summaryrefslogtreecommitdiff
path: root/kernel/sched/psi.c
blob: ec66b40bdd403ba2b4e9d5bb8c4459fc2a0f5ef2 (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
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
// SPDX-License-Identifier: GPL-2.0
/*
 * Pressure stall information for CPU, memory and IO
 *
 * Copyright (c) 2018 Facebook, Inc.
 * Author: Johannes Weiner <hannes@cmpxchg.org>
 *
 * Polling support by Suren Baghdasaryan <surenb@google.com>
 * Copyright (c) 2018 Google, Inc.
 *
 * When CPU, memory and IO are contended, tasks experience delays that
 * reduce throughput and introduce latencies into the workload. Memory
 * and IO contention, in addition, can cause a full loss of forward
 * progress in which the CPU goes idle.
 *
 * This code aggregates individual task delays into resource pressure
 * metrics that indicate problems with both workload health and
 * resource utilization.
 *
 *			Model
 *
 * The time in which a task can execute on a CPU is our baseline for
 * productivity. Pressure expresses the amount of time in which this
 * potential cannot be realized due to resource contention.
 *
 * This concept of productivity has two components: the workload and
 * the CPU. To measure the impact of pressure on both, we define two
 * contention states for a resource: SOME and FULL.
 *
 * In the SOME state of a given resource, one or more tasks are
 * delayed on that resource. This affects the workload's ability to
 * perform work, but the CPU may still be executing other tasks.
 *
 * In the FULL state of a given resource, all non-idle tasks are
 * delayed on that resource such that nobody is advancing and the CPU
 * goes idle. This leaves both workload and CPU unproductive.
 *
 *	SOME = nr_delayed_tasks != 0
 *	FULL = nr_delayed_tasks != 0 && nr_productive_tasks == 0
 *
 * What it means for a task to be productive is defined differently
 * for each resource. For IO, productive means a running task. For
 * memory, productive means a running task that isn't a reclaimer. For
 * CPU, productive means an oncpu task.
 *
 * Naturally, the FULL state doesn't exist for the CPU resource at the
 * system level, but exist at the cgroup level. At the cgroup level,
 * FULL means all non-idle tasks in the cgroup are delayed on the CPU
 * resource which is being used by others outside of the cgroup or
 * throttled by the cgroup cpu.max configuration.
 *
 * The percentage of wallclock time spent in those compound stall
 * states gives pressure numbers between 0 and 100 for each resource,
 * where the SOME percentage indicates workload slowdowns and the FULL
 * percentage indicates reduced CPU utilization:
 *
 *	%SOME = time(SOME) / period
 *	%FULL = time(FULL) / period
 *
 *			Multiple CPUs
 *
 * The more tasks and available CPUs there are, the more work can be
 * performed concurrently. This means that the potential that can go
 * unrealized due to resource contention *also* scales with non-idle
 * tasks and CPUs.
 *
 * Consider a scenario where 257 number crunching tasks are trying to
 * run concurrently on 256 CPUs. If we simply aggregated the task
 * states, we would have to conclude a CPU SOME pressure number of
 * 100%, since *somebody* is waiting on a runqueue at all
 * times. However, that is clearly not the amount of contention the
 * workload is experiencing: only one out of 256 possible execution
 * threads will be contended at any given time, or about 0.4%.
 *
 * Conversely, consider a scenario of 4 tasks and 4 CPUs where at any
 * given time *one* of the tasks is delayed due to a lack of memory.
 * Again, looking purely at the task state would yield a memory FULL
 * pressure number of 0%, since *somebody* is always making forward
 * progress. But again this wouldn't capture the amount of execution
 * potential lost, which is 1 out of 4 CPUs, or 25%.
 *
 * To calculate wasted potential (pressure) with multiple processors,
 * we have to base our calculation on the number of non-idle tasks in
 * conjunction with the number of available CPUs, which is the number
 * of potential execution threads. SOME becomes then the proportion of
 * delayed tasks to possible threads, and FULL is the share of possible
 * threads that are unproductive due to delays:
 *
 *	threads = min(nr_nonidle_tasks, nr_cpus)
 *	   SOME = min(nr_delayed_tasks / threads, 1)
 *	   FULL = (threads - min(nr_productive_tasks, threads)) / threads
 *
 * For the 257 number crunchers on 256 CPUs, this yields:
 *
 *	threads = min(257, 256)
 *	   SOME = min(1 / 256, 1)             = 0.4%
 *	   FULL = (256 - min(256, 256)) / 256 = 0%
 *
 * For the 1 out of 4 memory-delayed tasks, this yields:
 *
 *	threads = min(4, 4)
 *	   SOME = min(1 / 4, 1)               = 25%
 *	   FULL = (4 - min(3, 4)) / 4         = 25%
 *
 * [ Substitute nr_cpus with 1, and you can see that it's a natural
 *   extension of the single-CPU model. ]
 *
 *			Implementation
 *
 * To assess the precise time spent in each such state, we would have
 * to freeze the system on task changes and start/stop the state
 * clocks accordingly. Obviously that doesn't scale in practice.
 *
 * Because the scheduler aims to distribute the compute load evenly
 * among the available CPUs, we can track task state locally to each
 * CPU and, at much lower frequency, extrapolate the global state for
 * the cumulative stall times and the running averages.
 *
 * For each runqueue, we track:
 *
 *	   tSOME[cpu] = time(nr_delayed_tasks[cpu] != 0)
 *	   tFULL[cpu] = time(nr_delayed_tasks[cpu] && !nr_productive_tasks[cpu])
 *	tNONIDLE[cpu] = time(nr_nonidle_tasks[cpu] != 0)
 *
 * and then periodically aggregate:
 *
 *	tNONIDLE = sum(tNONIDLE[i])
 *
 *	   tSOME = sum(tSOME[i] * tNONIDLE[i]) / tNONIDLE
 *	   tFULL = sum(tFULL[i] * tNONIDLE[i]) / tNONIDLE
 *
 *	   %SOME = tSOME / period
 *	   %FULL = tFULL / period
 *
 * This gives us an approximation of pressure that is practical
 * cost-wise, yet way more sensitive and accurate than periodic
 * sampling of the aggregate task states would be.
 */

static int psi_bug __read_mostly;

