aboutsummaryrefslogtreecommitdiffstats
path: root/tools/perf/Documentation/intel-pt.txt
blob: b0b3007d3c9c0ff9b3288e8005d533efe48c0b57 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
Intel Processor Trace
=====================

Overview
========

Intel Processor Trace (Intel PT) is an extension of Intel Architecture that
collects information about software execution such as control flow, execution
modes and timings and formats it into highly compressed binary packets.
Technical details are documented in the Intel 64 and IA-32 Architectures
Software Developer Manuals, Chapter 36 Intel Processor Trace.

Intel PT is first supported in Intel Core M and 5th generation Intel Core
processors that are based on the Intel micro-architecture code name Broadwell.

Trace data is collected by 'perf record' and stored within the perf.data file.
See below for options to 'perf record'.

Trace data must be 'decoded' which involves walking the object code and matching
the trace data packets. For example a TNT packet only tells whether a
conditional branch was taken or not taken, so to make use of that packet the
decoder must know precisely which instruction was being executed.

Decoding is done on-the-fly.  The decoder outputs samples in the same format as
samples output by perf hardware events, for example as though the "instructions"
or "branches" events had been recorded.  Presently 3 tools support this:
'perf script', 'perf report' and 'perf inject'.  See below for more information
on using those tools.

The main distinguishing feature of Intel PT is that the decoder can determine
the exact flow of software execution.  Intel PT can be used to understand why
and how did software get to a certain point, or behave a certain way.  The
software does not have to be recompiled, so Intel PT works with debug or release
builds, however the executed images are needed - which makes use in JIT-compiled
environments, or with self-modified code, a challenge.  Also symbols need to be
provided to make sense of addresses.

A limitation of Intel PT is that it produces huge amounts of trace data
(hundreds of megabytes per second per core) which takes a long time to decode,
for example two or three orders of magnitude longer than it took to collect.
Another limitation is the performance impact of tracing, something that will
vary depending on the use-case and architecture.


Quickstart
==========

It is important to start small.  That is because it is easy to capture vastly
more data than can possibly be processed.

The simplest thing to do with Intel PT is userspace profiling of small programs.
Data is captured with 'perf record' e.g. to trace 'ls' userspace-only:

	perf record -e intel_pt//u ls

And profiled with 'perf report' e.g.

	perf report

To also trace kernel space presents a problem, namely kernel self-modifying
code.  A fairly good kernel image is available in /proc/kcore but to get an
accurate image a copy of /proc/kcore needs to be made under the same conditions
as the data capture.  A script perf-with-kcore can do that, but beware that the
script makes use of 'sudo' to copy /proc/kcore.  If you have perf installed
locally from the source tree you can do:

	~/libexec/perf-core/perf-with-kcore record pt_ls -e intel_pt// -- ls

which will create a directory named 'pt_ls' and put the perf.data file and
copies of /proc/kcore, /proc/kallsyms and /proc/modules into it.  Then to use
'perf report' becomes:

	~/libexec/perf-core/perf-with-kcore report pt_ls

Because samples are synthesized after-the-fact, the sampling period can be
selected for reporting. e.g. sample every microsecond

	~/libexec/perf-core/perf-with-kcore report pt_ls --itrace=i1usge

See the sections below for more information about the --itrace option.

Beware the smaller the period, the more samples that are produced, and the
longer it takes to process them.

Also note that the coarseness of Intel PT timing information will start to
distort the statistical value of the sampling as the sampling period becomes
smaller.

To represent software control flow, "branches" samples are produced.  By default
a branch sample is synthesized for every single branch.  To get an idea what
data is available you can use the 'perf script' tool with no parameters, which
will list all the samples.

	perf record -e intel_pt//u ls
	perf script

An interesting field that is not printed by default is 'flags' which can be
displayed as follows:

	perf script -Fcomm,tid,pid,time,cpu,event,trace,ip,sym,dso,addr,symoff,flags

The flags are "bcrosyiABEx" which stand for branch, call, return, conditional,
system, asynchronous, interrupt, transaction abort, trace begin, trace end, and
in transaction, respectively.

While it is possible to create scripts to analyze the data, an alternative
approach is available to export the data to a postgresql database.  Refer to
script export-to-postgresql.py for more details, and to script
call-graph-from-postgresql.py for an example of using the database.

As mentioned above, it is easy to capture too much data.  One way to limit the
data captured is to use 'snapshot' mode which is explained further below.
Refer to 'new snapshot option' and 'Intel PT modes of operation' further below.

Another problem that will be experienced is decoder errors.  They can be caused
by inability to access the executed image, self-modified or JIT-ed code, or the
inability to match side-band information (such as context switches and mmaps)
which results in the decoder not knowing what code was executed.

There is also the problem of perf not being able to copy the data fast enough,
resulting in data lost because the buffer was full.  See 'Buffer handling' below
for more details.


perf record
===========

new event
---------

The Intel PT kernel driver creates a new PMU for Intel PT.  PMU events are
selected by providing the PMU name followed by the "config" separated by slashes.
An enhancement has been made to allow default "config" e.g. the option

	-e intel_pt//

will use a default config value.  Currently that is the same as

	-e intel_pt/tsc,noretcomp=0/

which is the same as

	-e intel_pt/tsc=1,noretcomp=0/

Note there are now new config terms - see section 'config terms' further below.

The config terms are listed in /sys/devices/intel_pt/format.  They are bit
fields within the config member of the struct perf_event_attr which is
passed to the kernel by the perf_event_open system call.  They correspond to bit
fields in the IA32_RTIT_CTL MSR.  Here is a list of them and their definitions:

	$ grep -H . /sys/bus/event_source/devices/intel_pt/format/*
	/sys/bus/event_source/devices/intel_pt/format/cyc:config:1
	/sys/bus/event_source/devices/intel_pt/format/cyc_thresh:config:19-22
	/sys/bus/event_source/devices/intel_pt/format/mtc:config:9
	/sys/bus/event_source/devices/intel_pt/format/mtc_period:config:14-17
	/sys/bus/event_source/devices/intel_pt/format/noretcomp:config:11
	/sys/bus/event_source/devices/intel_pt/format/psb_period:config:24-27
	/sys/bus/event_source/devices/intel_pt/format/tsc:config:10

Note that the default config must be overridden for each term i.e.

	-e intel_pt/noretcomp=0/

is the same as:

	-e intel_pt/tsc=1,noretcomp=0/

So, to disable TSC packets use:

	-e intel_pt/tsc=0/

It is also possible to specify the config value explicitly:

	-e intel_pt/config=0x400/

Note that, as with all events, the event is suffixed with event modifiers:

	u	userspace
	k	kernel
	h	hypervisor
	G	guest
	H	host
	p	precise ip

'h', 'G' and 'H' are for virtualization which is not supported by Intel PT.
'p' is also not relevant to Intel PT.  So only options 'u' and 'k' are
meaningful for Intel PT.

perf_event_attr is displayed if the -vv option is used e.g.

	------------------------------------------------------------
	perf_event_attr:
	type                             6
	size                             112
	config                           0x400
	{ sample_period, sample_freq }   1
	sample_type                      IP|TID|TIME|CPU|IDENTIFIER
	read_format                      ID
	disabled                         1
	inherit                          1
	exclude_kernel                   1
	exclude_hv                       1
	enable_on_exec                   1
	sample_id_all                    1
	------------------------------------------------------------
	sys_perf_event_open: pid 31104  cpu 0  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 1  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 2  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 3  group_fd -1  flags 0x8
	------------------------------------------------------------


config terms
------------

The June 2015 version of Intel 64 and IA-32 Architectures Software Developer
Manuals, Chapter 36 Intel Processor Trace, defined new Intel PT features.
Some of the features are reflect in new config terms.  All the config terms are
described below.

tsc		Always supported.  Produces TSC timestamp packets to provide
		timing information.  In some cases it is possible to decode
		without timing information, for example a per-thread context
		that does not overlap executable memory maps.

