aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf/builtin-sched.c
blob: 4f9e943181a39ac40387c66c2f3b7db4dec1c6a9 (plain) (blame)
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
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
#include "builtin.h"

#include "util/util.h"
#include "util/cache.h"
#include "util/symbol.h"
#include "util/thread.h"
#include "util/header.h"

#include "util/parse-options.h"

#include "perf.h"
#include "util/debug.h"

#include "util/trace-event.h"
#include <sys/types.h>


#define MAX_CPUS 4096

static char			const *input_name = "perf.data";
static int			input;
static unsigned long		page_size;
static unsigned long		mmap_window = 32;

static unsigned long		total_comm = 0;

static struct rb_root		threads;
static struct thread		*last_match;

static struct perf_header	*header;
static u64			sample_type;

static int			replay_mode;
static int			lat_mode;


/*
 * Scheduler benchmarks
 */
#include <sys/resource.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/time.h>
#include <sys/prctl.h>

#include <linux/unistd.h>

#include <semaphore.h>
#include <pthread.h>
#include <signal.h>
#include <values.h>
#include <string.h>
#include <unistd.h>
#include <stdlib.h>
#include <assert.h>
#include <fcntl.h>
#include <time.h>
#include <math.h>

#include <stdio.h>

#define PR_SET_NAME	15               /* Set process name */

#define BUG_ON(x)	assert(!(x))

#define DEBUG		0

typedef unsigned long long nsec_t;

static nsec_t run_measurement_overhead;
static nsec_t sleep_measurement_overhead;

static nsec_t get_nsecs(void)
{
	struct timespec ts;

	clock_gettime(CLOCK_MONOTONIC, &ts);

	return ts.tv_sec * 1000000000ULL + ts.tv_nsec;
}

static void burn_nsecs(nsec_t nsecs)
{
	nsec_t T0 = get_nsecs(), T1;

	do {
		T1 = get_nsecs();
	} while (T1 + run_measurement_overhead < T0 + nsecs);
}

static void sleep_nsecs(nsec_t nsecs)
{
	struct timespec ts;

	ts.tv_nsec = nsecs % 999999999;
	ts.tv_sec = nsecs / 999999999;

	nanosleep(&ts, NULL);
}

static void calibrate_run_measurement_overhead(void)
{
	nsec_t T0, T1, delta, min_delta = 1000000000ULL;
	int i;

	for (i = 0; i < 10; i++) {
		T0 = get_nsecs();
		burn_nsecs(0);
		T1 = get_nsecs();
		delta = T1-T0;
		min_delta = min(min_delta, delta);
	}
	run_measurement_overhead = min_delta;

	printf("run measurement overhead: %Ld nsecs\n", min_delta);
}

static void calibrate_sleep_measurement_overhead(void)
{
	nsec_t T0, T1, delta, min_delta = 1000000000ULL;
	int i;

	for (i = 0; i < 10; i++) {
		T0 = get_nsecs();
		sleep_nsecs(10000);
		T1 = get_nsecs();
		delta = T1-T0;
		min_delta = min(min_delta, delta);
	}
	min_delta -= 10000;
	sleep_measurement_overhead = min_delta;

	printf("sleep measurement overhead: %Ld nsecs\n", min_delta);
}

#define COMM_LEN	20
#define SYM_LEN		129

#define MAX_PID		65536

static unsigned long nr_tasks;

struct sched_event;

struct task_desc {
	unsigned long		nr;
	unsigned long		pid;
	char			comm[COMM_LEN];

	unsigned long		nr_events;
	unsigned long		curr_event;
	struct sched_event	**events;

	pthread_t		thread;
	sem_t			sleep_sem;

	sem_t			ready_for_work;
	sem_t			work_done_sem;

	nsec_t			cpu_usage;
};

enum sched_event_type {
	SCHED_EVENT_RUN,
	SCHED_EVENT_SLEEP,
	SCHED_EVENT_WAKEUP,
};

struct sched_event {
	enum sched_event_type	type;
	nsec_t			timestamp;
	nsec_t			duration;
	unsigned long		nr;
	int			specific_wait;
	sem_t			*wait_sem;
	struct task_desc	*wakee;
};

static struct task_desc		*pid_to_task[MAX_PID];

static struct task_desc		**tasks;

static pthread_mutex_t		start_work_mutex = PTHREAD_MUTEX_INITIALIZER;
static nsec_t			start_time;

static pthread_mutex_t		work_done_wait_mutex = PTHREAD_MUTEX_INITIALIZER;

static unsigned long		nr_run_events;
static unsigned long		nr_sleep_events;
static unsigned long		nr_wakeup_events;

static unsigned long		nr_sleep_corrections;
static unsigned long		nr_run_events_optimized;

static struct sched_event *
get_new_event(struct task_desc *task, nsec_t timestamp)
{
	struct sched_event *event = calloc(1, sizeof(*event));
	unsigned long idx = task->nr_events;
	size_t size;

	event->timestamp = timestamp;
	event->nr = idx;

	task->nr_events++;
	size = sizeof(struct sched_event *) * task->nr_events;
	task->events = realloc(task->events, size);
	BUG_ON(!task->events);

	task->events[idx] = event;

	return event;
}

static struct sched_event *last_event(struct task_desc *task)
{
	if (!task->nr_events)
		return NULL;

	return task->events[task->nr_events - 1];
}

static void
add_sched_event_run(struct task_desc *task, nsec_t timestamp, u64 duration)
{
	struct sched_event *event, *curr_event = last_event(task);