DEFINE_STATIC_KEY_FALSE(psi_disabled);
DEFINE_STATIC_KEY_TRUE(psi_cgroups_enabled);

#ifdef CONFIG_PSI_DEFAULT_DISABLED
static bool psi_enable;
#else
static bool psi_enable = true;
#endif
static int __init setup_psi(char *str)
{
	return kstrtobool(str, &psi_enable) == 0;
}
__setup("psi=", setup_psi);

/* Running averages - we need to be higher-res than loadavg */
#define PSI_FREQ	(2*HZ+1)	/* 2 sec intervals */
#define EXP_10s		1677		/* 1/exp(2s/10s) as fixed-point */
#define EXP_60s		1981		/* 1/exp(2s/60s) */
#define EXP_300s	2034		/* 1/exp(2s/300s) */

/* PSI trigger definitions */
#define WINDOW_MIN_US 500000	/* Min window size is 500ms */
#define WINDOW_MAX_US 10000000	/* Max window size is 10s */
#define UPDATES_PER_WINDOW 10	/* 10 updates per window */

/* Sampling frequency in nanoseconds */
static u64 psi_period __read_mostly;

/* System-level pressure and stall tracking */
static DEFINE_PER_CPU(struct psi_group_cpu, system_group_pcpu);
struct psi_group psi_system = {
	.pcpu = &system_group_pcpu,
};

static void psi_avgs_work(struct work_struct *work);

static void poll_timer_fn(struct timer_list *t);

static void group_init(struct psi_group *group)
{
	int cpu;

	for_each_possible_cpu(cpu)
		seqcount_init(&per_cpu_ptr(group->pcpu, cpu)->seq);
	group->avg_last_update = sched_clock();
	group->avg_next_update = group->avg_last_update + psi_period;
	INIT_DELAYED_WORK(&group->avgs_work, psi_avgs_work);
	mutex_init(&group->avgs_lock);
	/* Init trigger-related members */
	mutex_init(&group->trigger_lock);
	INIT_LIST_HEAD(&group->triggers);
	memset(group->nr_triggers, 0, sizeof(group->nr_triggers));
	group->poll_states = 0;
	group->poll_min_period = U32_MAX;
	memset(group->polling_total, 0, sizeof(group->polling_total));
	group->polling_next_update = ULLONG_MAX;
	group->polling_until = 0;
	init_waitqueue_head(&group->poll_wait);
	timer_setup(&group->poll_timer, poll_timer_fn, 0);
	rcu_assign_pointer(group->poll_task, NULL);
}

void __init psi_init(void)
{
	if (!psi_enable) {
		static_branch_enable(&psi_disabled);
		return;
	}

	if (!cgroup_psi_enabled())
		static_branch_disable(&psi_cgroups_enabled);

	psi_period = jiffies_to_nsecs(PSI_FREQ);
	group_init(&psi_system);
}

static bool test_state(unsigned int *tasks, enum psi_states state)
{
	switch (state) {
	case PSI_IO_SOME:
		return unlikely(tasks[NR_IOWAIT]);
	case PSI_IO_FULL:
		return unlikely(tasks[NR_IOWAIT] && !tasks[NR_RUNNING]);
	case PSI_MEM_SOME:
		return unlikely(tasks[NR_MEMSTALL]);
	case PSI_MEM_FULL:
		return unlikely(tasks[NR_MEMSTALL] &&
			tasks[NR_RUNNING] == tasks[NR_MEMSTALL_RUNNING]);
	case PSI_CPU_SOME:
		return unlikely(tasks[NR_RUNNING] > tasks[NR_ONCPU]);
	case PSI_CPU_FULL:
		return unlikely(tasks[NR_RUNNING] && !tasks[NR_ONCPU]);
	case PSI_NONIDLE:
		return tasks[NR_IOWAIT] || tasks[NR_MEMSTALL] ||
			tasks[NR_RUNNING];
	default:
		return false;
	}
}

static void get_recent_times(struct psi_group *group, int cpu,
			     enum psi_aggregators aggregator, u32 *times,
			     u32 *pchanged_states)
{
	struct psi_group_cpu *groupc = per_cpu_ptr(group->pcpu, cpu);
	u64 now, state_start;
	enum psi_states s;
	unsigned int seq;
	u32 state_mask;

	*pchanged_states = 0;

	/* Snapshot a coherent view of the CPU state */
	do {
		seq = read_seqcount_begin(&groupc->seq);
		now = cpu_clock(cpu);
		memcpy(times, groupc->times, sizeof(groupc->times));
		state_mask = groupc->state_mask;
		state_start = groupc->state_start;
	} while (read_seqcount_retry(&groupc->seq, seq));

	/* Calculate state time deltas against the previous snapshot */
	for (s = 0; s < NR_PSI_STATES; s++) {
		u32 delta;
		/*
		 * In addition to already concluded states, we also
		 * incorporate currently active states on the CPU,
		 * since states may last for many sampling periods.
		 *
		 * This way we keep our delta sampling buckets small
		 * (u32) and our reported pressure close to what's
		 * actually happening.
		 */
		if (state_mask & (1 << s))
			times[s] += now - state_start;

		delta = times[s] - groupc->times_prev[aggregator][s];
		groupc->times_prev[aggregator][s] = times[s];

		times[s] = delta;
		if (delta)
			*pchanged_states |= (1 << s);
	}
}

static void calc_avgs(unsigned long avg[3], int missed_periods,
		      u64 time, u64 period)
{
	unsigned long pct;

