RE: Regression in reading /proc/stat in the newer kernels withlarge SMP and NUMA configurations

[Oberman, Laurence (HAS GSE)]

Hi Eric,
I am preparing for that test. Need to stage a full RHEL6. I had only updated the kernel on an existing SLES10.
Standby and I will run the tests for you.

Thanks
Laurence

-----Original Message-----
From: Eric Dumazet [mailto:eric.dumazet@xxxxxxxxx] 
Sent: Friday, October 14, 2011 9:47 AM
To: Seger, Mark
Cc: Oberman, Laurence (HAS GSE); linux-kernel@xxxxxxxxxxxxxxx; Cabaniols, Sebastien
Subject: RE: Regression in reading /proc/stat in the newer kernels with large SMP and NUMA configurations

Le vendredi 14 octobre 2011 Ã 14:36 +0100, Seger, Mark a Ãcrit :

> For me the easiest reproducer, which admittedly doesn't show any
> deeper analysis, is to just run "cat /proc/stat>/dev/null" in a loop
> and time it.

OK, but you already provided a global number.

We would like to check where in kernel cpu time is consumed.
Maybe something really obvious could pop out.

"perf" is provided in kernel sources, and really is a piece of cake.

cd tools/perf ; make 



2x4x2 means : Two sockets, 4 cores per physical package, 2 threads per
core.



èº{.nÇ+‰·Ÿ®‰­†+%ŠËlzwm…ébëæìr¸›zX§»®w¥Š{ayºÊÚë,j­¢f£¢·hš‹àz¹®w¥¢¸¢·¦j:+v‰¨ŠwèjØm¶Ÿÿ¾«‘êçzZ+ƒùšŽŠÝj"ú!¶iO•æ¬z·švØ^¶m§ÿðÃnÆàþY&—