-
Notifications
You must be signed in to change notification settings - Fork 12
/
task-state.c
1054 lines (927 loc) · 36.3 KB
/
task-state.c
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
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <monitor.h>
#include <dlfcn.h>
#include <errno.h>
#include <linux/rblist.h>
#include <linux/thread_map.h>
#include <monitor.h>
#include <trace_helpers.h>
#include <stack_helpers.h>
#include <latency_helpers.h>
#include <tp_struct.h>
#define TASK_RUNNING 0
#define TASK_INTERRUPTIBLE 1
#define TASK_UNINTERRUPTIBLE 2
#define __TASK_STOPPED 4
#define __TASK_TRACED 8
/* in tsk->exit_state */
#define EXIT_ZOMBIE 16
#define EXIT_DEAD 32
/* in tsk->state again */
#define TASK_DEAD 64
#define TASK_WAKEKILL 128
#define TASK_WAKING 256
#define TASK_PARKED 512
#define TASK_STATE_MAX 1024
/* Convenience macros for the sake of set_task_state */
#define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
#define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
#define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
#if 0 // LINUX 4.14
/* Used in tsk->state: */
#define TASK_RUNNING 0x0000
#define TASK_INTERRUPTIBLE 0x0001
#define TASK_UNINTERRUPTIBLE 0x0002
#define __TASK_STOPPED 0x0004
#define __TASK_TRACED 0x0008
/* Used in tsk->exit_state: */
#define EXIT_DEAD 0x0010
#define EXIT_ZOMBIE 0x0020
#define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
/* Used in tsk->state again: */
#define TASK_PARKED 0x0040
#define TASK_DEAD 0x0080
/* get_task_state(): */
#define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
__TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
TASK_PARKED)
#define TASK_REPORT_IDLE (TASK_REPORT + 1)
#define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
#endif
#define TASK_REPORT (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | __TASK_TRACED)
#define RUNDELAY (TASK_STATE_MAX << 1)
#define TASK_REPORT_IDLE 0x80 // kernel 4.14 and later.
#define TASK_REPORT_MAX 0x100 // kernel 4.14 and later.
struct task_state_ctx {
struct callchain_ctx *cc;
struct flame_graph *flame;
struct perf_thread_map *thread_map;
union {
struct perf_evsel *sched_switch;
struct perf_evsel *sched_switch_prev;
};
struct perf_evsel *sched_switch_next;
struct perf_evsel *sched_wakeup;
struct perf_evsel *sched_wakeup_new;
struct tp_matcher *matcher_switch, *matcher_wakeup, *matcher_wakeup_new;
struct rblist task_states;
struct latency_dist *lat_dist;
struct comm_notify notify;
int state_dead, report_max; // Compatible with different kernel release.
int task_report;
union {
int mode;
struct {
int filter:1;
int SD:1;
};
};
// lost
struct list_head lost_list; // struct task_lost_node
// minevtime
u64 recent_time;
// stat
struct __dup_stat {
u64 sampled;
u64 freed;
u64 mem_bytes;
} stat;
};
struct task_state_node {
struct rb_node rbnode;
int pid;
int state;
u64 time;
union perf_event *event;
};
struct task_lost_node {
struct list_head lost_link;
int ins;
bool reclaim;
u64 start_time;
u64 end_time;
};
union sched_event {
unsigned short common_type;// offset:0; size:2; signed:0;
struct sched_wakeup sched_wakeup;
struct sched_wakeup_no_success sched_wakeup_no_success;
struct sched_switch sched_switch;
};
// in linux/perf_event.h
// PERF_SAMPLE_TID | PERF_SAMPLE_TIME | PERF_SAMPLE_ID | PERF_SAMPLE_CPU | PERF_SAMPLE_RAW
struct sample_type_header {
struct {
__u32 pid;
__u32 tid;
} tid_entry;
__u64 time;
__u64 id;
struct {
__u32 cpu;
__u32 reserved;
} cpu_entry;
};
struct sample_type_callchain {
struct sample_type_header h;
struct callchain callchain;
};
struct sample_type_raw {
struct sample_type_header h;
struct {
__u32 size;
__u8 data[0];
} raw;
};
static void task_state_fork(void *opaque, void *raw);
