On Fri, Aug 21, 2015 at 09:14:58AM -0700, Eric Dumazet wrote:
[...]
> > diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
> > index 1fb65f9..ab614ee 100644
> > --- a/net/sched/sch_generic.c
> > +++ b/net/sched/sch_generic.c
> > @@ -634,6 +634,11 @@ struct Qdisc *qdisc_create_dflt(struct netdev_queue 
> > *dev_queue,
> >     if (IS_ERR(sch))
> >             goto errout;
> >     sch->parent = parentid;
> > +#ifdef CONFIG_NET_SCHED
> > +   sch->handle = qdisc_alloc_handle(dev_queue->dev);
> > +   if (!sch->handle)
> > +           goto errout;
> > +#endif
> >  
> >     if (!ops->init || ops->init(sch, NULL) == 0)
> >             return sch;
> 
> This might break HTB setups with more than 32768 classes ?

Urgh. Thanks for noticing this!

> The pfifo qdisc that gets attached had no handle.

Yes, looks like I need to leave qdisc_create_dflt() alone. It is
possible, by doing the above twice in sch_generic.c (once in
attach_one_default_qdisc(), and in attach_default_qdiscs() as well).

> qdisc_alloc_handle() has a limited range.

Yes, I noticed this. Handles aren't reused in a running system either,
which might contribute to this problem in other situations.

V2 will follow, thanks again.

Cheers, Phil
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to