		The default config selects tsc (i.e. tsc=1).

noretcomp	Always supported.  Disables "return compression" so a TIP packet
		is produced when a function returns.  Causes more packets to be
		produced but might make decoding more reliable.

		The default config does not select noretcomp (i.e. noretcomp=0).

psb_period	Allows the frequency of PSB packets to be specified.

		The PSB packet is a synchronization packet that provides a
		starting point for decoding or recovery from errors.

		Support for psb_period is indicated by:

			/sys/bus/event_source/devices/intel_pt/caps/psb_cyc

		which contains "1" if the feature is supported and "0"
		otherwise.

		Valid values are given by:

			/sys/bus/event_source/devices/intel_pt/caps/psb_periods

		which contains a hexadecimal value, the bits of which represent
		valid values e.g. bit 2 set means value 2 is valid.

		The psb_period value is converted to the approximate number of
		trace bytes between PSB packets as:

			2 ^ (value + 11)

		e.g. value 3 means 16KiB bytes between PSBs

		If an invalid value is entered, the error message
		will give a list of valid values e.g.

			$ perf record -e intel_pt/psb_period=15/u uname
			Invalid psb_period for intel_pt. Valid values are: 0-5

		If MTC packets are selected, the default config selects a value
		of 3 (i.e. psb_period=3) or the nearest lower value that is
		supported (0 is always supported).  Otherwise the default is 0.

		If decoding is expected to be reliable and the buffer is large
		then a large PSB period can be used.

		Because a TSC packet is produced with PSB, the PSB period can
		also affect the granularity to timing information in the absence
		of MTC or CYC.

mtc		Produces MTC timing packets.

		MTC packets provide finer grain timestamp information than TSC
		packets.  MTC packets record time using the hardware crystal
		clock (CTC) which is related to TSC packets using a TMA packet.

		Support for this feature is indicated by:

			/sys/bus/event_source/devices/intel_pt/caps/mtc

		which contains "1" if the feature is supported and
		"0" otherwise.

		The frequency of MTC packets can also be specified - see
		mtc_period below.

mtc_period	Specifies how frequently MTC packets are produced - see mtc
		above for how to determine if MTC packets are supported.

		Valid values are given by:

			/sys/bus/event_source/devices/intel_pt/caps/mtc_periods

		which contains a hexadecimal value, the bits of which represent
		valid values e.g. bit 2 set means value 2 is valid.

		The mtc_period value is converted to the MTC frequency as:

			CTC-frequency / (2 ^ value)

		e.g. value 3 means one eighth of CTC-frequency

		Where CTC is the hardware crystal clock, the frequency of which
		can be related to TSC via values provided in cpuid leaf 0x15.

		If an invalid value is entered, the error message
		will give a list of valid values e.g.

			$ perf record -e intel_pt/mtc_period=15/u uname
			Invalid mtc_period for intel_pt. Valid values are: 0,3,6,9

		The default value is 3 or the nearest lower value
		that is supported (0 is always supported).

cyc		Produces CYC timing packets.

		CYC packets provide even finer grain timestamp information than
		MTC and TSC packets.  A CYC packet contains the number of CPU
		cycles since the last CYC packet. Unlike MTC and TSC packets,
		CYC packets are only sent when another packet is also sent.

		Support for this feature is indicated by:

			/sys/bus/event_source/devices/intel_pt/caps/psb_cyc

		which contains "1" if the feature is supported and
		"0" otherwise.

		The number of CYC packets produced can be reduced by specifying
		a threshold - see cyc_thresh below.

cyc_thresh	Specifies how frequently CYC packets are produced - see cyc
		above for how to determine if CYC packets are supported.