static void task_state_hangup(void *opaque);
static int task_state_node_cmp(struct rb_node *rbn, const void *entry)
{
struct task_state_node *b = container_of(rbn, struct task_state_node, rbnode);
const struct task_state_node *e = entry;
return b->pid - e->pid;
}
static struct rb_node *task_state_node_new(struct rblist *rlist, const void *new_entry)
{
struct task_state_node *b = malloc(sizeof(*b));
if (b) {
b->pid = -1;
b->time = 0;
b->event = NULL;
RB_CLEAR_NODE(&b->rbnode);
return &b->rbnode;
} else
return NULL;
}
static void task_state_node_delete(struct rblist *rblist, struct rb_node *rb_node)
{
struct task_state_ctx *ctx = container_of(rblist, struct task_state_ctx, task_states);
struct task_state_node *b = container_of(rb_node, struct task_state_node, rbnode);
if (b->event) {
ctx->stat.mem_bytes -= b->event->header.size;
ctx->stat.freed++;
free(b->event);
}
free(b);
}
static void monitor_ctx_exit(struct prof_dev *dev);
static int monitor_ctx_init(struct prof_dev *dev)
{
struct env *env = dev->env;
struct task_state_ctx *ctx = zalloc(sizeof(*ctx));
if (!ctx)
return -1;
dev->private = ctx;
INIT_LIST_HEAD(&ctx->lost_list);
tep__ref();
if (env->callchain) {
if (!env->flame_graph)
ctx->cc = callchain_ctx_new(callchain_flags(dev, CALLCHAIN_KERNEL | CALLCHAIN_USER), stdout);
else
ctx->flame = flame_graph_open(callchain_flags(dev, CALLCHAIN_KERNEL | CALLCHAIN_USER), env->flame_graph);
dev->pages *= 2;
}
rblist__init(&ctx->task_states);
ctx->task_states.node_cmp = task_state_node_cmp;
ctx->task_states.node_new = task_state_node_new;
ctx->task_states.node_delete = task_state_node_delete;
if (prof_dev_is_cloned(dev)) {
struct task_state_ctx *pctx = prof_dev_is_cloned(dev)->private;
ctx->lat_dist = latency_dist_ref(pctx->lat_dist);
} else {
ctx->lat_dist = latency_dist_new_quantile(env->perins, true, 0);
if (!ctx->lat_dist)
goto failed;
}
return 0;
failed:
monitor_ctx_exit(dev);
return -1;
}
static void monitor_ctx_exit(struct prof_dev *dev)
{
struct task_state_ctx *ctx = dev->private;
struct task_lost_node *lost, *next;
list_for_each_entry_safe(lost, next, &ctx->lost_list, lost_link)
free(lost);
perf_thread_map__put(ctx->thread_map);
rblist__exit(&ctx->task_states);
if (dev->env->callchain) {
if (!dev->env->flame_graph)
callchain_ctx_free(ctx->cc);
else {
flame_graph_output(ctx->flame);
flame_graph_close(ctx->flame);
}
}
latency_dist_free(ctx->lat_dist);
if ((ctx->mode == 1 && dev->env->filter) || ctx->mode == 3)
global_comm_unregister_notify(&ctx->notify);
tep__unref();
free(ctx);
}
static int task_state_notify(struct comm_notify *notify, int pid, int state, u64 free_time)
{
if (state == NOTIFY_COMM_DELETE) {
struct task_state_ctx *ctx = container_of(notify, struct task_state_ctx, notify);
struct task_state_node *task, tmp;
struct rb_node *rbn;
tmp.pid = pid;
rbn = rblist__find(&ctx->task_states, &tmp);
if (rbn) {
task = rb_entry(rbn, struct task_state_node, rbnode);
if (task->time < free_time)
rblist__remove_node(&ctx->task_states, rbn);
}
}
return 0;
}
static int task_state_init(struct prof_dev *dev)
{
struct perf_evlist *evlist = dev->evlist;
struct env *env = dev->env;
struct task_state_ctx *ctx;
struct perf_event_attr attr = {
.type = PERF_TYPE_TRACEPOINT,
.config = 0,
.size = sizeof(struct perf_event_attr),
.sample_period = 1,
.sample_type = PERF_SAMPLE_TID | PERF_SAMPLE_TIME | PERF_SAMPLE_ID | PERF_SAMPLE_CPU | PERF_SAMPLE_RAW |
(env->callchain ? PERF_SAMPLE_CALLCHAIN : 0),
.read_format = PERF_FORMAT_ID,
.pinned = 1,
.disabled = 1,
.watermark = 1,
.exclude_callchain_user = exclude_callchain_user(dev, CALLCHAIN_KERNEL | CALLCHAIN_USER),
.exclude_callchain_kernel = exclude_callchain_kernel(dev, CALLCHAIN_KERNEL | CALLCHAIN_USER),
};
struct perf_evsel *evsel;
int id;