	/* Fill in zeroes for periods of no activity */
	if (missed_periods) {
		avg[0] = calc_load_n(avg[0], EXP_10s, 0, missed_periods);
		avg[1] = calc_load_n(avg[1], EXP_60s, 0, missed_periods);
		avg[2] = calc_load_n(avg[2], EXP_300s, 0, missed_periods);
	}

	/* Sample the most recent active period */
	pct = div_u64(time * 100, period);
	pct *= FIXED_1;
	avg[0] = calc_load(avg[0], EXP_10s, pct);
	avg[1] = calc_load(avg[1], EXP_60s, pct);
	avg[2] = calc_load(avg[2], EXP_300s, pct);
}

static void collect_percpu_times(struct psi_group *group,
				 enum psi_aggregators aggregator,
				 u32 *pchanged_states)
{
	u64 deltas[NR_PSI_STATES - 1] = { 0, };
	unsigned long nonidle_total = 0;
	u32 changed_states = 0;
	int cpu;
	int s;

	/*
	 * Collect the per-cpu time buckets and average them into a
	 * single time sample that is normalized to wallclock time.
	 *
	 * For averaging, each CPU is weighted by its non-idle time in
	 * the sampling period. This eliminates artifacts from uneven
	 * loading, or even entirely idle CPUs.
	 */
	for_each_possible_cpu(cpu) {
		u32 times[NR_PSI_STATES];
		u32 nonidle;
		u32 cpu_changed_states;

		get_recent_times(group, cpu, aggregator, times,
				&cpu_changed_states);
		changed_states |= cpu_changed_states;

		nonidle = nsecs_to_jiffies(times[PSI_NONIDLE]);
		nonidle_total += nonidle;

		for (s = 0; s < PSI_NONIDLE; s++)
			deltas[s] += (u64)times[s] * nonidle;
	}

	/*
	 * Integrate the sample into the running statistics that are
	 * reported to userspace: the cumulative stall times and the
	 * decaying averages.
	 *
	 * Pressure percentages are sampled at PSI_FREQ. We might be
	 * called more often when the user polls more frequently than
	 * that; we might be called less often when there is no task
	 * activity, thus no data, and clock ticks are sporadic. The
	 * below handles both.
	 */

	/* total= */
	for (s = 0; s < NR_PSI_STATES - 1; s++)
		group->total[aggregator][s] +=
				div_u64(deltas[s], max(nonidle_total, 1UL));

	if (pchanged_states)
		*pchanged_states = changed_states;
}

static u64 update_averages(struct psi_group *group, u64 now)
{
	unsigned long missed_periods = 0;
	u64 expires, period;
	u64 avg_next_update;
	int s;

	/* avgX= */
	expires = group->avg_next_update;
	if (now - expires >= psi_period)
		missed_periods = div_u64(now - expires, psi_period);

	/*
	 * The periodic clock tick can get delayed for various
	 * reasons, especially on loaded systems. To avoid clock
	 * drift, we schedule the clock in fixed psi_period intervals.
	 * But the deltas we sample out of the per-cpu buckets above
	 * are based on the actual time elapsing between clock ticks.
	 */
	avg_next_update = expires + ((1 + missed_periods) * psi_period);
	period = now - (group->avg_last_update + (missed_periods * psi_period));
	group->avg_last_update = now;

	for (s = 0; s < NR_PSI_STATES - 1; s++) {
		u32 sample;

		sample = group->total[PSI_AVGS][s] - group->avg_total[s];
		/*
		 * Due to the lockless sampling of the time buckets,
		 * recorded time deltas can slip into the next period,
		 * which under full pressure can result in samples in
		 * excess of the period length.
		 *
		 * We don't want to report non-sensical pressures in
		 * excess of 100%, nor do we want to drop such events
		 * on the floor. Instead we punt any overage into the
		 * future until pressure subsides. By doing this we
		 * don't underreport the occurring pressure curve, we
		 * just report it delayed by one period length.
		 *
		 * The error isn't cumulative. As soon as another
		 * delta slips from a period P to P+1, by definition
		 * it frees up its time T in P.
		 */
		if (sample > period)
			sample = period;
		group->avg_total[s] += sample;
		calc_avgs(group->avg[s], missed_periods, sample, period);
	}

	return avg_next_update;
}

static void psi_avgs_work(struct work_struct *work)
{
	struct delayed_work *dwork;
	struct psi_group *group;
	u32 changed_states;
	bool nonidle;
	u64 now;

	dwork = to_delayed_work(work);
	group = container_of(dwork, struct psi_group, avgs_work);

	mutex_lock(&group->avgs_lock);

	now = sched_clock();

	collect_percpu_times(group, PSI_AVGS, &changed_states);
	nonidle = changed_states & (1 << PSI_NONIDLE);
	/*
	 * If there is task activity, periodically fold the per-cpu
	 * times and feed samples into the running averages. If things
	 * are idle and there is no data to process, stop the clock.
	 * Once restarted, we'll catch up the running averages in one
	 * go - see calc_avgs() and missed_periods.
	 */
	if (now >= group->avg_next_update)
		group->avg_next_update = update_averages(group, now);

	if (nonidle) {
		schedule_delayed_work(dwork, nsecs_to_jiffies(
				group->avg_next_update - now) + 1);
	}

	mutex_unlock(&group->avgs_lock);
}

/* Trigger tracking window manipulations */
static void window_reset(struct psi_window *win, u64 now, u64 value,
			 u64 prev_growth)
{
	win->start_time = now;
	win->start_value = value;
	win->prev_growth = prev_growth;
}