	/*
	 * optimize an existing RUN event by merging this one
	 * to it:
	 */
	if (curr_event && curr_event->type == SCHED_EVENT_RUN) {
		nr_run_events_optimized++;
		curr_event->duration += duration;
		return;
	}

	event = get_new_event(task, timestamp);

	event->type = SCHED_EVENT_RUN;
	event->duration = duration;

	nr_run_events++;
}

static unsigned long targetless_wakeups;
static unsigned long multitarget_wakeups;

static void
add_sched_event_wakeup(struct task_desc *task, nsec_t timestamp,
		       struct task_desc *wakee)
{
	struct sched_event *event, *wakee_event;

	event = get_new_event(task, timestamp);
	event->type = SCHED_EVENT_WAKEUP;
	event->wakee = wakee;

	wakee_event = last_event(wakee);
	if (!wakee_event || wakee_event->type != SCHED_EVENT_SLEEP) {
		targetless_wakeups++;
		return;
	}
	if (wakee_event->wait_sem) {
		multitarget_wakeups++;
		return;
	}

	wakee_event->wait_sem = calloc(1, sizeof(*wakee_event->wait_sem));
	sem_init(wakee_event->wait_sem, 0, 0);
	wakee_event->specific_wait = 1;
	event->wait_sem = wakee_event->wait_sem;

	nr_wakeup_events++;
}

static void
add_sched_event_sleep(struct task_desc *task, nsec_t timestamp,
		      u64 task_state __used)
{
	struct sched_event *event = get_new_event(task, timestamp);

	event->type = SCHED_EVENT_SLEEP;

	nr_sleep_events++;
}

static struct task_desc *register_pid(unsigned long pid, const char *comm)
{
	struct task_desc *task;

	BUG_ON(pid >= MAX_PID);

	task = pid_to_task[pid];

	if (task)
		return task;

	task = calloc(1, sizeof(*task));
	task->pid = pid;
	task->nr = nr_tasks;
	strcpy(task->comm, comm);
	/*
	 * every task starts in sleeping state - this gets ignored
	 * if there's no wakeup pointing to this sleep state:
	 */
	add_sched_event_sleep(task, 0, 0);

	pid_to_task[pid] = task;
	nr_tasks++;
	tasks = realloc(tasks, nr_tasks*sizeof(struct task_task *));
	BUG_ON(!tasks);
	tasks[task->nr] = task;

	if (verbose)
		printf("registered task #%ld, PID %ld (%s)\n", nr_tasks, pid, comm);

	return task;
}


static void print_task_traces(void)
{
	struct task_desc *task;
	unsigned long i;

	for (i = 0; i < nr_tasks; i++) {
		task = tasks[i];
		printf("task %6ld (%20s:%10ld), nr_events: %ld\n",
			task->nr, task->comm, task->pid, task->nr_events);
	}
}

static void add_cross_task_wakeups(void)
{
	struct task_desc *task1, *task2;
	unsigned long i, j;

	for (i = 0; i < nr_tasks; i++) {
		task1 = tasks[i];
		j = i + 1;
		if (j == nr_tasks)
			j = 0;
		task2 = tasks[j];
		add_sched_event_wakeup(task1, 0, task2);
	}
}

static void
process_sched_event(struct task_desc *this_task __used, struct sched_event *event)
{
	int ret = 0;
	nsec_t now;
	long long delta;

	now = get_nsecs();
	delta = start_time + event->timestamp - now;

	switch (event->type) {
		case SCHED_EVENT_RUN:
			burn_nsecs(event->duration);
			break;
		case SCHED_EVENT_SLEEP:
			if (event->wait_sem)
				ret = sem_wait(event->wait_sem);
			BUG_ON(ret);
			break;
		case SCHED_EVENT_WAKEUP:
			if (event->wait_sem)
				ret = sem_post(event->wait_sem);
			BUG_ON(ret);
			break;
		default:
			BUG_ON(1);
	}
}

static nsec_t get_cpu_usage_nsec_parent(void)
{
	struct rusage ru;
	nsec_t sum;
	int err;

	err = getrusage(RUSAGE_SELF, &ru);
	BUG_ON(err);

	sum =  ru.ru_utime.tv_sec*1e9 + ru.ru_utime.tv_usec*1e3;
	sum += ru.ru_stime.tv_sec*1e9 + ru.ru_stime.tv_usec*1e3;

	return sum;
}

static nsec_t get_cpu_usage_nsec_self(void)
{
	char filename [] = "/proc/1234567890/sched";
	unsigned long msecs, nsecs;
	char *line = NULL;
	nsec_t total = 0;
	size_t len = 0;
	ssize_t chars;
	FILE *file;
	int ret;

