1 ================
2 Event Histograms
3 ================
4
5 Documentation written by Tom Zanussi
6
7 1. Introduction
8 ===============
9
10 Histogram triggers are special event trigger
11 aggregate trace event data into histograms.
12 trace events and event triggers, see Documen
13
14
15 2. Histogram Trigger Command
16 ============================
17
18 A histogram trigger command is an event trig
19 aggregates event hits into a hash table keye
20 event format fields (or stacktrace) and a se
21 derived from one or more trace event format
22 counts (hitcount).
23
24 The format of a hist trigger is as follows::
25
26 hist:keys=<field1[,field2,...]>[:value
27 [:sort=<field1[,field2,...]>][:size=
28 [:clear][:name=histname1][:nohitcoun
29
30 When a matching event is hit, an entry is ad
31 using the key(s) and value(s) named. Keys a
32 fields in the event's format description. V
33 numeric fields - on an event hit, the value(
34 sum kept for that field. The special string
35 in place of an explicit value field - this i
36 event hits. If 'values' isn't specified, an
37 value will be automatically created and used
38 Keys can be any field, or the special string
39 will use the event's kernel stacktrace as th
40 'keys' or 'key' can be used to specify keys,
41 'values', 'vals', or 'val' can be used to sp
42 keys consisting of up to three fields can be
43 keyword. Hashing a compound key produces a
44 table for each unique combination of compone
45 useful for providing more fine-grained summa
46 Additionally, sort keys consisting of up to
47 specified by the 'sort' keyword. If more th
48 specified, the result will be a 'sort within
49 is taken to be the primary sort key and the
50 key. If a hist trigger is given a name usin
51 its histogram data will be shared with other
52 name, and trigger hits will update this comm
53 with 'compatible' fields can be combined in
54 'compatible' if the fields named in the trig
55 number and type of fields and those fields a
56 Note that any two events always share the co
57 'common_stacktrace' fields and can therefore
58 fields, however pointless that may be.
59
60 'hist' triggers add a 'hist' file to each ev
61 Reading the 'hist' file for the event will d
62 its entirety to stdout. If there are multip
63 attached to an event, there will be a table
64 output. The table displayed for a named tri
65 any other instance having the same name. Eac
66 entry is a simple list of the keys and value
67 keys are printed first and are delineated by
68 followed by the set of value fields for the
69 numeric fields are displayed as base-10 inte
70 modified by appending any of the following m
71 name:
72
73 ============= =======================
74 .hex display a number as a h
75 .sym display an address as a
76 .sym-offset display an address as a
77 .syscall display a syscall id as
78 .execname display a common_pid as
79 .log2 display log2 value rath
80 .buckets=size display grouping of val
81 .usecs display a common_timest
82 .percent display a number of per
83 .graph display a bar-graph of
84 .stacktrace display as a stacktrace
85 ============= =======================
86
87 Note that in general the semantics of a give
88 interpreted when applying a modifier to it,
89 restrictions to be aware of in this regard:
90
91 - only the 'hex' modifier can be used for
92 are essentially sums, and the other modi
93 in that context).
94 - the 'execname' modifier can only be used
95 reason for this is that the execname is
96 saved for the 'current' process when an
97 which is the same as the common_pid valu
98 tracing code. Trying to apply that comm
99 values wouldn't be correct, and typicall
100 pid-specific comm fields in the event it
101
102 A typical usage scenario would be the follow
103 trigger, read its current contents, and then
104
105 # echo 'hist:keys=skbaddr.hex:vals=len' >
106 /sys/kernel/tracing/events/net/netif_rx/
107
108 # cat /sys/kernel/tracing/events/net/netif
109
110 # echo '!hist:keys=skbaddr.hex:vals=len' >
111 /sys/kernel/tracing/events/net/netif_rx/
112
113 The trigger file itself can be read to show
114 currently attached hist trigger. This infor
115 at the top of the 'hist' file when read.
116
117 By default, the size of the hash table is 20
118 parameter can be used to specify more or few
119 are in terms of hashtable entries - if a run
120 specified, the results will show the number
121 of hits that were ignored. The size should
122 128 and 131072 (any non- power-of-2 number s
123 up).
124
125 The 'sort' parameter can be used to specify
126 on. The default if unspecified is 'hitcount
127 order is 'ascending'. To sort in the opposi
128 .descending' to the sort key.
129
130 The 'pause' parameter can be used to pause a
131 or to start a hist trigger but not log any e
132 so. 'continue' or 'cont' can be used to sta
133 hist trigger.
134
135 The 'clear' parameter will clear the content
136 trigger and leave its current paused/active
137
138 Note that the 'pause', 'cont', and 'clear' p
139 applied using 'append' shell operator ('>>')
140 existing trigger, rather than via the '>' op
141 the trigger to be removed through truncation
142
143 The 'nohitcount' (or NOHC) parameter will su
144 raw hitcount in the histogram. This option r
145 value field which is not a 'raw hitcount'. F
146 'hist:...:vals=hitcount:nohitcount' is rejec
147 'hist:...:vals=hitcount.percent:nohitcount'
148
149 - enable_hist/disable_hist
150
151 The enable_hist and disable_hist triggers ca
152 event conditionally start and stop another e
153 hist trigger. Any number of enable_hist and
154 can be attached to a given event, allowing t
155 and stop aggregations on a host of other eve
156
157 The format is very similar to the enable/dis
158
159 enable_hist:<system>:<event>[:count]
160 disable_hist:<system>:<event>[:count]
161
162 Instead of enabling or disabling the tracing
163 into the trace buffer as the enable/disable_
164 enable/disable_hist triggers enable or disab
165 the target event into a hash table.
166
167 A typical usage scenario for the enable_hist
168 would be to first set up a paused hist trigg
169 followed by an enable_hist/disable_hist pair
170 aggregation on and off when conditions of in
171
172 # echo 'hist:keys=skbaddr.hex:vals=len:paus
173 /sys/kernel/tracing/events/net/netif_rec
174
175 # echo 'enable_hist:net:netif_receive_skb
176 /sys/kernel/tracing/events/sched/sched_p
177
178 # echo 'disable_hist:net:netif_receive_skb
179 /sys/kernel/tracing/events/sched/sched_p
180
181 The above sets up an initially paused hist t
182 and starts aggregating events when a given p
183 which stops aggregating when the process exi
184 is paused again.
185
186 The examples below provide a more concrete i
187 concepts and typical usage patterns discusse
188
189 'special' event fields
190 ------------------------
191
192 There are a number of 'special event fields'
193 keys or values in a hist trigger. These loo
194 they were actual event fields, but aren't re
195 field definition or format file. They are h
196 event, and can be used anywhere an actual ev
197 They are:
198
199 ====================== ==== ==============
200 common_timestamp u64 timestamp (fro
201 with the event
202 modified by .u
203 interpreted as
204 common_cpu int the cpu on whi
205 ====================== ==== ==============
206
207 Extended error information
208 --------------------------
209
210 For some error conditions encountered when i
211 command, extended error information is avail
212 tracing/error_log file. See Error Condition
213 :file:`Documentation/trace/ftrace.rst` for d
214
215 6.2 'hist' trigger examples
216 ---------------------------
217
218 The first set of examples creates aggregatio
219 event. The fields that can be used for the
220 in the kmalloc event's format file::
221
222 # cat /sys/kernel/tracing/events/kmem/kmal
223 name: kmalloc
224 ID: 374
225 format:
226 field:unsigned short common_type;
227 field:unsigned char common_flags;
228 field:unsigned char common_preempt_cou
229 field:int common_pid;
230
231 field:unsigned long call_site;
232 field:const void * ptr;
233 field:size_t bytes_req;
234 field:size_t bytes_alloc;
235 field:gfp_t gfp_flags;
236
237 We'll start by creating a hist trigger that
238 that lists the total number of bytes request
239 the kernel that made one or more calls to km
240
241 # echo 'hist:key=call_site:val=bytes_req.b
242 /sys/kernel/tracing/events/kmem/km
243
244 This tells the tracing system to create a 'h
245 call_site field of the kmalloc event as the
246 just means that each unique call_site addres
247 created for it in the table. The 'val=bytes
248 the hist trigger that for each unique entry
249 table, it should keep a running total of the
250 requested by that call_site.
251
252 We'll let it run for awhile and then dump th
253 file in the kmalloc event's subdirectory (fo
254 of entries have been omitted)::
255
256 # cat /sys/kernel/tracing/events/kmem/kmal
257 # trigger info: hist:keys=call_site:vals=b
258
259 { call_site: 18446744072106379007 } hitcou
260 { call_site: 18446744071579557049 } hitcou
261 { call_site: 18446744071580608289 } hitcou
262 { call_site: 18446744071581827654 } hitcou
263 { call_site: 18446744071580700980 } hitcou
264 { call_site: 18446744071579359876 } hitcou
265 { call_site: 18446744071580795365 } hitcou
266 { call_site: 18446744071581303129 } hitcou
267 { call_site: 18446744071580713234 } hitcou
268 { call_site: 18446744071580933750 } hitcou
269 .
270 .
271 .
272 { call_site: 18446744072106047046 } hitcou
273 { call_site: 18446744071582116407 } hitcou
274 { call_site: 18446744072106054684 } hitcou
275 { call_site: 18446744072106224230 } hitcou
276 { call_site: 18446744072106078074 } hitcou
277 { call_site: 18446744072106062406 } hitcou
278 { call_site: 18446744071582507929 } hitcou
279 { call_site: 18446744072102520590 } hitcou
280 { call_site: 18446744071582143559 } hitcou
281 { call_site: 18446744072106465852 } hitcou
282 { call_site: 18446744072102523378 } hitcou
283 { call_site: 18446744072099568646 } hitcou
284
285 Totals:
286 Hits: 4610
287 Entries: 45
288 Dropped: 0
289
290 The output displays a line for each entry, b
291 specified in the trigger, followed by the va
292 the trigger. At the beginning of the output
293 the trigger info, which can also be displaye
294 'trigger' file::
295
296 # cat /sys/kernel/tracing/events/kmem/kmal
297 hist:keys=call_site:vals=bytes_req:sort=hi
298
299 At the end of the output are a few lines tha
300 totals for the run. The 'Hits' field shows
301 times the event trigger was hit, the 'Entrie
302 number of used entries in the hash table, an
303 shows the number of hits that were dropped b
304 used entries for the run exceeded the maximu
305 allowed for the table (normally 0, but if no
306 want to increase the size of the table using
307
308 Notice in the above output that there's an e
309 which wasn't specified in the trigger. Also
310 trigger info output, there's a parameter, 's
311 wasn't specified in the trigger either. The
312 every trigger implicitly keeps a count of th
313 attributed to a given entry, called the 'hit
314 information is explicitly displayed in the o
315 absence of a user-specified sort parameter,
316 sort field.
317
318 The value 'hitcount' can be used in place of
319 the 'values' parameter if you don't really n
320 particular field summed and are mainly inter
321 frequencies.
322
323 To turn the hist trigger off, simply call up
324 command history and re-execute it with a '!'
325
326 # echo '!hist:key=call_site:val=bytes_req'
327 /sys/kernel/tracing/events/kmem/kma
328
329 Finally, notice that the call_site as displa
330 isn't really very useful. It's an address,
331 are displayed in hex. To have a numeric fie
332 value, simply append '.hex' to the field nam
333
334 # echo 'hist:key=call_site.hex:val=bytes_r
335 /sys/kernel/tracing/events/kmem/kma
336
337 # cat /sys/kernel/tracing/events/kmem/kmal
338 # trigger info: hist:keys=call_site.hex:va
339
340 { call_site: ffffffffa026b291 } hitcount:
341 { call_site: ffffffffa07186ff } hitcount:
342 { call_site: ffffffff811ae721 } hitcount:
343 { call_site: ffffffff811c5134 } hitcount:
344 { call_site: ffffffffa04a9ebb } hitcount:
345 { call_site: ffffffff8122e0a6 } hitcount:
346 { call_site: ffffffff8107da84 } hitcount:
347 { call_site: ffffffff812d8246 } hitcount:
348 { call_site: ffffffff811dc1e5 } hitcount:
349 { call_site: ffffffffa02515e8 } hitcount:
350 { call_site: ffffffff81258159 } hitcount:
351 { call_site: ffffffff811c80f4 } hitcount:
352 .
353 .
354 .
355 { call_site: ffffffffa06c7646 } hitcount:
356 { call_site: ffffffffa06cb246 } hitcount:
357 { call_site: ffffffffa06cef7a } hitcount:
358 { call_site: ffffffff8137e399 } hitcount:
359 { call_site: ffffffffa06c941c } hitcount:
360 { call_site: ffffffffa06f2a66 } hitcount:
361 { call_site: ffffffffa036a70e } hitcount:
362 { call_site: ffffffff81325447 } hitcount:
363 { call_site: ffffffffa072da3c } hitcount:
364 { call_site: ffffffffa036b1f2 } hitcount:
365 { call_site: ffffffffa0099c06 } hitcount:
366
367 Totals:
368 Hits: 4775
369 Entries: 46
370 Dropped: 0
371
372 Even that's only marginally more useful - wh
373 more like addresses, what users are typicall
374 when looking at text addresses are the corre
375 instead. To have an address displayed as sy
376 simply append '.sym' or '.sym-offset' to the
377 trigger::
378
379 # echo 'hist:key=call_site.sym:val=bytes_r
380 /sys/kernel/tracing/events/kmem/kma
381
382 # cat /sys/kernel/tracing/events/kmem/kmal
383 # trigger info: hist:keys=call_site.sym:va
384
385 { call_site: [ffffffff810adcb9] syslog_pri
386 { call_site: [ffffffff8154bc62] usb_contro
387 { call_site: [ffffffffa00bf6fe] hidraw_sen
388 { call_site: [ffffffff8154acbe] usb_alloc_
389 { call_site: [ffffffffa00bf1ca] hidraw_rep
390 { call_site: [ffffffff811e3a25] __seq_open
391 { call_site: [ffffffff8109524a] alloc_fair
392 { call_site: [ffffffff811febd5] fsnotify_a
393 { call_site: [ffffffff81440f58] __tty_buff
394 { call_site: [ffffffff81200ba6] inotify_ne
395 { call_site: [ffffffffa05e19af] ieee80211_
396 { call_site: [ffffffff81672406] tcp_get_me
397 { call_site: [ffffffff81097ec2] alloc_rt_s
398 { call_site: [ffffffff81089b05] sched_crea
399 .
400 .
401 .
402 { call_site: [ffffffffa04a580c] intel_crtc
403 { call_site: [ffffffffa0287592] drm_mode_p
404 { call_site: [ffffffffa04c4a3c] intel_plan
405 { call_site: [ffffffff812891ca] ext4_find_
406 { call_site: [ffffffffa029070e] drm_vma_no
407 { call_site: [ffffffffa045e7c4] i915_gem_d
408 { call_site: [ffffffffa02911f2] drm_modese
409 { call_site: [ffffffffa0489a66] intel_ring
410 { call_site: [ffffffffa046041c] i915_gem_e
411 { call_site: [ffffffff81325607] apparmor_f
412 { call_site: [ffffffffa00b7c06] hid_report
413 { call_site: [ffffffff8131ebf7] aa_alloc_t
414 { call_site: [ffffffff8125847d] ext4_htree
415
416 Totals:
417 Hits: 109928
418 Entries: 71
419 Dropped: 0
420
421 Because the default sort key above is 'hitco
422 the list of call_sites by increasing hitcoun
423 we see the functions that made the most kmal
424 run. If instead we wanted to see the top km
425 terms of the number of bytes requested rathe
426 calls, and we wanted the top caller to appea
427 the 'sort' parameter, along with the 'descen
428
429 # echo 'hist:key=call_site.sym:val=bytes_r
430 /sys/kernel/tracing/events/kmem/kma
431
432 # cat /sys/kernel/tracing/events/kmem/kmal
433 # trigger info: hist:keys=call_site.sym:va
434
435 { call_site: [ffffffffa046041c] i915_gem_e
436 { call_site: [ffffffffa045e7c4] i915_gem_d
437 { call_site: [ffffffff8125847d] ext4_htree
438 { call_site: [ffffffff811e2a1b] seq_buf_al
439 { call_site: [ffffffffa0489a66] intel_ring
440 { call_site: [ffffffff812891ca] ext4_find_
441 { call_site: [ffffffff811ae8e1] __kmalloc
442 { call_site: [ffffffffa04c4a3c] intel_plan
443 { call_site: [ffffffffa02911f2] drm_modese
444 { call_site: [ffffffffa04a580c] intel_crtc
445 { call_site: [ffffffffa0287592] drm_mode_p
446 { call_site: [ffffffffa00b7c06] hid_report
447 { call_site: [ffffffff8137e559] sg_kmalloc
448 .