/*
 * PSI growth tracking window update and growth calculation routine.
 *
 * This approximates a sliding tracking window by interpolating
 * partially elapsed windows using historical growth data from the
 * previous intervals. This minimizes memory requirements (by not storing
 * all the intermediate values in the previous window) and simplifies
 * the calculations. It works well because PSI signal changes only in
 * positive direction and over relatively small window sizes the growth
 * is close to linear.
 */
static u64 window_update(struct psi_window *win, u64 now, u64 value)
{
	u64 elapsed;
	u64 growth;

	elapsed = now - win->start_time;
	growth = value - win->start_value;
	/*
	 * After each tracking window passes win->start_value and
	 * win->start_time get reset and win->prev_growth stores
	 * the average per-window growth of the previous window.
	 * win->prev_growth is then used to interpolate additional
	 * growth from the previous window assuming it was linear.
	 */
	if (elapsed > win->size)
		window_reset(win, now, value, growth);
	else {
		u32 remaining;

		remaining = win->size - elapsed;
		growth += div64_u64(win->prev_growth * remaining, win->size);
	}

	return growth;
}

static void init_triggers(struct psi_group *group, u64 now)
{
	struct psi_trigger *t;

	list_for_each_entry(t, &group->triggers, node)
		window_reset(&t->win, now,
				group->total[PSI_POLL][t->state], 0);
	memcpy(group->polling_total, group->total[PSI_POLL],
		   sizeof(group->polling_total));
	group->polling_next_update = now + group->poll_min_period;
}

static u64 update_triggers(struct psi_group *group, u64 now)
{
	struct psi_trigger *t;
	bool update_total = false;
	u64 *total = group->total[PSI_POLL];

	/*
	 * On subsequent updates, calculate growth deltas and let
	 * watchers know when their specified thresholds are exceeded.
	 */
	list_for_each_entry(t, &group->triggers, node) {
		u64 growth;
		bool new_stall;

		new_stall = group->polling_total[t->state] != total[t->state];

		/* Check for stall activity or a previous threshold breach */
		if (!new_stall && !t->pending_event)
			continue;
		/*
		 * Check for new stall activity, as well as deferred
		 * events that occurred in the last window after the
		 * trigger had already fired (we want to ratelimit
		 * events without dropping any).
		 */
		if (new_stall) {
			/*
			 * Multiple triggers might be looking at the same state,
			 * remember to update group->polling_total[] once we've
			 * been through all of them. Also remember to extend the
			 * polling time if we see new stall activity.
			 */
			update_total = true;

			/* Calculate growth since last update */
			growth = window_update(&t->win, now, total[t->state]);
			if (growth < t->threshold)
				continue;

			t->pending_event = true;
		}
		/* Limit event signaling to once per window */
		if (now < t->last_event_time + t->win.size)
			continue;

		/* Generate an event */
		if (cmpxchg(&t->event, 0, 1) == 0)
			wake_up_interruptible(&t->event_wait);
		t->last_event_time = now;
		/* Reset threshold breach flag once event got generated */
		t->pending_event = false;
	}

	if (update_total)
		memcpy(group->polling_total, total,
				sizeof(group->polling_total));

	return now + group->poll_min_period;
}

/* Schedule polling if it's not already scheduled. */
static void psi_schedule_poll_work(struct psi_group *group, unsigned long delay)
{
	struct task_struct *task;

	/*
	 * Do not reschedule if already scheduled.
	 * Possible race with a timer scheduled after this check but before
	 * mod_timer below can be tolerated because group->polling_next_update
	 * will keep updates on schedule.
	 */
	if (timer_pending(&group->poll_timer))
		return;

	rcu_read_lock();

	task = rcu_dereference(group->poll_task);
	/*
	 * kworker might be NULL in case psi_trigger_destroy races with
	 * psi_task_change (hotpath) which can't use locks
	 */
	if (likely(task))
		mod_timer(&group->poll_timer, jiffies + delay);

	rcu_read_unlock();
}

static void psi_poll_work(struct psi_group *group)
{
	u32 changed_states;
	u64 now;

	mutex_lock(&group->trigger_lock);

	now = sched_clock();

	collect_percpu_times(group, PSI_POLL, &changed_states);

	if (changed_states & group->poll_states) {
		/* Initialize trigger windows when entering polling mode */
		if (now > group->polling_until)
			init_triggers(group, now);

		/*
		 * Keep the monitor active for at least the duration of the
		 * minimum tracking window as long as monitor states are
		 * changing.
		 */
		group->polling_until = now +
			group->poll_min_period * UPDATES_PER_WINDOW;
	}

	if (now > group->polling_until) {
		group->polling_next_update = ULLONG_MAX;
		goto out;
	}

	if (now >= group->polling_next_update)
		group->polling_next_update = update_triggers(group, now);

	psi_schedule_poll_work(group,
		nsecs_to_jiffies(group->polling_next_update - now) + 1);

out:
	mutex_unlock(&group->trigger_lock);
}

static int psi_poll_worker(void *data)
{
	struct psi_group *group = (struct psi_group *)data;

	sched_set_fifo_low(current);

	while (true) {
		wait_event_interruptible(group->poll_wait,
				atomic_cmpxchg(&group->poll_wakeup, 1, 0) ||
				kthread_should_stop());
		if (kthread_should_stop())
			break;

		psi_poll_work(group);
	}
	return 0;
}

static void poll_timer_fn(struct timer_list *t)
{
	struct psi_group *group = from_timer(group, t, poll_timer);

	atomic_set(&group->poll_wakeup, 1);
	wake_up_interruptible(&group->poll_wait);
}

static void record_times(struct psi_group_cpu *groupc, u64 now)
{
	u32 delta;

	delta = now - groupc->state_start;
	groupc->state_start = now;

	if (groupc->state_mask & (1 << PSI_IO_SOME)) {
		groupc->times[PSI_IO_SOME] += delta;
		if (groupc->state_mask & (1 << PSI_IO_FULL))
			groupc->times[PSI_IO_FULL] += delta;
	}