		Valid cyc_thresh values are given by:

			/sys/bus/event_source/devices/intel_pt/caps/cycle_thresholds

		which contains a hexadecimal value, the bits of which represent
		valid values e.g. bit 2 set means value 2 is valid.

		The cyc_thresh value represents the minimum number of CPU cycles
		that must have passed before a CYC packet can be sent.  The
		number of CPU cycles is:

			2 ^ (value - 1)

		e.g. value 4 means 8 CPU cycles must pass before a CYC packet
		can be sent.  Note a CYC packet is still only sent when another
		packet is sent, not at, e.g. every 8 CPU cycles.

		If an invalid value is entered, the error message
		will give a list of valid values e.g.

			$ perf record -e intel_pt/cyc,cyc_thresh=15/u uname
			Invalid cyc_thresh for intel_pt. Valid values are: 0-12

		CYC packets are not requested by default.


new snapshot option
-------------------

The difference between full trace and snapshot from the kernel's perspective is
that in full trace we don't overwrite trace data that the user hasn't collected
yet (and indicated that by advancing aux_tail), whereas in snapshot mode we let
the trace run and overwrite older data in the buffer so that whenever something
interesting happens, we can stop it and grab a snapshot of what was going on
around that interesting moment.

To select snapshot mode a new option has been added:

	-S

Optionally it can be followed by the snapshot size e.g.

	-S0x100000

The default snapshot size is the auxtrace mmap size.  If neither auxtrace mmap size
nor snapshot size is specified, then the default is 4MiB for privileged users
(or if /proc/sys/kernel/perf_event_paranoid < 0), 128KiB for unprivileged users.
If an unprivileged user does not specify mmap pages, the mmap pages will be
reduced as described in the 'new auxtrace mmap size option' section below.

The snapshot size is displayed if the option -vv is used e.g.

	Intel PT snapshot size: %zu


new auxtrace mmap size option
---------------------------

Intel PT buffer size is specified by an addition to the -m option e.g.

	-m,16

selects a buffer size of 16 pages i.e. 64KiB.

Note that the existing functionality of -m is unchanged.  The auxtrace mmap size
is specified by the optional addition of a comma and the value.

The default auxtrace mmap size for Intel PT is 4MiB/page_size for privileged users
(or if /proc/sys/kernel/perf_event_paranoid < 0), 128KiB for unprivileged users.
If an unprivileged user does not specify mmap pages, the mmap pages will be
reduced from the default 512KiB/page_size to 256KiB/page_size, otherwise the
user is likely to get an error as they exceed their mlock limit (Max locked
memory as shown in /proc/self/limits).  Note that perf does not count the first
512KiB (actually /proc/sys/kernel/perf_event_mlock_kb minus 1 page) per cpu
against the mlock limit so an unprivileged user is allowed 512KiB per cpu plus
their mlock limit (which defaults to 64KiB but is not multiplied by the number
of cpus).

In full-trace mode, powers of two are allowed for buffer size, with a minimum
size of 2 pages.  In snapshot mode, it is the same but the minimum size is
1 page.

The mmap size and auxtrace mmap size are displayed if the -vv option is used e.g.

	mmap length 528384
	auxtrace mmap length 4198400


Intel PT modes of operation
---------------------------

Intel PT can be used in 2 modes:
	full-trace mode
	snapshot mode

Full-trace mode traces continuously e.g.

	perf record -e intel_pt//u uname

Snapshot mode captures the available data when a signal is sent e.g.

	perf record -v -e intel_pt//u -S ./loopy 1000000000 &
	[1] 11435
	kill -USR2 11435
	Recording AUX area tracing snapshot

Note that the signal sent is SIGUSR2.
Note that "Recording AUX area tracing snapshot" is displayed because the -v
option is used.

The 2 modes cannot be used together.