449 .
450 .
451 { call_site: [ffffffff81095225] alloc_fair
452 { call_site: [ffffffff81097ec2] alloc_rt_s
453 { call_site: [ffffffff812d8406] copy_semun
454 { call_site: [ffffffff81200ba6] inotify_ne
455 { call_site: [ffffffffa027121a] drm_getmag
456 { call_site: [ffffffff811e3a25] __seq_open
457 { call_site: [ffffffff811c52f4] bprm_chang
458 { call_site: [ffffffff8154bc62] usb_contro
459 { call_site: [ffffffffa00bf1ca] hidraw_rep
460 { call_site: [ffffffffa00bf6fe] hidraw_sen
461
462 Totals:
463 Hits: 32133
464 Entries: 81
465 Dropped: 0
466
467 To display the offset and size information i
468 name, just use 'sym-offset' instead::
469
470 # echo 'hist:key=call_site.sym-offset:val=
471 /sys/kernel/tracing/events/kmem/kma
472
473 # cat /sys/kernel/tracing/events/kmem/kmal
474 # trigger info: hist:keys=call_site.sym-of
475
476 { call_site: [ffffffffa046041c] i915_gem_e
477 { call_site: [ffffffffa0489a66] intel_ring
478 { call_site: [ffffffffa045e7c4] i915_gem_d
479 { call_site: [ffffffffa045e646] i915_gem_d
480 { call_site: [ffffffff811e2a1b] seq_buf_al
481 { call_site: [ffffffffa04a580c] intel_crtc
482 { call_site: [ffffffff811ae8e1] __kmalloc+
483 { call_site: [ffffffffa0287592] drm_mode_p
484 { call_site: [ffffffffa02911f2] drm_modese
485 { call_site: [ffffffffa04c4a3c] intel_plan
486 { call_site: [ffffffffa029070e] drm_vma_no
487 { call_site: [ffffffff815f8d7b] sk_prot_al
488 .
489 .
490 .
491 { call_site: [ffffffff8109524a] alloc_fair
492 { call_site: [ffffffffa027b921] drm_vm_ope
493 { call_site: [ffffffff8122e266] proc_self_
494 { call_site: [ffffffff81213e80] load_elf_b
495 { call_site: [ffffffff8154bc62] usb_contro
496 { call_site: [ffffffffa00bf6fe] hidraw_sen
497 { call_site: [ffffffffa00bf1ca] hidraw_rep
498
499 Totals:
500 Hits: 26098
501 Entries: 64
502 Dropped: 0
503
504 We can also add multiple fields to the 'valu
505 example, we might want to see the total numb
506 alongside bytes requested, and display the r
507 allocated in a descending order::
508
509 # echo 'hist:keys=call_site.sym:values=byt
510 /sys/kernel/tracing/events/kmem/kma
511
512 # cat /sys/kernel/tracing/events/kmem/kmal
513 # trigger info: hist:keys=call_site.sym:va
514
515 { call_site: [ffffffffa046041c] i915_gem_e
516 { call_site: [ffffffff811e2a1b] seq_buf_al
517 { call_site: [ffffffffa0489a66] intel_ring
518 { call_site: [ffffffffa045e7c4] i915_gem_d
519 { call_site: [ffffffff8125847d] ext4_htree
520 { call_site: [ffffffffa045e646] i915_gem_d
521 { call_site: [ffffffffa04c4a3c] intel_plan
522 { call_site: [ffffffffa02911f2] drm_modese
523 { call_site: [ffffffff815f8d7b] sk_prot_al
524 { call_site: [ffffffff8137e559] sg_kmalloc
525 { call_site: [ffffffffa00b7c06] hid_report
526 { call_site: [ffffffffa04a580c] intel_crtc
527 .
528 .
529 .
530 { call_site: [ffffffff8109bd3b] sched_auto
531 { call_site: [ffffffff81097ee8] alloc_rt_s
532 { call_site: [ffffffff8109524a] alloc_fair
533 { call_site: [ffffffff81095225] alloc_fair
534 { call_site: [ffffffff81097ec2] alloc_rt_s
535 { call_site: [ffffffff81213e80] load_elf_b
536 { call_site: [ffffffff81079a2e] kthread_cr
537 { call_site: [ffffffffa00bf6fe] hidraw_sen
538 { call_site: [ffffffff8154bc62] usb_contro
539 { call_site: [ffffffffa00bf1ca] hidraw_rep
540
541 Totals:
542 Hits: 66598
543 Entries: 65
544 Dropped: 0
545
546 Finally, to finish off our kmalloc example,
547 the hist trigger display symbolic call_sites
548 trigger additionally display the complete se
549 that led to each call_site. To do that, we
550 value 'common_stacktrace' for the key parame
551
552 # echo 'hist:keys=common_stacktrace:values
553 /sys/kernel/tracing/events/kmem/kma
554
555 The above trigger will use the kernel stack
556 event is triggered as the key for the hash t
557 enumeration of every kernel callpath that le
558 event, along with a running total of any of
559 that event. Here we tally bytes requested a
560 every callpath in the system that led up to
561 every callpath to a kmalloc for a kernel com
562
563 # cat /sys/kernel/tracing/events/kmem/kmal
564 # trigger info: hist:keys=common_stacktrac
565
566 { common_stacktrace:
567 __kmalloc_track_caller+0x10b/0x1a0
568 kmemdup+0x20/0x50
569 hidraw_report_event+0x8a/0x120 [hid]
570 hid_report_raw_event+0x3ea/0x440 [hid
571 hid_input_report+0x112/0x190 [hid]
572 hid_irq_in+0xc2/0x260 [usbhid]
573 __usb_hcd_giveback_urb+0x72/0x120
574 usb_giveback_urb_bh+0x9e/0xe0
575 tasklet_hi_action+0xf8/0x100
576 __do_softirq+0x114/0x2c0
577 irq_exit+0xa5/0xb0
578 do_IRQ+0x5a/0xf0
579 ret_from_intr+0x0/0x30
580 cpuidle_enter+0x17/0x20
581 cpu_startup_entry+0x315/0x3e0
582 rest_init+0x7c/0x80
583 } hitcount: 3 bytes_req:
584 { common_stacktrace:
585 __kmalloc_track_caller+0x10b/0x1a0
586 kmemdup+0x20/0x50
587 hidraw_report_event+0x8a/0x120 [hid]
588 hid_report_raw_event+0x3ea/0x440 [hid
589 hid_input_report+0x112/0x190 [hid]
590 hid_irq_in+0xc2/0x260 [usbhid]
591 __usb_hcd_giveback_urb+0x72/0x120
592 usb_giveback_urb_bh+0x9e/0xe0
593 tasklet_hi_action+0xf8/0x100
594 __do_softirq+0x114/0x2c0
595 irq_exit+0xa5/0xb0
596 do_IRQ+0x5a/0xf0
597 ret_from_intr+0x0/0x30
598 } hitcount: 3 bytes_req:
599 { common_stacktrace:
600 kmem_cache_alloc_trace+0xeb/0x150
601 aa_alloc_task_context+0x27/0x40
602 apparmor_cred_prepare+0x1f/0x50
603 security_prepare_creds+0x16/0x20
604 prepare_creds+0xdf/0x1a0
605 SyS_capset+0xb5/0x200
606 system_call_fastpath+0x12/0x6a
607 } hitcount: 1 bytes_req:
608 .
609 .
610 .
611 { common_stacktrace:
612 __kmalloc+0x11b/0x1b0
613 i915_gem_execbuffer2+0x6c/0x2c0 [i915
614 drm_ioctl+0x349/0x670 [drm]
615 do_vfs_ioctl+0x2f0/0x4f0
616 SyS_ioctl+0x81/0xa0
617 system_call_fastpath+0x12/0x6a
618 } hitcount: 17726 bytes_req: 13944
619 { common_stacktrace:
620 __kmalloc+0x11b/0x1b0
621 load_elf_phdrs+0x76/0xa0
622 load_elf_binary+0x102/0x1650
623 search_binary_handler+0x97/0x1d0
624 do_execveat_common.isra.34+0x551/0x6e
625 SyS_execve+0x3a/0x50
626 return_from_execve+0x0/0x23
627 } hitcount: 33348 bytes_req: 17152
628 { common_stacktrace:
629 kmem_cache_alloc_trace+0xeb/0x150
630 apparmor_file_alloc_security+0x27/0x4
631 security_file_alloc+0x16/0x20
632 get_empty_filp+0x93/0x1c0
633 path_openat+0x31/0x5f0
634 do_filp_open+0x3a/0x90
635 do_sys_open+0x128/0x220
636 SyS_open+0x1e/0x20
637 system_call_fastpath+0x12/0x6a
638 } hitcount: 4766422 bytes_req: 9532
639 { common_stacktrace:
640 __kmalloc+0x11b/0x1b0
641 seq_buf_alloc+0x1b/0x50
642 seq_read+0x2cc/0x370
643 proc_reg_read+0x3d/0x80
644 __vfs_read+0x28/0xe0
645 vfs_read+0x86/0x140
646 SyS_read+0x46/0xb0
647 system_call_fastpath+0x12/0x6a
648 } hitcount: 19133 bytes_req: 78368
649
650 Totals:
651 Hits: 6085872
652 Entries: 253
653 Dropped: 0
654
655 If you key a hist trigger on common_pid, in
656 gather and display sorted totals for each pr
657 special .execname modifier to display the ex
658 processes in the table rather than raw pids.
659 keeps a per-process sum of total bytes read:
660
661 # echo 'hist:key=common_pid.execname:val=c
662 /sys/kernel/tracing/events/syscalls
663
664 # cat /sys/kernel/tracing/events/syscalls/
665 # trigger info: hist:keys=common_pid.execn
666
667 { common_pid: gnome-terminal [ 3196]
668 { common_pid: Xorg [ 1309]
669 { common_pid: compiz [ 2889]
670 { common_pid: bash [ 8710]
671 { common_pid: dbus-daemon-lau [ 8703]
672 { common_pid: irqbalance [ 1252]
673 { common_pid: 01ifupdown [ 8705]
674 { common_pid: dbus-daemon [ 772]
675 { common_pid: Socket Thread [ 8342]
676 { common_pid: nm-dhcp-client. [ 8701]
677 { common_pid: gmain [ 1315]
678 .
679 .
680 .
681 { common_pid: postgres [ 1892]
682 { common_pid: postgres [ 1891]
683 { common_pid: gmain [ 8704]
684 { common_pid: upstart-dbus-br [ 2740]
685 { common_pid: nm-dispatcher.a [ 8696]
686 { common_pid: indicator-datet [ 2904]
687 { common_pid: gdbus [ 2998]
688 { common_pid: rtkit-daemon [ 2052]
689 { common_pid: init [ 1]
690
691 Totals:
692 Hits: 2116
693 Entries: 51
694 Dropped: 0
695
696 Similarly, if you key a hist trigger on sysc
697 gather and display a list of systemwide sysc
698 the special .syscall modifier to display the
699 than raw ids. The example below keeps a run
700 counts for the system during the run::
701
702 # echo 'hist:key=id.syscall:val=hitcount'
703 /sys/kernel/tracing/events/raw_sysc
704
705 # cat /sys/kernel/tracing/events/raw_sysca
706 # trigger info: hist:keys=id.syscall:vals=
707
708 { id: sys_fsync [ 74]
709 { id: sys_newuname [ 63]
710 { id: sys_prctl [157]
711 { id: sys_statfs [137]
712 { id: sys_symlink [ 88]
713 { id: sys_sendmmsg [307]
714 { id: sys_semctl [ 66]
715 { id: sys_readlink [ 89]
716 { id: sys_bind [ 49]
717 { id: sys_getsockname [ 51]
718 { id: sys_unlink [ 87]
719 { id: sys_rename [ 82]
720 { id: unknown_syscall [ 58]
721 { id: sys_connect [ 42]
722 { id: sys_getpid [ 39]
723 .
724 .
725 .
726 { id: sys_rt_sigprocmask [ 14]
727 { id: sys_futex [202]
728 { id: sys_write [ 1]
729 { id: sys_setitimer [ 38]
730 { id: sys_read [ 0]
731 { id: sys_select [ 23]
732 { id: sys_writev [ 20]
733 { id: sys_poll [ 7]
734 { id: sys_recvmsg [ 47]
735 { id: sys_ioctl [ 16]
736
737 Totals:
738 Hits: 67612
739 Entries: 72
740 Dropped: 0
741
742 The syscall counts above provide a rough ove
743 call activity on the system; we can see for
744 popular system call on this system was the '
745
746 We can use 'compound' keys to refine that nu
747 further insight as to which processes exactl
748 overall ioctl count.
749
750 The command below keeps a hitcount for every
751 system call id and pid - the end result is e
752 that keeps a per-pid sum of system call hits
753 sorted using the system call id as the prima
754 hitcount sum as the secondary key::
755
756 # echo 'hist:key=id.syscall,common_pid.exe
757 /sys/kernel/tracing/events/raw_sysc
758
759 # cat /sys/kernel/tracing/events/raw_sysca
760 # trigger info: hist:keys=id.syscall,commo
761
762 { id: sys_read [ 0],
763 { id: sys_read [ 0],
764 { id: sys_read [ 0],
765 { id: sys_read [ 0],
766 { id: sys_read [ 0],
767 { id: sys_read [ 0],
768 { id: sys_read [ 0],
769 { id: sys_read [ 0],
770 { id: sys_read [ 0],
771 { id: sys_read [ 0],
772 .
773 .
774 .
775 { id: sys_ioctl [ 16],
776 { id: sys_ioctl [ 16],
777 { id: sys_ioctl [ 16],
778 { id: sys_ioctl [ 16],
779 { id: sys_ioctl [ 16],
780 .
781 .
782 .
783 { id: sys_waitid [247],
784 { id: sys_waitid [247],
785 { id: sys_inotify_add_watch [254],
786 { id: sys_inotify_add_watch [254],
787 { id: sys_inotify_add_watch [254],
788 { id: sys_inotify_add_watch [254],
789 { id: sys_inotify_add_watch [254],
790 { id: sys_inotify_add_watch [254],
791 { id: sys_openat [257],
792 { id: sys_eventfd2 [290],
793 { id: sys_eventfd2 [290],
794
795 Totals:
796 Hits: 31536
797 Entries: 323
798 Dropped: 0
799
800 The above list does give us a breakdown of t
801 pid, but it also gives us quite a bit more t
802 don't really care about at the moment. Sinc
803 id for sys_ioctl (16, displayed next to the
804 can use that to filter out all the other sys
805
806 # echo 'hist:key=id.syscall,common_pid.exe
807 /sys/kernel/tracing/events/raw_sysc
808
809 # cat /sys/kernel/tracing/events/raw_sysca
810 # trigger info: hist:keys=id.syscall,commo
811
812 { id: sys_ioctl [ 16],
813 { id: sys_ioctl [ 16],
814 { id: sys_ioctl [ 16],
815 { id: sys_ioctl [ 16],
816 { id: sys_ioctl [ 16],
817 { id: sys_ioctl [ 16],
818 { id: sys_ioctl [ 16],
819 { id: sys_ioctl [ 16],
820 { id: sys_ioctl [ 16],
821 .