if (monitor_ctx_init(dev) < 0)
return -1;
ctx = dev->private;
reduce_wakeup_times(dev, &attr);
/**
* sched:sched_switch and sched:sched_wakeup are not suitable for binding to threads
**/
if (!prof_dev_ins_oncpu(dev)) {
ctx->thread_map = dev->threads;
perf_cpu_map__put(dev->cpus);
dev->cpus = perf_cpu_map__new(NULL);
dev->threads = perf_thread_map__new_dummy();
}
ctx->task_report = TASK_REPORT;
if (env->interruptible_set && !env->interruptible)
ctx->task_report &= ~TASK_INTERRUPTIBLE;
if (kernel_release() >= KERNEL_VERSION(4, 14, 0)) {
ctx->state_dead = EXIT_ZOMBIE|EXIT_DEAD;
ctx->report_max = TASK_REPORT_MAX;
ctx->task_report |= TASK_REPORT_IDLE;
} else {
ctx->state_dead = EXIT_ZOMBIE|EXIT_DEAD|TASK_DEAD;
ctx->report_max = TASK_STATE_MAX;
}
if (env->greater_than && using_order(dev))
dev->dup = true;
/* | mode |
* | filter | event
* | S/D pid/comm | sched_switch sched_switch sched_wakeup sched_wakeup_new
* 0 0 // none // //
* 0 1 /prev_pid==xx/ /next_pid==xx/ /pid==xx/ /pid==xx/
* 1 0 /prev_state==xx/ none // none
* 1 1 /prev_state==xx && prev_pid==xx/ none /pid==xx/ none
*/
ctx->SD = !!(env->interruptible || env->uninterruptible);
ctx->filter = !!(ctx->thread_map || env->filter);
/*
* Mode 1 filtering comm will be affected by task_rename.
* --filter "sh"
* sched_switch: next_comm=sh next_pid=1045
* task_rename: pid=1045 oldcomm=sh newcomm=awk
* pid=1045 will remain on the ctx->task_states list until the pid is reused.
*
* Mode 3 is affected by task_switch.
* -S --filter "sh"
* 311612.197341 sched:sched_switch: prev_pid=91319 prev_comm=sh prev_state=S
* 311612.197354 sched:sched_switch: next_pid=91319 next_comm=sh
* pid=91319 sleeps, but there is no sched_wakeup and starts running directly.
* After this, pid=91319 will remain on the ctx->task_states list until the pid is reused.
*
* These two modes track the process free and delete the corresponding task_state_node from
* the ctx->task_states list. This doesn't completely solve the problem, just alleviates it.
*/
if ((ctx->mode == 1 && env->filter) || ctx->mode == 3) {
ctx->notify.notify = task_state_notify;
global_comm_register_notify(&ctx->notify);
}
// sched:sched_switch//
// sched:sched_switch/prev_pid==xx/
// sched:sched_switch/prev_comm==xx/
// sched:sched_switch/prev_state==xx/
// sched:sched_switch/prev_state==xx && prev_pid==xx/
// sched:sched_switch/prev_state==xx && prev_comm==xx/
id = tep__event_id("sched", "sched_switch");
if (id < 0)
goto failed;
attr.config = id;
evsel = ctx->sched_switch = perf_evsel__new(&attr);
if (!evsel)
goto failed;
perf_evlist__add(evlist, evsel);
ctx->matcher_switch = tp_matcher_find("sched", "sched_switch");
// sched:sched_switch/next_pid==xx/
// sched:sched_switch/next_comm==xx/
if (ctx->mode == 1) {
evsel = ctx->sched_switch_next = perf_evsel__new(&attr);
if (!evsel)
goto failed;
perf_evlist__add(evlist, evsel);
} else
ctx->sched_switch_next = NULL;
// sched:sched_wakeup//
// sched:sched_wakeup/pid==xx/
id = tep__event_id("sched", "sched_wakeup");
if (id < 0)
goto failed;
attr.config = id;
evsel = ctx->sched_wakeup = perf_evsel__new(&attr);
if (!evsel)
goto failed;
perf_evlist__add(evlist, evsel);
ctx->matcher_wakeup = tp_matcher_find("sched", "sched_wakeup");
// sched:sched_wakeup_new//
// sched:sched_wakeup_new/pid==xx/
if (ctx->mode == 0 || (ctx->mode == 1 && env->filter)) {
id = tep__event_id("sched", "sched_wakeup_new");
if (id < 0)
goto failed;
attr.config = id;
evsel = ctx->sched_wakeup_new = perf_evsel__new(&attr);
if (!evsel)
goto failed;
perf_evlist__add(evlist, evsel);
ctx->matcher_wakeup_new = tp_matcher_find("sched", "sched_wakeup_new");
} else
ctx->sched_wakeup_new = NULL;
/*
* mode 1, mode 3, !-p pid, !-t tids, only for workload and cloned prof_dev.
*
* For trace_dev_open(), if the event is attached to `thread_map' and sched
* switching is frequent, the CPU utilization of these threads will increase.
* So, for workload only, trace sched_process_fork.
*/
if (ctx->thread_map && !env->filter && !env->pids && !env->tids) {
// sched:sched_process_fork
trace_dev_open("sched:sched_process_fork", NULL, ctx->thread_map, dev,
task_state_fork, task_state_hangup);
}
return 0;
failed:
monitor_ctx_exit(dev);
return -1;
}
static int task_state_filter(struct prof_dev *dev)
{
struct perf_evlist *evlist = dev->evlist;
struct env *env = dev->env;
struct task_state_ctx *ctx = dev->private;
char filter[4096];
struct perf_evsel *evsel;
int err = 0;
perf_evlist__for_each_evsel(evlist, evsel) {
if (evsel == ctx->sched_switch) {
struct tp_filter *prev_filter = NULL;
prev_filter = tp_filter_new(ctx->thread_map, "prev_pid", env->filter, "prev_comm");
if (env->interruptible && env->uninterruptible) {
if (prev_filter) {
snprintf(filter, sizeof(filter), "(prev_state==%d || prev_state==%d || prev_state==%d) && (%s)",
TASK_INTERRUPTIBLE, TASK_UNINTERRUPTIBLE, TASK_KILLABLE, prev_filter->filter);
} else
snprintf(filter, sizeof(filter), "prev_state==%d || prev_state==%d || prev_state==%d",
TASK_INTERRUPTIBLE, TASK_UNINTERRUPTIBLE, TASK_KILLABLE);
} else if (env->interruptible) {
if (prev_filter)
snprintf(filter, sizeof(filter), "prev_state==%d && (%s)",
TASK_INTERRUPTIBLE, prev_filter->filter);
else
snprintf(filter, sizeof(filter), "prev_state==%d", TASK_INTERRUPTIBLE);
} else if (env->uninterruptible) {
if (prev_filter)
snprintf(filter, sizeof(filter), "(prev_state==%d || prev_state==%d) && (%s)",
TASK_UNINTERRUPTIBLE, TASK_KILLABLE, prev_filter->filter);
else
snprintf(filter, sizeof(filter), "prev_state==%d || prev_state==%d",
TASK_UNINTERRUPTIBLE, TASK_KILLABLE);
} else if (prev_filter) {
snprintf(filter, sizeof(filter), "%s", prev_filter->filter);
} else {
filter[0] = '\0';
}
tp_filter_free(prev_filter);
if (filter[0])
err = perf_evsel__apply_filter(evsel, filter);
if (err < 0 || env->verbose >= VERBOSE_NOTICE)
fprintf(err < 0 ? stderr : stdout, "sched:sched_switch filter \"%s\"\n", filter);
if (err < 0) return err;
} else if (evsel == ctx->sched_switch_next) {
struct tp_filter *next_filter = NULL;
next_filter = tp_filter_new(ctx->thread_map, "next_pid", env->filter, "next_comm");
if (next_filter)
err = perf_evsel__apply_filter(evsel, next_filter->filter);
if (err < 0 || env->verbose >= VERBOSE_NOTICE)
fprintf(err < 0 ? stderr : stdout, "sched:sched_switch filter \"%s\"\n", next_filter ? next_filter->filter : "");
tp_filter_free(next_filter);
if (err < 0) return err;
} else if (evsel == ctx->sched_wakeup || evsel == ctx->sched_wakeup_new) {
struct tp_filter *tp_filter = NULL;
tp_filter = tp_filter_new(ctx->thread_map, "pid", env->filter, "comm");
if (tp_filter)
err = perf_evsel__apply_filter(evsel, tp_filter->filter);
if (err < 0 || env->verbose >= VERBOSE_NOTICE)
fprintf(err < 0 ? stderr : stdout, "sched:sched_wakeup%s filter \"%s\"\n",
evsel == ctx->sched_wakeup ? "" : "_new", tp_filter ? tp_filter->filter : "");