	if (groupc->state_mask & (1 << PSI_MEM_SOME)) {
		groupc->times[PSI_MEM_SOME] += delta;
		if (groupc->state_mask & (1 << PSI_MEM_FULL))
			groupc->times[PSI_MEM_FULL] += delta;
	}

	if (groupc->state_mask & (1 << PSI_CPU_SOME)) {
		groupc->times[PSI_CPU_SOME] += delta;
		if (groupc->state_mask & (1 << PSI_CPU_FULL))
			groupc->times[PSI_CPU_FULL] += delta;
	}

	if (groupc->state_mask & (1 << PSI_NONIDLE))
		groupc->times[PSI_NONIDLE] += delta;
}

static void psi_group_change(struct psi_group *group, int cpu,
			     unsigned int clear, unsigned int set, u64 now,
			     bool wake_clock)
{
	struct psi_group_cpu *groupc;
	u32 state_mask = 0;
	unsigned int t, m;
	enum psi_states s;

	groupc = per_cpu_ptr(group->pcpu, cpu);

	/*
	 * First we assess the aggregate resource states this CPU's
	 * tasks have been in since the last change, and account any
	 * SOME and FULL time these may have resulted in.
	 *
	 * Then we update the task counts according to the state
	 * change requested through the @clear and @set bits.
	 */
	write_seqcount_begin(&groupc->seq);

	record_times(groupc, now);

	for (t = 0, m = clear; m; m &= ~(1 << t), t++) {
		if (!(m & (1 << t)))
			continue;
		if (groupc->tasks[t]) {
			groupc->tasks[t]--;
		} else if (!psi_bug) {
			printk_deferred(KERN_ERR "psi: task underflow! cpu=%d t=%d tasks=[%u %u %u %u %u] clear=%x set=%x\n",
					cpu, t, groupc->tasks[0],
					groupc->tasks[1], groupc->tasks[2],
					groupc->tasks[3], groupc->tasks[4],
					clear, set);
			psi_bug = 1;
		}
	}

	for (t = 0; set; set &= ~(1 << t), t++)
		if (set & (1 << t))
			groupc->tasks[t]++;

	/* Calculate state mask representing active states */
	for (s = 0; s < NR_PSI_STATES; s++) {
		if (test_state(groupc->tasks, s))
			state_mask |= (1 << s);
	}

	/*
	 * Since we care about lost potential, a memstall is FULL
	 * when there are no other working tasks, but also when
	 * the CPU is actively reclaiming and nothing productive
	 * could run even if it were runnable. So when the current
	 * task in a cgroup is in_memstall, the corresponding groupc
	 * on that cpu is in PSI_MEM_FULL state.
	 */
	if (unlikely(groupc->tasks[NR_ONCPU] && cpu_curr(cpu)->in_memstall))
		state_mask |= (1 << PSI_MEM_FULL);

	groupc->state_mask = state_mask;

	write_seqcount_end(&groupc->seq);

	if (state_mask & group->poll_states)
		psi_schedule_poll_work(group, 1);

	if (wake_clock && !delayed_work_pending(&group->avgs_work))
		schedule_delayed_work(&group->avgs_work, PSI_FREQ);
}

static struct psi_group *iterate_groups(struct task_struct *task, void **iter)
{
	if (*iter == &psi_system)
		return NULL;

#ifdef CONFIG_CGROUPS
	if (static_branch_likely(&psi_cgroups_enabled)) {
		struct cgroup *cgroup = NULL;

		if (!*iter)
			cgroup = task->cgroups->dfl_cgrp;
		else
			cgroup = cgroup_parent(*iter);

		if (cgroup && cgroup_parent(cgroup)) {
			*iter = cgroup;
			return cgroup_psi(cgroup);
		}
	}
#endif
	*iter = &psi_system;
	return &psi_system;
}

static void psi_flags_change(struct task_struct *task, int clear, int set)
{
	if (((task->psi_flags & set) ||
	     (task->psi_flags & clear) != clear) &&
	    !psi_bug) {
		printk_deferred(KERN_ERR "psi: inconsistent task state! task=%d:%s cpu=%d psi_flags=%x clear=%x set=%x\n",
				task->pid, task->comm, task_cpu(task),
				task->psi_flags, clear, set);
		psi_bug = 1;
	}

	task->psi_flags &= ~clear;
	task->psi_flags |= set;
}

void psi_task_change(struct task_struct *task, int clear, int set)
{
	int cpu = task_cpu(task);
	struct psi_group *group;
	bool wake_clock = true;
	void *iter = NULL;
	u64 now;

	if (!task->pid)
		return;

	psi_flags_change(task, clear, set);

	now = cpu_clock(cpu);
	/*
	 * Periodic aggregation shuts off if there is a period of no
	 * task changes, so we wake it back up if necessary. However,
	 * don't do this if the task change is the aggregation worker
	 * itself going to sleep, or we'll ping-pong forever.
	 */
	if (unlikely((clear & TSK_RUNNING) &&
		     (task->flags & PF_WQ_WORKER) &&
		     wq_worker_last_func(task) == psi_avgs_work))
		wake_clock = false;

	while ((group = iterate_groups(task, &iter)))
		psi_group_change(group, cpu, clear, set, now, wake_clock);
}

void psi_task_switch(struct task_struct *prev, struct task_struct *next,
		     bool sleep)
{
	struct psi_group *group, *common = NULL;
	int cpu = task_cpu(prev);
	void *iter;
	u64 now = cpu_clock(cpu);

	if (next->pid) {
		bool identical_state;

		psi_flags_change(next, 0, TSK_ONCPU);
		/*
		 * When switching between tasks that have an identical
		 * runtime state, the cgroup that contains both tasks
		 * we reach the first common ancestor. Iterate @next's
		 * ancestors only until we encounter @prev's ONCPU.
		 */
		identical_state = prev->psi_flags == next->psi_flags;
		iter = NULL;
		while ((group = iterate_groups(next, &iter))) {
			if (identical_state &&
			    per_cpu_ptr(group->pcpu, cpu)->tasks[NR_ONCPU]) {
				common = group;
				break;
			}

			psi_group_change(group, cpu, 0, TSK_ONCPU, now, true);
		}
	}

	if (prev->pid) {
		int clear = TSK_ONCPU, set = 0;

