On Sun, 2005-01-30 at 19:31, Tommy Christensen wrote:
> jamal wrote:
> > Tommy,
> >
> > I wasnt sure why you did:
> >
> > dev->qdisc = &noop_qdisc;
> >
> > You should probably save the old qdisc in qdisc_sleeping instead
> > and restore it on wakeup - otherwise you always end with default qdisc.
>
> I think normally we will have qdisc == qdisc_sleeping. At least this
> is how I read the code in dev_graft_qdisc(). When can they differ?
You are correct.
> > check_carrier should probably just call dev_activate which does all you
> > wanted, no?
>
> I copied the logic from dev_deactivate(), but I didn't want to include
> the waiting parts, since this is not in process context. OK?
Same here.
It does look pretty sane.. Tested?
cheers,
jamal
|