linux/tools/perf/util
Arnaldo Carvalho de Melo 401b8e1317 perf tools: Fix thread_map event synthesizing in top and record
Jeff Moyer reported these messages:

  Warning:  ... trying to fall back to cpu-clock-ticks

couldn't open /proc/-1/status
couldn't open /proc/-1/maps
[ls output]
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.008 MB perf.data (~363 samples) ]

That lead me and David Ahern to see that something was fishy on the thread
synthesizing routines, at least for the case where the workload is started
from 'perf record', as -1 is the default for target_tid in 'perf record --tid'
parameter, so somehow we were trying to synthesize the PERF_RECORD_MMAP and
PERF_RECORD_COMM events for the thread -1, a bug.

So I investigated this and noticed that when we introduced support for
recording a process and its threads using --pid some bugs were introduced and
that the way to fix it was to instead of passing the target_tid to the event
synthesizing routines we should better pass the thread_map that has the list of
threads for a --pid or just the single thread for a --tid.

Checked in the following ways:

On a 8-way machine run cyclictest:

[root@emilia ~]# perf record cyclictest -a -t -n -p99 -i100 -d50
policy: fifo: loadavg: 0.00 0.13 0.31 2/139 28798

T: 0 (28791) P:99 I:100 C:  25072 Min:      4 Act:    5 Avg:    6 Max:     122
T: 1 (28792) P:98 I:150 C:  16715 Min:      4 Act:    6 Avg:    5 Max:      27
T: 2 (28793) P:97 I:200 C:  12534 Min:      4 Act:    5 Avg:    4 Max:       8
T: 3 (28794) P:96 I:250 C:  10028 Min:      4 Act:    5 Avg:    5 Max:      96
T: 4 (28795) P:95 I:300 C:   8357 Min:      5 Act:    6 Avg:    5 Max:      12
T: 5 (28796) P:94 I:350 C:   7163 Min:      5 Act:    6 Avg:    5 Max:      12
T: 6 (28797) P:93 I:400 C:   6267 Min:      4 Act:    5 Avg:    5 Max:       9
T: 7 (28798) P:92 I:450 C:   5571 Min:      4 Act:    5 Avg:    5 Max:       9
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.108 MB perf.data (~4719 samples) ]

[root@emilia ~]#

This will create one extra thread per CPU:

[root@emilia ~]# tuna -t cyclictest -CP
                      thread       ctxt_switches
    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
 28825   OTHER     0     0xff      2169          671      cyclictest
  28832   FIFO    93        6     52338            1      cyclictest
  28833   FIFO    92        7     46524            1      cyclictest
  28826   FIFO    99        0    209360            1      cyclictest
  28827   FIFO    98        1    139577            1      cyclictest
  28828   FIFO    97        2    104686            0      cyclictest
  28829   FIFO    96        3     83751            1      cyclictest
  28830   FIFO    95        4     69794            1      cyclictest
  28831   FIFO    94        5     59825            1      cyclictest
[root@emilia ~]#

So we should expect only samples for the above 9 threads when using the
--dump-raw-trace|-D perf report switch to look at the column with the tid:

[root@emilia ~]# perf report -D | grep RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort | uniq -c
    629 28825
    110 28826
    491 28827
    308 28828
    198 28829
    621 28830
    225 28831
    203 28832
     89 28833
[root@emilia ~]#

So for workloads started by 'perf record' seems to work, now for existing workloads,
just run cyclictest first, without 'perf record':

[root@emilia ~]# tuna -t cyclictest -CP
                      thread       ctxt_switches
    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
 28859   OTHER     0     0xff       594          200      cyclictest
  28864   FIFO    95        4     16587            1      cyclictest
  28865   FIFO    94        5     14219            1      cyclictest
  28866   FIFO    93        6     12443            0      cyclictest
  28867   FIFO    92        7     11062            1      cyclictest
  28860   FIFO    99        0     49779            1      cyclictest
  28861   FIFO    98        1     33190            1      cyclictest
  28862   FIFO    97        2     24895            1      cyclictest
  28863   FIFO    96        3     19918            1      cyclictest
[root@emilia ~]#

and then later did:

