Re: Rescheduling interrupts

From: Vegard Nossum
Date: Tue May 13 2008 - 15:04:22 EST


On Tue, May 13, 2008 at 8:03 PM, Vegard Nossum <vegard.nossum@xxxxxxxxx> wrote:
> Hi,
>
>
> On Tue, Jan 22, 2008 at 6:05 PM, Ingo Molnar <mingo@xxxxxxx> wrote:
> >
> > * S.ÃaÄlar Onur <caglar@xxxxxxxxxxxxx> wrote:
> >
> >
> > > > My theory is that for whatever reason we get "repeat" IPIs: multiple
> > > > reschedule IPIs although the other CPU only initiated one.
> > >
> > > Ok, please see http://cekirdek.pardus.org.tr/~caglar/dmesg.3rd :)
> >
> > hm, the IPI sending and receiving is nicely paired up:
> >
> > [ 625.795008] IPI (@smp_reschedule_interrupt) from task swapper:0 on CPU#1:
> > [ 625.795223] IPI (@native_smp_send_reschedule) from task amarokapp:2882 on CPU#1:
> >
> > amarokapp does wake up threads every 20 microseconds - that could
> > explain it. It's probably Xorg running on one core, amarokapp on the
> > other core. That's already 100 reschedules/sec.
>
> Was this IPI thing ever resolved/fixed? I'm seeing the exact same
> thing on my newest Fedora 8 kernel (2.6.24.5-85.fc8) now:
>
> 33.3% (334.0) <kernel IPI> : Rescheduling interrupts
>
> And I know for sure that this did not show up a couple of months back.
>
> If the cause of this is known, please let me know about it. If it
> isn't, I will start a bisect run now :-)

Ok -- just a confirmation. I have this on v2.6.26-rc1:

Running only init=/bin/sh (from busybox) and powertop shows:
kernel IPI : Rescheduling interrupts

completely idle box (except powertop): 1 wakeups/sec
one busy-loop[1]: 2 wakeups/sec
two busy-loops: 500 wakeups/sec

So it seems like a real bug to me.


Vegard

[1]: while true; do echo hi; done

--
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
-- E. W. Dijkstra, EWD1036
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§ÿðà nÆàþY&—