On Tue, Mar 25, 2008 at 07:19:30AM -0700, Arjan van de Ven wrote:
> Colin Ian King wrote:
> > Hi,
> >
> > I've noticed that there a quite a few users reporting excessive amounts
> > of "Rescheduling Interrupts" from multi core machines, and it seems
> > concerting that the scheduler is pulling the processors out of C3 states
> > rather a lot. In some cases, there are reports of >10,000 Rescheduling
> > Interrupts a second, which seems excessive.
> >
> > The following rune apparently should help, but it does not make (much)
> > of a significant change:
> >
> > echo 1 > /sys/devices/system/cpu/sched_mc_power_savings
>
> this only does something on multi-socket systems, not on dual core single
> socket systems.
>
> >
> > Since it may be draining excessive amounts of power are there ideas if
> > this a scheduling problem and how to resolve this?
> > Thanks,
>
>
> it's a combination of two things:
> 1) An application or interrupt IS waking things up at that frequency.
> It's currently hard to track down which one; it may need kernel changes
> to be able to find out.
> 2) The scheduler lately seems very trigger happy to do a remote wakeup rather
> than
> local wakeups for the wakups done by 1)
And also, Colin, which kernel is this?
There were some recent bug fixes in mainline kernel.
For example.
commit 62fb185130e4d420f71a30ff59d8b16b74ef5d2b
Author: Peter Zijlstra <[EMAIL PROTECTED]>
Date: Mon Feb 25 17:34:02 2008 +0100
sched: revert load_balance_monitor() changes
The following commits cause a number of regressions:
commit 58e2d4ca581167c2a079f4ee02be2f0bc52e8729
Author: Srivatsa Vaddagiri <[EMAIL PROTECTED]>
Date: Fri Jan 25 21:08:00 2008 +0100
sched: group scheduling, change how cpu load is calculated
commit 6b2d7700266b9402e12824e11e0099ae6a4a6a79
Author: Srivatsa Vaddagiri <[EMAIL PROTECTED]>
Date: Fri Jan 25 21:08:00 2008 +0100
sched: group scheduler, fix fairness of cpu bandwidth allocation for task
groups
Namely:
- very frequent wakeups on SMP, reported by PowerTop users.
- cacheline trashing on (large) SMP
- some latencies larger than 500ms
While there is a mergeable patch to fix the latter, the former issues
are not fixable in a manner suitable for .25 (we're at -rc3 now).
Hence we revert them and try again in v2.6.26.
Signed-off-by: Peter Zijlstra <[EMAIL PROTECTED]>
CC: Srivatsa Vaddagiri <[EMAIL PROTECTED]>
Tested-by: Alexey Zaytsev <[EMAIL PROTECTED]>
Signed-off-by: Ingo Molnar <[EMAIL PROTECTED]>
_______________________________________________
Discuss mailing list
[email protected]
http://mail.lesswatts.org/mailman/listinfo/discuss