[root@emilia ~]# perf record --pid 28859 sleep 3
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.027 MB perf.data (~1195 samples) ]
[root@emilia ~]#

To collect 3 seconds worth of samples for pid 28859 and its children:

[root@emilia ~]# perf report -D | grep RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort | uniq -c
     15 28859
     33 28860
     19 28861
     13 28862
     13 28863
     10 28864
     11 28865
      9 28866
    255 28867
[root@emilia ~]#

Works, last thing is to check if looking at just one of those threads also works:

[root@emilia ~]# perf record --tid 28866 sleep 3
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.006 MB perf.data (~242 samples) ]
[root@emilia ~]# perf report -D | grep RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort | uniq -c
      3 28866
[root@emilia ~]#

Works too.

Reported-by: Jeff Moyer <jmoyer@redhat.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-02-10 12:52:47 -02:00
..
include perf tools: Add missing header, fixes build 2011-01-22 19:15:39 -02:00
scripting-engines perf: Rename 'perf trace' to 'perf script' 2010-11-16 19:37:44 +01:00
ui perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
abspath.c perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
alias.c perf_counter tools: Add more warnings and fix/annotate them 2009-07-01 12:49:48 +02:00
bitmap.c perf tools: Don't use code surrounded by __KERNEL__ 2010-05-02 12:00:44 -03:00
build-id.c perf session: Parse sample earlier 2010-12-04 23:05:19 -02:00
build-id.h perf annotate: Use build-ids to find the right DSO 2010-05-20 12:15:33 -03:00
cache.h perf tools: add test for strlcpy() 2010-08-21 11:22:47 -03:00
callchain.c perf: Support for callchains merge 2010-08-22 21:10:35 +02:00
callchain.h Merge branch 'perf/urgent' into perf/core 2010-08-27 02:30:07 +02:00
color.c perf hist: Replace ->print() routines by ->snprintf() equivalents 2010-04-02 16:28:15 -03:00
color.h perf hist: Replace ->print() routines by ->snprintf() equivalents 2010-04-02 16:28:15 -03:00
config.c perf buildid: add perfconfig option to specify buildid cache dir 2010-06-05 09:34:04 -03:00
cpumap.c perf tools: Refactor cpumap to hold nr and the map 2011-01-04 00:23:55 -02:00
cpumap.h perf tools: Refactor cpumap to hold nr and the map 2011-01-04 00:23:55 -02:00
ctype.c perf tools: Move graph_line and graph_dotted_line from top 2009-11-23 21:55:20 +01:00
debug.c perf debug: Simplify trace_event 2010-11-30 20:58:42 -02:00
debug.h perf tools: Fix lost and unknown events handling 2010-11-27 02:41:01 -02:00
debugfs.c perf trace: Clean up find_debugfs() 2009-12-28 10:36:36 +01:00
debugfs.h perf tools: Mount debugfs automatically 2009-12-28 10:36:36 +01:00
environment.c
event.c perf tools: Fix thread_map event synthesizing in top and record 2011-02-10 12:52:47 -02:00
event.h perf tools: Fix thread_map event synthesizing in top and record 2011-02-10 12:52:47 -02:00
evsel.c perf stat: Fix aggreate counter reading accounting 2011-02-03 17:26:06 -02:00
evsel.h perf tools: Pass whole attr to event selectors 2011-01-07 01:44:36 -02:00
exec_cmd.c perf tools: remove xstrndup, xmalloc, xzalloc 2010-05-18 23:05:28 -03:00
exec_cmd.h perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
generate-cmdlist.sh
header.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
header.h perf tools: Introduce event selectors 2011-01-03 16:39:04 -02:00
help.c perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
help.h perf tools: Protect header files with a consistent style 2009-09-24 21:27:51 +02:00
hist.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
hist.h perf session: Parse sample earlier 2010-12-04 23:05:19 -02:00
hweight.c perf tools: Don't use code surrounded by __KERNEL__ 2010-05-02 12:00:44 -03:00
levenshtein.c
levenshtein.h perf tools: Protect header files with a consistent style 2009-09-24 21:27:51 +02:00
map.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