tp_filter_free(tp_filter);
if (err < 0) return err;
}
}
return 0;
}
static void task_state_enabled(struct prof_dev *dev)
{
if (prof_dev_is_cloned(dev)) {
struct task_state_ctx *ctx = dev->private;
int idx, pid;
if (!ctx->thread_map)
return;
perf_thread_map__for_each_thread(pid, idx, ctx->thread_map) {
if (kill(pid, 0) == 0)
return;
}
if (dev->env->verbose) {
pid = perf_thread_map__pid(ctx->thread_map, 0);
print_time(stdout);
printf("%s close %d in task_state_enabled()\n", dev->prof->name, pid);
}
prof_dev_close(dev);
}
}
static void task_print_node(void *opaque, struct latency_node *node)
{
struct prof_dev *dev = opaque;
double p50 = tdigest_quantile(node->td, 0.50);
double p95 = tdigest_quantile(node->td, 0.95);
double p99 = tdigest_quantile(node->td, 0.99);
const char *state = NULL;
switch (node->key) {
case TASK_RUNNING: state = "R "; break;
case TASK_INTERRUPTIBLE: state = "S "; break;
case TASK_UNINTERRUPTIBLE: state = "D "; break;
case __TASK_STOPPED: state = "T "; break;
case __TASK_TRACED: state = "t "; break;
case TASK_REPORT_IDLE: state = "I "; break;
case RUNDELAY: state = "RD"; break;
default: return;
}
if (dev->env->perins) {
int pid = (int)node->instance;
printf("%6d %-16s ", pid, tep__pid_to_comm(pid));
}
printf("%s %8lu %16.3f %12.3f %12.3f %12.3f %12.3f %12.3f\n", state,
node->n, node->sum/1000.0, node->min/1000.0, p50/1000.0, p95/1000.0, p99/1000.0, node->max/1000.0);
}
static void task_state_interval(struct prof_dev *dev)
{
struct task_state_ctx *ctx = dev->private;
if (!prof_dev_at_top(dev))
return;
if (latency_dist_empty(ctx->lat_dist))
return;
print_time(stdout);
printf("\n");
if (dev->env->perins)
printf("thread %-*s ", 16, "comm");
printf("St %8s %16s %12s %12s %12s %12s %12s\n", "calls", "total(us)", "min(us)", "p50(us)",
"p95(us)", "p99(us)", "max(us)");
if (dev->env->perins)
printf("------ ---------------- ");
printf("-- %8s %16s %12s %12s %12s %12s %12s\n",
"--------", "----------------", "------------", "------------", "------------",
"------------", "------------");
latency_dist_print(ctx->lat_dist, task_print_node, dev);
}
static void task_state_deinit(struct prof_dev *dev)
{
task_state_interval(dev);
monitor_ctx_exit(dev);
}
static u64 task_state_minevtime(struct prof_dev *dev)
{
struct task_state_ctx *ctx = dev->private;
struct env *env = dev->env;
u64 minevtime = ULLONG_MAX;
/*
* The processes on the ctx->task_states rblist are all alive, and
* there is no need to obtain their minevtime.
*/
if (env->perins) {
u64 mintime = 0;
if (env->interval && ctx->recent_time > env->interval * NSEC_PER_MSEC)
mintime = ctx->recent_time - env->interval * NSEC_PER_MSEC;
if (mintime < minevtime)
minevtime = mintime;
}
return minevtime;
}
static void task_state_lost(struct prof_dev *dev, union perf_event *event, int ins, u64 lost_start, u64 lost_end)
{
struct task_state_ctx *ctx = dev->private;
struct task_lost_node *pos;
struct task_lost_node *lost;
print_lost_fn(dev, event, ins);
// task-state serves as the forwarding source device.
if (unlikely(!prof_dev_is_final(dev)))
return;
// Order is enabled by default.