	sprintf(filename, "/proc/%d/sched", getpid());
	file = fopen(filename, "r");
	BUG_ON(!file);

	while ((chars = getline(&line, &len, file)) != -1) {
		ret = sscanf(line, "se.sum_exec_runtime : %ld.%06ld\n",
			&msecs, &nsecs);
		if (ret == 2) {
			total = msecs*1e6 + nsecs;
			break;
		}
	}
	if (line)
		free(line);
	fclose(file);

	return total;
}

static void *thread_func(void *ctx)
{
	struct task_desc *this_task = ctx;
	nsec_t cpu_usage_0, cpu_usage_1;
	unsigned long i, ret;
	char comm2[22];

	sprintf(comm2, ":%s", this_task->comm);
	prctl(PR_SET_NAME, comm2);

again:
	ret = sem_post(&this_task->ready_for_work);
	BUG_ON(ret);
	ret = pthread_mutex_lock(&start_work_mutex);
	BUG_ON(ret);
	ret = pthread_mutex_unlock(&start_work_mutex);
	BUG_ON(ret);

	cpu_usage_0 = get_cpu_usage_nsec_self();

	for (i = 0; i < this_task->nr_events; i++) {
		this_task->curr_event = i;
		process_sched_event(this_task, this_task->events[i]);
	}

	cpu_usage_1 = get_cpu_usage_nsec_self();
	this_task->cpu_usage = cpu_usage_1 - cpu_usage_0;

	ret = sem_post(&this_task->work_done_sem);
	BUG_ON(ret);

	ret = pthread_mutex_lock(&work_done_wait_mutex);
	BUG_ON(ret);
	ret = pthread_mutex_unlock(&work_done_wait_mutex);
	BUG_ON(ret);

	goto again;
}

static void create_tasks(void)
{
	struct task_desc *task;
	pthread_attr_t attr;
	unsigned long i;
	int err;

	err = pthread_attr_init(&attr);
	BUG_ON(err);
	err = pthread_attr_setstacksize(&attr, (size_t)(16*1024));
	BUG_ON(err);
	err = pthread_mutex_lock(&start_work_mutex);
	BUG_ON(err);
	err = pthread_mutex_lock(&work_done_wait_mutex);
	BUG_ON(err);
	for (i = 0; i < nr_tasks; i++) {
		task = tasks[i];
		sem_init(&task->sleep_sem, 0, 0);
		sem_init(&task->ready_for_work, 0, 0);
		sem_init(&task->work_done_sem, 0, 0);
		task->curr_event = 0;
		err = pthread_create(&task->thread, &attr, thread_func, task);
		BUG_ON(err);
	}
}

static nsec_t cpu_usage;
static nsec_t runavg_cpu_usage;
static nsec_t parent_cpu_usage;
static nsec_t runavg_parent_cpu_usage;

static void wait_for_tasks(void)
{
	nsec_t cpu_usage_0, cpu_usage_1;
	struct task_desc *task;
	unsigned long i, ret;

	start_time = get_nsecs();
	cpu_usage = 0;
	pthread_mutex_unlock(&work_done_wait_mutex);

	for (i = 0; i < nr_tasks; i++) {
		task = tasks[i];
		ret = sem_wait(&task->ready_for_work);
		BUG_ON(ret);
		sem_init(&task->ready_for_work, 0, 0);
	}
	ret = pthread_mutex_lock(&work_done_wait_mutex);
	BUG_ON(ret);

	cpu_usage_0 = get_cpu_usage_nsec_parent();

	pthread_mutex_unlock(&start_work_mutex);

	for (i = 0; i < nr_tasks; i++) {
		task = tasks[i];
		ret = sem_wait(&task->work_done_sem);
		BUG_ON(ret);
		sem_init(&task->work_done_sem, 0, 0);
		cpu_usage += task->cpu_usage;
		task->cpu_usage = 0;
	}

	cpu_usage_1 = get_cpu_usage_nsec_parent();
	if (!runavg_cpu_usage)
		runavg_cpu_usage = cpu_usage;
	runavg_cpu_usage = (runavg_cpu_usage*9 + cpu_usage)/10;

	parent_cpu_usage = cpu_usage_1 - cpu_usage_0;
	if (!runavg_parent_cpu_usage)
		runavg_parent_cpu_usage = parent_cpu_usage;
	runavg_parent_cpu_usage = (runavg_parent_cpu_usage*9 +
				   parent_cpu_usage)/10;

	ret = pthread_mutex_lock(&start_work_mutex);
	BUG_ON(ret);

	for (i = 0; i < nr_tasks; i++) {
		task = tasks[i];
		sem_init(&task->sleep_sem, 0, 0);
		task->curr_event = 0;
	}
}

static int __cmd_sched(void);

static void parse_trace(void)
{
	__cmd_sched();

	printf("nr_run_events:        %ld\n", nr_run_events);
	printf("nr_sleep_events:      %ld\n", nr_sleep_events);
	printf("nr_wakeup_events:     %ld\n", nr_wakeup_events);

	if (targetless_wakeups)
		printf("target-less wakeups:  %ld\n", targetless_wakeups);
	if (multitarget_wakeups)
		printf("multi-target wakeups: %ld\n", multitarget_wakeups);
	if (nr_run_events_optimized)
		printf("run events optimized: %ld\n",
			nr_run_events_optimized);
}

static unsigned long nr_runs;
static nsec_t sum_runtime;
static nsec_t sum_fluct;
static nsec_t run_avg;

static void run_one_test(void)
{
	nsec_t T0, T1, delta, avg_delta, fluct, std_dev;

	T0 = get_nsecs();
	wait_for_tasks();
	T1 = get_nsecs();

	delta = T1 - T0;
	sum_runtime += delta;
	nr_runs++;

	avg_delta = sum_runtime / nr_runs;
	if (delta < avg_delta)
		fluct = avg_delta - delta;
	else
		fluct = delta - avg_delta;
	sum_fluct += fluct;
	std_dev = sum_fluct / nr_runs / sqrt(nr_runs);
	if (!run_avg)
		run_avg = delta;
	run_avg = (run_avg*9 + delta)/10;

	printf("#%-3ld: %0.3f, ",
		nr_runs, (double)delta/1000000.0);

#if 0
	printf("%0.2f +- %0.2f, ",
		(double)avg_delta/1e6, (double)std_dev/1e6);
#endif
	printf("ravg: %0.2f, ",
		(double)run_avg/1e6);

	printf("cpu: %0.2f / %0.2f",
		(double)cpu_usage/1e6, (double)runavg_cpu_usage/1e6);

#if 0
	/*
	 * rusage statistics done by the parent, these are less
	 * accurate than the sum_exec_runtime based statistics:
	 */
	printf(" [%0.2f / %0.2f]",
		(double)parent_cpu_usage/1e6,
		(double)runavg_parent_cpu_usage/1e6);
#endif

	printf("\n");

	if (nr_sleep_corrections)
		printf(" (%ld sleep corrections)\n", nr_sleep_corrections);
	nr_sleep_corrections = 0;
}

static void test_calibrations(void)
{
	nsec_t T0, T1;

	T0 = get_nsecs();
	burn_nsecs(1e6);
	T1 = get_nsecs();

	printf("the run test took %Ld nsecs\n", T1-T0);

	T0 = get_nsecs();
	sleep_nsecs(1e6);
	T1 = get_nsecs();

	printf("the sleep test took %Ld nsecs\n", T1-T0);
}

static int
process_comm_event(event_t *event, unsigned long offset, unsigned long head)
{
	struct thread *thread;

	thread = threads__findnew(event->comm.pid, &threads, &last_match);

	dump_printf("%p [%p]: PERF_EVENT_COMM: %s:%d\n",
		(void *)(offset + head),
		(void *)(long)(event->header.size),
		event->comm.comm, event->comm.pid);

	if (thread == NULL ||
	    thread__set_comm(thread, event->comm.comm)) {
		dump_printf("problem processing PERF_EVENT_COMM, skipping event.\n");
		return -1;
	}
	total_comm++;

	return 0;
}


struct raw_event_sample {
	u32 size;
	char data[0];
};

#define FILL_FIELD(ptr, field, event, data)	\
	ptr.field = (typeof(ptr.field)) raw_field_value(event, #field, data)

#define FILL_ARRAY(ptr, array, event, data)			\
do {								\
	void *__array = raw_field_ptr(event, #array, data);	\
	memcpy(ptr.array, __array, sizeof(ptr.array));	\
} while(0)

#define FILL_COMMON_FIELDS(ptr, event, data)			\
do {								\
	FILL_FIELD(ptr, common_type, event, data);		\
	FILL_FIELD(ptr, common_flags, event, data);		\
	FILL_FIELD(ptr, common_preempt_count, event, data);	\
	FILL_FIELD(ptr, common_pid, event, data);		\
	FILL_FIELD(ptr, common_tgid, event, data);		\
} while (0)



struct trace_switch_event {
	u32 size;

	u16 common_type;
	u8 common_flags;
	u8 common_preempt_count;
	u32 common_pid;
	u32 common_tgid;

	char prev_comm[16];
	u32 prev_pid;
	u32 prev_prio;
	u64 prev_state;
	char next_comm[16];
	u32 next_pid;
	u32 next_prio;
};


struct trace_wakeup_event {
	u32 size;

	u16 common_type;
	u8 common_flags;
	u8 common_preempt_count;
	u32 common_pid;
	u32 common_tgid;

	char comm[16];
	u32 pid;

	u32 prio;
	u32 success;
	u32 cpu;
};

struct trace_fork_event {
	u32 size;

	u16 common_type;
	u8 common_flags;
	u8 common_preempt_count;
	u32 common_pid;
	u32 common_tgid;

	char parent_comm[16];
	u32 parent_pid;
	char child_comm[16];
	u32 child_pid;
};