map.h perf probe: Add basic module support 2010-10-21 16:11:44 -02:00
pager.c perf_counter tools: Remove dead code 2009-06-27 06:06:39 +02:00
parse-events.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
parse-events.h perf tools: Fix build when using gcc 3.4.6 2011-01-22 19:15:39 -02:00
parse-options.c perf options: Type check all the remaining OPT_ variants 2010-05-17 16:22:41 -03:00
parse-options.h perf options: add OPT_CALLBACK_DEFAULT_NOOPT 2010-12-06 15:33:29 -02:00
path.c perf tools: add test for strlcpy() 2010-08-21 11:22:47 -03:00
PERF-VERSION-GEN perf: Version String fix, for fallback if not from git 2010-07-05 10:42:58 +02:00
probe-event.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
probe-event.h perf probe: Add basic module support 2010-10-21 16:11:44 -02:00
probe-finder.c Merge commit 'v2.6.37-rc8' into perf/core 2011-01-04 08:08:54 +01:00
probe-finder.h perf tools: Remove hardcoded include paths for elfutils 2010-11-19 16:38:04 -02:00
pstack.c perf newt: Make <- zoom out filters 2010-05-14 20:05:21 -03:00
pstack.h perf ui: Move hists browser to util/ui/browsers/ 2010-08-10 16:11:08 -03:00
quote.c perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
quote.h perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
run-command.c perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
run-command.h perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
session.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
session.h perf util: Move do_read from session to util 2011-01-03 16:50:55 -02:00
sigchain.c perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
sigchain.h perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
sort.c perf hist: Better displaying of unresolved DSOs and symbols 2010-12-06 15:12:34 -02:00
sort.h perf: Keep track of the max depth of a callchain 2010-08-22 20:43:17 +02:00
strbuf.c perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
strbuf.h perf tools: Remove some unused functions 2010-05-18 23:03:35 -03:00
string.c perf tools: Fix lazy wildcard matching 2010-12-21 19:15:42 -02:00
strlist.c perf probe: Fix --del to update current event list 2009-12-15 20:22:01 +01:00
strlist.h perf tools: Add for_each macros for strlist 2009-12-15 20:22:02 +01:00
svghelper.c perf tools: Fix time function double declaration with glibc 2011-01-22 19:53:00 -02:00
svghelper.h perf tools: Protect header files with a consistent style 2009-09-24 21:27:51 +02:00
symbol.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
symbol.h Merge commit 'v2.6.37-rc8' into perf/core 2011-01-04 08:08:54 +01:00
thread.c perf tools: Refactor all_tids to hold nr and the map 2011-01-04 00:24:16 -02:00
thread.h perf tools: Refactor all_tids to hold nr and the map 2011-01-04 00:24:16 -02:00
trace-event-info.c perf tools: Introduce event selectors 2011-01-03 16:39:04 -02:00
trace-event-parse.c Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip 2010-05-18 08:19:03 -07:00
trace-event-read.c perf: Use read() instead of lseek() in trace_event_read.c:skip() 2010-05-20 08:37:17 +02:00
trace-event-scripting.c perf trace scripting: Fix extern struct definitions 2010-10-04 12:24:51 -03:00
trace-event.h perf tools: Introduce event selectors 2011-01-03 16:39:04 -02:00
types.h perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
usage.c perf top: Fall back to cpu-clock-tick hrtimer sampling if no cycle counter available 2009-06-07 17:31:52 +02:00
util.c perf util: Move do_read from session to util 2011-01-03 16:50:55 -02:00
util.h perf util: Move do_read from session to util 2011-01-03 16:50:55 -02:00
values.c perf tools: Fix 64 bit integer format strings 2011-01-22 23:41:57 -02:00
values.h perf tools: Protect header files with a consistent style 2009-09-24 21:27:51 +02:00
wrapper.c perf tools: remove xstrndup, xmalloc, xzalloc 2010-05-18 23:05:28 -03:00
xyarray.c perf tools: Introduce event selectors 2011-01-03 16:39:04 -02:00
xyarray.h perf tools: Introduce event selectors 2011-01-03 16:39:04 -02:00