822 .
823 .
824 { id: sys_ioctl [ 16],
825 { id: sys_ioctl [ 16],
826 { id: sys_ioctl [ 16],
827 { id: sys_ioctl [ 16],
828 { id: sys_ioctl [ 16],
829 { id: sys_ioctl [ 16],
830
831 Totals:
832 Hits: 101162
833 Entries: 103
834 Dropped: 0
835
836 The above output shows that 'compiz' and 'Xo
837 the heaviest ioctl callers (which might lead
838 whether they really need to be making all th
839 possible avenues for further investigation.)
840
841 The compound key examples used a key and a s
842 sort the output, but we can just as easily u
843 Here's an example where we use a compound ke
844 common_pid and size event fields. Sorting w
845 key and 'size' as the secondary key allows u
846 ordered summary of the recvfrom sizes, with
847 each process::
848
849 # echo 'hist:key=common_pid.execname,size:
850 /sys/kernel/tracing/events/syscalls
851
852 # cat /sys/kernel/tracing/events/syscalls/
853 # trigger info: hist:keys=common_pid.execn
854
855 { common_pid: smbd [ 784]
856 { common_pid: dnsmasq [ 1412]
857 { common_pid: postgres [ 1796]
858 { common_pid: postgres [ 1867]
859 { common_pid: bamfdaemon [ 2787]
860 { common_pid: bamfdaemon [ 2787]
861 { common_pid: compiz [ 2994]
862 { common_pid: compiz [ 2994]
863 { common_pid: gnome-terminal [ 3199]
864 { common_pid: firefox [ 8817]
865 { common_pid: firefox [ 8817]
866 { common_pid: firefox [ 8817]
867 { common_pid: firefox [ 8817]
868 { common_pid: firefox [ 8817]
869 { common_pid: firefox [ 8817]
870 { common_pid: firefox [ 8822]
871 { common_pid: firefox [ 8822]
872 { common_pid: firefox [ 8822]
873 { common_pid: firefox [ 8822]
874 .
875 .
876 .
877 { common_pid: pool [ 8923]
878 { common_pid: pool [ 8923]
879 { common_pid: pool [ 8924]
880 { common_pid: pool [ 8924]
881 { common_pid: pool [ 8928]
882 { common_pid: pool [ 8928]
883 { common_pid: pool [ 8928]
884 { common_pid: pool [ 8929]
885 { common_pid: pool [ 8929]
886
887 Totals:
888 Hits: 2016
889 Entries: 224
890 Dropped: 0
891
892 The above example also illustrates the fact
893 key is treated as a single entity for hashin
894 it's composed of can be accessed independent
895
896 The next example uses a string field as the
897 demonstrates how you can manually pause and
898 In this example, we'll aggregate fork counts
899 large number of entries in the hash table, s
900 much smaller number, say 256::
901
902 # echo 'hist:key=child_comm:val=hitcount:s
903 /sys/kernel/tracing/events/sched/sc
904
905 # cat /sys/kernel/tracing/events/sched/sch
906 # trigger info: hist:keys=child_comm:vals=
907
908 { child_comm: dconf worker
909 { child_comm: ibus-daemon
910 { child_comm: whoopsie
911 { child_comm: smbd
912 { child_comm: gdbus
913 { child_comm: kthreadd
914 { child_comm: dconf worker
915 { child_comm: evolution-alarm
916 { child_comm: Socket Thread
917 { child_comm: postgres
918 { child_comm: bash
919 { child_comm: compiz
920 { child_comm: evolution-sourc
921 { child_comm: dhclient
922 { child_comm: pool
923 { child_comm: nm-dispatcher.a
924 { child_comm: firefox
925 { child_comm: dbus-daemon
926 { child_comm: glib-pacrunner
927 { child_comm: evolution
928
929 Totals:
930 Hits: 89
931 Entries: 20
932 Dropped: 0
933
934 If we want to pause the hist trigger, we can
935 the command that started the trigger. Notic
936 displays as [paused]::
937
938 # echo 'hist:key=child_comm:val=hitcount:s
939 /sys/kernel/tracing/events/sched/sc
940
941 # cat /sys/kernel/tracing/events/sched/sch
942 # trigger info: hist:keys=child_comm:vals=
943
944 { child_comm: dconf worker
945 { child_comm: kthreadd
946 { child_comm: dconf worker
947 { child_comm: gdbus
948 { child_comm: ibus-daemon
949 { child_comm: Socket Thread
950 { child_comm: evolution-alarm
951 { child_comm: smbd
952 { child_comm: bash
953 { child_comm: whoopsie
954 { child_comm: compiz
955 { child_comm: evolution-sourc
956 { child_comm: pool
957 { child_comm: postgres
958 { child_comm: firefox
959 { child_comm: dhclient
960 { child_comm: emacs
961 { child_comm: dbus-daemon
962 { child_comm: nm-dispatcher.a
963 { child_comm: evolution
964 { child_comm: glib-pacrunner
965
966 Totals:
967 Hits: 199
968 Entries: 21
969 Dropped: 0
970
971 To manually continue having the trigger aggr
972 :cont instead. Notice that the trigger info
973 again, and the data has changed::
974
975 # echo 'hist:key=child_comm:val=hitcount:s
976 /sys/kernel/tracing/events/sched/sc
977
978 # cat /sys/kernel/tracing/events/sched/sch
979 # trigger info: hist:keys=child_comm:vals=
980
981 { child_comm: dconf worker
982 { child_comm: dconf worker
983 { child_comm: kthreadd
984 { child_comm: gdbus
985 { child_comm: ibus-daemon
986 { child_comm: Socket Thread
987 { child_comm: evolution-alarm
988 { child_comm: smbd
989 { child_comm: whoopsie
990 { child_comm: compiz
991 { child_comm: evolution-sourc
992 { child_comm: bash
993 { child_comm: pool
994 { child_comm: postgres
995 { child_comm: firefox
996 { child_comm: dhclient
997 { child_comm: emacs
998 { child_comm: dbus-daemon
999 { child_comm: nm-dispatcher.a
1000 { child_comm: evolution
1001 { child_comm: glib-pacrunner
1002
1003 Totals:
1004 Hits: 206
1005 Entries: 21
1006 Dropped: 0
1007
1008 The previous example showed how to start an
1009 appending 'pause' and 'continue' to the his
1010 hist trigger can also be started in a pause
1011 starting the trigger with ':pause' appended
1012 start the trigger only when you're ready to
1013 and not before. For example, you could sta
1014 paused state, then unpause it and do someth
1015 then pause the trigger again when done.
1016
1017 Of course, doing this manually can be diffi
1018 it is possible to automatically start and s
1019 on some condition, via the enable_hist and
1020
1021 For example, suppose we wanted to take a lo
1022 weights in terms of skb length for each cal
1023 netif_receive_skb event when downloading a
1024 wget.
1025
1026 First we set up an initially paused stacktr
1027 netif_receive_skb event::
1028
1029 # echo 'hist:key=common_stacktrace:vals=l
1030 /sys/kernel/tracing/events/net/net
1031
1032 Next, we set up an 'enable_hist' trigger on
1033 event, with an 'if filename==/usr/bin/wget'
1034 this new trigger is that it will 'unpause'
1035 set up on netif_receive_skb if and only if
1036 sched_process_exec event with a filename of
1037 that happens, all netif_receive_skb events
1038 hash table keyed on stacktrace::
1039
1040 # echo 'enable_hist:net:netif_receive_skb
1041 /sys/kernel/tracing/events/sched/s
1042
1043 The aggregation continues until the netif_r
1044 again, which is what the following disable_
1045 creating a similar setup on the sched_proce
1046 filter 'comm==wget'::
1047
1048 # echo 'disable_hist:net:netif_receive_sk
1049 /sys/kernel/tracing/events/sched/s
1050
1051 Whenever a process exits and the comm field
1052 trigger filter matches 'comm==wget', the ne
1053 trigger is disabled.
1054
1055 The overall effect is that netif_receive_sk
1056 into the hash table for only the duration o
1057 wget command and then listing the 'hist' fi
1058 output generated by the wget command::
1059
1060 $ wget https://www.kernel.org/pub/linux/k
1061
1062 # cat /sys/kernel/tracing/events/net/neti
1063 # trigger info: hist:keys=common_stacktra
1064
1065 { common_stacktrace:
1066 __netif_receive_skb_core+0x46d/0x990
1067 __netif_receive_skb+0x18/0x60
1068 netif_receive_skb_internal+0x23/0x90
1069 napi_gro_receive+0xc8/0x100
1070 ieee80211_deliver_skb+0xd6/0x270 [ma
1071 ieee80211_rx_handlers+0xccf/0x22f0 [
1072 ieee80211_prepare_and_rx_handle+0x4e
1073 ieee80211_rx+0x31d/0x900 [mac80211]
1074 iwlagn_rx_reply_rx+0x3db/0x6f0 [iwld
1075 iwl_rx_dispatch+0x8e/0xf0 [iwldvm]
1076 iwl_pcie_irq_handler+0xe3c/0x12f0 [i
1077 irq_thread_fn+0x20/0x50
1078 irq_thread+0x11f/0x150
1079 kthread+0xd2/0xf0
1080 ret_from_fork+0x42/0x70
1081 } hitcount: 85 len: 28884
1082 { common_stacktrace:
1083 __netif_receive_skb_core+0x46d/0x990
1084 __netif_receive_skb+0x18/0x60
1085 netif_receive_skb_internal+0x23/0x90
1086 napi_gro_complete+0xa4/0xe0
1087 dev_gro_receive+0x23a/0x360
1088 napi_gro_receive+0x30/0x100
1089 ieee80211_deliver_skb+0xd6/0x270 [ma
1090 ieee80211_rx_handlers+0xccf/0x22f0 [
1091 ieee80211_prepare_and_rx_handle+0x4e
1092 ieee80211_rx+0x31d/0x900 [mac80211]
1093 iwlagn_rx_reply_rx+0x3db/0x6f0 [iwld
1094 iwl_rx_dispatch+0x8e/0xf0 [iwldvm]
1095 iwl_pcie_irq_handler+0xe3c/0x12f0 [i
1096 irq_thread_fn+0x20/0x50
1097 irq_thread+0x11f/0x150
1098 kthread+0xd2/0xf0
1099 } hitcount: 98 len: 664329
1100 { common_stacktrace:
1101 __netif_receive_skb_core+0x46d/0x990
1102 __netif_receive_skb+0x18/0x60
1103 process_backlog+0xa8/0x150
1104 net_rx_action+0x15d/0x340
1105 __do_softirq+0x114/0x2c0
1106 do_softirq_own_stack+0x1c/0x30
1107 do_softirq+0x65/0x70
1108 __local_bh_enable_ip+0xb5/0xc0
1109 ip_finish_output+0x1f4/0x840
1110 ip_output+0x6b/0xc0
1111 ip_local_out_sk+0x31/0x40
1112 ip_send_skb+0x1a/0x50
1113 udp_send_skb+0x173/0x2a0
1114 udp_sendmsg+0x2bf/0x9f0
1115 inet_sendmsg+0x64/0xa0
1116 sock_sendmsg+0x3d/0x50
1117 } hitcount: 115 len: 13030
1118 { common_stacktrace:
1119 __netif_receive_skb_core+0x46d/0x990
1120 __netif_receive_skb+0x18/0x60
1121 netif_receive_skb_internal+0x23/0x90
1122 napi_gro_complete+0xa4/0xe0
1123 napi_gro_flush+0x6d/0x90
1124 iwl_pcie_irq_handler+0x92a/0x12f0 [i
1125 irq_thread_fn+0x20/0x50
1126 irq_thread+0x11f/0x150
1127 kthread+0xd2/0xf0
1128 ret_from_fork+0x42/0x70
1129 } hitcount: 934 len: 5512212
1130
1131 Totals:
1132 Hits: 1232
1133 Entries: 4
1134 Dropped: 0
1135
1136 The above shows all the netif_receive_skb c
1137 lengths for the duration of the wget comman
1138
1139 The 'clear' hist trigger param can be used
1140 Suppose we wanted to try another run of the
1141 this time also wanted to see the complete l
1142 into the histogram. In order to avoid havi
1143 again, we can just clear the histogram firs
1144
1145 # echo 'hist:key=common_stacktrace:vals=l
1146 /sys/kernel/tracing/events/net/net
1147
1148 Just to verify that it is in fact cleared,
1149 the hist file::
1150
1151 # cat /sys/kernel/tracing/events/net/neti
1152 # trigger info: hist:keys=common_stacktra
1153
1154 Totals:
1155 Hits: 0
1156 Entries: 0
1157 Dropped: 0
1158
1159 Since we want to see the detailed list of e
1160 event occurring during the new run, which a
1161 events being aggregated into the hash table
1162 'enable_event' events to the triggering sch
1163 sched_process_exit events as such::
1164
1165 # echo 'enable_event:net:netif_receive_sk
1166 /sys/kernel/tracing/events/sched/s
1167
1168 # echo 'disable_event:net:netif_receive_s
1169 /sys/kernel/tracing/events/sched/s
1170
1171 If you read the trigger files for the sched
1172 sched_process_exit triggers, you should see
1173 one enabling/disabling the hist aggregation
1174 enabling/disabling the logging of events::
1175
1176 # cat /sys/kernel/tracing/events/sched/sc
1177 enable_event:net:netif_receive_skb:unlimi
1178 enable_hist:net:netif_receive_skb:unlimit
1179
1180 # cat /sys/kernel/tracing/events/sched/sc
1181 enable_event:net:netif_receive_skb:unlimi
1182 disable_hist:net:netif_receive_skb:unlimi
1183
1184 In other words, whenever either of the sche
1185 sched_process_exit events is hit and matche
1186 disables both the histogram and the event l
1187 with is a hash table and set of events just
1188 duration. Run the wget command again::
1189
1190 $ wget https://www.kernel.org/pub/linux/k
1191
1192 Displaying the 'hist' file should show some
1193 saw in the last run, but this time you shou
1194 individual events in the trace file::
1195
1196 # cat /sys/kernel/tracing/trace
1197
1198 # tracer: nop
1199 #
1200 # entries-in-buffer/entries-written: 183/
1201 #
1202 # _-----=> i
1203 # / _----=> n
1204 # | / _---=> h
1205 # || / _--=> p
1206 # ||| / de
1207 # TASK-PID CPU# |||| TIME
1208 # | | | |||| |
1209 wget-15108 [000] ..s1 31769.6
1210 wget-15108 [000] ..s1 31769.6
1211 dnsmasq-1382 [000] ..s1 31769.6
1212 dnsmasq-1382 [000] ..s1 31769.6
1213 ##### CPU 2 buffer started ####
1214 irq/29-iwlwifi-559 [002] ..s. 31772.0
1215 irq/29-iwlwifi-559 [002] ..s. 31772.0
1216 irq/29-iwlwifi-559 [002] ..s. 31772.0
1217 irq/29-iwlwifi-559 [002] ..s. 31772.0
1218 irq/29-iwlwifi-559 [002] ..s. 31772.0
1219 .
