_nmi_watchdog, NULL, 1);
			BUG_ON(atomic_read(&nmi_active) != 0);
		}
		atomic_inc(&active_events);
	}
	mutex_unlock(&pmc_grab_mutex);
}

void perf_event_release_pmc(void)
{
	if (atomic_dec_and_mutex_lock(&active_events, &pmc_grab_mutex)) {
		if (atomic_read(&nmi_active) == 0)
			on_each_cpu(start_nmi_watchdog, NULL, 1);
		mutex_unlock(&pmc_grab_mutex);
	}
}

static const struct perf_event_map *sparc_map_cache_event(u64 config)
{
	unsigned int cache_type, cache_op, cache_result;
	const struct perf_event_map *pmap;

	if (!sparc_pmu->cache_map)
		return ERR_PTR(-ENOENT);

	cache_type = (config >>  0) & 0xff;
	if (cache_type >= PERF_COUNT_HW_CACHE_MAX)
		return ERR_PTR(-EINVAL);

	cache_op = (config >>  8) & 0xff;
	if (cache_op >= PERF_COUNT_HW_CACHE_OP_MAX)
		return ERR_PTR(-EINVAL);

	cache_result = (config >> 16) & 0xff;
	if (cache_result >= PERF_COUNT_HW_CACHE_RESULT_MAX)
		return ERR_PTR(-EINVAL);

	pmap = &((*sparc_pmu->cache_map)[cache_type][cache_op][cache_result]);

	if (pmap->encoding == CACHE_OP_UNSUPPORTED)
		return ERR_PTR(-ENOENT);

	if (pmap->encoding == CACHE_OP_NONSENSE)
		return ERR_PTR(-EINVAL);

	return pmap;
}

static void hw_perf_event_destroy(struct perf_event *event)
{
	perf_event_release_pmc();
}

/* Make sure all events can be scheduled into the hardware at
 * the same time.  This is simplified by the fact that we only
 * need to support 2 simultaneous HW events.
 */
static int sparc_check_constraints(unsigned long *events, int n_ev)
{
	if (n_ev <= perf_max_events) {
		u8 msk1, msk2;
		u16 dummy;

		if (n_ev == 1)
			return 0;
		BUG_ON(n_ev != 2);
		perf_event_decode(events[0], &dummy, &msk1);
		perf_event_decode(events[1], &dummy, &msk2);

		/* If both events can go on any counter, OK.  */
		if (msk1 == (PIC_UPPER | PIC_LOWER) &&
		    msk2 == (PIC_UPPER | PIC_LOWER))
			return 0;

		/* If one event is limited to a specific counter,
		 * and the other can go on both, OK.
		 */
		if ((msk1 == PIC_UPPER || msk1 == PIC_LOWER) &&
		    msk2 == (PIC_UPPER | PIC_LOWER))
			return 0;
		if ((msk2 == PIC_UPPER || msk2 == PIC_LOWER) &&
		    msk1 == (PIC_UPPER | PIC_LOWER))
			return 0;

		/* If the events are fixed to different counters, OK.  */
		if ((msk1 == PIC_UPPER && msk2 == PIC_LOWER) ||
		    (msk1 == PIC_LOWER && msk2 == PIC_UPPER))
			return 0;

		/* Otherwise, there is a conflict.  */
	}

	return -1;
}

static int check_excludes(struct perf_event **evts, int n_prev, int n_new)
{
	int eu = 0, ek = 0, eh = 0;
	struct perf_event *event;
	int i, n, first;

	n = n_prev + n_new;
	if (n <= 1)
		return 0;

	first = 1;
	for (i = 0; i < n; i++) {
		event = evts[i];
		if (first) {
			eu = event->attr.exclude_user;
			ek = event->attr.exclude_kernel;
			eh = event->attr.exclude_hv;
			first = 0;
		} else if (event->attr.exclude_user != eu ||
			   event->attr.exclude_kernel != ek ||
			   event->attr.exclude_hv != eh) {
			return -EAGAIN;
		}
	}

	return 0;
}

static int collect_events(struct perf_event *group, int max_count,
			  struct perf_event *evts[], unsigned long *events)
{
	struct perf_event *event;
	int n = 0;

	if (!is_software_event(group)) {
		if (n >= max_count)
			return -1;
		evts[n] = group;
		events[n++] = group->hw.event_base;
	}
	list_for_each_entry(event, &group->sibling_list, group_entry) {
		if (!is_software_event(event) &&
		    event->state != PERF_EVENT_STATE_OFF) {
			if (n >= max_count)
				return -1;
			evts[n] = event;
			events[n++] = event->hw.event_base;
		}
	}
	return n;
}

static int __hw_perf_event_init(struct perf_event *event)
{
	struct perf_event_attr *attr = &event->attr;
	struct perf_event *evts[MAX_HWEVENTS];
	struct hw_perf_event *hwc = &event->hw;
	unsigned long events[MAX_HWEVENTS];
	const struct perf_event_map *pmap;
	u64 enc;
	int n;

	if (atomic_read(&nmi_active) < 0)
		return -ENODEV;

	if (attr->type == PERF_TYPE_HARDWARE) {
		if (attr->config >= sparc_pmu->max_events)
			return -EINVAL;
		pmap = sparc_pmu->event_map(attr->config);
	} else if (attr->type == PERF_TYPE_HW_CACHE) {
		pmap = sparc_map_cache_event(attr->config);
		if (IS_ERR(pmap))
			return PTR_ERR(pmap);
	} else
		return -EOPNOTSUPP;