struct trace_sched_handler {
	void (*switch_event)(struct trace_switch_event *,
			     struct event *,
			     int cpu,
			     u64 timestamp,
			     struct thread *thread);

	void (*wakeup_event)(struct trace_wakeup_event *,
			     struct event *,
			     int cpu,
			     u64 timestamp,
			     struct thread *thread);

	void (*fork_event)(struct trace_fork_event *,
			   struct event *,
			   int cpu,
			   u64 timestamp,
			   struct thread *thread);
};


static void
replay_wakeup_event(struct trace_wakeup_event *wakeup_event,
		    struct event *event,
		    int cpu __used,
		    u64 timestamp __used,
		    struct thread *thread __used)
{
	struct task_desc *waker, *wakee;

	if (verbose) {
		printf("sched_wakeup event %p\n", event);

		printf(" ... pid %d woke up %s/%d\n",
			wakeup_event->common_pid,
			wakeup_event->comm,
			wakeup_event->pid);
	}

	waker = register_pid(wakeup_event->common_pid, "<unknown>");
	wakee = register_pid(wakeup_event->pid, wakeup_event->comm);

	add_sched_event_wakeup(waker, timestamp, wakee);
}

static unsigned long cpu_last_switched[MAX_CPUS];

static void
replay_switch_event(struct trace_switch_event *switch_event,
		    struct event *event,
		    int cpu,
		    u64 timestamp,
		    struct thread *thread __used)
{
	struct task_desc *prev, *next;
	u64 timestamp0;
	s64 delta;

	if (verbose)
		printf("sched_switch event %p\n", event);

	if (cpu >= MAX_CPUS || cpu < 0)
		return;

	timestamp0 = cpu_last_switched[cpu];
	if (timestamp0)
		delta = timestamp - timestamp0;
	else
		delta = 0;

	if (delta < 0)
		die("hm, delta: %Ld < 0 ?\n", delta);

	if (verbose) {
		printf(" ... switch from %s/%d to %s/%d [ran %Ld nsecs]\n",
			switch_event->prev_comm, switch_event->prev_pid,
			switch_event->next_comm, switch_event->next_pid,
			delta);
	}

	prev = register_pid(switch_event->prev_pid, switch_event->prev_comm);
	next = register_pid(switch_event->next_pid, switch_event->next_comm);

	cpu_last_switched[cpu] = timestamp;

	add_sched_event_run(prev, timestamp, delta);
	add_sched_event_sleep(prev, timestamp, switch_event->prev_state);
}


static void
replay_fork_event(struct trace_fork_event *fork_event,
		  struct event *event,
		  int cpu __used,
		  u64 timestamp __used,
		  struct thread *thread __used)
{
	if (verbose) {
		printf("sched_fork event %p\n", event);
		printf("... parent: %s/%d\n", fork_event->parent_comm, fork_event->parent_pid);
		printf("...  child: %s/%d\n", fork_event->child_comm, fork_event->child_pid);
	}
	register_pid(fork_event->parent_pid, fork_event->parent_comm);
	register_pid(fork_event->child_pid, fork_event->child_comm);
}

static struct trace_sched_handler replay_ops  = {
	.wakeup_event = replay_wakeup_event,
	.switch_event = replay_switch_event,
	.fork_event = replay_fork_event,
};

#define TASK_STATE_TO_CHAR_STR "RSDTtZX"

enum thread_state {
	THREAD_SLEEPING,
	THREAD_WAKED_UP,
	THREAD_SCHED_IN,
	THREAD_IGNORE
};

struct lat_snapshot {
	struct list_head	list;
	enum thread_state	state;
	u64			wake_up_time;
	u64			sched_in_time;
};

struct thread_latency {
	struct list_head	snapshot_list;
	struct thread		*thread;
	struct rb_node		node;
};

static struct rb_root lat_snapshot_root;

static struct thread_latency *
thread_latency_search(struct rb_root *root, struct thread *thread)
{
	struct rb_node *node = root->rb_node;

	while (node) {
		struct thread_latency *lat;

		lat = container_of(node, struct thread_latency, node);
		if (thread->pid < lat->thread->pid)
			node = node->rb_left;
		else if (thread->pid > lat->thread->pid)
			node = node->rb_right;
		else {
			return lat;
		}
	}
	return NULL;
}

static void
__thread_latency_insert(struct rb_root *root, struct thread_latency *data)
{
	struct rb_node **new = &(root->rb_node), *parent = NULL;

	while (*new) {
		struct thread_latency *this;

		this = container_of(*new, struct thread_latency, node);
		parent = *new;
		if (data->thread->pid < this->thread->pid)
			new = &((*new)->rb_left);
		else if (data->thread->pid > this->thread->pid)
			new = &((*new)->rb_right);
		else
			die("Double thread insertion\n");
	}

	rb_link_node(&data->node, parent, new);
	rb_insert_color(&data->node, root);
}

static void thread_latency_insert(struct thread *thread)
{
	struct thread_latency *lat;
	lat = calloc(sizeof(*lat), 1);
	if (!lat)
		die("No memory");