Buffer handling
---------------

There may be buffer limitations (i.e. single ToPa entry) which means that actual
buffer sizes are limited to powers of 2 up to 4MiB (MAX_ORDER).  In order to
provide other sizes, and in particular an arbitrarily large size, multiple
buffers are logically concatenated.  However an interrupt must be used to switch
between buffers.  That has two potential problems:
	a) the interrupt may not be handled in time so that the current buffer
	becomes full and some trace data is lost.
	b) the interrupts may slow the system and affect the performance
	results.

If trace data is lost, the driver sets 'truncated' in the PERF_RECORD_AUX event
which the tools report as an error.

In full-trace mode, the driver waits for data to be copied out before allowing
the (logical) buffer to wrap-around.  If data is not copied out quickly enough,
again 'truncated' is set in the PERF_RECORD_AUX event.  If the driver has to
wait, the intel_pt event gets disabled.  Because it is difficult to know when
that happens, perf tools always re-enable the intel_pt event after copying out
data.


Intel PT and build ids
----------------------

By default "perf record" post-processes the event stream to find all build ids
for executables for all addresses sampled.  Deliberately, Intel PT is not
decoded for that purpose (it would take too long).  Instead the build ids for
all executables encountered (due to mmap, comm or task events) are included
in the perf.data file.

To see buildids included in the perf.data file use the command:

	perf buildid-list

If the perf.data file contains Intel PT data, that is the same as:

	perf buildid-list --with-hits


Snapshot mode and event disabling
---------------------------------

In order to make a snapshot, the intel_pt event is disabled using an IOCTL,
namely PERF_EVENT_IOC_DISABLE.  However doing that can also disable the
collection of side-band information.  In order to prevent that,  a dummy
software event has been introduced that permits tracking events (like mmaps) to
continue to be recorded while intel_pt is disabled.  That is important to ensure
there is complete side-band information to allow the decoding of subsequent
snapshots.

A test has been created for that.  To find the test:

	perf test list
	...
	23: Test using a dummy software event to keep tracking

To run the test:

	perf test 23
	23: Test using a dummy software event to keep tracking     : Ok


perf record modes (nothing new here)
------------------------------------

perf record essentially operates in one of three modes:
	per thread
	per cpu
	workload only

"per thread" mode is selected by -t or by --per-thread (with -p or -u or just a
workload).
"per cpu" is selected by -C or -a.
"workload only" mode is selected by not using the other options but providing a
command to run (i.e. the workload).

In per-thread mode an exact list of threads is traced.  There is no inheritance.
Each thread has its own event buffer.

In per-cpu mode all processes (or processes from the selected cgroup i.e. -G
option, or processes selected with -p or -u) are traced.  Each cpu has its own
buffer. Inheritance is allowed.

In workload-only mode, the workload is traced but with per-cpu buffers.
Inheritance is allowed.  Note that you can now trace a workload in per-thread
mode by using the --per-thread option.


Privileged vs non-privileged users
----------------------------------

Unless /proc/sys/kernel/perf_event_paranoid is set to -1, unprivileged users
have memory limits imposed upon them.  That affects what buffer sizes they can
have as outlined above.

The v4.2 kernel introduced support for a context switch metadata event,
PERF_RECORD_SWITCH, which allows unprivileged users to see when their processes
are scheduled out and in, just not by whom, which is left for the
PERF_RECORD_SWITCH_CPU_WIDE, that is only accessible in system wide context,
which in turn requires CAP_SYS_ADMIN.

Please see the 45ac1403f564 ("perf: Add PERF_RECORD_SWITCH to indicate context
switches") commit, that introduces these metadata events for further info.

When working with kernels < v4.2, the following considerations must be taken,
as the sched:sched_switch tracepoints will be used to receive such information:

Unless /proc/sys/kernel/perf_event_paranoid is set to -1, unprivileged users are
not permitted to use tracepoints which means there is insufficient side-band
information to decode Intel PT in per-cpu mode, and potentially workload-only
mode too if the workload creates new processes.