1220 .
1221 .
1222
1223 The following example demonstrates how mult
1224 attached to a given event. This capability
1225 creating a set of different summaries deriv
1226 events, or for comparing the effects of dif
1227 other things::
1228
1229 # echo 'hist:keys=skbaddr.hex:vals=len if
1230 /sys/kernel/tracing/events/net/net
1231 # echo 'hist:keys=skbaddr.hex:vals=len if
1232 /sys/kernel/tracing/events/net/net
1233 # echo 'hist:keys=skbaddr.hex:vals=len if
1234 /sys/kernel/tracing/events/net/net
1235 # echo 'hist:keys=skbaddr.hex:vals=len' >
1236 /sys/kernel/tracing/events/net/net
1237 # echo 'hist:keys=len:vals=common_preempt
1238 /sys/kernel/tracing/events/net/net
1239
1240 The above set of commands create four trigg
1241 their filters, along with a completely diff
1242 nonsensical trigger. Note that in order to
1243 triggers to the same file, you should use t
1244 append them ('>' will also add the new hist
1245 any existing hist triggers beforehand).
1246
1247 Displaying the contents of the 'hist' file
1248 contents of all five histograms::
1249
1250 # cat /sys/kernel/tracing/events/net/neti
1251
1252 # event histogram
1253 #
1254 # trigger info: hist:keys=len:vals=hitcou
1255 #
1256
1257 { len: 176 } hitcount: 1
1258 { len: 223 } hitcount: 1
1259 { len: 4854 } hitcount: 1
1260 { len: 395 } hitcount: 1
1261 { len: 177 } hitcount: 1
1262 { len: 446 } hitcount: 1
1263 { len: 1601 } hitcount: 1
1264 .
1265 .
1266 .
1267 { len: 1280 } hitcount: 66
1268 { len: 116 } hitcount: 81
1269 { len: 708 } hitcount: 112
1270 { len: 46 } hitcount: 221
1271 { len: 1264 } hitcount: 458
1272
1273 Totals:
1274 Hits: 1428
1275 Entries: 147
1276 Dropped: 0
1277
1278
1279 # event histogram
1280 #
1281 # trigger info: hist:keys=skbaddr.hex:val
1282 #
1283
1284 { skbaddr: ffff8800baee5e00 } hitcount:
1285 { skbaddr: ffff88005f3d5600 } hitcount:
1286 { skbaddr: ffff88005f3d4900 } hitcount:
1287 { skbaddr: ffff88009fed6300 } hitcount:
1288 { skbaddr: ffff88009fe0ad00 } hitcount:
1289 { skbaddr: ffff88008cdb1900 } hitcount:
1290 { skbaddr: ffff880064b5ef00 } hitcount:
1291 { skbaddr: ffff880044e3c700 } hitcount:
1292 { skbaddr: ffff880100065900 } hitcount:
1293 { skbaddr: ffff8800d46bd500 } hitcount:
1294 { skbaddr: ffff88005f3d5f00 } hitcount:
1295 { skbaddr: ffff880100064700 } hitcount:
1296 { skbaddr: ffff8800badb6f00 } hitcount:
1297 .
1298 .
1299 .
1300 { skbaddr: ffff88009fe0be00 } hitcount:
1301 { skbaddr: ffff88009fe0a400 } hitcount:
1302 { skbaddr: ffff88009fe0b700 } hitcount:
1303 { skbaddr: ffff88009fe0b600 } hitcount:
1304 { skbaddr: ffff88006a462800 } hitcount:
1305 { skbaddr: ffff88006a463700 } hitcount:
1306 { skbaddr: ffff88006a462b00 } hitcount:
1307 { skbaddr: ffff88006a463600 } hitcount:
1308 { skbaddr: ffff88006a462200 } hitcount:
1309
1310 Totals:
1311 Hits: 1451
1312 Entries: 318
1313 Dropped: 0
1314
1315
1316 # event histogram
1317 #
1318 # trigger info: hist:keys=skbaddr.hex:val
1319 #
1320
1321
1322 Totals:
1323 Hits: 0
1324 Entries: 0
1325 Dropped: 0
1326
1327
1328 # event histogram
1329 #
1330 # trigger info: hist:keys=skbaddr.hex:val
1331 #
1332
1333 { skbaddr: ffff88009fd2c300 } hitcount:
1334 { skbaddr: ffff8800d2bcce00 } hitcount:
1335 { skbaddr: ffff8800d2bcd700 } hitcount:
1336 { skbaddr: ffff8800d2bcda00 } hitcount:
1337 { skbaddr: ffff8800ae2e2d00 } hitcount:
1338 { skbaddr: ffff8800d2bcdb00 } hitcount:
1339 { skbaddr: ffff88006a4df500 } hitcount:
1340 { skbaddr: ffff88008ce47b00 } hitcount:
1341 { skbaddr: ffff8800ae2e2200 } hitcount:
1342 { skbaddr: ffff88005f3e1000 } hitcount:
1343 { skbaddr: ffff8800d2bcdc00 } hitcount:
1344 { skbaddr: ffff8800d2bcc200 } hitcount:
1345
1346 Totals:
1347 Hits: 14
1348 Entries: 12
1349 Dropped: 0
1350
1351
1352 # event histogram
1353 #
1354 # trigger info: hist:keys=skbaddr.hex:val
1355 #
1356
1357
1358 Totals:
1359 Hits: 0
1360 Entries: 0
1361 Dropped: 0
1362
1363 Named triggers can be used to have triggers
1364 histogram data. This capability is mostly
1365 output of events generated by tracepoints c
1366 functions, but names can be used in a hist
1367 For example, these two triggers when hit wi
1368 field in the shared 'foo' histogram data::
1369
1370 # echo 'hist:name=foo:keys=skbaddr.hex:va
1371 /sys/kernel/tracing/events/net/net
1372 # echo 'hist:name=foo:keys=skbaddr.hex:va
1373 /sys/kernel/tracing/events/net/net
1374
1375 You can see that they're updating common hi
1376 each event's hist files at the same time::
1377
1378 # cat /sys/kernel/tracing/events/net/neti
1379 cat /sys/kernel/tracing/events/net/neti
1380
1381 # event histogram
1382 #
1383 # trigger info: hist:name=foo:keys=skbadd
1384 #
1385
1386 { skbaddr: ffff88000ad53500 } hitcount:
1387 { skbaddr: ffff8800af5a1500 } hitcount:
1388 { skbaddr: ffff8800d62a1900 } hitcount:
1389 { skbaddr: ffff8800d2bccb00 } hitcount:
1390 { skbaddr: ffff8800d3c69900 } hitcount:
1391 { skbaddr: ffff88009ff09100 } hitcount:
1392 { skbaddr: ffff88010f13ab00 } hitcount:
1393 { skbaddr: ffff88006a54f400 } hitcount:
1394 { skbaddr: ffff8800d2bcc500 } hitcount:
1395 { skbaddr: ffff880064505000 } hitcount:
1396 { skbaddr: ffff8800baf24e00 } hitcount:
1397 { skbaddr: ffff88009fe0ad00 } hitcount:
1398 { skbaddr: ffff8800d3edff00 } hitcount:
1399 { skbaddr: ffff88009fe0b400 } hitcount:
1400 { skbaddr: ffff8800a1c55a00 } hitcount:
1401 { skbaddr: ffff8800d2bcd100 } hitcount:
1402 { skbaddr: ffff880064505f00 } hitcount:
1403 { skbaddr: ffff8800a8bff200 } hitcount:
1404 { skbaddr: ffff880044e3cc00 } hitcount:
1405 { skbaddr: ffff8800a8bfe700 } hitcount:
1406 { skbaddr: ffff8800d2bcdc00 } hitcount:
1407 { skbaddr: ffff8800a1f64800 } hitcount:
1408 { skbaddr: ffff8800d2bcde00 } hitcount:
1409 { skbaddr: ffff88006a5dea00 } hitcount:
1410 { skbaddr: ffff88002e37a200 } hitcount:
1411 { skbaddr: ffff8800a1f32c00 } hitcount:
1412 { skbaddr: ffff88000ad52600 } hitcount:
1413 { skbaddr: ffff8800a1f91e00 } hitcount:
1414 { skbaddr: ffff8800af5a0200 } hitcount:
1415 { skbaddr: ffff8800d2bcc600 } hitcount:
1416 { skbaddr: ffff8800ba36f500 } hitcount:
1417 { skbaddr: ffff8800d021f800 } hitcount:
1418 { skbaddr: ffff8800a1f33600 } hitcount:
1419 { skbaddr: ffff8800a8bfff00 } hitcount:
1420 { skbaddr: ffff8800d62a1300 } hitcount:
1421 { skbaddr: ffff88002e37a100 } hitcount:
1422 { skbaddr: ffff880064504400 } hitcount:
1423 { skbaddr: ffff8800a8bfec00 } hitcount:
1424 { skbaddr: ffff88000ad53700 } hitcount:
1425 { skbaddr: ffff8800d2bcdb00 } hitcount:
1426 { skbaddr: ffff8800a1f90000 } hitcount:
1427 { skbaddr: ffff88006a54f900 } hitcount:
1428
1429 Totals:
1430 Hits: 81
1431 Entries: 42
1432 Dropped: 0
1433 # event histogram
1434 #
1435 # trigger info: hist:name=foo:keys=skbadd
1436 #
1437
1438 { skbaddr: ffff88000ad53500 } hitcount:
1439 { skbaddr: ffff8800af5a1500 } hitcount:
1440 { skbaddr: ffff8800d62a1900 } hitcount:
1441 { skbaddr: ffff8800d2bccb00 } hitcount:
1442 { skbaddr: ffff8800d3c69900 } hitcount:
1443 { skbaddr: ffff88009ff09100 } hitcount:
1444 { skbaddr: ffff88010f13ab00 } hitcount:
1445 { skbaddr: ffff88006a54f400 } hitcount:
1446 { skbaddr: ffff8800d2bcc500 } hitcount:
1447 { skbaddr: ffff880064505000 } hitcount:
1448 { skbaddr: ffff8800baf24e00 } hitcount:
1449 { skbaddr: ffff88009fe0ad00 } hitcount:
1450 { skbaddr: ffff8800d3edff00 } hitcount:
1451 { skbaddr: ffff88009fe0b400 } hitcount:
1452 { skbaddr: ffff8800a1c55a00 } hitcount:
1453 { skbaddr: ffff8800d2bcd100 } hitcount:
1454 { skbaddr: ffff880064505f00 } hitcount:
1455 { skbaddr: ffff8800a8bff200 } hitcount:
1456 { skbaddr: ffff880044e3cc00 } hitcount:
1457 { skbaddr: ffff8800a8bfe700 } hitcount:
1458 { skbaddr: ffff8800d2bcdc00 } hitcount:
1459 { skbaddr: ffff8800a1f64800 } hitcount:
1460 { skbaddr: ffff8800d2bcde00 } hitcount:
1461 { skbaddr: ffff88006a5dea00 } hitcount:
1462 { skbaddr: ffff88002e37a200 } hitcount:
1463 { skbaddr: ffff8800a1f32c00 } hitcount:
1464 { skbaddr: ffff88000ad52600 } hitcount:
1465 { skbaddr: ffff8800a1f91e00 } hitcount:
1466 { skbaddr: ffff8800af5a0200 } hitcount:
1467 { skbaddr: ffff8800d2bcc600 } hitcount:
1468 { skbaddr: ffff8800ba36f500 } hitcount:
1469 { skbaddr: ffff8800d021f800 } hitcount:
1470 { skbaddr: ffff8800a1f33600 } hitcount:
1471 { skbaddr: ffff8800a8bfff00 } hitcount:
1472 { skbaddr: ffff8800d62a1300 } hitcount:
1473 { skbaddr: ffff88002e37a100 } hitcount:
1474 { skbaddr: ffff880064504400 } hitcount:
1475 { skbaddr: ffff8800a8bfec00 } hitcount:
1476 { skbaddr: ffff88000ad53700 } hitcount:
1477 { skbaddr: ffff8800d2bcdb00 } hitcount:
1478 { skbaddr: ffff8800a1f90000 } hitcount:
1479 { skbaddr: ffff88006a54f900 } hitcount:
1480
1481 Totals:
1482 Hits: 81
1483 Entries: 42
1484 Dropped: 0
1485
1486 And here's an example that shows how to com
1487 any two events even if they don't share any
1488 other than 'hitcount' and 'common_stacktrac
1489 couple of triggers named 'bar' using those
1490
1491 # echo 'hist:name=bar:key=common_stacktra
1492 /sys/kernel/tracing/events/sched/s
1493 # echo 'hist:name=bar:key=common_stacktra
1494 /sys/kernel/tracing/events/net/neti
1495
1496 And displaying the output of either shows s
1497 somewhat confusing output::
1498
1499 # cat /sys/kernel/tracing/events/sched/sc
1500 # cat /sys/kernel/tracing/events/net/neti
1501
1502 # event histogram
1503 #
1504 # trigger info: hist:name=bar:keys=common
1505 #
1506
1507 { common_stacktrace:
1508 kernel_clone+0x18e/0x330
1509 kernel_thread+0x29/0x30
1510 kthreadd+0x154/0x1b0
1511 ret_from_fork+0x3f/0x70
1512 } hitcount: 1
1513 { common_stacktrace:
1514 netif_rx_internal+0xb2/0xd0
1515 netif_rx_ni+0x20/0x70
1516 dev_loopback_xmit+0xaa/0xd0
1517 ip_mc_output+0x126/0x240
1518 ip_local_out_sk+0x31/0x40
1519 igmp_send_report+0x1e9/0x230
1520 igmp_timer_expire+0xe9/0x120
1521 call_timer_fn+0x39/0xf0
1522 run_timer_softirq+0x1e1/0x290
1523 __do_softirq+0xfd/0x290
1524 irq_exit+0x98/0xb0
1525 smp_apic_timer_interrupt+0x4a/0x
1526 apic_timer_interrupt+0x6d/0x80
1527 cpuidle_enter+0x17/0x20
1528 call_cpuidle+0x3b/0x60
1529 cpu_startup_entry+0x22d/0x310
1530 } hitcount: 1
1531 { common_stacktrace:
1532 netif_rx_internal+0xb2/0xd0
1533 netif_rx_ni+0x20/0x70
1534 dev_loopback_xmit+0xaa/0xd0
1535 ip_mc_output+0x17f/0x240
1536 ip_local_out_sk+0x31/0x40
1537 ip_send_skb+0x1a/0x50
1538 udp_send_skb+0x13e/0x270
1539 udp_sendmsg+0x2bf/0x980
1540 inet_sendmsg+0x67/0xa0
1541 sock_sendmsg+0x38/0x50
1542 SYSC_sendto+0xef/0x170
1543 SyS_sendto+0xe/0x10
1544 entry_SYSCALL_64_fastpath+0x12/0
1545 } hitcount: 2
1546 { common_stacktrace:
1547 netif_rx_internal+0xb2/0xd0
1548 netif_rx+0x1c/0x60
1549 loopback_xmit+0x6c/0xb0
1550 dev_hard_start_xmit+0x219/0x3a0
1551 __dev_queue_xmit+0x415/0x4f0
1552 dev_queue_xmit_sk+0x13/0x20
1553 ip_finish_output2+0x237/0x340
1554 ip_finish_output+0x113/0x1d0
1555 ip_output+0x66/0xc0
1556 ip_local_out_sk+0x31/0x40
1557 ip_send_skb+0x1a/0x50
1558 udp_send_skb+0x16d/0x270
1559 udp_sendmsg+0x2bf/0x980
1560 inet_sendmsg+0x67/0xa0
1561 sock_sendmsg+0x38/0x50
1562 ___sys_sendmsg+0x14e/0x270
1563 } hitcount: 76
1564 { common_stacktrace:
1565 netif_rx_internal+0xb2/0xd0
1566 netif_rx+0x1c/0x60
1567 loopback_xmit+0x6c/0xb0
1568 dev_hard_start_xmit+0x219/0x3a0
1569 __dev_queue_xmit+0x415/0x4f0
1570 dev_queue_xmit_sk+0x13/0x20
1571 ip_finish_output2+0x237/0x340
1572 ip_finish_output+0x113/0x1d0
1573 ip_output+0x66/0xc0
1574 ip_local_out_sk+0x31/0x40
1575 ip_send_skb+0x1a/0x50
1576 udp_send_skb+0x16d/0x270
1577 udp_sendmsg+0x2bf/0x980
1578 inet_sendmsg+0x67/0xa0
1579 sock_sendmsg+0x38/0x50
1580 ___sys_sendmsg+0x269/0x270
1581 } hitcount: 77
1582 { common_stacktrace:
1583 netif_rx_internal+0xb2/0xd0
1584 netif_rx+0x1c/0x60
1585 loopback_xmit+0x6c/0xb0
1586 dev_hard_start_xmit+0x219/0x3a0
1587 __dev_queue_xmit+0x415/0x4f0
1588 dev_queue_xmit_sk+0x13/0x20
1589 ip_finish_output2+0x237/0x340
1590 ip_finish_output+0x113/0x1d0
1591 ip_output+0x66/0xc0
1592 ip_local_out_sk+0x31/0x40
1593 ip_send_skb+0x1a/0x50
1594 udp_send_skb+0x16d/0x270
1595 udp_sendmsg+0x2bf/0x980
1596 inet_sendmsg+0x67/0xa0
1597 sock_sendmsg+0x38/0x50
1598 SYSC_sendto+0xef/0x170
1599 } hitcount: 88
1600 { common_stacktrace:
1601 kernel_clone+0x18e/0x330
1602 SyS_clone+0x19/0x20
1603 entry_SYSCALL_64_fastpath+0x12/0
1604 } hitcount: 244
1605
1606 Totals:
1607 Hits: 489
1608 Entries: 7
1609 Dropped: 0
1610
1611 2.2 Inter-event hist triggers
1612 -----------------------------
1613
1614 Inter-event hist triggers are hist triggers t
1615 one or more other events and create a histogr
1616 from an inter-event histogram can in turn bec
1617 further combined histograms, thus providing a
1618 histograms, which is important for some appli
1619
1620 The most important example of an inter-event
1621 in this manner is latency, which is simply a
1622 between two events. Although latency is the
1623 inter-event quantity, note that because the s
1624 general across the trace event subsystem, any
1625 in an inter-event quantity.