		/*
		 * When we're going to sleep, psi_dequeue() lets us
		 * handle TSK_RUNNING, TSK_MEMSTALL_RUNNING and
		 * TSK_IOWAIT here, where we can combine it with
		 * TSK_ONCPU and save walking common ancestors twice.
		 */
		if (sleep) {
			clear |= TSK_RUNNING;
			if (prev->in_memstall)
				clear |= TSK_MEMSTALL_RUNNING;
			if (prev->in_iowait)
				set |= TSK_IOWAIT;
		}

		psi_flags_change(prev, clear, set);

		iter = NULL;
		while ((group = iterate_groups(prev, &iter)) && group != common)
			psi_group_change(group, cpu, clear, set, now, true);

		/*
		 * TSK_ONCPU is handled up to the common ancestor. If we're tasked
		 * with dequeuing too, finish that for the rest of the hierarchy.
		 */
		if (sleep) {
			clear &= ~TSK_ONCPU;
			for (; group; group = iterate_groups(prev, &iter))
				psi_group_change(group, cpu, clear, set, now, true);
		}
	}
}

/**
 * psi_memstall_enter - mark the beginning of a memory stall section
 * @flags: flags to handle nested sections
 *
 * Marks the calling task as being stalled due to a lack of memory,
 * such as waiting for a refault or performing reclaim.
 */
void psi_memstall_enter(unsigned long *flags)
{
	struct rq_flags rf;
	struct rq *rq;

	if (static_branch_likely(&psi_disabled))
		return;

	*flags = current->in_memstall;
	if (*flags)
		return;
	/*
	 * in_memstall setting & accounting needs to be atomic wrt
	 * changes to the task's scheduling state, otherwise we can
	 * race with CPU migration.
	 */
	rq = this_rq_lock_irq(&rf);

	current->in_memstall = 1;
	psi_task_change(current, 0, TSK_MEMSTALL | TSK_MEMSTALL_RUNNING);

	rq_unlock_irq(rq, &rf);
}

/**
 * psi_memstall_leave - mark the end of an memory stall section
 * @flags: flags to handle nested memdelay sections
 *
 * Marks the calling task as no longer stalled due to lack of memory.
 */
void psi_memstall_leave(unsigned long *flags)
{
	struct rq_flags rf;
	struct rq *rq;

	if (static_branch_likely(&psi_disabled))
		return;

	if (*flags)
		return;
	/*
	 * in_memstall clearing & accounting needs to be atomic wrt
	 * changes to the task's scheduling state, otherwise we could
	 * race with CPU migration.
	 */
	rq = this_rq_lock_irq(&rf);

	current->in_memstall = 0;
	psi_task_change(current, TSK_MEMSTALL | TSK_MEMSTALL_RUNNING, 0);

	rq_unlock_irq(rq, &rf);
}

#ifdef CONFIG_CGROUPS
int psi_cgroup_alloc(struct cgroup *cgroup)
{
	if (static_branch_likely(&psi_disabled))
		return 0;

	cgroup->psi = kmalloc(sizeof(struct psi_group), GFP_KERNEL);
	if (!cgroup->psi)
		return -ENOMEM;

	cgroup->psi->pcpu = alloc_percpu(struct psi_group_cpu);
	if (!cgroup->psi->pcpu) {
		kfree(cgroup->psi);
		return -ENOMEM;
	}
	group_init(cgroup->psi);
	return 0;
}

void psi_cgroup_free(struct cgroup *cgroup)
{
	if (static_branch_likely(&psi_disabled))
		return;

	cancel_delayed_work_sync(&cgroup->psi->avgs_work);
	free_percpu(cgroup->psi->pcpu);
	/* All triggers must be removed by now */
	WARN_ONCE(cgroup->psi->poll_states, "psi: trigger leak\n");
	kfree(cgroup->psi);
}

/**
 * cgroup_move_task - move task to a different cgroup
 * @task: the task
 * @to: the target css_set
 *
 * Move task to a new cgroup and safely migrate its associated stall
 * state between the different groups.
 *
 * This function acquires the task's rq lock to lock out concurrent
 * changes to the task's scheduling state and - in case the task is
 * running - concurrent changes to its stall state.
 */
void cgroup_move_task(struct task_struct *task, struct css_set *to)
{
	unsigned int task_flags;
	struct rq_flags rf;
	struct rq *rq;

	if (static_branch_likely(&psi_disabled)) {
		/*
		 * Lame to do this here, but the scheduler cannot be locked
		 * from the outside, so we move cgroups from inside sched/.
		 */
		rcu_assign_pointer(task->cgroups, to);
		return;
	}

	rq = task_rq_lock(task, &rf);

	/*
	 * We may race with schedule() dropping the rq lock between
	 * deactivating prev and switching to next. Because the psi
	 * updates from the deactivation are deferred to the switch
	 * callback to save cgroup tree updates, the task's scheduling
	 * state here is not coherent with its psi state:
	 *
	 * schedule()                   cgroup_move_task()
	 *   rq_lock()
	 *   deactivate_task()
	 *     p->on_rq = 0
	 *     psi_dequeue() // defers TSK_RUNNING & TSK_IOWAIT updates
	 *   pick_next_task()
	 *     rq_unlock()
	 *                                rq_lock()
	 *                                psi_task_change() // old cgroup
	 *                                task->cgroups = to
	 *                                psi_task_change() // new cgroup
	 *                                rq_unlock()
	 *     rq_lock()
	 *   psi_sched_switch() // does deferred updates in new cgroup
	 *
	 * Don't rely on the scheduling state. Use psi_flags instead.
	 */
	task_flags = task->psi_flags;

	if (task_flags)
		psi_task_change(task, task_flags, 0);