Note also, that to use tracepoints, read-access to debugfs is required.  So if
debugfs is not mounted or the user does not have read-access, it will again not
be possible to decode Intel PT in per-cpu mode.


sched_switch tracepoint
-----------------------

The sched_switch tracepoint is used to provide side-band data for Intel PT
decoding in kernels where the PERF_RECORD_SWITCH metadata event isn't
available.

The sched_switch events are automatically added. e.g. the second event shown
below:

	$ perf record -vv -e intel_pt//u uname
	------------------------------------------------------------
	perf_event_attr:
	type                             6
	size                             112
	config                           0x400
	{ sample_period, sample_freq }   1
	sample_type                      IP|TID|TIME|CPU|IDENTIFIER
	read_format                      ID
	disabled                         1
	inherit                          1
	exclude_kernel                   1
	exclude_hv                       1
	enable_on_exec                   1
	sample_id_all                    1
	------------------------------------------------------------
	sys_perf_event_open: pid 31104  cpu 0  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 1  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 2  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 3  group_fd -1  flags 0x8
	------------------------------------------------------------
	perf_event_attr:
	type                             2
	size                             112
	config                           0x108
	{ sample_period, sample_freq }   1
	sample_type                      IP|TID|TIME|CPU|PERIOD|RAW|IDENTIFIER
	read_format                      ID
	inherit                          1
	sample_id_all                    1
	exclude_guest                    1
	------------------------------------------------------------
	sys_perf_event_open: pid -1  cpu 0  group_fd -1  flags 0x8
	sys_perf_event_open: pid -1  cpu 1  group_fd -1  flags 0x8
	sys_perf_event_open: pid -1  cpu 2  group_fd -1  flags 0x8
	sys_perf_event_open: pid -1  cpu 3  group_fd -1  flags 0x8
	------------------------------------------------------------
	perf_event_attr:
	type                             1
	size                             112
	config                           0x9
	{ sample_period, sample_freq }   1
	sample_type                      IP|TID|TIME|IDENTIFIER
	read_format                      ID
	disabled                         1
	inherit                          1
	exclude_kernel                   1
	exclude_hv                       1
	mmap                             1
	comm                             1
	enable_on_exec                   1
	task                             1
	sample_id_all                    1
	mmap2                            1
	comm_exec                        1
	------------------------------------------------------------
	sys_perf_event_open: pid 31104  cpu 0  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 1  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 2  group_fd -1  flags 0x8
	sys_perf_event_open: pid 31104  cpu 3  group_fd -1  flags 0x8
	mmap size 528384B
	AUX area mmap length 4194304
	perf event ring buffer mmapped per cpu
	Synthesizing auxtrace information
	Linux
	[ perf record: Woken up 1 times to write data ]
	[ perf record: Captured and wrote 0.042 MB perf.data ]

Note, the sched_switch event is only added if the user is permitted to use it
and only in per-cpu mode.

Note also, the sched_switch event is only added if TSC packets are requested.
That is because, in the absence of timing information, the sched_switch events
cannot be matched against the Intel PT trace.


perf script
===========

By default, perf script will decode trace data found in the perf.data file.
This can be further controlled by new option --itrace.


New --itrace option
-------------------

Having no option is the same as

	--itrace

which, in turn, is the same as

	--itrace=ibxe

The letters are:

	i	synthesize "instructions" events
	b	synthesize "branches" events
	x	synthesize "transactions" events
	c	synthesize branches events (calls only)
	r	synthesize branches events (returns only)
	e	synthesize tracing error events
	d	create a debug log
	g	synthesize a call chain (use with i or x)
	l	synthesize last branch entries (use with i or x)
	s	skip initial number of events

"Instructions" events look like they were recorded by "perf record -e
instructions".

"Branches" events look like they were recorded by "perf record -e branches". "c"
and "r" can be combined to get calls and returns.

"Transactions" events correspond to the start or end of transactions. The
'flags' field can be used in perf script to determine whether the event is a
tranasaction start, commit or abort.

