[PATCH 22/50] AP: MLO: Process EHT ML element from association request

[Otcheretianski, Andrei]

> Instead of manually building elements, can we add logic to intersect and add
> non inherited info alone in link sta profile.

I couldn't understand what you mean here.. Can you please better explain your comment?

> > +
> > +	/*
> > +	 * Get the AID from the station on which the association was
> > preformed,
> > +	 * and mark it as used.
> > +	 */
> > +	sta->aid = origin_sta->aid;
> > +	hapd->sta_aid[(sta->aid - 1) / 32] |= BIT((sta->aid - 1) % 32);
> AID generation logic is incomplete, need to consider if the aid is available in all
> the hapd links for which the association is requested.
> Current implementation may overwrite aid that is allocated to a different sta
> in link hapd.

You're right, this is a very good point.. We will need to adjust hostapd_get_aid() to properly handle MLD case.

> > +	sta->listen_interval = origin_sta->listen_interval;
> > +	update_ht_state(hapd, sta);
> > +
> > +	/* WPA authenticator should always be the one on the original
> > station */
> > +	wpa_auth_sta_deinit(sta->wpa_sm);
> > +	sta->wpa_sm = NULL;

> Should we restrict __check_assoc_ies to not create state machine for link
> stations instead?

Right. This makes sense. Will change.

> > +							      link-
> > >resp_sta_profile,
> > +							      sizeof(link-
> > >resp_sta_profile));
> > +		} else {
> > +			ieee80211_ml_process_link(iface->bss[0],
> > +						  sta, link,
> > +						  ies, ies_len, reassoc);
> > +		}
> > +	}
> > +#endif /* CONFIG_IEEE80211BE */
> >  }
> >
> >
> > @@ -5135,6 +5391,9 @@ static void handle_assoc(struct hostapd_data
> > *hapd,
> >  	    add_associated_sta(hapd, sta, reassoc))
> Shouldn't we add link Stations before moving ML station to ASSOC in kernel,
> that way the drivers will learn all partner link STA info before ML Station
> moves to associated state.
Sure.. Since the MLD station was added already in auth processing, it's possible to switch the order here.

Andrei