1626
1627 An example of a histogram that combines data
1628 into a useful chain would be a 'wakeupswitch
1629 combines a 'wakeup latency' histogram and a '
1630 histogram.
1631
1632 Normally, a hist trigger specification consis
1633 compound) key along with one or more numeric
1634 continually updated sums associated with that
1635 specification in this case consists of indivi
1636 specifications that refer to trace event fiel
1637 single event type.
1638
1639 The inter-event hist trigger extension allows
1640 events to be referenced and combined into a m
1641 specification. In support of this overall go
1642 features have been added to the hist trigger
1643
1644 - In order to compute an inter-event quanti
1645 event needs to saved and then referenced
1646 requires the introduction of support for
1647
1648 - The computation of inter-event quantities
1649 require some minimal amount of support fo
1650 expressions to variables (+ and -).
1651
1652 - A histogram consisting of inter-event qua
1653 histogram on either event (so having the
1654 event host the histogram output doesn't r
1655 address the idea that the histogram is as
1656 combination of events, support is added a
1657 'synthetic' events that are events derive
1658 These synthetic events are full-fledged e
1659 and can be used as such, as for instance
1660 'combination' histograms mentioned previo
1661
1662 - A set of 'actions' can be associated with
1663 these can be used to generate the previou
1664 events, but can also be used for other pu
1665 example saving context when a 'max' laten
1666
1667 - Trace events don't have a 'timestamp' ass
1668 there is an implicit timestamp saved alon
1669 underlying ftrace ring buffer. This time
1670 a synthetic field named 'common_timestamp
1671 histograms as if it were any other event
1672 field in the trace format but rather is a
1673 nonetheless can be used as if it were an
1674 it is in units of nanoseconds; appending
1675 common_timestamp field changes the units
1676
1677 A note on inter-event timestamps: If common_t
1678 histogram, the trace buffer is automatically
1679 absolute timestamps and the "global" trace cl
1680 bogus timestamp differences with other clocks
1681 across CPUs. This can be overridden by speci
1682 trace clocks instead, using the "clock=XXX" h
1683 where XXX is any of the clocks listed in the
1684 pseudo-file.
1685
1686 These features are described in more detail i
1687
1688 2.2.1 Histogram Variables
1689 -------------------------
1690
1691 Variables are simply named locations used for
1692 values between matching events. A 'matching'
1693 event that has a matching key - if a variable
1694 entry corresponding to that key, any subseque
1695 key can access that variable.
1696
1697 A variable's value is normally available to a
1698 it is set to something else by a subsequent e
1699 to that rule is that any variable used in an
1700 'read-once' - once it's used by an expression
1701 it's reset to its 'unset' state, which means
1702 unless it's set again. This ensures not only
1703 use an uninitialized variable in a calculatio
1704 is used only once and not for any unrelated s
1705
1706 The basic syntax for saving a variable is to
1707 variable name not corresponding to any keywor
1708 to any event field.
1709
1710 Either keys or values can be saved and retrie
1711 creates a variable named 'ts0' for a histogra
1712 'next_pid'::
1713
1714 # echo 'hist:keys=next_pid:vals=$ts0:ts0=co
1715 event/trigger
1716
1717 The ts0 variable can be accessed by any subse
1718 same pid as 'next_pid'.
1719
1720 Variable references are formed by prepending
1721 the '$' sign. Thus for example, the ts0 vari
1722 referenced as '$ts0' in expressions.
1723
1724 Because 'vals=' is used, the common_timestamp
1725 will also be summed as a normal histogram val
1726 timestamp it makes little sense).
1727
1728 The below shows that a key value can also be
1729
1730 # echo 'hist:timer_pid=common_pid:key=timer
1731
1732 If a variable isn't a key variable or prefixe
1733 associated event field will be saved in a var
1734 as a value::
1735
1736 # echo 'hist:keys=next_pid:ts1=common_times
1737
1738 Multiple variables can be assigned at the sam
1739 result in both ts0 and b being created as var
1740 common_timestamp and field1 additionally bein
1741
1742 # echo 'hist:keys=pid:vals=$ts0,$b:ts0=comm
1743 event/trigger
1744
1745 Note that variable assignments can appear eit
1746 following their use. The command below behav
1747 command above::
1748
1749 # echo 'hist:keys=pid:ts0=common_timestamp,
1750 event/trigger
1751
1752 Any number of variables not bound to a 'vals=
1753 assigned by simply separating them with colon
1754 thing but without the values being summed in
1755
1756 # echo 'hist:keys=pid:ts0=common_timestamp:
1757
1758 Variables set as above can be referenced and
1759 another event.
1760
1761 For example, here's how a latency can be calc
1762
1763 # echo 'hist:keys=pid,prio:ts0=common_times
1764 # echo 'hist:keys=next_pid:wakeup_lat=commo
1765
1766 In the first line above, the event's timestam
1767 variable ts0. In the next line, ts0 is subtr
1768 event's timestamp to produce the latency, whi
1769 yet another variable, 'wakeup_lat'. The hist
1770 makes use of the wakeup_lat variable to compu
1771 using the same key and variable from yet anot
1772
1773 # echo 'hist:key=pid:wakeupswitch_lat=$wake
1774
1775 Expressions support the use of addition, subt
1776 division operators (+-\*/).
1777
1778 Note if division by zero cannot be detected a
1779 divisor is not a constant), the result will b
1780
1781 Numeric constants can also be used directly i
1782
1783 # echo 'hist:keys=next_pid:timestamp_secs=c
1784
1785 or assigned to a variable and referenced in a
1786
1787 # echo 'hist:keys=next_pid:us_per_sec=10000
1788 # echo 'hist:keys=next_pid:timestamp_secs=c
1789
1790 Variables can even hold stacktraces, which ar
1791
1792 2.2.2 Synthetic Events
1793 ----------------------
1794
1795 Synthetic events are user-defined events gene
1796 variables or fields associated with one or mo
1797 purpose is to provide a mechanism for display
1798 multiple events consistent with the existing
1799 usage for normal events.
1800
1801 To define a synthetic event, the user writes
1802 consisting of the name of the new event along
1803 variables and their types, which can be any v
1804 separated by semicolons, to the tracing/synth
1805
1806 See synth_field_size() for available types.
1807
1808 If field_name contains [n], the field is cons
1809
1810 If field_names contains[] (no subscript), the
1811 be a dynamic array, which will only take as m
1812 is required to hold the array.
1813
1814 A string field can be specified using either
1815
1816 char name[32];
1817
1818 Or the dynamic:
1819
1820 char name[];
1821
1822 The size limit for either is 256.
1823
1824 For instance, the following creates a new eve
1825 with 3 fields: lat, pid, and prio. Each of t
1826 variable reference to a variable on another e
1827
1828 # echo 'wakeup_latency \
1829 u64 lat; \
1830 pid_t pid; \
1831 int prio' >> \
1832 /sys/kernel/tracing/synthetic_event
1833
1834 Reading the tracing/synthetic_events file lis
1835 defined synthetic events, in this case the ev
1836
1837 # cat /sys/kernel/tracing/synthetic_events
1838 wakeup_latency u64 lat; pid_t pid; int pr
1839
1840 An existing synthetic event definition can be
1841 the command that defined it with a '!'::
1842
1843 # echo '!wakeup_latency u64 lat pid_t pid i
1844 /sys/kernel/tracing/synthetic_events
1845
1846 At this point, there isn't yet an actual 'wak
1847 instantiated in the event subsystem - for thi
1848 trigger action' needs to be instantiated and
1849 and variables defined on other events (see Se
1850 how that is done using hist trigger 'onmatch'
1851 done, the 'wakeup_latency' synthetic event in
1852
1853 The new event is created under the tracing/ev
1854 and looks and behaves just like any other eve
1855
1856 # ls /sys/kernel/tracing/events/synthetic/w
1857 enable filter format hist id tri
1858
1859 A histogram can now be defined for the new sy
1860
1861 # echo 'hist:keys=pid,prio,lat.log2:sort=la
1862 /sys/kernel/tracing/events/synthetic/
1863
1864 The above shows the latency "lat" in a power
1865
1866 Like any other event, once a histogram is ena
1867 output can be displayed by reading the event'
1868
1869 # cat /sys/kernel/tracing/events/synthetic/
1870
1871 # event histogram
1872 #
1873 # trigger info: hist:keys=pid,prio,lat.log2
1874 #
1875
1876 { pid: 2035, prio: 9, lat: ~
1877 { pid: 2034, prio: 9, lat: ~
1878 { pid: 2029, prio: 9, lat: ~
1879 { pid: 2034, prio: 120, lat: ~
1880 { pid: 2033, prio: 120, lat: ~
1881 { pid: 2030, prio: 9, lat: ~
1882 { pid: 2030, prio: 120, lat: ~
1883 { pid: 2032, prio: 120, lat: ~
1884 { pid: 2035, prio: 120, lat: ~
1885 { pid: 2031, prio: 9, lat: ~
1886 { pid: 2028, prio: 120, lat: ~
1887 { pid: 2033, prio: 9, lat: ~
1888 { pid: 2032, prio: 9, lat: ~
1889 { pid: 2029, prio: 120, lat: ~
1890 { pid: 2031, prio: 120, lat: ~
1891 { pid: 2029, prio: 120, lat: ~
1892 { pid: 2035, prio: 9, lat: ~
1893 { pid: 2030, prio: 120, lat: ~
1894 { pid: 2032, prio: 9, lat: ~
1895 { pid: 2031, prio: 9, lat: ~
1896 { pid: 2034, prio: 9, lat: ~
1897 { pid: 2030, prio: 9, lat: ~
1898 { pid: 2033, prio: 9, lat: ~
1899 { pid: 2029, prio: 9, lat: ~
1900 { pid: 2028, prio: 120, lat: ~
1901 { pid: 2031, prio: 120, lat: ~
1902 { pid: 2028, prio: 120, lat: ~
1903 { pid: 2029, prio: 9, lat: ~
1904 { pid: 2031, prio: 120, lat: ~
1905 { pid: 2032, prio: 120, lat: ~
1906
1907 Totals:
1908 Hits: 2122
1909 Entries: 30
1910 Dropped: 0
1911
1912
1913 The latency values can also be grouped linear
1914 the ".buckets" modifier and specify a size (i
1915
1916 # echo 'hist:keys=pid,prio,lat.buckets=10:s
1917 /sys/kernel/tracing/events/synthetic/
1918
1919 # event histogram
1920 #
1921 # trigger info: hist:keys=pid,prio,lat.buck
1922 #
1923
1924 { pid: 2067, prio: 9, lat: ~
1925 { pid: 2068, prio: 9, lat: ~
1926 { pid: 2070, prio: 9, lat: ~
1927 { pid: 2067, prio: 120, lat: ~
1928 { pid: 2065, prio: 120, lat: ~
1929 { pid: 2066, prio: 120, lat: ~
1930 { pid: 2069, prio: 9, lat: ~
1931 { pid: 2069, prio: 120, lat: ~
1932 { pid: 2070, prio: 120, lat: ~
1933 { pid: 2068, prio: 120, lat: ~
1934 { pid: 2066, prio: 9, lat: ~
1935 { pid: 2064, prio: 120, lat: ~
1936 { pid: 2065, prio: 9, lat: ~
1937 { pid: 2071, prio: 9, lat: ~
1938 { pid: 2065, prio: 9, lat: ~
1939 { pid: 2064, prio: 120, lat: ~
1940
1941 Totals:
1942 Hits: 2112
1943 Entries: 16
1944 Dropped: 0
1945
1946 To save stacktraces, create a synthetic event
1947 or even just "long[]". For example, to see ho
1948 uninterruptible state::
1949
1950 # cd /sys/kernel/tracing
1951 # echo 's:block_lat pid_t pid; u64 delta; u
1952 # echo 'hist:keys=next_pid:ts=common_timest
1953 # echo 'hist:keys=prev_pid:delta=common_tim
1954 # echo 1 > events/synthetic/block_lat/enabl
1955 # cat trace
1956
1957 # tracer: nop
1958 #
1959 # entries-in-buffer/entries-written: 2/2
1960 #
1961 # _-----=> i
1962 # / _----=> n
1963 # | / _---=> h
1964 # || / _--=> p
1965 # ||| / _-=> m
1966 # |||| / d
1967 # TASK-PID CPU# ||||| TIMES
1968 # | | | ||||| |
1969 <idle>-0 [005] d..4. 521.
1970 => __schedule+0x448/0x7b0
1971 => schedule+0x5a/0xb0
1972 => io_schedule+0x42/0x70
1973 => bit_wait_io+0xd/0x60
1974 => __wait_on_bit+0x4b/0x140
1975 => out_of_line_wait_on_bit+0x91/0xb0
1976 => jbd2_journal_commit_transaction+0x1679/0
1977 => kjournald2+0xa9/0x280
1978 => kthread+0xe9/0x110
1979 => ret_from_fork+0x2c/0x50
1980
1981 <...>-2 [004] d..4. 525.
