All of lore.kernel.org
 help / color / mirror / Atom feed
* Uncore event for one process
@ 2015-04-07  5:57 Tianwei Zhang
  2015-04-10 12:14 ` Andi Kleen
  0 siblings, 1 reply; 2+ messages in thread
From: Tianwei Zhang @ 2015-04-07  5:57 UTC (permalink / raw)
  To: linux-perf-users

Hi,

I want to count the uncore events on a specified process id (Intel Xeon E5-2667
machine). I tried the following perf command to count the CAS_COUNT.RD for process 3460, but it gets stuck with the following output:

#perf stat -e uncore_imc_0/event=0x4,umask=0x3/,uncore_imc_1/event=0x4,umask=0x3/,uncore_imc_2/event=0x4,umask=0x3/,uncore_imc_3/event=0x4,umask=0x3/ -p 3460 sleep 1

Performance counter stats for process id '3460':

                 0 uncore_imc_0/event=0x4,umask=0x3/
                 0 uncore_imc_1/event=0x4,umask=0x3/
                 0 uncore_imc_2/event=0x4,umask=0x3/
                 0 uncore_imc_3/event=0x4,umask=0x3/

       1.000750918 seconds time elapsed

*** glibc detected *** /usr/lib/linux-tools/3.13.0-32-generic/perf: free(): invalid next size (fast): 0x0000000000aaee20 ***
*** glibc detected *** /usr/lib/linux-tools/3.13.0-32-generic/perf: malloc(): memory corruption: 0x0000000000aaf3d0 ***

However, if I switched the perf to the system-wide counting mode, it works:

#perf stat -e uncore_imc_0/event=0x4,umask=0x3/,uncore_imc_1/event=0x4,umask=0x3/,uncore_imc_2/event=0x4,umask=0x3/,uncore_imc_3/event=0x4,umask=0x3/ -a sleep 1

 Performance counter stats for 'system wide':

             4,487 uncore_imc_0/event=0x4,umask=0x3/
             3,721 uncore_imc_1/event=0x4,umask=0x3/
             4,885 uncore_imc_2/event=0x4,umask=0x3/
           212,513 uncore_imc_3/event=0x4,umask=0x3/

       1.000788447 seconds time elapsed

Can anyone help me with this problem?

Thank you very much.

--Tianwei

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

* Re: Uncore event for one process
  2015-04-07  5:57 Uncore event for one process Tianwei Zhang
@ 2015-04-10 12:14 ` Andi Kleen
  0 siblings, 0 replies; 2+ messages in thread
From: Andi Kleen @ 2015-04-10 12:14 UTC (permalink / raw)
  To: Tianwei Zhang; +Cc: linux-perf-users

Tianwei Zhang <artdream07@gmail.com> writes:
>
> Can anyone help me with this problem?

uncore monitoring is only supported socket wide.

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only

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

end of thread, other threads:[~2015-04-10 12:14 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-04-07  5:57 Uncore event for one process Tianwei Zhang
2015-04-10 12:14 ` Andi Kleen

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.