Hi Ivo,

quoted text
> Overall the rt2x00 changes are good, but I have a few suggestions :)

quoted text
> >  	struct rt2x00_dev *rt2x00dev =3D hw->priv;
> >  	struct ieee80211_hdr *ieee80211hdr =3D (struct ieee80211_hdr *)skb->d=
ata;
quoted text
> > -	enum data_queue_qid qid =3D mac80211_queue_to_qid(control->queue);
> > +	enum data_queue_qid qid =3D skb_get_queue_mapping(skb);
> >  	struct data_queue *queue;

quoted text
> >  		if (rt2x00queue_available(queue) <=3D 1) {
> > -			ieee80211_stop_queue(rt2x00dev->hw, control->queue);
> > +			ieee80211_stop_queue(
> > +				rt2x00dev->hw, skb_get_queue_mapping(skb));
>=20
> ieee80211_stop_queue(rt2x00dev->hw, qid);
> The result of skb_get_queue_mapping(skb) was already stored in the
> 'qid' variable at the start of the function. And since this is a 1-1 mapp=
ing
quoted text
> from skb queue to rt2x00 queue, it is safe to use 'qid' directly in all i=
nstances.

Heh, good point, changed that. Will probably generate the same code
since get_queue_mapping is a static inline, but hey, much nicer to read
that way :)

johannes