1982 => __schedule+0x448/0x7b0
1983 => schedule+0x5a/0xb0
1984 => schedule_timeout+0x11a/0x150
1985 => wait_for_completion_killable+0x144/0x1f0
1986 => __kthread_create_on_node+0xe7/0x1e0
1987 => kthread_create_on_node+0x51/0x70
1988 => create_worker+0xcc/0x1a0
1989 => worker_thread+0x2ad/0x380
1990 => kthread+0xe9/0x110
1991 => ret_from_fork+0x2c/0x50
1992
1993 A synthetic event that has a stacktrace field
1994 histogram::
1995
1996 # echo 'hist:keys=delta.buckets=100,stack.s
1997 # cat events/synthetic/block_lat/hist
1998
1999 # event histogram
2000 #
2001 # trigger info: hist:keys=delta.buckets=100
2002 #
2003 { delta: ~ 0-99, stack.stacktrace _
2004 schedule+0x6b/0x110
2005 io_schedule+0x46/0x80
2006 bit_wait_io+0x11/0x80
2007 __wait_on_bit+0x4e/0x120
2008 out_of_line_wait_on_bit+0x8d/0xb0
2009 __wait_on_buffer+0x33/0x40
2010 jbd2_journal_commit_transaction+0x15
2011 kjournald2+0xab/0x270
2012 kthread+0xfa/0x130
2013 ret_from_fork+0x29/0x50
2014 } hitcount: 1
2015 { delta: ~ 0-99, stack.stacktrace _
2016 schedule+0x6b/0x110
2017 io_schedule+0x46/0x80
2018 rq_qos_wait+0xd0/0x170
2019 wbt_wait+0x9e/0xf0
2020 __rq_qos_throttle+0x25/0x40
2021 blk_mq_submit_bio+0x2c3/0x5b0
2022 __submit_bio+0xff/0x190
2023 submit_bio_noacct_nocheck+0x25b/0x2b
2024 submit_bio_noacct+0x20b/0x600
2025 submit_bio+0x28/0x90
2026 ext4_bio_write_page+0x1e0/0x8c0
2027 mpage_submit_page+0x60/0x80
2028 mpage_process_page_bufs+0x16c/0x180
2029 mpage_prepare_extent_to_map+0x23f/0x
2030 } hitcount: 1
2031 { delta: ~ 0-99, stack.stacktrace _
2032 schedule+0x6b/0x110
2033 schedule_hrtimeout_range_clock+0x97/
2034 schedule_hrtimeout_range+0x13/0x20
2035 usleep_range_state+0x65/0x90
2036 __intel_wait_for_register+0x1c1/0x23
2037 intel_psr_wait_for_idle_locked+0x171
2038 intel_pipe_update_start+0x169/0x360
2039 intel_update_crtc+0x112/0x490 [i915]
2040 skl_commit_modeset_enables+0x199/0x6
2041 intel_atomic_commit_tail+0x7c4/0x108
2042 intel_atomic_commit_work+0x12/0x20 [
2043 process_one_work+0x21c/0x3f0
2044 worker_thread+0x50/0x3e0
2045 kthread+0xfa/0x130
2046 } hitcount: 3
2047 { delta: ~ 0-99, stack.stacktrace _
2048 schedule+0x6b/0x110
2049 schedule_timeout+0x11e/0x160
2050 __wait_for_common+0x8f/0x190
2051 wait_for_completion+0x24/0x30
2052 __flush_work.isra.0+0x1cc/0x360
2053 flush_work+0xe/0x20
2054 drm_mode_rmfb+0x18b/0x1d0 [drm]
2055 drm_mode_rmfb_ioctl+0x10/0x20 [drm]
2056 drm_ioctl_kernel+0xb8/0x150 [drm]
2057 drm_ioctl+0x243/0x560 [drm]
2058 __x64_sys_ioctl+0x92/0xd0
2059 do_syscall_64+0x59/0x90
2060 entry_SYSCALL_64_after_hwframe+0x72/
2061 } hitcount: 1
2062 { delta: ~ 0-99, stack.stacktrace _
2063 schedule+0x6b/0x110
2064 schedule_timeout+0x87/0x160
2065 __wait_for_common+0x8f/0x190
2066 wait_for_completion_timeout+0x1d/0x3
2067 drm_atomic_helper_wait_for_flip_done
2068 intel_atomic_commit_tail+0x8ce/0x108
2069 intel_atomic_commit_work+0x12/0x20 [
2070 process_one_work+0x21c/0x3f0
2071 worker_thread+0x50/0x3e0
2072 kthread+0xfa/0x130
2073 ret_from_fork+0x29/0x50
2074 } hitcount: 1
2075 { delta: ~ 100-199, stack.stacktrace
2076 schedule+0x6b/0x110
2077 schedule_hrtimeout_range_clock+0x97/
2078 schedule_hrtimeout_range+0x13/0x20
2079 usleep_range_state+0x65/0x90
2080 pci_set_low_power_state+0x17f/0x1f0
2081 pci_set_power_state+0x49/0x250
2082 pci_finish_runtime_suspend+0x4a/0x90
2083 pci_pm_runtime_suspend+0xcb/0x1b0
2084 __rpm_callback+0x48/0x120
2085 rpm_callback+0x67/0x70
2086 rpm_suspend+0x167/0x780
2087 rpm_idle+0x25a/0x380
2088 pm_runtime_work+0x93/0xc0
2089 process_one_work+0x21c/0x3f0
2090 } hitcount: 1
2091
2092 Totals:
2093 Hits: 10
2094 Entries: 7
2095 Dropped: 0
2096
2097 2.2.3 Hist trigger 'handlers' and 'actions'
2098 -------------------------------------------
2099
2100 A hist trigger 'action' is a function that's
2101 conditionally) whenever a histogram entry is
2102
2103 When a histogram entry is added or updated, a
2104 is what decides whether the corresponding act
2105 or not.
2106
2107 Hist trigger handlers and actions are paired
2108 form:
2109
2110 <handler>.<action>
2111
2112 To specify a handler.action pair for a given
2113 that handler.action pair between colons in th
2114 specification.
2115
2116 In theory, any handler can be combined with a
2117 practice, not every handler.action combinatio
2118 if a given handler.action combination isn't s
2119 trigger will fail with -EINVAL;
2120
2121 The default 'handler.action' if none is expli
2122 always has been, to simply update the set of
2123 entry. Some applications, however, may want
2124 actions at that point, such as generate anoth
2125 save a maximum.
2126
2127 The supported handlers and actions are listed
2128 described in more detail in the following par
2129 of descriptions of some common and useful han
2130
2131 The available handlers are:
2132
2133 - onmatch(matching.event) - invoke actio
2134 - onmax(var) - invoke actio
2135 - onchange(var) - invoke actio
2136
2137 The available actions are:
2138
2139 - trace(<synthetic_event_name>,param list)
2140 - save(field,...)
2141 - snapshot()
2142
2143 The following commonly-used handler.action pa
2144
2145 - onmatch(matching.event).trace(<synthetic_
2146
2147 The 'onmatch(matching.event).trace(<synth
2148 list)' hist trigger action is invoked whe
2149 and the histogram entry would be added or
2150 named synthetic event to be generated wit
2151 'param list'. The result is the generati
2152 that consists of the values contained in
2153 time the invoking event was hit. For exa
2154 event name is 'wakeup_latency', a wakeup_
2155 generated using onmatch(event).trace(wake
2156
2157 There is also an equivalent alternative f
2158 generating synthetic events. In this for
2159 name is used as if it were a function nam
2160 the 'wakeup_latency' synthetic event name
2161 wakeup_latency event would be generated b
2162 were a function call, with the event fiel
2163 arguments: onmatch(event).wakeup_latency(
2164 for this form is:
2165
2166 onmatch(matching.event).<synthetic_even
2167
2168 In either case, the 'param list' consists
2169 parameters which may be either variables
2170 either the 'matching.event' or the target
2171 fields specified in the param list may be
2172 or unqualified. If a variable is specifi
2173 must be unique between the two events. A
2174 param can be unqualified if it refers to
2175 must be fully qualified if it refers to t
2176 fully-qualified name is of the form 'syst
2177 or 'system.event_name.field'.
2178
2179 The 'matching.event' specification is sim
2180 event name of the event that matches the
2181 onmatch() functionality, in the form 'sys
2182 keys of both events are compared to find
2183 multiple histogram keys are used, they al
2184 order.
2185
2186 Finally, the number and type of variables
2187 list' must match the number and types of
2188 synthetic event being generated.
2189
2190 As an example the below defines a simple
2191 a variable defined on the sched_wakeup_ne
2192 when invoking the synthetic event. Here
2193 event::
2194
2195 # echo 'wakeup_new_test pid_t pid' >> \
2196 /sys/kernel/tracing/synthetic_ev
2197
2198 # cat /sys/kernel/tracing/synthetic_eve
2199 wakeup_new_test pid_t pid
2200
2201 The following hist trigger both defines t
2202 variable and specifies an onmatch() actio
2203 wakeup_new_test synthetic event whenever
2204 occurs, which because of the 'if comm ==
2205 happens when the executable is cyclictest
2206
2207 # echo 'hist:keys=$testpid:testpid=pid:
2208 wakeup_new_test($testpid) if co
2209 /sys/kernel/tracing/events/sche
2210
2211 Or, equivalently, using the 'trace' keywo
2212
2213 # echo 'hist:keys=$testpid:testpid=pid:
2214 trace(wakeup_new_test,$testpid)
2215 /sys/kernel/tracing/events/sche
2216
2217 Creating and displaying a histogram based
2218 just a matter of using the fields and new
2219 tracing/events/synthetic directory, as us
2220
2221 # echo 'hist:keys=pid:sort=pid' >> \
2222 /sys/kernel/tracing/events/synth
2223
2224 Running 'cyclictest' should cause wakeup_
2225 wakeup_new_test synthetic events which sh
2226 output in the wakeup_new_test event's his
2227
2228 # cat /sys/kernel/tracing/events/synthe
2229
2230 A more typical usage would be to use two
2231 latency. The following example uses a se
2232 produce a 'wakeup_latency' histogram.
2233
2234 First, we define a 'wakeup_latency' synth
2235
2236 # echo 'wakeup_latency u64 lat; pid_t p
2237 /sys/kernel/tracing/synthetic_e
2238
2239 Next, we specify that whenever we see a s
2240 cyclictest thread, save the timestamp in
2241
2242 # echo 'hist:keys=$saved_pid:saved_pid=
2243 if comm=="cyclictest"' >> \
2244 /sys/kernel/tracing/events/sche
2245
2246 Then, when the corresponding thread is ac
2247 CPU by a sched_switch event (saved_pid ma
2248 the latency and use that along with anoth
2249 to generate a wakeup_latency synthetic ev
2250
2251 # echo 'hist:keys=next_pid:wakeup_lat=c
2252 onmatch(sched.sched_waking).wak
2253 $saved_pid,next_prio) i
2254 /sys/kernel/tracing/events/sche
2255
2256 We also need to create a histogram on the
2257 event in order to aggregate the generated
2258
2259 # echo 'hist:keys=pid,prio,lat:sort=pid
2260 /sys/kernel/tracing/events/synt
2261
2262 Finally, once we've run cyclictest to act
2263 events, we can see the output by looking
2264 synthetic event's hist file::
2265
2266 # cat /sys/kernel/tracing/events/synthe
2267
2268 - onmax(var).save(field,.. .)
2269
2270 The 'onmax(var).save(field,...)' hist tri
2271 whenever the value of 'var' associated wi
2272 exceeds the current maximum contained in
2273
2274 The end result is that the trace event fi
2275 onmax.save() params will be saved if 'var
2276 maximum for that hist trigger entry. Thi
2277 event that exhibited the new maximum to b
2278 reference. When the histogram is display
2279 displaying the saved values will be print
2280
2281 As an example the below defines a couple
2282 sched_waking and another for sched_switch
2283 a sched_waking occurs, the timestamp is s
2284 corresponding to the current pid, and whe
2285 back to that pid, the timestamp differenc
2286 resulting latency, stored in wakeup_lat,
2287 maximum latency, the values specified in
2288 recorded::
2289
2290 # echo 'hist:keys=pid:ts0=common_timest
2291 if comm=="cyclictest"' >> \
2292 /sys/kernel/tracing/events/sche
2293
2294 # echo 'hist:keys=next_pid:\
2295 wakeup_lat=common_timestamp.use
2296 onmax($wakeup_lat).save(next_co
2297 if next_comm=="cyclictest"' >>
2298 /sys/kernel/tracing/events/sche
2299
2300 When the histogram is displayed, the max
2301 values corresponding to the max are displ
2302 of the fields::
2303
2304 # cat /sys/kernel/tracing/events/sched/
2305 { next_pid: 2255 } hitcount:
2306 common_timestamp-ts0: 0
2307 max: 27
2308 next_comm: cyclictest
2309 prev_pid: 0 prev_prio:
2310
2311 { next_pid: 2256 } hitcount:
2312 common_timestamp-ts0: 0
2313 max: 49 next_comm: cyclict
2314 prev_pid: 0 prev_prio:
2315
2316 Totals:
2317 Hits: 12970
2318 Entries: 2
2319 Dropped: 0
2320
2321 - onmax(var).snapshot()
2322
2323 The 'onmax(var).snapshot()' hist trigger
2324 whenever the value of 'var' associated wi
2325 exceeds the current maximum contained in
2326
2327 The end result is that a global snapshot
2328 be saved in the tracing/snapshot file if
2329 maximum for any hist trigger entry.
2330
2331 Note that in this case the maximum is a g
2332 current trace instance, which is the maxi
2333 the histogram. The key of the specific t
2334 the global maximum and the global maximum
2335 along with a message stating that a snaps
2336 where to find it. The user can use the k
2337 to locate the corresponding bucket in the
2338 detail.
2339
2340 As an example the below defines a couple
2341 sched_waking and another for sched_switch
2342 a sched_waking event occurs, the timestam
2343 corresponding to the current pid, and whe
2344 back to that pid, the timestamp differenc
2345 resulting latency, stored in wakeup_lat,
2346 maximum latency, a snapshot is taken. As
2347 the scheduler events are also enabled, wh
2348 will show up in the snapshot when it is t
2349
2350 # echo 1 > /sys/kernel/tracing/events/s
2351
2352 # echo 'hist:keys=pid:ts0=common_timest
2353 if comm=="cyclictest"' >> \
2354 /sys/kernel/tracing/events/sche
2355
2356 # echo 'hist:keys=next_pid:wakeup_lat=c
2357 onmax($wakeup_lat).save(next_pr
2358 prev_comm):onmax($wakeup_lat).s
2359 if next_comm=="cyclictest"' >>
2360 /sys/kernel/tracing/events/sche
2361
2362 When the histogram is displayed, for each
2363 and the saved values corresponding to the
2364 following the rest of the fields.
2365
2366 If a snapshot was taken, there is also a
2367 along with the value and event that trigg
2368
2369 # cat /sys/kernel/tracing/events/sched/
2370 { next_pid: 2101 } hitcount:
2371 max: 52 next_prio:
2372 prev_pid: 0 prev_prio:
2373
2374 { next_pid: 2103 } hitcount:
2375 max: 572 next_prio:
2376 prev_pid: 0 prev_prio:
2377
2378 { next_pid: 2102 } hitcount:
2379 max: 74 next_prio:
2380 prev_pid: 0 prev_prio:
2381
2382 Snapshot taken (see tracing/snapshot).
2383 triggering value { onmax($wakeup_la
2384 triggered by event with key: { next
2385
2386 Totals:
2387 Hits: 3508
2388 Entries: 3
2389 Dropped: 0
2390
2391 In the above case, the event that trigger
2392 the key with next_pid == 2103. If you lo
2393 2103 as the key, you'll find the addition
2394 with the local maximum for that bucket, w
2395 as the global maximum (since that was the
2396 triggered the global snapshot).
