Re: [PATCH V2 01/10] nl80211: add basic multiple bssid support

[Johannes Berg <johannes@sipsolutions.net>]

On Mon, 2020-07-06 at 13:52 +0200, John Crispin wrote:

>  /**
> @@ -5109,6 +5111,9 @@ struct cfg80211_cqm_config;
>   * @pmsr_free_wk: (private) peer measurements cleanup work
>   * @unprot_beacon_reported: (private) timestamp of last
>   *	unprotected beacon report
> + * @multi_bssid_mode: Is this a legacy, transmitted or non-transmitted bssid
> + * @multi_bssid_parent: a non-transmitted bssid has a transmitted parent
> + * @multi_bssid_list: linked list for tracking parent - child relations.
>   */
>  struct wireless_dev {
>  	struct wiphy *wiphy;
> @@ -5188,6 +5193,10 @@ struct wireless_dev {
>  	struct work_struct pmsr_free_wk;
>  
>  	unsigned long unprot_beacon_reported;
> +
> +	enum nl80211_multi_bssid_mode multi_bssid_mode;
> +	struct wireless_dev *multi_bssid_parent;
> +	struct list_head multi_bssid_list;

Ah, I forgot - this was where I thought it'd be better to change ...
after reading the later patches.

Or let me just ask: why is that here? You're passing it down in the
params (code I removed above), and then you're effectively making it the
driver's (mac80211's) responsibility to track it.

So why should it be here? cfg80211 isn't really using this much. I saw
there's a small validation thing here, but that could also be pushed
down, and then this can live in mac80211?

Or do you have any future plans for using this in cfg80211?

johannes