Error events are new.  They show where the decoder lost the trace.  Error events
are quite important.  Users must know if what they are seeing is a complete
picture or not.

The "d" option will cause the creation of a file "intel_pt.log" containing all
decoded packets and instructions.  Note that this option slows down the decoder
and that the resulting file may be very large.

In addition, the period of the "instructions" event can be specified. e.g.

	--itrace=i10us

sets the period to 10us i.e. one  instruction sample is synthesized for each 10
microseconds of trace.  Alternatives to "us" are "ms" (milliseconds),
"ns" (nanoseconds), "t" (TSC ticks) or "i" (instructions).

"ms", "us" and "ns" are converted to TSC ticks.

The timing information included with Intel PT does not give the time of every
instruction.  Consequently, for the purpose of sampling, the decoder estimates
the time since the last timing packet based on 1 tick per instruction.  The time
on the sample is *not* adjusted and reflects the last known value of TSC.

For Intel PT, the default period is 100us.

Setting it to a zero period means "as often as possible".

In the case of Intel PT that is the same as a period of 1 and a unit of
'instructions' (i.e. --itrace=i1i).

Also the call chain size (default 16, max. 1024) for instructions or
transactions events can be specified. e.g.

	--itrace=ig32
	--itrace=xg32

Also the number of last branch entries (default 64, max. 1024) for instructions or
transactions events can be specified. e.g.

       --itrace=il10
       --itrace=xl10

Note that last branch entries are cleared for each sample, so there is no overlap
from one sample to the next.

To disable trace decoding entirely, use the option --no-itrace.

It is also possible to skip events generated (instructions, branches, transactions)
at the beginning. This is useful to ignore initialization code.

	--itrace=i0nss1000000

skips the first million instructions.

dump option
-----------

perf script has an option (-D) to "dump" the events i.e. display the binary
data.

When -D is used, Intel PT packets are displayed.  The packet decoder does not
pay attention to PSB packets, but just decodes the bytes - so the packets seen
by the actual decoder may not be identical in places where the data is corrupt.
One example of that would be when the buffer-switching interrupt has been too
slow, and the buffer has been filled completely.  In that case, the last packet
in the buffer might be truncated and immediately followed by a PSB as the trace
continues in the next buffer.

To disable the display of Intel PT packets, combine the -D option with
--no-itrace.


perf report
===========

By default, perf report will decode trace data found in the perf.data file.
This can be further controlled by new option --itrace exactly the same as
perf script, with the exception that the default is --itrace=igxe.


perf inject
===========

perf inject also accepts the --itrace option in which case tracing data is
removed and replaced with the synthesized events. e.g.

	perf inject --itrace -i perf.data -o perf.data.new

Below is an example of using Intel PT with autofdo.  It requires autofdo
(https://github.com/google/autofdo) and gcc version 5.  The bubble
sort example is from the AutoFDO tutorial (https://gcc.gnu.org/wiki/AutoFDO/Tutorial)
amended to take the number of elements as a parameter.

	$ gcc-5 -O3 sort.c -o sort_optimized
	$ ./sort_optimized 30000
	Bubble sorting array of 30000 elements
	2254 ms

	$ cat ~/.perfconfig
	[intel-pt]
		mispred-all

	$ perf record -e intel_pt//u ./sort 3000
	Bubble sorting array of 3000 elements
	58 ms
	[ perf record: Woken up 2 times to write data ]
	[ perf record: Captured and wrote 3.939 MB perf.data ]
	$ perf inject -i perf.data -o inj --itrace=i100usle --strip
	$ ./create_gcov --binary=./sort --profile=inj --gcov=sort.gcov -gcov_version=1
	$ gcc-5 -O3 -fauto-profile=sort.gcov sort.c -o sort_autofdo
	$ ./sort_autofdo 30000
	Bubble sorting array of 30000 elements
	2155 ms

Note there is currently no advantage to using Intel PT instead of LBR, but
that may change in the future if greater use is made of the data.