// When order is enabled, event loss will be sensed in advance, but it
// needs to be processed later.
lost = malloc(sizeof(*lost));
if (lost) {
lost->ins = ins;
lost->reclaim = false;
lost->start_time = lost_start;
lost->end_time = lost_end;
list_for_each_entry(pos, &ctx->lost_list, lost_link) {
if (pos->start_time > lost_start)
break;
}
list_add_tail(&lost->lost_link, &pos->lost_link);
}
}
static void __raw_size(struct prof_dev *dev, union perf_event *event, void **praw, int *psize)
{
if (dev->env->callchain) {
struct sample_type_callchain *data = (void *)event->sample.array;
struct {
__u32 size;
__u8 data[0];
} *raw = (void *)data->callchain.ips + data->callchain.nr * sizeof(__u64);
*praw = raw->data;
*psize = raw->size;
} else {
struct sample_type_raw *raw = (void *)event->sample.array;
*praw = raw->raw.data;
*psize = raw->raw.size;
}
}
static inline void __print_callchain(struct prof_dev *dev, union perf_event *event)
{
struct task_state_ctx *ctx = dev->private;
struct sample_type_callchain *data = (void *)event->sample.array;
if (dev->env->callchain) {
if (!dev->env->flame_graph)
print_callchain_common(ctx->cc, &data->callchain, data->h.tid_entry.pid);
else {
const char *comm = tep__pid_to_comm((int)data->h.tid_entry.pid);
flame_graph_add_callchain(ctx->flame, &data->callchain, data->h.tid_entry.pid, !strcmp(comm, "<...>") ? NULL : comm);
}
}
}
static void task_state_print_event(struct prof_dev *dev, union perf_event *event)
{
struct sample_type_header *data = (void *)event->sample.array;
void *raw;
int size;
__raw_size(dev, event, &raw, &size);
if (dev->print_title) prof_dev_print_time(dev, data->time, stdout);
tep__print_event(data->time, data->cpu_entry.cpu, raw, size);
__print_callchain(dev, event);
}
static inline int task_state_event_lost(struct prof_dev *dev, union perf_event *event)
{
struct task_state_ctx *ctx = dev->private;
struct sample_type_header *data = (void *)event->sample.array;
struct task_lost_node *lost, *next;
if (likely(list_empty(&ctx->lost_list)))
return 0;
list_for_each_entry_safe(lost, next, &ctx->lost_list, lost_link) {
// Events before lost->start_time are processed normally.
if (data->time <= lost->start_time)
return 0;
/*
* Not sure which events are lost, we can only delete all process states
* in `ctx->task_states'. Restart collection after lost.
*/
if (!lost->reclaim) {
rblist__exit(&ctx->task_states);
lost->reclaim = true;
}
// Within the lost range, new events are also unsafe.
if (data->time < lost->end_time) {
return -1;
} else {
// Re-process subsequent events normally.
list_del(&lost->lost_link);
free(lost);
}
}
return 0;
}
static void task_state_fork(void *opaque, void *raw)
{
struct prof_dev *dev = opaque;
struct sched_process_fork *sched_fork = raw;
struct perf_thread_map *map;
if (kill(sched_fork->child_pid, 0) < 0) return;
map = thread_map__new_by_tid(sched_fork->child_pid);
if (!map) return;
dev = prof_dev_clone(dev, NULL, map);
perf_thread_map__put(map);
}
static void task_state_hangup(void *opaque)
{
struct prof_dev *dev = opaque;
prof_dev_close(dev);
}
static void task_state_sample(struct prof_dev *dev, union perf_event *event, int instance)
{
struct env *env = dev->env;
struct task_state_ctx *ctx = dev->private;
// in linux/perf_event.h
// PERF_SAMPLE_TID | PERF_SAMPLE_TIME | PERF_SAMPLE_CPU | PERF_SAMPLE_RAW
struct sample_type_header *data = (void *)event->sample.array;
struct task_state_node *task, tmp;
union sched_event *sched_event;
struct sched_switch *sw = NULL;
struct perf_evsel *evsel;
struct rb_node *rbn;
void *raw;
int size;
bool keep = false;
if (dev->dup)
ctx->stat.sampled ++;
if (data->time > ctx->recent_time)
ctx->recent_time = data->time;
if (unlikely(!prof_dev_is_final(dev))) {
// When task-state is used as a forwarding device, it only prints out the event.
task_state_print_event(dev, event);
goto free_event;
}
if (unlikely(env->verbose >= VERBOSE_EVENT))
task_state_print_event(dev, event);
if (unlikely(task_state_event_lost(dev, event) < 0))
goto free_event;
evsel = perf_evlist__id_to_evsel(dev->evlist, data->id, NULL);
if (!evsel)
goto free_event;
__raw_size(dev, event, &raw, &size);
sched_event = raw;
/* | mode |
* | filter | event
* | S/D pid/comm | sched_switch | sched_switch | sched_wakeup | sched_wakeup_new
* - - - - - - - - | - - - - - - - - - - - - - - - - - | - - - - - - - - | - - - - - - - | - - - - - - - - -
* 0 0 | // | none | // | //
* | RUNNING | | S/D/T/t | to RUNDELAY
* | to S/D/T/t | | to RUNDELAY |
* | RUNDELAY | | |
* | to RUNNING | | |
* | | | |
* - - - - - - - - | - - - - - - - - - - - - - - - - - | - - - - - - - - | - - - - - - - | - - - - - - - - -
* 0 1 | /prev_pid==xx/ | /next_pid==xx/ | /pid==xx/ | /pid==xx/
* | RUNNING | RUNDELAY | S/D/T/t | to RUNDELAY
* | to S/D/T/t | to RUNNING | to RUNDELAY |
* | | | |
* - - - - - - - - | - - - - - - - - - - - - - - - - - | - - - - - - - - | - - - - - - - | - - - - - - - - -
* 1 0 | /prev_state==xx/ | none | // | none
* | to S/D | | S/D |
* | | | |
* - - - - - - - - | - - - - - - - - - - - - - - - - - | - - - - - - - - | - - - - - - - | - - - - - - - - -
* 1 1 | /prev_state==xx && prev_pid==xx/ | none | /pid==xx/ | none
* | to S/D | | S/D |
*/
if (evsel == ctx->sched_switch) {
sw = &sched_event->sched_switch;
if (sw->prev_pid > 0) {
tmp.pid = sw->prev_pid;
rbn = rblist__findnew(&ctx->task_states, &tmp);
task = rb_entry_safe(rbn, struct task_state_node, rbnode);
if (task) {
if (task->pid != -1 && data->time > task->time) {
// RUNNING
if (task->state == TASK_RUNNING) {
latency_dist_input(ctx->lat_dist, task->pid, TASK_RUNNING, data->time - task->time, env->greater_than);
}
}
// to INTERRUPTIBLE/UNINTERRUPTIBLE/STOPPED/TRACED
// to S/D/T/t
task->pid = sw->prev_pid;
task->state = sw->prev_state == ctx->report_max ? TASK_RUNNING : sw->prev_state;
task->time = data->time;
if (sw->prev_state & ctx->state_dead)
rblist__remove_node(&ctx->task_states, rbn);
else if (dev->dup) {
if (task->event) {
ctx->stat.mem_bytes -= task->event->header.size;
ctx->stat.freed++;
free(task->event);
}
task->event = event;
keep = true;
}
}
}
if (ctx->mode == 0)
goto parse_next;
} else if (evsel == ctx->sched_switch_next) {
sw = &sched_event->sched_switch;
parse_next:
if (sw->next_pid > 0) {
tmp.pid = sw->next_pid;
rbn = rblist__findnew(&ctx->task_states, &tmp);
task = rb_entry_safe(rbn, struct task_state_node, rbnode);
if (task) {
if (task->pid != -1 && data->time > task->time) {
// RUNDELAY: sched_wakeup -> sched_switch
if (task->state == TASK_RUNNING) {
u64 delta = data->time - task->time;
latency_dist_input(ctx->lat_dist, task->pid, RUNDELAY, delta, env->greater_than);
if (env->greater_than && delta > env->greater_than &&
task->event) {
if (dev->print_title) print_time(stdout);
printf(" task-state: %d %s RUNDELAY %lu ms\n", task->pid, tep__pid_to_comm(task->pid), delta/NSEC_PER_MSEC);
task_state_print_event(dev, task->event);
task_state_print_event(dev, event);
}
}
}
// to RUNNING
task->pid = sw->next_pid;
task->state = TASK_RUNNING;
task->time = data->time;
if (dev->dup) {
if (task->event) {
ctx->stat.mem_bytes -= task->event->header.size;
ctx->stat.freed++;
free(task->event);
}
task->event = NULL;
}
// goto free_event;
}
}
} else if (evsel == ctx->sched_wakeup || evsel == ctx->sched_wakeup_new) {
