linux-perf-users.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* toplev would hang forever when analyzing a program with PID
@ 2017-03-26 14:07 Yijie Shen
  2017-03-27 20:35 ` Andi Kleen
  0 siblings, 1 reply; 2+ messages in thread
From: Yijie Shen @ 2017-03-26 14:07 UTC (permalink / raw)
  To: linux-perf-users

Hi all,

I'm planning to do some exploration on my program performance using
toplev, but faced with these problems:

when I called toplev.py -l1 ls, it stoped gracefully with the following output:

Will measure complete system.
Using level 1.
Nodes Backend_Bound Frontend_Bound have errata HSD135
perf stat -x\; -e
'{cpu/event=0x3c,umask=0x0,any=1/,cpu/event=0xe,umask=0x1/,cpu/event=0x9c,umask=0x1/,cpu/event=0xd,umask=0x3,any=1,cmask=1/,cpu/event=0xc2,umask=0x2/}'
-A -a ls

S0-C0    FE      Frontend_Bound:                51.94 % Slots
[100.00%] This category represents slots fraction where the
processor's Frontend undersupplies its Backend...
S0-C0    BE      Backend_Bound:                 24.73 % Slots
[100.00%] This category represents slots fraction where no uops are
being delivered due to a lack of required resources for
accepting new uops in the Backend...
S0-C1    FE      Frontend_Bound:                37.89 % Slots
[100.00%]S0-C1    BE      Backend_Bound:                 33.91 % Slots
    [100.00%]S1-C4    FE      Frontend_Bound:                82.60 %
Slots     [100.00%]S1-C4    RET     Retiring:
88.53 % Slots     [100.00%]


but toplev would hang forever when it was called to analyze a program using pid:

toplev.py -l1 -- --pid 29297
Will measure complete system.
Using level 1.
Nodes Backend_Bound Frontend_Bound have errata HSD135
perf stat -x\; -e
'{cpu/event=0x3c,umask=0x0,any=1/,cpu/event=0xe,umask=0x1/,cpu/event=0x9c,umask=0x1/,cpu/event=0xd,umask=0x3,any=1,cmask=1/,cpu/event=0xc2,umask=0x2/}'
-A -a --pid 29297


this would hang forever even if the process with such pid end, this
happens whatever the pid is pointing at is, a Java process or a Linux
tool such as top.


Am I doing something wrong? What does the right command look like to
analyze a program using PID?


Thanks in advance.

Yours,
Yijie

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: toplev would hang forever when analyzing a program with PID
  2017-03-26 14:07 toplev would hang forever when analyzing a program with PID Yijie Shen
@ 2017-03-27 20:35 ` Andi Kleen
  0 siblings, 0 replies; 2+ messages in thread
From: Andi Kleen @ 2017-03-27 20:35 UTC (permalink / raw)
  To: Yijie Shen; +Cc: linux-perf-users

Yijie Shen <henry.yijieshen@gmail.com> writes:
>
>
> this would hang forever even if the process with such pid end, this
> happens whatever the pid is pointing at is, a Java process or a Linux
> tool such as top.
>
>
> Am I doing something wrong? What does the right command look like to
> analyze a program using PID?

You're using SMT, so toplev has to enable global monitoring (-a),
but that does not interact well with --pid in perf.

That's more a perf bug .

It would work in single thread mode or without SMT, or using intervals.

toplev should probably print a warning in this case.

-Andi

>

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2017-03-27 20:35 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-03-26 14:07 toplev would hang forever when analyzing a program with PID Yijie Shen
2017-03-27 20:35 ` Andi Kleen

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).