On Tue, 2017-06-20 at 23:09 +0300, Sergey Matyukevich wrote:
> >
> > > + if (sme->channel) {
> > > + /* FIXME: need to set proper nl80211_channel_type
> > > value */
> > > + cfg80211_chandef_create(&chandef, sme->channel,
> > > + NL80211_CHAN_HT20);
> > > + /* fall-back to minimal safe chandef description */
> > > + if (!cfg80211_chandef_valid(&chandef))
> > > + cfg80211_chandef_create(&chandef, sme-
> > > > channel,
> > >
> > > + NL80211_CHAN_HT20);
> > >
> >
> > This seems odd since you just do the same thing over again? Not
> > that I
> > could see how it would be invalid anyway.
>
> The first call of cfg80211_chandef_create will be replaced by proper
> chandef calculation based on current h/w channel settings. This piece
> is still in work. NL80211_CHAN_HT20 is going to be used as a safe
> fallback when channel info turns out to be inconsistent.
Yeah, ok. I guess I'd advocate doing that when the code is actually
there, but I suppose it doesn't really matter much.
johannes