struct sched_wakeup *wakeup = &sched_event->sched_wakeup;
tmp.pid = wakeup->pid;
if (ctx->mode == 2 || ctx->mode == 3)
rbn = rblist__find(&ctx->task_states, &tmp);
else rbn = rblist__findnew(&ctx->task_states, &tmp);
task = rb_entry_safe(rbn, struct task_state_node, rbnode);
if (task) {
if (task->pid != -1 && data->time > task->time) {
// S/D/T/t/I
int state = task->state & ctx->task_report;
if (state) {
u64 delta = data->time - task->time;
latency_dist_input(ctx->lat_dist, task->pid, state, delta, env->greater_than);
if (env->greater_than && delta > env->greater_than &&
task->event) {
if (dev->print_title) print_time(stdout);
printf(" task-state: %d %s WAIT %lu ms\n", task->pid, tep__pid_to_comm(task->pid), delta/NSEC_PER_MSEC);
task_state_print_event(dev, task->event);
task_state_print_event(dev, event);
}
}
}
if (ctx->mode == 2 || ctx->mode == 3) {
rblist__remove_node(&ctx->task_states, rbn);
goto free_event;
}
// to RUNDELAY
if (task->state != TASK_RUNNING || task->pid == -1 || evsel == ctx->sched_wakeup_new)
task->time = data->time;
task->pid = wakeup->pid;
task->state = TASK_RUNNING;
if (dev->dup) {
if (task->event) {
ctx->stat.mem_bytes -= task->event->header.size;
ctx->stat.freed++;
free(task->event);
}
task->event = event;
keep = true;
}
}
}
free_event:
if (keep) ctx->stat.mem_bytes += event->header.size;
if (dev->dup && !keep) {
ctx->stat.freed++;
free(event);
}
}
static void task_state_sigusr(struct prof_dev *dev, int signum)
{
struct task_state_ctx *ctx = dev->private;
if (signum == SIGUSR1) {
print_time(stdout);
printf("task-state\n");
printf(" sampled: %lu\n"
" freed: %lu\n"
" mem_bytes: %lu\n"
" tasks %d\n",
ctx->stat.sampled, ctx->stat.freed, ctx->stat.mem_bytes,
rblist__nr_entries(&ctx->task_states));
}
}
static const char *task_state_desc[] = PROFILER_DESC("task-state",
"[OPTION...] [-S] [-D] [--than ns] [--filter comm] [--perins] [-g [--flame-graph file]]",
"Trace task state, wakeup, switch, INTERRUPTIBLE, UNINTERRUPTIBLE.", "",
"TRACEPOINT",
" sched:sched_switch, sched:sched_wakeup, sched:sched_wakeup_new", "",
"EXAMPLES",
" "PROGRAME" task-state -i 1000 --no-interruptible",
" "PROGRAME" task-state -p 2347 -SD --than 20ms -g",
" "PROGRAME" task-state --filter 'java,python*' -S --than 100ms -g",
" "PROGRAME" task-state -- ip link show eth0");
static const char *task_state_argv[] = PROFILER_ARGV("task-state",
PROFILER_ARGV_OPTION,
PROFILER_ARGV_CALLCHAIN_FILTER,
PROFILER_ARGV_PROFILER, "interruptible", "uninterruptible", "than", "filter", "perins", "call-graph", "flame-graph");
struct monitor task_state = {
.name = "task-state",
.desc = task_state_desc,
.argv = task_state_argv,
.pages = 8,
.order = 1,
.init = task_state_init,
.filter = task_state_filter,
.enabled = task_state_enabled,
.deinit = task_state_deinit,
.sigusr = task_state_sigusr,
.interval = task_state_interval,
.minevtime = task_state_minevtime,
.lost = task_state_lost,
.sample = task_state_sample,
};
MONITOR_REGISTER(task_state);
struct matcher_result {
struct tp_matcher *matcher;
void *true_raw;
int true_size;
};
/*
* task-state itself does not use tp. But when it is used as a forwarding source,
* tp will be assigned to it in the forwarding target device.
*/
static void task_state_matcher(struct tp *tp, void *raw, int size, struct matcher_result *result)
{
union perf_event *event = raw;
struct prof_dev *dev;
struct task_state_ctx *ctx;