	/* We save the enable bits in the config_base.  So to
	 * turn off sampling just write 'config', and to enable
	 * things write 'config | config_base'.
	 */
	hwc->config_base = sparc_pmu->irq_bit;
	if (!attr->exclude_user)
		hwc->config_base |= PCR_UTRACE;
	if (!attr->exclude_kernel)
		hwc->config_base |= PCR_STRACE;
	if (!attr->exclude_hv)
		hwc->config_base |= sparc_pmu->hv_bit;

	hwc->event_base = perf_event_encode(pmap);

	enc = pmap->encoding;

	n = 0;
	if (event->group_leader != event) {
		n = collect_events(event->group_leader,
				   perf_max_events - 1,
				   evts, events);
		if (n < 0)
			return -EINVAL;
	}
	events[n] = hwc->event_base;
	evts[n] = event;

	if (check_excludes(evts, n, 1))
		return -EINVAL;

	if (sparc_check_constraints(events, n + 1))
		return -EINVAL;

	/* Try to do all error checking before this point, as unwinding
	 * state after grabbing the PMC is difficult.
	 */
	perf_event_grab_pmc();
	event->destroy = hw_perf_event_destroy;

	if (!hwc->sample_period) {
		hwc->sample_period = MAX_PERIOD;
		hwc->last_period = hwc->sample_period;
		atomic64_set(&hwc->period_left, hwc->sample_period);
	}

	if (pmap->pic_mask & PIC_UPPER) {
		hwc->idx = PIC_UPPER_INDEX;
		enc <<= sparc_pmu->upper_shift;
	} else {
		hwc->idx = PIC_LOWER_INDEX;
		enc <<= sparc_pmu->lower_shift;
	}

	hwc->config |= enc;
	return 0;
}

static const struct pmu pmu = {
	.enable		= sparc_pmu_enable,
	.disable	= sparc_pmu_disable,
	.read		= sparc_pmu_read,
	.unthrottle	= sparc_pmu_unthrottle,
};

const struct pmu *hw_perf_event_init(struct perf_event *event)
{
	int err = __hw_perf_event_init(event);

	if (err)
		return ERR_PTR(err);
	return &pmu;
}

void perf_event_print_debug(void)
{
	unsigned long flags;
	u64 pcr, pic;
	int cpu;

	if (!sparc_pmu)
		return;

	local_irq_save(flags);

	cpu = smp_processor_id();

	pcr = pcr_ops->read();
	read_pic(pic);

	pr_info("\n");
	pr_info("CPU#%d: PCR[%016llx] PIC[%016llx]\n",
		cpu, pcr, pic);

	local_irq_restore(flags);
}

static int __kprobes perf_event_nmi_handler(struct notifier_block *self,
					    unsigned long cmd, void *__args)
{
	struct die_args *args = __args;
	struct perf_sample_data data;
	struct cpu_hw_events *cpuc;
	struct pt_regs *regs;
	int idx;

	if (!atomic_read(&active_events))
		return NOTIFY_DONE;

	switch (cmd) {
	case DIE_NMI:
		break;

	default:
		return NOTIFY_DONE;
	}

	regs = args->regs;

	data.addr = 0;

	cpuc = &__get_cpu_var(cpu_hw_events);
	for (idx = 0; idx < MAX_HWEVENTS; idx++) {
		struct perf_event *event = cpuc->events[idx];
		struct hw_perf_event *hwc;
		u64 val;

		if (!test_bit(idx, cpuc->active_mask))
			continue;
		hwc = &event->hw;
		val = sparc_perf_event_update(event, hwc, idx);
		if (val & (1ULL << 31))
			continue;

		data.period = event->hw.last_period;
		if (!sparc_perf_event_set_period(event, hwc, idx))
			continue;

		if (perf_event_overflow(event, 1, &data, regs))
			sparc_pmu_disable_event(cpuc, hwc, idx);
	}

	return NOTIFY_STOP;
}

static __read_mostly struct notifier_block perf_event_nmi_notifier = {
	.notifier_call		= perf_event_nmi_handler,
};

static bool __init supported_pmu(void)
{
	if (!strcmp(sparc_pmu_type, "ultra3") ||
	    !strcmp(sparc_pmu_type, "ultra3+") ||
	    !strcmp(sparc_pmu_type, "ultra3i") ||
	    !strcmp(sparc_pmu_type, "ultra4+")) {
		sparc_pmu = &ultra3_pmu;
		return true;
	}
	if (!strcmp(sparc_pmu_type, "niagara")) {
		sparc_pmu = &niagara1_pmu;
		return true;
	}
	if (!strcmp(sparc_pmu_type, "niagara2")) {
		sparc_pmu = &niagara2_pmu;
		return true;
	}
	return false;
}

void __init init_hw_perf_events(void)
{
	pr_info("Performance events: ");

	if (!supported_pmu()) {
		pr_cont("No support for PMU type '%s'\n", sparc_pmu_type);
		return;
	}

	pr_cont("Supported PMU type is '%s'\n", sparc_pmu_type);

	/* All sparc64 PMUs currently have 2 events.  But this simple
	 * driver only supports one active event at a time.
	 */
	perf_max_events = 1;

	register_die_notifier(&perf_event_nmi_notifier);
}