lkml.org 
[lkml]   [2008]   [Jul]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH] sched: do not stop ticks when cpu is not idle

* Peter Zijlstra <a.p.zijlstra@chello.nl> wrote:

> On Fri, 2008-07-18 at 12:54 +0200, Ingo Molnar wrote:
> > * Ingo Molnar <mingo@elte.hu> wrote:
> >
> > > --- a/kernel/sched.c
> > > +++ b/kernel/sched.c
> > > @@ -4446,7 +4446,8 @@ need_resched_nonpreemptible:
> > > rq->nr_switches++;
> > > rq->curr = next;
> > > ++*switch_count;
> > > -
> > > + if (rq->curr != rq->idle)
> > > + tick_nohz_restart_sched_tick();
> > > context_switch(rq, prev, next); /* unlocks the rq */
> >
> > hm, one problem i can see is lock dependencies. This code is executed
> > with the rq lock while tick_nohz_restart_sched_tick() takes hr locks =>
> > not good. So i havent applied this just yet - this needs to be solved
> > differently.
>
> Actually, that should work these days...
>
> Also, I assume Eric actually tested this with lockdep enabled (right,
> Eric?) and that'll shout - or rather, lockup hard in this case - if
> you got it wrong.

nope:

[ 0.188011] =================================
[ 0.188011] [ INFO: inconsistent lock state ]
[ 0.188011] 2.6.26-tip-03835-g9d964b9-dirty #20198
[ 0.188011] ---------------------------------
[ 0.188011] inconsistent {in-hardirq-W} -> {hardirq-on-W} usage.
[ 0.188011] swapper/0 [HC0[0]:SC0[0]:HE1:SE1] takes:
[ 0.188011] (&rq->rq_lock_key){+...}, at: [<ffffffff816a4b35>] schedule+0x191/0x900
[ 0.188011] {in-hardirq-W} state was registered at:
[ 0.188011] [<ffffffffffffffff>] 0xffffffffffffffff
Ingo


\
 
 \ /
  Last update: 2008-07-19 00:31    [from the cache]
©2003-2011 Jasper Spaans