	/* See comment above */
	rcu_assign_pointer(task->cgroups, to);

	if (task_flags)
		psi_task_change(task, 0, task_flags);

	task_rq_unlock(rq, task, &rf);
}
#endif /* CONFIG_CGROUPS */

int psi_show(struct seq_file *m, struct psi_group *group, enum psi_res res)
{
	int full;
	u64 now;

	if (static_branch_likely(&psi_disabled))
		return -EOPNOTSUPP;

	/* Update averages before reporting them */
	mutex_lock(&group->avgs_lock);
	now = sched_clock();
	collect_percpu_times(group, PSI_AVGS, NULL);
	if (now >= group->avg_next_update)
		group->avg_next_update = update_averages(group, now);
	mutex_unlock(&group->avgs_lock);

	for (full = 0; full < 2; full++) {
		unsigned long avg[3] = { 0, };
		u64 total = 0;
		int w;

		/* CPU FULL is undefined at the system level */
		if (!(group == &psi_system && res == PSI_CPU && full)) {
			for (w = 0; w < 3; w++)
				avg[w] = group->avg[res * 2 + full][w];
			total = div_u64(group->total[PSI_AVGS][res * 2 + full],
					NSEC_PER_USEC);
		}

		seq_printf(m, "%s avg10=%lu.%02lu avg60=%lu.%02lu avg300=%lu.%02lu total=%llu\n",
			   full ? "full" : "some",
			   LOAD_INT(avg[0]), LOAD_FRAC(avg[0]),
			   LOAD_INT(avg[1]), LOAD_FRAC(avg[1]),
			   LOAD_INT(avg[2]), LOAD_FRAC(avg[2]),
			   total);
	}

	return 0;
}

struct psi_trigger *psi_trigger_create(struct psi_group *group,
			char *buf, size_t nbytes, enum psi_res res)
{
	struct psi_trigger *t;
	enum psi_states state;
	u32 threshold_us;
	u32 window_us;

	if (static_branch_likely(&psi_disabled))
		return ERR_PTR(-EOPNOTSUPP);

	if (sscanf(buf, "some %u %u", &threshold_us, &window_us) == 2)
		state = PSI_IO_SOME + res * 2;
	else if (sscanf(buf, "full %u %u", &threshold_us, &window_us) == 2)
		state = PSI_IO_FULL + res * 2;
	else
		return ERR_PTR(-EINVAL);

	if (state >= PSI_NONIDLE)
		return ERR_PTR(-EINVAL);

	if (window_us < WINDOW_MIN_US ||
		window_us > WINDOW_MAX_US)
		return ERR_PTR(-EINVAL);

	/* Check threshold */
	if (threshold_us == 0 || threshold_us > window_us)
		return ERR_PTR(-EINVAL);

	t = kmalloc(sizeof(*t), GFP_KERNEL);
	if (!t)
		return ERR_PTR(-ENOMEM);

	t->group = group;
	t->state = state;
	t->threshold = threshold_us * NSEC_PER_USEC;
	t->win.size = window_us * NSEC_PER_USEC;
	window_reset(&t->win, sched_clock(),
			group->total[PSI_POLL][t->state], 0);

	t->event = 0;
	t->last_event_time = 0;
	init_waitqueue_head(&t->event_wait);
	t->pending_event = false;

	mutex_lock(&group->trigger_lock);

	if (!rcu_access_pointer(group->poll_task)) {
		struct task_struct *task;

		task = kthread_create(psi_poll_worker, group, "psimon");
		if (IS_ERR(task)) {
			kfree(t);
			mutex_unlock(&group->trigger_lock);
			return ERR_CAST(task);
		}
		atomic_set(&group->poll_wakeup, 0);
		wake_up_process(task);
		rcu_assign_pointer(group->poll_task, task);
	}

	list_add(&t->node, &group->triggers);
	group->poll_min_period = min(group->poll_min_period,
		div_u64(t->win.size, UPDATES_PER_WINDOW));
	group->nr_triggers[t->state]++;
	group->poll_states |= (1 << t->state);

	mutex_unlock(&group->trigger_lock);

	return t;
}

void psi_trigger_destroy(struct psi_trigger *t)
{
	struct psi_group *group;
	struct task_struct *task_to_destroy = NULL;

	/*
	 * We do not check psi_disabled since it might have been disabled after
	 * the trigger got created.
	 */
	if (!t)
		return;

	group = t->group;
	/*
	 * Wakeup waiters to stop polling. Can happen if cgroup is deleted
	 * from under a polling process.
	 */
	wake_up_interruptible(&t->event_wait);

	mutex_lock(&group->trigger_lock);

	if (!list_empty(&t->node)) {
		struct psi_trigger *tmp;
		u64 period = ULLONG_MAX;

		list_del(&t->node);
		group->nr_triggers[t->state]--;
		if (!group->nr_triggers[t->state])
			group->poll_states &= ~(1 << t->state);
		/* reset min update period for the remaining triggers */
		list_for_each_entry(tmp, &group->triggers, node)
			period = min(period, div_u64(tmp->win.size,
					UPDATES_PER_WINDOW));
		group->poll_min_period = period;
		/* Destroy poll_task when the last trigger is destroyed */
		if (group->poll_states == 0) {
			group->polling_until = 0;
			task_to_destroy = rcu_dereference_protected(
					group->poll_task,
					lockdep_is_held(&group->trigger_lock));
			rcu_assign_pointer(group->poll_task, NULL);
			del_timer(&group->poll_timer);
		}
	}

	mutex_unlock(&group->trigger_lock);

