On Mon, Aug 10, 2009 at 10:28:43AM -0700, Paul E. McKenney wrote:
> On Mon, Aug 10, 2009 at 06:16:39PM +0400, Dmitry Eremin-Solenikov wrote:
> > Some of available devices are just dump radios implementing IEEE 802.15.4
> > PHY layer. This commit adds a common library that acts like an intermediate
> > layer between our socket family and drivers for those dumb devices.
> >
> > Currently this is data-only part (no commands, no beacons). Control
> > interfaces will follow up shortly.
> >
> > Note this implementaion is neither certified, nor feature complete!
>
> One question below, otherwise looks plausible. (I am not entirely sure
> which lists are which...)
>
> Thanx, Paul
[skipped]
>
> > +
> > +/*
> > + * This is for hw unregistration only, as it doesn't do RCU locking
>
> So this list is different than the RCU-protected one, and readers
> always hold locks when traversing it?
>
> I am not familiar with this code, so might be missing something, but
> it looks to me like the same list that RCU readers traverse. If so,
> need list_del_rcu() and synchronize_rcu().
On the first glance, yes. On the second glance, list_del instead of
list_del_rcu should be safe as there should be now other list traversals
at the same time. I'll think about it however.
>
> > + */
> > +void ieee802154_drop_slaves(struct ieee802154_dev *hw)
> > +{
> > + struct ieee802154_priv *priv = ieee802154_to_priv(hw);
> > + struct ieee802154_sub_if_data *sdata, *next;
> > +
> > + ASSERT_RTNL();
> > +
> > + list_for_each_entry_safe(sdata, next, &priv->slaves, list) {
> > + mutex_lock(&sdata->hw->slaves_mtx);
> > + list_del(&sdata->list);
> > + mutex_unlock(&sdata->hw->slaves_mtx);
> > +
> > + dev_put(sdata->hw->netdev);
> > +
> > + unregister_netdevice(sdata->dev);
> > + }
> > +}
--
With best wishes
Dmitry
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