	lat->thread = thread;
	INIT_LIST_HEAD(&lat->snapshot_list);
	__thread_latency_insert(&lat_snapshot_root, lat);
}

static void
latency_fork_event(struct trace_fork_event *fork_event __used,
		   struct event *event __used,
		   int cpu __used,
		   u64 timestamp __used,
		   struct thread *thread __used)
{
	/* should insert the newcomer */
}

static char sched_out_state(struct trace_switch_event *switch_event)
{
	const char *str = TASK_STATE_TO_CHAR_STR;

	return str[switch_event->prev_state];
}

static void
lat_sched_out(struct thread_latency *lat,
	     struct trace_switch_event *switch_event)
{
	struct lat_snapshot *snapshot;

	if (sched_out_state(switch_event) == 'R')
		return;

	snapshot = calloc(sizeof(*snapshot), 1);
	if (!snapshot)
		die("Non memory");

	list_add_tail(&snapshot->list, &lat->snapshot_list);
}

static void
lat_sched_in(struct thread_latency *lat, u64 timestamp)
{
	struct lat_snapshot *snapshot;

	if (list_empty(&lat->snapshot_list))
		return;

	snapshot = list_entry(lat->snapshot_list.prev, struct lat_snapshot,
			      list);

	if (snapshot->state != THREAD_WAKED_UP)
		return;

	if (timestamp < snapshot->wake_up_time) {
		snapshot->state = THREAD_IGNORE;
		return;
	}

	snapshot->state = THREAD_SCHED_IN;
	snapshot->sched_in_time = timestamp;
}


static void
latency_switch_event(struct trace_switch_event *switch_event,
		     struct event *event __used,
		     int cpu __used,
		     u64 timestamp,
		     struct thread *thread __used)
{
	struct thread_latency *out_lat, *in_lat;
	struct thread *sched_out, *sched_in;

	sched_out = threads__findnew(switch_event->prev_pid, &threads, &last_match);
	sched_in = threads__findnew(switch_event->next_pid, &threads, &last_match);

	in_lat = thread_latency_search(&lat_snapshot_root, sched_in);
	if (!in_lat) {
		thread_latency_insert(sched_in);
		in_lat = thread_latency_search(&lat_snapshot_root, sched_in);
		if (!in_lat)
			die("Internal latency tree error");
	}

	out_lat = thread_latency_search(&lat_snapshot_root, sched_out);
	if (!out_lat) {
		thread_latency_insert(sched_out);
		out_lat = thread_latency_search(&lat_snapshot_root, sched_out);
		if (!out_lat)
			die("Internal latency tree error");
	}

	lat_sched_in(in_lat, timestamp);
	lat_sched_out(out_lat, switch_event);
}

static void
latency_wakeup_event(struct trace_wakeup_event *wakeup_event,
		     struct event *event __used,
		     int cpu __used,
		     u64 timestamp,
		     struct thread *thread __used)
{
	struct thread_latency *lat;
	struct lat_snapshot *snapshot;
	struct thread *wakee;

	/* Note for later, it may be interesting to observe the failing cases */
	if (!wakeup_event->success)
		return;

	wakee = threads__findnew(wakeup_event->pid, &threads, &last_match);
	lat = thread_latency_search(&lat_snapshot_root, wakee);
	if (!lat) {
		thread_latency_insert(wakee);
		return;
	}

	if (list_empty(&lat->snapshot_list))
		return;

	snapshot = list_entry(lat->snapshot_list.prev, struct lat_snapshot,
			      list);

	if (snapshot->state != THREAD_SLEEPING)
		return;

	snapshot->state = THREAD_WAKED_UP;
	snapshot->wake_up_time = timestamp;
}

static struct trace_sched_handler lat_ops  = {
	.wakeup_event = latency_wakeup_event,
	.switch_event = latency_switch_event,
	.fork_event = latency_fork_event,
};

static void output_lat_thread(struct thread_latency *lat)
{
	struct lat_snapshot *shot;
	int count = 0;
	int i;
	int ret;
	u64 max = 0, avg;
	u64 total = 0, delta;

	list_for_each_entry(shot, &lat->snapshot_list, list) {
		if (shot->state != THREAD_SCHED_IN)
			continue;

		count++;

		delta = shot->sched_in_time - shot->wake_up_time;
		if (delta > max)
			max = delta;
		total += delta;
	}

	if (!count)
		return;

	ret = printf("%s", lat->thread->comm);

	for (i = 0; i < 25 - ret; i++)
		printf(" ");

	avg = total / count;

	printf("%5d        %10llu       %10llu      %10llu\n", count, total, avg, max);
}

static void output_lat_results(void)
{
	struct rb_node *next;

	printf(" Tasks");
	printf("                     count");
	printf("          total");
	printf("              avg");
	printf("            max\n\n");

	next = rb_first(&lat_snapshot_root);

	while (next) {
		struct thread_latency *lat;

		lat = rb_entry(next, struct thread_latency, node);
		output_lat_thread(lat);
		next = rb_next(next);
	}
}

static struct trace_sched_handler *trace_handler;

static void
process_sched_wakeup_event(struct raw_event_sample *raw,
			   struct event *event,
			   int cpu __used,
			   u64 timestamp __used,
			   struct thread *thread __used)
{
	struct trace_wakeup_event wakeup_event;

	FILL_COMMON_FIELDS(wakeup_event, event, raw->data);

	FILL_ARRAY(wakeup_event, comm, event, raw->data);
	FILL_FIELD(wakeup_event, pid, event, raw->data);
	FILL_FIELD(wakeup_event, prio, event, raw->data);
	FILL_FIELD(wakeup_event, success, event, raw->data);
	FILL_FIELD(wakeup_event, cpu, event, raw->data);

	trace_handler->wakeup_event(&wakeup_event, event, cpu, timestamp, thread);
}

static void
process_sched_switch_event(struct raw_event_sample *raw,
			   struct event *event,
			   int cpu __used,
			   u64 timestamp __used,
			   struct thread *thread __used)
{
	struct trace_switch_event switch_event;

	FILL_COMMON_FIELDS(switch_event, event, raw->data);

	FILL_ARRAY(switch_event, prev_comm, event, raw->data);
	FILL_FIELD(switch_event, prev_pid, event, raw->data);
	FILL_FIELD(switch_event, prev_prio, event, raw->data);
	FILL_FIELD(switch_event, prev_state, event, raw->data);
	FILL_ARRAY(switch_event, next_comm, event, raw->data);
	FILL_FIELD(switch_event, next_pid, event, raw->data);
	FILL_FIELD(switch_event, next_prio, event, raw->data);

	trace_handler->switch_event(&switch_event, event, cpu, timestamp, thread);
}

static void
process_sched_fork_event(struct raw_event_sample *raw,
			 struct event *event,
			 int cpu __used,
			 u64 timestamp __used,
			 struct thread *thread __used)
{
	struct trace_fork_event fork_event;

	FILL_COMMON_FIELDS(fork_event, event, raw->data);

	FILL_ARRAY(fork_event, parent_comm, event, raw->data);
	FILL_FIELD(fork_event, parent_pid, event, raw->data);
	FILL_ARRAY(fork_event, child_comm, event, raw->data);
	FILL_FIELD(fork_event, child_pid, event, raw->data);

	trace_handler->fork_event(&fork_event, event, cpu, timestamp, thread);
}

static void
process_sched_exit_event(struct event *event,
			 int cpu __used,
			 u64 timestamp __used,
			 struct thread *thread __used)
{
	if (verbose)
		printf("sched_exit event %p\n", event);
}

static void
process_raw_event(event_t *raw_event __used, void *more_data,
		  int cpu, u64 timestamp, struct thread *thread)
{
	struct raw_event_sample *raw = more_data;
	struct event *event;
	int type;

	type = trace_parse_common_type(raw->data);
	event = trace_find_event(type);

	if (!strcmp(event->name, "sched_switch"))
		process_sched_switch_event(raw, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "sched_wakeup"))
		process_sched_wakeup_event(raw, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "sched_wakeup_new"))
		process_sched_wakeup_event(raw, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "sched_process_fork"))
		process_sched_fork_event(raw, event, cpu, timestamp, thread);
	if (!strcmp(event->name, "sched_process_exit"))
		process_sched_exit_event(event, cpu, timestamp, thread);
}

static int
process_sample_event(event_t *event, unsigned long offset, unsigned long head)
{
	char level;
	int show = 0;
	struct dso *dso = NULL;
	struct thread *thread;
	u64 ip = event->ip.ip;
	u64 timestamp = -1;
	u32 cpu = -1;
	u64 period = 1;
	void *more_data = event->ip.__more_data;
	int cpumode;

	thread = threads__findnew(event->ip.pid, &threads, &last_match);

	if (sample_type & PERF_SAMPLE_TIME) {
		timestamp = *(u64 *)more_data;
		more_data += sizeof(u64);
	}

	if (sample_type & PERF_SAMPLE_CPU) {
		cpu = *(u32 *)more_data;
		more_data += sizeof(u32);
		more_data += sizeof(u32); /* reserved */
	}

	if (sample_type & PERF_SAMPLE_PERIOD) {
		period = *(u64 *)more_data;
		more_data += sizeof(u64);
	}

	dump_printf("%p [%p]: PERF_EVENT_SAMPLE (IP, %d): %d/%d: %p period: %Ld\n",
		(void *)(offset + head),
		(void *)(long)(event->header.size),
		event->header.misc,
		event->ip.pid, event->ip.tid,
		(void *)(long)ip,
		(long long)period);

	dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);

	if (thread == NULL) {
		eprintf("problem processing %d event, skipping it.\n",
			event->header.type);
		return -1;
	}

	cpumode = event->header.misc & PERF_EVENT_MISC_CPUMODE_MASK;

	if (cpumode == PERF_EVENT_MISC_KERNEL) {
		show = SHOW_KERNEL;
		level = 'k';

		dso = kernel_dso;

		dump_printf(" ...... dso: %s\n", dso->name);

	} else if (cpumode == PERF_EVENT_MISC_USER) {

		show = SHOW_USER;
		level = '.';

	} else {
		show = SHOW_HV;
		level = 'H';

		dso = hypervisor_dso;

		dump_printf(" ...... dso: [hypervisor]\n");
	}

	if (sample_type & PERF_SAMPLE_RAW)
		process_raw_event(event, more_data, cpu, timestamp, thread);

	return 0;
}

static int
process_event(event_t *event, unsigned long offset, unsigned long head)
{
	trace_event(event);

	switch (event->header.type) {
	case PERF_EVENT_MMAP ... PERF_EVENT_LOST:
		return 0;

	case PERF_EVENT_COMM:
		return process_comm_event(event, offset, head);

	case PERF_EVENT_EXIT ... PERF_EVENT_READ:
		return 0;

	case PERF_EVENT_SAMPLE:
		return process_sample_event(event, offset, head);

	case PERF_EVENT_MAX:
	default:
		return -1;
	}

	return 0;
}

static int __cmd_sched(void)
{
	int ret, rc = EXIT_FAILURE;
	unsigned long offset = 0;
	unsigned long head = 0;
	struct stat perf_stat;
	event_t *event;
	uint32_t size;
	char *buf;

	trace_report();
	register_idle_thread(&threads, &last_match);

	input = open(input_name, O_RDONLY);
	if (input < 0) {
		perror("failed to open file");
		exit(-1);
	}

	ret = fstat(input, &perf_stat);
	if (ret < 0) {
		perror("failed to stat file");
		exit(-1);
	}

	if (!perf_stat.st_size) {
		fprintf(stderr, "zero-sized file, nothing to do!\n");
		exit(0);
	}
	header = perf_header__read(input);
	head = header->data_offset;
	sample_type = perf_header__sample_type(header);

	if (!(sample_type & PERF_SAMPLE_RAW))
		die("No trace sample to read. Did you call perf record "
		    "without -R?");

	if (load_kernel() < 0) {
		perror("failed to load kernel symbols");
		return EXIT_FAILURE;
	}

remap:
	buf = (char *)mmap(NULL, page_size * mmap_window, PROT_READ,
			   MAP_SHARED, input, offset);
	if (buf == MAP_FAILED) {
		perror("failed to mmap file");
		exit(-1);
	}

more:
	event = (event_t *)(buf + head);

	size = event->header.size;
	if (!size)
		size = 8;

	if (head + event->header.size >= page_size * mmap_window) {
		unsigned long shift = page_size * (head / page_size);
		int res;

		res = munmap(buf, page_size * mmap_window);
		assert(res == 0);

		offset += shift;
		head -= shift;
		goto remap;
	}

	size = event->header.size;


	if (!size || process_event(event, offset, head) < 0) {

		/*
		 * assume we lost track of the stream, check alignment, and
		 * increment a single u64 in the hope to catch on again 'soon'.
		 */

		if (unlikely(head & 7))
			head &= ~7ULL;

		size = 8;
	}

	head += size;

	if (offset + head < (unsigned long)perf_stat.st_size)
		goto more;

	rc = EXIT_SUCCESS;
	close(input);

	return rc;
}

static const char * const annotate_usage[] = {
	"perf trace [<options>] <command>",
	NULL
};

static const struct option options[] = {
	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
		    "dump raw trace in ASCII"),
	OPT_BOOLEAN('r', "replay", &replay_mode,
		    "replay sched behaviour from traces"),
	OPT_BOOLEAN('l', "latency", &lat_mode,
		    "measure various latencies"),
	OPT_BOOLEAN('v', "verbose", &verbose,
		    "be more verbose (show symbol address, etc)"),
	OPT_END()
};

int cmd_sched(int argc, const char **argv, const char *prefix __used)
{
	long nr_iterations = 10, i;

	symbol__init();
	page_size = getpagesize();

	argc = parse_options(argc, argv, options, annotate_usage, 0);
	if (argc) {
		/*
		 * Special case: if there's an argument left then assume tha
		 * it's a symbol filter:
		 */
		if (argc > 1)
			usage_with_options(annotate_usage, options);
	}

//	setup_pager();

	if (replay_mode)
		trace_handler = &replay_ops;
	else if (lat_mode)
		trace_handler = &lat_ops;
	else /* We may need a default subcommand (perf trace?) */
		die("Please select a sub command (-r)\n");

	if (replay_mode) {
		calibrate_run_measurement_overhead();
		calibrate_sleep_measurement_overhead();

		test_calibrations();

		parse_trace();
		print_task_traces();
		add_cross_task_wakeups();

		create_tasks();
		printf("------------------------------------------------------------\n");
		for (i = 0; i < nr_iterations; i++)
			run_one_test();
	} else if (lat_mode) {
		setup_pager();
		__cmd_sched();
		output_lat_results();
	}

	return 0;
}