2397
2398 And finally, looking at the snapshot data
2399 the end the event that triggered the snap
2400 can verify the timestamps between the sch
2401 sched_switch events, which should match t
2402 global maximum)::
2403
2404 # cat /sys/kernel/tracing/snapshot
2405
2406 <...>-2103 [005] d..3 309.873125:
2407 <idle>-0 [005] d.h3 309.873611
2408 <idle>-0 [005] dNh4 309.873613
2409 <idle>-0 [005] d..3 309.873616
2410 <...>-2102 [005] d..3 309.873625:
2411 <idle>-0 [005] d.h3 309.874624
2412 <idle>-0 [005] dNh4 309.874626
2413 <idle>-0 [005] dNh3 309.874628
2414 <idle>-0 [005] dNh4 309.874630
2415 <idle>-0 [005] d..3 309.874633
2416 <idle>-0 [004] d.h3 309.874757
2417 <idle>-0 [004] dNh4 309.874762
2418 <idle>-0 [004] d..3 309.874766
2419 gnome-terminal--1699 [004] d.h2 309.8
2420 <idle>-0 [003] d.s4 309.874956
2421 <idle>-0 [003] d.s5 309.874960
2422 <idle>-0 [003] d.s5 309.874961
2423 <idle>-0 [007] d..3 309.874963
2424 rcu_sched-9 [007] d..3 309.874973
2425 rcu_sched-9 [007] d..3 309.874978
2426 <...>-2102 [005] d..4 309.874994
2427 <...>-2102 [005] d..4 309.875185
2428 <idle>-0 [001] d..3 309.875200
2429
2430 - onchange(var).save(field,.. .)
2431
2432 The 'onchange(var).save(field,...)' hist
2433 whenever the value of 'var' associated wi
2434 changes.
2435
2436 The end result is that the trace event fi
2437 onchange.save() params will be saved if '
2438 hist trigger entry. This allows context
2439 changed the value to be saved for later r
2440 histogram is displayed, additional fields
2441 values will be printed.
2442
2443 - onchange(var).snapshot()
2444
2445 The 'onchange(var).snapshot()' hist trigg
2446 whenever the value of 'var' associated wi
2447 changes.
2448
2449 The end result is that a global snapshot
2450 be saved in the tracing/snapshot file if
2451 hist trigger entry.
2452
2453 Note that in this case the changed value
2454 associated with current trace instance.
2455 trace event that caused the value to chan
2456 itself are displayed, along with a messag
2457 has been taken and where to find it. The
2458 information displayed to locate the corre
2459 histogram for even more detail.
2460
2461 As an example the below defines a hist tr
2462 event, keyed on dport. Whenever a tcp_pr
2463 cwnd field is checked against the current
2464 $cwnd variable. If the value has changed
2465 As part of the setup, all the scheduler a
2466 enabled, which are the events that will s
2467 when it is taken at some point::
2468
2469 # echo 1 > /sys/kernel/tracing/events/s
2470 # echo 1 > /sys/kernel/tracing/events/t
2471
2472 # echo 'hist:keys=dport:cwnd=snd_cwnd:
2473 onchange($cwnd).save(snd_wnd,sr
2474 onchange($cwnd).snapshot()' >>
2475 /sys/kernel/tracing/events/tcp/
2476
2477 When the histogram is displayed, for each
2478 and the saved values corresponding to tha
2479 following the rest of the fields.
2480
2481 If a snapshot was taken, there is also a
2482 along with the value and event that trigg
2483
2484 # cat /sys/kernel/tracing/events/tcp/tc
2485
2486 { dport: 1521 } hitcount:
2487 changed: 10 snd_wnd: 35
2488
2489 { dport: 80 } hitcount:
2490 changed: 10 snd_wnd: 28
2491
2492 { dport: 9001 } hitcount:
2493 changed: 10 snd_wnd: 48
2494
2495 { dport: 443 } hitcount:
2496 changed: 10 snd_wnd: 26
2497
2498 Snapshot taken (see tracing/snapshot).
2499
2500 triggering value { onchange($cwnd)
2501 triggered by event with key: { dpor
2502
2503 Totals:
2504 Hits: 414
2505 Entries: 4
2506 Dropped: 0
2507
2508 In the above case, the event that trigger
2509 key with dport == 80. If you look at the
2510 the key, you'll find the additional value
2511 changed value for that bucket, which shou
2512 global changed value (since that was the
2513 the global snapshot).
2514
2515 And finally, looking at the snapshot data
2516 the end the event that triggered the snap
2517
2518 # cat /sys/kernel/tracing/snapshot
2519
2520 gnome-shell-1261 [006] dN.3 49.8
2521 kworker/u16:4-773 [003] d..3 49.8
2522 gnome-shell-1261 [006] d..3 49.8
2523 kworker/3:2-135 [003] d..3 49.8
2524 kworker/6:2-387 [006] d..3 49.8
2525 kworker/6:2-387 [006] d..3 49.8
2526 kworker/3:2-135 [003] d..3 49.8
2527 <idle>-0 [004] ..s7 49.8
2528
2529 3. User space creating a trigger
2530 --------------------------------
2531
2532 Writing into /sys/kernel/tracing/trace_marker
2533 ring buffer. This can also act like an event,
2534 file located in /sys/kernel/tracing/events/ft
2535
2536 Modifying cyclictest to write into the trace_
2537 and after it wakes up, something like this::
2538
2539 static void traceputs(char *str)
2540 {
2541 /* tracemark_fd is the trace_marker f
2542 if (tracemark_fd < 0)
2543 return;
2544 /* write the tracemark message */
2545 write(tracemark_fd, str, strlen(str))
2546 }
2547
2548 And later add something like::
2549
2550 traceputs("start");
2551 clock_nanosleep(...);
2552 traceputs("end");
2553
2554 We can make a histogram from this::
2555
2556 # cd /sys/kernel/tracing
2557 # echo 'latency u64 lat' > synthetic_events
2558 # echo 'hist:keys=common_pid:ts0=common_time
2559 # echo 'hist:keys=common_pid:lat=common_time
2560 # echo 'hist:keys=lat,common_pid:sort=lat' >
2561
2562 The above created a synthetic event called "l
2563 against the trace_marker, one gets triggered
2564 trace_marker file and the other when "end" is
2565 it will call the "latency" synthetic event wi
2566 parameter. Finally, a histogram is added to t
2567 record the calculated latency along with the
2568
2569 Now running cyclictest with::
2570
2571 # ./cyclictest -p80 -d0 -i250 -n -a -t --tra
2572
2573 -p80 : run threads at priority 80
2574 -d0 : have all threads run at the same int
2575 -i250 : start the interval at 250 microsecon
2576 -n : sleep with nanosleep
2577 -a : affine all threads to a separate CPU
2578 -t : one thread per available CPU
2579 --tracemark : enable trace mark writing
2580 -b 1000 : stop if any latency is greater tha
2581
2582 Note, the -b 1000 is used just to make --trac
2583
2584 Then we can see the histogram created by this
2585
2586 # cat events/synthetic/latency/hist
2587 # event histogram
2588 #
2589 # trigger info: hist:keys=lat,common_pid:val
2590 #
2591
2592 { lat: 107, common_pid: 2039 }
2593 { lat: 122, common_pid: 2041 }
2594 { lat: 166, common_pid: 2039 }
2595 { lat: 174, common_pid: 2039 }
2596 { lat: 194, common_pid: 2041 }
2597 { lat: 196, common_pid: 2036 }
2598 { lat: 197, common_pid: 2038 }
2599 { lat: 198, common_pid: 2039 }
2600 { lat: 199, common_pid: 2039 }
2601 { lat: 200, common_pid: 2041 }
2602 { lat: 201, common_pid: 2039 }
2603 { lat: 202, common_pid: 2038 }
2604 { lat: 202, common_pid: 2043 }
2605 { lat: 203, common_pid: 2039 }
2606 { lat: 203, common_pid: 2036 }
2607 { lat: 203, common_pid: 2041 }
2608 { lat: 206, common_pid: 2038 }
2609 { lat: 207, common_pid: 2039 }
2610 { lat: 207, common_pid: 2036 }
2611 { lat: 208, common_pid: 2040 }
2612 { lat: 209, common_pid: 2043 }
2613 { lat: 210, common_pid: 2039 }
2614 { lat: 211, common_pid: 2039 }
2615 { lat: 212, common_pid: 2043 }
2616 { lat: 212, common_pid: 2039 }
2617 { lat: 213, common_pid: 2039 }
2618 { lat: 214, common_pid: 2038 }
2619 { lat: 214, common_pid: 2039 }
2620 { lat: 214, common_pid: 2042 }
2621 { lat: 215, common_pid: 2039 }
2622 { lat: 217, common_pid: 2036 }
2623 { lat: 217, common_pid: 2040 }
2624 { lat: 217, common_pid: 2039 }
2625 { lat: 218, common_pid: 2039 }
2626 { lat: 219, common_pid: 2039 }
2627 { lat: 220, common_pid: 2039 }
2628 { lat: 221, common_pid: 2039 }
2629 { lat: 221, common_pid: 2042 }
2630 { lat: 222, common_pid: 2039 }
2631 { lat: 223, common_pid: 2036 }
2632 { lat: 223, common_pid: 2039 }
2633 { lat: 224, common_pid: 2039 }
2634 { lat: 224, common_pid: 2037 }
2635 { lat: 224, common_pid: 2036 }
2636 { lat: 225, common_pid: 2039 }
2637 { lat: 225, common_pid: 2042 }
2638 { lat: 226, common_pid: 2039 }
2639 { lat: 226, common_pid: 2036 }
2640 { lat: 227, common_pid: 2039 }
2641 { lat: 227, common_pid: 2036 }
2642 { lat: 227, common_pid: 2043 }
2643 { lat: 228, common_pid: 2039 }
2644 { lat: 228, common_pid: 2036 }
2645 { lat: 229, common_pid: 2039 }
2646 { lat: 229, common_pid: 2036 }
2647 { lat: 229, common_pid: 2038 }
2648 { lat: 230, common_pid: 2039 }
2649 { lat: 230, common_pid: 2036 }
2650 { lat: 230, common_pid: 2043 }
2651 { lat: 230, common_pid: 2042 }
2652 { lat: 231, common_pid: 2041 }
2653 { lat: 231, common_pid: 2036 }
2654 { lat: 231, common_pid: 2043 }
2655 { lat: 231, common_pid: 2039 }
2656 { lat: 232, common_pid: 2037 }
2657 { lat: 232, common_pid: 2039 }
2658 { lat: 232, common_pid: 2040 }
2659 { lat: 232, common_pid: 2036 }
2660 { lat: 232, common_pid: 2043 }
2661 { lat: 233, common_pid: 2036 }
2662 { lat: 233, common_pid: 2039 }
2663 { lat: 234, common_pid: 2039 }
2664 { lat: 234, common_pid: 2038 }
2665 { lat: 234, common_pid: 2043 }
2666 { lat: 234, common_pid: 2036 }
2667 { lat: 234, common_pid: 2040 }
2668 { lat: 235, common_pid: 2037 }
2669 { lat: 235, common_pid: 2036 }
2670 { lat: 235, common_pid: 2043 }
2671 { lat: 235, common_pid: 2039 }
2672 { lat: 235, common_pid: 2042 }
2673 { lat: 235, common_pid: 2040 }
2674 { lat: 235, common_pid: 2041 }
2675 { lat: 236, common_pid: 2036 }
2676 { lat: 236, common_pid: 2037 }
2677 { lat: 236, common_pid: 2041 }
2678 { lat: 236, common_pid: 2039 }
2679 { lat: 236, common_pid: 2043 }
2680 { lat: 236, common_pid: 2040 }
2681 { lat: 237, common_pid: 2037 }
2682 { lat: 237, common_pid: 2040 }
2683 { lat: 237, common_pid: 2036 }
2684 { lat: 237, common_pid: 2039 }
2685 { lat: 237, common_pid: 2043 }
2686 { lat: 237, common_pid: 2042 }
2687 { lat: 237, common_pid: 2041 }
2688 { lat: 238, common_pid: 2043 }
2689 { lat: 238, common_pid: 2040 }
2690 { lat: 238, common_pid: 2037 }
2691 { lat: 238, common_pid: 2038 }
2692 { lat: 238, common_pid: 2039 }
2693 { lat: 238, common_pid: 2042 }
2694 { lat: 238, common_pid: 2036 }
2695 { lat: 239, common_pid: 2041 }
2696 { lat: 239, common_pid: 2043 }
2697 { lat: 239, common_pid: 2037 }
2698 { lat: 239, common_pid: 2038 }
2699 { lat: 239, common_pid: 2036 }
2700 { lat: 239, common_pid: 2040 }
2701 { lat: 239, common_pid: 2042 }
2702 { lat: 240, common_pid: 2037 }
2703 { lat: 240, common_pid: 2043 }
2704 { lat: 240, common_pid: 2040 }
2705 { lat: 240, common_pid: 2039 }
2706 { lat: 240, common_pid: 2041 }
2707 { lat: 240, common_pid: 2038 }
2708 { lat: 240, common_pid: 2036 }
2709 { lat: 240, common_pid: 2042 }
2710 { lat: 241, common_pid: 2036 }
2711 { lat: 241, common_pid: 2041 }
2712 { lat: 241, common_pid: 2037 }
2713 { lat: 241, common_pid: 2042 }
2714 { lat: 241, common_pid: 2040 }
2715 { lat: 241, common_pid: 2039 }
2716 { lat: 241, common_pid: 2038 }
2717 { lat: 241, common_pid: 2043 }
2718 { lat: 242, common_pid: 2040 }
2719 { lat: 242, common_pid: 2041 }
2720 { lat: 242, common_pid: 2039 }
2721 { lat: 242, common_pid: 2042 }
2722 { lat: 242, common_pid: 2037 }
2723 { lat: 242, common_pid: 2043 }
2724 { lat: 242, common_pid: 2036 }
2725 { lat: 242, common_pid: 2038 }
2726 { lat: 243, common_pid: 2037 }
2727 { lat: 243, common_pid: 2039 }
2728 { lat: 243, common_pid: 2042 }
2729 { lat: 243, common_pid: 2041 }
2730 { lat: 243, common_pid: 2036 }
2731 { lat: 243, common_pid: 2038 }
2732 { lat: 243, common_pid: 2040 }
2733 { lat: 243, common_pid: 2043 }
2734 { lat: 244, common_pid: 2043 }
2735 { lat: 244, common_pid: 2042 }
2736 { lat: 244, common_pid: 2039 }
2737 { lat: 244, common_pid: 2040 }
2738 { lat: 244, common_pid: 2038 }
2739 { lat: 244, common_pid: 2037 }
2740 { lat: 244, common_pid: 2036 }
