On Fri, Mar 28, 2008 at 11:56:29AM +0100, Ingo Molnar wrote:

hm, what's the context of this discussion? The call chain looks ok, 
that's how we preempt tasks from the timer tick. But other code besides 
the scheduler shouldnt do this.

The code under discussion is __qdisc_run from net/sched/sch_generic.c.

It runs with BH off from either process context or softirq context.
As it is it can keep running forever.  We were discussing adding
a need_resched check in there.  So the question is would need_resched
ever get updated while BH is disabled?

Anyway, I've just realised that even if it does get updated, we still
need a better bound to avoid starving other local softirq events so
this is probably moot.

Thanks,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt