#systemtap on 2017-07-07 — irc logs at freenode.irclog.whitequark.org

2015-11-12 23:18 fche changed the topic of #systemtap to: http://sourceware.org/systemtap; email systemtap@sourceware.org if answers here not timely, conversations may be logged

00:50 pwithnall_ has quit [Quit: pwithnall_]

01:40 hpt has joined #systemtap

04:35 hpt_ has joined #systemtap

04:38 hpt has quit [Ping timeout: 260 seconds]

04:43 _whitelogger has joined #systemtap

05:04 sanoj has joined #systemtap

05:32 sanoj has quit [Ping timeout: 240 seconds]

05:32 naveen_ has joined #systemtap

05:43 sanoj has joined #systemtap

05:55 irker094 has quit [Quit: transmission timeout]

06:38 atomlin has quit [Ping timeout: 240 seconds]

06:38 orivej has quit [Read error: Connection reset by peer]

06:38 orivej has joined #systemtap

06:39 atomlin has joined #systemtap

06:53 naveen__ has joined #systemtap

06:55 naveen_ has quit [Ping timeout: 255 seconds]

07:54 mjw has joined #systemtap

08:31 pwithnall has joined #systemtap

08:41 pwithnall has quit [Quit: pwithnall]

08:45 orivej has quit [Ping timeout: 260 seconds]

08:47 orivej has joined #systemtap

09:12 hpt_ has quit [Quit: Lost terminal]

10:09 pwithnall_ has joined #systemtap

10:11 pwithnall_ has quit [Client Quit]

10:13 pwithnall has joined #systemtap

10:23 sanoj has quit [Ping timeout: 248 seconds]

10:27 sanoj has joined #systemtap

10:37 CME has quit [Read error: Connection reset by peer]

10:39 CME has joined #systemtap

10:46 orivej has quit [Ping timeout: 248 seconds]

10:57 CME has quit [Ping timeout: 260 seconds]

11:00 CME has joined #systemtap

11:09 nkambo_ has quit [Ping timeout: 240 seconds]

11:28 pwithnall_ has joined #systemtap

11:31 naveen_ has joined #systemtap

11:34 naveen__ has quit [Ping timeout: 276 seconds]

11:37 pwithnall_ has quit [Quit: pwithnall_]

11:45 naveen__ has joined #systemtap

11:48 naveen_ has quit [Ping timeout: 240 seconds]

12:18 orivej has joined #systemtap

12:21 sanoj has quit [Quit: Leaving]

12:27 nkambo_ has joined #systemtap

12:29 naveen__ has quit [Quit: WeeChat 1.9]

12:46 brolley has joined #systemtap

12:46 wcohen has quit [Ping timeout: 260 seconds]

12:50 drsmith_away is now known as drsmith

13:02 mbenitez has joined #systemtap

13:02 mbenitez has quit [Changing host]

13:07 gpiccoli has joined #systemtap

13:31 tromey has joined #systemtap

13:40 wcohen has joined #systemtap

13:40 mjw has quit [Quit: Leaving]

13:56 nkambo__ has joined #systemtap

13:58 nkambo_ has quit [Ping timeout: 260 seconds]

14:21 mjw has joined #systemtap

14:26 CME has quit [Quit: ZNC - http://znc.sourceforge.net]

14:26 CME has joined #systemtap

14:28 eichiro has quit [Ping timeout: 260 seconds]

14:30 slowfranklin has joined #systemtap

14:30 <slowfranklin> Hey folks!

14:31 <slowfranklin> Another question: how can I probe static functions by name if there are multiple versions and I only want to probe one?

14:31 <slowfranklin> Ie I have:

14:31 <slowfranklin> (gdb) l query_user_list

14:31 <slowfranklin> file: "../source3/winbindd/winbindd_ads.c", line number: 295

14:31 <slowfranklin> file: "../source3/winbindd/winbindd_reconnect.c", line number: 87

14:31 <slowfranklin> file: "../source3/winbindd/winbindd_reconnect_ads.c", line number: 38

14:32 <slowfranklin> I tried

14:32 <slowfranklin> probe process("winbindd").function("query_user_list@../source3/winbindd/winbindd_ads.c")

14:32 <fche> tried probe foo.function("bar@file.c") ?

14:32 <slowfranklin> but that doesn't seem to work

14:32 <fche> hmm, it should

14:32 <slowfranklin> really? ok. I'll try again.

14:33 <slowfranklin> There's a chance that the functions are just not called in my scenario.

14:33 eichiro has joined #systemtap

14:34 <fche> there may be a bug that prevents proper filtering from working, but yeah it should work that way

14:36 <mjw> slowfranklin, If you use stap -L 'process("winbindd").function("query_user_list@*")' it should show you the variants stap can see/find.

14:36 <slowfranklin> ahhh, more magic. Awesome! ;)

14:36 <mjw> of -l if you don't care about the arguments

14:36 * slowfranklin checking...

14:38 <slowfranklin> https://cpaste.org/?92b2490f3a20bb8d#cfzOgAXYk8O4TJXu0r5XyfSc2mOoU5aApDVe+ElwtAI=

14:38 <slowfranklin> Has the line number in the probe eg

14:38 <slowfranklin> process("/home/slow/git/samba/scratch/bin/default/source3/winbindd/winbindd").function("query_user_list@../source3/winbindd/winbindd_ads.c:292")

14:39 <slowfranklin> I wanted to avoid the line numer so the script works regardless of source code changes...