2741 { lat: 244, common_pid: 2041 }
2742 { lat: 245, common_pid: 2036 }
2743 { lat: 245, common_pid: 2041 }
2744 { lat: 245, common_pid: 2037 }
2745 { lat: 245, common_pid: 2039 }
2746 { lat: 245, common_pid: 2038 }
2747 { lat: 245, common_pid: 2042 }
2748 { lat: 245, common_pid: 2040 }
2749 { lat: 245, common_pid: 2043 }
2750 { lat: 246, common_pid: 2038 }
2751 { lat: 246, common_pid: 2041 }
2752 { lat: 246, common_pid: 2037 }
2753 { lat: 246, common_pid: 2040 }
2754 { lat: 246, common_pid: 2043 }
2755 { lat: 246, common_pid: 2039 }
2756 { lat: 246, common_pid: 2042 }
2757 { lat: 246, common_pid: 2036 }
2758 { lat: 247, common_pid: 2042 }
2759 { lat: 247, common_pid: 2043 }
2760 { lat: 247, common_pid: 2041 }
2761 { lat: 247, common_pid: 2038 }
2762 { lat: 247, common_pid: 2040 }
2763 { lat: 247, common_pid: 2036 }
2764 { lat: 247, common_pid: 2039 }
2765 { lat: 247, common_pid: 2037 }
2766 { lat: 248, common_pid: 2037 }
2767 { lat: 248, common_pid: 2042 }
2768 { lat: 248, common_pid: 2038 }
2769 { lat: 248, common_pid: 2036 }
2770 { lat: 248, common_pid: 2039 }
2771 { lat: 248, common_pid: 2041 }
2772 { lat: 248, common_pid: 2043 }
2773 { lat: 248, common_pid: 2040 }
2774 { lat: 249, common_pid: 2037 }
2775 { lat: 249, common_pid: 2038 }
2776 { lat: 249, common_pid: 2043 }
2777 { lat: 249, common_pid: 2039 }
2778 { lat: 249, common_pid: 2042 }
2779 { lat: 249, common_pid: 2040 }
2780 { lat: 249, common_pid: 2041 }
2781 { lat: 249, common_pid: 2036 }
2782 { lat: 250, common_pid: 2038 }
2783 { lat: 250, common_pid: 2036 }
2784 { lat: 250, common_pid: 2040 }
2785 { lat: 250, common_pid: 2043 }
2786 { lat: 250, common_pid: 2042 }
2787 { lat: 250, common_pid: 2041 }
2788 { lat: 250, common_pid: 2039 }
2789 { lat: 250, common_pid: 2037 }
2790 { lat: 251, common_pid: 2037 }
2791 { lat: 251, common_pid: 2039 }
2792 { lat: 251, common_pid: 2036 }
2793 { lat: 251, common_pid: 2042 }
2794 { lat: 251, common_pid: 2041 }
2795 { lat: 251, common_pid: 2043 }
2796 { lat: 251, common_pid: 2040 }
2797 { lat: 252, common_pid: 2040 }
2798 { lat: 252, common_pid: 2036 }
2799 { lat: 252, common_pid: 2037 }
2800 { lat: 252, common_pid: 2043 }
2801 { lat: 253, common_pid: 2037 }
2802 { lat: 253, common_pid: 2039 }
2803 { lat: 253, common_pid: 2036 }
2804 { lat: 253, common_pid: 2043 }
2805 { lat: 253, common_pid: 2040 }
2806 { lat: 254, common_pid: 2036 }
2807 { lat: 254, common_pid: 2043 }
2808 { lat: 254, common_pid: 2041 }
2809 { lat: 254, common_pid: 2042 }
2810 { lat: 254, common_pid: 2039 }
2811 { lat: 254, common_pid: 2037 }
2812 { lat: 255, common_pid: 2043 }
2813 { lat: 255, common_pid: 2037 }
2814 { lat: 255, common_pid: 2036 }
2815 { lat: 255, common_pid: 2039 }
2816 { lat: 256, common_pid: 2043 }
2817 { lat: 256, common_pid: 2036 }
2818 { lat: 256, common_pid: 2039 }
2819 { lat: 257, common_pid: 2039 }
2820 { lat: 257, common_pid: 2036 }
2821 { lat: 258, common_pid: 2039 }
2822 { lat: 258, common_pid: 2036 }
2823 { lat: 259, common_pid: 2036 }
2824 { lat: 259, common_pid: 2039 }
2825 { lat: 260, common_pid: 2036 }
2826 { lat: 260, common_pid: 2039 }
2827 { lat: 261, common_pid: 2036 }
2828 { lat: 261, common_pid: 2039 }
2829 { lat: 262, common_pid: 2039 }
2830 { lat: 262, common_pid: 2036 }
2831 { lat: 263, common_pid: 2039 }
2832 { lat: 263, common_pid: 2036 }
2833 { lat: 264, common_pid: 2039 }
2834 { lat: 264, common_pid: 2036 }
2835 { lat: 265, common_pid: 2036 }
2836 { lat: 265, common_pid: 2039 }
2837 { lat: 266, common_pid: 2036 }
2838 { lat: 266, common_pid: 2039 }
2839 { lat: 267, common_pid: 2036 }
2840 { lat: 267, common_pid: 2039 }
2841 { lat: 268, common_pid: 2036 }
2842 { lat: 268, common_pid: 2039 }
2843 { lat: 269, common_pid: 2036 }
2844 { lat: 269, common_pid: 2043 }
2845 { lat: 269, common_pid: 2039 }
2846 { lat: 270, common_pid: 2040 }
2847 { lat: 270, common_pid: 2039 }
2848 { lat: 271, common_pid: 2041 }
2849 { lat: 271, common_pid: 2039 }
2850 { lat: 272, common_pid: 2039 }
2851 { lat: 273, common_pid: 2039 }
2852 { lat: 274, common_pid: 2039 }
2853 { lat: 275, common_pid: 2039 }
2854 { lat: 276, common_pid: 2039 }
2855 { lat: 276, common_pid: 2037 }
2856 { lat: 276, common_pid: 2038 }
2857 { lat: 277, common_pid: 2039 }
2858 { lat: 277, common_pid: 2042 }
2859 { lat: 278, common_pid: 2039 }
2860 { lat: 279, common_pid: 2039 }
2861 { lat: 279, common_pid: 2043 }
2862 { lat: 280, common_pid: 2039 }
2863 { lat: 283, common_pid: 2036 }
2864 { lat: 284, common_pid: 2039 }
2865 { lat: 284, common_pid: 2043 }
2866 { lat: 288, common_pid: 2039 }
2867 { lat: 289, common_pid: 2039 }
2868 { lat: 300, common_pid: 2039 }
2869 { lat: 384, common_pid: 2039 }
2870
2871 Totals:
2872 Hits: 67625
2873 Entries: 278
2874 Dropped: 0
2875
2876 Note, the writes are around the sleep, so ide
2877 microseconds. If you are wondering how there
2878 250 microseconds, that is because the way cyc
2879 iteration comes in late, the next one will se
2880 250. That is, if an iteration came in 50 micr
2881 will be at 200 microseconds.
2882
2883 But this could easily be done in userspace. T
2884 interesting, we can mix the histogram between
2885 kernel with trace_marker::
2886
2887 # cd /sys/kernel/tracing
2888 # echo 'latency u64 lat' > synthetic_events
2889 # echo 'hist:keys=pid:ts0=common_timestamp.u
2890 # echo 'hist:keys=common_pid:lat=common_time
2891 # echo 'hist:keys=lat,common_pid:sort=lat' >
2892
2893 The difference this time is that instead of u
2894 the latency, the sched_waking event is used,
2895 trace_marker write with the pid that is being
2896
2897 After running cyclictest again with the same
2898
2899 # cat events/synthetic/latency/hist
2900 # event histogram
2901 #
2902 # trigger info: hist:keys=lat,common_pid:val
2903 #
2904
2905 { lat: 7, common_pid: 2302 }
2906 { lat: 7, common_pid: 2299 }
2907 { lat: 7, common_pid: 2303 }
2908 { lat: 7, common_pid: 2305 }
2909 { lat: 7, common_pid: 2306 }
2910 { lat: 7, common_pid: 2301 }
2911 { lat: 7, common_pid: 2300 }
2912 { lat: 8, common_pid: 2303 }
2913 { lat: 8, common_pid: 2304 }
2914 { lat: 8, common_pid: 2305 }
2915 { lat: 8, common_pid: 2301 }
2916 { lat: 8, common_pid: 2306 }
2917 { lat: 8, common_pid: 2302 }
2918 { lat: 8, common_pid: 2299 }
2919 { lat: 8, common_pid: 2300 }
2920 { lat: 9, common_pid: 2305 }
2921 { lat: 9, common_pid: 2299 }
2922 { lat: 9, common_pid: 2303 }
2923 { lat: 9, common_pid: 2301 }
2924 { lat: 9, common_pid: 2304 }
2925 { lat: 9, common_pid: 2302 }
2926 { lat: 9, common_pid: 2300 }
2927 { lat: 9, common_pid: 2306 }
2928 { lat: 10, common_pid: 2299 }
2929 { lat: 10, common_pid: 2306 }
2930 { lat: 10, common_pid: 2303 }
2931 { lat: 10, common_pid: 2304 }
2932 { lat: 10, common_pid: 2302 }
2933 { lat: 10, common_pid: 2301 }
2934 { lat: 10, common_pid: 2300 }
2935 { lat: 10, common_pid: 2305 }
2936 { lat: 11, common_pid: 2303 }
2937 { lat: 11, common_pid: 2305 }
2938 { lat: 11, common_pid: 2306 }
2939 { lat: 11, common_pid: 2302 }
2940 { lat: 11, common_pid: 2304 }
2941 { lat: 11, common_pid: 2300 }
2942 { lat: 11, common_pid: 2299 }
2943 { lat: 12, common_pid: 2305 }
2944 { lat: 12, common_pid: 2300 }
2945 { lat: 12, common_pid: 2306 }
2946 { lat: 12, common_pid: 2302 }
2947 { lat: 12, common_pid: 2303 }
2948 { lat: 12, common_pid: 2304 }
2949 { lat: 13, common_pid: 2300 }
2950 { lat: 13, common_pid: 2301 }
2951 { lat: 13, common_pid: 2306 }
2952 { lat: 13, common_pid: 2302 }
2953 { lat: 13, common_pid: 2305 }
2954 { lat: 13, common_pid: 2303 }
2955 { lat: 13, common_pid: 2304 }
2956 { lat: 14, common_pid: 2303 }
2957 { lat: 14, common_pid: 2306 }
2958 { lat: 14, common_pid: 2305 }
2959 { lat: 14, common_pid: 2304 }
2960 { lat: 14, common_pid: 2302 }
2961 { lat: 14, common_pid: 2300 }
2962 { lat: 14, common_pid: 2299 }
2963 { lat: 14, common_pid: 2301 }
2964 { lat: 15, common_pid: 2305 }
2965 { lat: 15, common_pid: 2302 }
2966 { lat: 15, common_pid: 2300 }
2967 { lat: 15, common_pid: 2299 }
2968 { lat: 15, common_pid: 2301 }
2969 { lat: 15, common_pid: 2304 }
2970 { lat: 15, common_pid: 2303 }
2971 { lat: 15, common_pid: 2306 }
2972 { lat: 16, common_pid: 2302 }
2973 { lat: 16, common_pid: 2306 }
2974 { lat: 16, common_pid: 2300 }
2975 { lat: 17, common_pid: 2302 }
2976 { lat: 17, common_pid: 2303 }
2977 { lat: 18, common_pid: 2304 }
2978 { lat: 18, common_pid: 2302 }
2979 { lat: 18, common_pid: 2299 }
2980 { lat: 18, common_pid: 2301 }
2981 { lat: 19, common_pid: 2303 }
2982 { lat: 19, common_pid: 2304 }
2983 { lat: 19, common_pid: 2302 }
2984 { lat: 19, common_pid: 2299 }
2985 { lat: 19, common_pid: 2306 }
2986 { lat: 19, common_pid: 2300 }
2987 { lat: 19, common_pid: 2305 }
2988 { lat: 20, common_pid: 2299 }
2989 { lat: 20, common_pid: 2302 }
2990 { lat: 20, common_pid: 2305 }
2991 { lat: 20, common_pid: 2300 }
2992 { lat: 20, common_pid: 2301 }
2993 { lat: 20, common_pid: 2303 }
2994 { lat: 21, common_pid: 2305 }
2995 { lat: 21, common_pid: 2299 }
2996 { lat: 21, common_pid: 2303 }
2997 { lat: 21, common_pid: 2302 }
2998 { lat: 21, common_pid: 2300 }
2999 { lat: 21, common_pid: 2301 }
3000 { lat: 21, common_pid: 2304 }
3001 { lat: 22, common_pid: 2302 }
3002 { lat: 22, common_pid: 2303 }
3003 { lat: 22, common_pid: 2306 }
3004 { lat: 22, common_pid: 2301 }
3005 { lat: 22, common_pid: 2300 }
3006 { lat: 22, common_pid: 2299 }
3007 { lat: 22, common_pid: 2305 }
3008 { lat: 22, common_pid: 2304 }
3009 { lat: 23, common_pid: 2299 }
3010 { lat: 23, common_pid: 2306 }
3011 { lat: 23, common_pid: 2302 }
3012 { lat: 24, common_pid: 2302 }
3013 { lat: 24, common_pid: 2300 }
3014 { lat: 24, common_pid: 2306 }
3015 { lat: 24, common_pid: 2305 }
3016 { lat: 24, common_pid: 2299 }
3017 { lat: 25, common_pid: 2300 }
3018 { lat: 25, common_pid: 2302 }
3019 { lat: 26, common_pid: 2302 }
3020 { lat: 27, common_pid: 2305 }
3021 { lat: 27, common_pid: 2300 }
3022 { lat: 27, common_pid: 2302 }
3023 { lat: 28, common_pid: 2306 }
3024 { lat: 28, common_pid: 2302 }
3025 { lat: 29, common_pid: 2302 }
3026 { lat: 29, common_pid: 2300 }
3027 { lat: 29, common_pid: 2306 }
3028 { lat: 29, common_pid: 2304 }
3029 { lat: 30, common_pid: 2302 }
3030 { lat: 31, common_pid: 2302 }
3031 { lat: 32, common_pid: 2302 }
3032 { lat: 33, common_pid: 2299 }
3033 { lat: 33, common_pid: 2302 }
3034 { lat: 34, common_pid: 2302 }
3035 { lat: 35, common_pid: 2302 }
3036 { lat: 35, common_pid: 2304 }
3037 { lat: 36, common_pid: 2302 }
3038 { lat: 37, common_pid: 2302 }
3039 { lat: 38, common_pid: 2302 }
3040 { lat: 39, common_pid: 2302 }
3041 { lat: 39, common_pid: 2304 }
3042 { lat: 40, common_pid: 2304 }
3043 { lat: 40, common_pid: 2302 }
3044 { lat: 41, common_pid: 2304 }
3045 { lat: 41, common_pid: 2302 }
3046 { lat: 42, common_pid: 2302 }
3047 { lat: 42, common_pid: 2304 }
3048 { lat: 43, common_pid: 2302 }
3049 { lat: 43, common_pid: 2304 }
3050 { lat: 44, common_pid: 2302 }
3051 { lat: 45, common_pid: 2302 }
3052 { lat: 46, common_pid: 2302 }
3053 { lat: 47, common_pid: 2302 }
3054 { lat: 48, common_pid: 2301 }
3055 { lat: 48, common_pid: 2302 }
3056 { lat: 49, common_pid: 2302 }
3057 { lat: 50, common_pid: 2302 }
3058 { lat: 50, common_pid: 2301 }
3059 { lat: 51, common_pid: 2302 }
3060 { lat: 51, common_pid: 2301 }
3061 { lat: 61, common_pid: 2302 }
3062 { lat: 110, common_pid: 2302 }
3063
3064 Totals:
3065 Hits: 89565
3066 Entries: 158
3067 Dropped: 0
3068
3069 This doesn't tell us any information about ho
3070 woken up, but it does show us a nice histogra
3071 the time that cyclictest was woken to the tim
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.