	/*
	 * Wait for psi_schedule_poll_work RCU to complete its read-side
	 * critical section before destroying the trigger and optionally the
	 * poll_task.
	 */
	synchronize_rcu();
	/*
	 * Stop kthread 'psimon' after releasing trigger_lock to prevent a
	 * deadlock while waiting for psi_poll_work to acquire trigger_lock
	 */
	if (task_to_destroy) {
		/*
		 * After the RCU grace period has expired, the worker
		 * can no longer be found through group->poll_task.
		 */
		kthread_stop(task_to_destroy);
	}
	kfree(t);
}

__poll_t psi_trigger_poll(void **trigger_ptr,
				struct file *file, poll_table *wait)
{
	__poll_t ret = DEFAULT_POLLMASK;
	struct psi_trigger *t;

	if (static_branch_likely(&psi_disabled))
		return DEFAULT_POLLMASK | EPOLLERR | EPOLLPRI;

	t = smp_load_acquire(trigger_ptr);
	if (!t)
		return DEFAULT_POLLMASK | EPOLLERR | EPOLLPRI;

	poll_wait(file, &t->event_wait, wait);

	if (cmpxchg(&t->event, 1, 0) == 1)
		ret |= EPOLLPRI;

	return ret;
}

#ifdef CONFIG_PROC_FS
static int psi_io_show(struct seq_file *m, void *v)
{
	return psi_show(m, &psi_system, PSI_IO);
}

static int psi_memory_show(struct seq_file *m, void *v)
{
	return psi_show(m, &psi_system, PSI_MEM);
}

static int psi_cpu_show(struct seq_file *m, void *v)
{
	return psi_show(m, &psi_system, PSI_CPU);
}

static int psi_open(struct file *file, int (*psi_show)(struct seq_file *, void *))
{
	if (file->f_mode & FMODE_WRITE && !capable(CAP_SYS_RESOURCE))
		return -EPERM;

	return single_open(file, psi_show, NULL);
}

static int psi_io_open(struct inode *inode, struct file *file)
{
	return psi_open(file, psi_io_show);
}

static int psi_memory_open(struct inode *inode, struct file *file)
{
	return psi_open(file, psi_memory_show);
}

static int psi_cpu_open(struct inode *inode, struct file *file)
{
	return psi_open(file, psi_cpu_show);
}

static ssize_t psi_write(struct file *file, const char __user *user_buf,
			 size_t nbytes, enum psi_res res)
{
	char buf[32];
	size_t buf_size;
	struct seq_file *seq;
	struct psi_trigger *new;

	if (static_branch_likely(&psi_disabled))
		return -EOPNOTSUPP;

	if (!nbytes)
		return -EINVAL;

	buf_size = min(nbytes, sizeof(buf));
	if (copy_from_user(buf, user_buf, buf_size))
		return -EFAULT;

	buf[buf_size - 1] = '\0';

	seq = file->private_data;

	/* Take seq->lock to protect seq->private from concurrent writes */
	mutex_lock(&seq->lock);

	/* Allow only one trigger per file descriptor */
	if (seq->private) {
		mutex_unlock(&seq->lock);
		return -EBUSY;
	}

	new = psi_trigger_create(&psi_system, buf, nbytes, res);
	if (IS_ERR(new)) {
		mutex_unlock(&seq->lock);
		return PTR_ERR(new);
	}

	smp_store_release(&seq->private, new);
	mutex_unlock(&seq->lock);

	return nbytes;
}

static ssize_t psi_io_write(struct file *file, const char __user *user_buf,
			    size_t nbytes, loff_t *ppos)
{
	return psi_write(file, user_buf, nbytes, PSI_IO);
}

static ssize_t psi_memory_write(struct file *file, const char __user *user_buf,
				size_t nbytes, loff_t *ppos)
{
	return psi_write(file, user_buf, nbytes, PSI_MEM);
}

static ssize_t psi_cpu_write(struct file *file, const char __user *user_buf,
			     size_t nbytes, loff_t *ppos)
{
	return psi_write(file, user_buf, nbytes, PSI_CPU);
}

static __poll_t psi_fop_poll(struct file *file, poll_table *wait)
{
	struct seq_file *seq = file->private_data;

	return psi_trigger_poll(&seq->private, file, wait);
}

static int psi_fop_release(struct inode *inode, struct file *file)
{
	struct seq_file *seq = file->private_data;

	psi_trigger_destroy(seq->private);
	return single_release(inode, file);
}

static const struct proc_ops psi_io_proc_ops = {
	.proc_open	= psi_io_open,
	.proc_read	= seq_read,
	.proc_lseek	= seq_lseek,
	.proc_write	= psi_io_write,
	.proc_poll	= psi_fop_poll,
	.proc_release	= psi_fop_release,
};

static const struct proc_ops psi_memory_proc_ops = {
	.proc_open	= psi_memory_open,
	.proc_read	= seq_read,
	.proc_lseek	= seq_lseek,
	.proc_write	= psi_memory_write,
	.proc_poll	= psi_fop_poll,
	.proc_release	= psi_fop_release,
};

static const struct proc_ops psi_cpu_proc_ops = {
	.proc_open	= psi_cpu_open,
	.proc_read	= seq_read,
	.proc_lseek	= seq_lseek,
	.proc_write	= psi_cpu_write,
	.proc_poll	= psi_fop_poll,
	.proc_release	= psi_fop_release,
};

static int __init psi_proc_init(void)
{
	if (psi_enable) {
		proc_mkdir("pressure", NULL);
		proc_create("pressure/io", 0666, NULL, &psi_io_proc_ops);
		proc_create("pressure/memory", 0666, NULL, &psi_memory_proc_ops);
		proc_create("pressure/cpu", 0666, NULL, &psi_cpu_proc_ops);
	}
	return 0;
}
module_init(psi_proc_init);

#endif /* CONFIG_PROC_FS */