14:39 <mjw> I don't think the :line is required

14:39 <mjw> if it is you can surely use :*

14:41 <slowfranklin> ah...

14:41 <mjw> I think you can use lots of wildcards, so just query_user_list@*windbindd_ads* should also work I think

14:41 <mjw> I think a lot apparently...

14:41 * slowfranklin trying to use * for the line numbers

14:43 <slowfranklin> Damn :)

14:43 <slowfranklin> WARNING: For probing a particular line, use a .statement() probe, not .function(): keyword at script/winbindd.stp:570:1

14:43 <slowfranklin> source: probe process("winbindd").function("query_user_list@../source3/winbindd/winbindd_ads.c:*") {

14:43 <slowfranklin> ^

14:43 <fche> ignore it, trying to be helpful

14:44 <fche> other than the warning, is the output the same with/without the :* ?

14:44 * slowfranklin checking...

14:45 <slowfranklin> Hm, looks like the probe points still aren't triggered.

14:46 <slowfranklin> But let me do some sanity checks first, maybe I'm doing something wrong somewhere else.

14:47 <fche> but stap is identifying the probe point OK? (stap -t can help verify)

14:48 <slowfranklin> Will check

14:48 <slowfranklin> Fwiw, the script is here:

14:48 <slowfranklin> https://git.samba.org/?p=slow/samba.git;a=shortlog;h=refs/heads/perf

14:48 <slowfranklin> https://git.samba.org/?p=slow/samba.git;a=blob;f=script/winbindd.stp;h=6a7dfdede45f394c325bdbf4de974b17ff9a1c21;hb=962460a578d0149f620ecedf4417731d16111da2

14:48 <slowfranklin> Actually generated via a shell script:

14:48 <slowfranklin> https://git.samba.org/?p=slow/samba.git;a=blob;f=script/generate-winbindd.stp.sh;h=8379fe057f0a68836778ebffa975b1b3a0382811;hb=962460a578d0149f620ecedf4417731d16111da2

15:10 orivej has quit [Ping timeout: 258 seconds]

15:26 orivej has joined #systemtap

15:46 naveen has joined #systemtap

15:50 nkambo__ has quit [Ping timeout: 240 seconds]

16:02 nkambo__ has joined #systemtap

16:12 orivej has quit [Remote host closed the connection]

16:13 mjw has quit [Quit: Leaving]

16:13 orivej has joined #systemtap

16:15 mjw has joined #systemtap

16:21 slowfranklin has quit [Quit: slowfranklin]

16:26 slowfranklin has joined #systemtap

16:28 eichiro has quit [Ping timeout: 260 seconds]

16:29 eichiro has joined #systemtap

16:34 orivej has quit [Ping timeout: 260 seconds]

17:04 naveen_ has joined #systemtap

17:06 naveen has quit [Ping timeout: 240 seconds]

17:07 * slowfranklin facepalm

17:07 <slowfranklin> Wrong variable in one place... It works now. :)

17:08 <slowfranklin> Another question:

17:09 <slowfranklin> how to interpret the timing output from -t to infer the overhead associated with instrumenting a process?

17:10 <slowfranklin> Example: https://cpaste.org/?004932552d43c07e#9Wz68l8mNRRcPDT2lCGai9eJVCbJqL8aU1Zexql1l8E=

17:14 <slowfranklin> What unit? Hopefully not microseconds. :O

17:16 <mjw> It is machine cycles

17:17 brolley has left #systemtap [#systemtap]

17:18 <mjw> And no, I don't fully understand what that means either

17:18 <slowfranklin> rofl, I was just about to ask

17:21 <mjw> There is lots of voodoo in runtime/time.c

17:23 <mjw> I believe the idea is that cpus/cores change frequency so often that you don't want to report something in n/m-secs since that is completely dependent on which core took the hit and how "hot" it was.

17:24 <slowfranklin> So it's basically TSC and I do the math based on clock frequency to get some numbers

17:26 slowfranklin has quit [Quit: slowfranklin]

17:29 slowfranklin has joined #systemtap

17:30 <slowfranklin> huh, sorry, lunch is ready. :) Thanks and cu!

17:30 slowfranklin has quit [Client Quit]

17:31 Humble has quit [Read error: Connection reset by peer]

17:32 Humble has joined #systemtap

17:49 orivej has joined #systemtap

17:57 pwithnall has quit [Ping timeout: 268 seconds]

18:02 slowfranklin has joined #systemtap

18:07 slowfranklin has quit [Quit: slowfranklin]

18:11 slowfranklin has joined #systemtap

18:56 orivej has quit [Ping timeout: 255 seconds]

18:57 brolley has joined #systemtap

19:03 naveen_ has quit [Quit: WeeChat 1.9]

19:07 slowfranklin has quit [Quit: slowfranklin]

19:56 slowfranklin has joined #systemtap

19:57 gromero has joined #systemtap

20:18 drsmith is now known as drsmith_away

20:21 slowfranklin has quit [Quit: slowfranklin]

20:24 wcohen has quit [Ping timeout: 240 seconds]

20:25 gila has joined #systemtap

20:26 brolley has left #systemtap [#systemtap]

20:34 scox has quit [Ping timeout: 248 seconds]

21:03 tromey has quit [Quit: ERC (IRC client for Emacs 26.0.50)]

21:04 mbenitez has quit [Quit: Leaving]

21:05 wcohen has joined #systemtap

22:01 gpiccoli has quit [Quit: Leaving]