Re: [PATCH wireless] wifi: mac80211: abort chanswitch when leaving an IBSS

["Luka Gejak" <luka.gejak@linux.dev>]

Hi Mehmet,

Thanks for picking this up. One comment below.

October 25, 2026 at 8:36 PM, "Mehmet Fide" <mehmet.fide@gmail.com> wrote:

> diff --git a/net/mac80211/ibss.c b/net/mac80211/ibss.c
> index d74b66426349..b9ab4ceb0332 100644
> --- a/net/mac80211/ibss.c
> +++ b/net/mac80211/ibss.c
> @@ -669,6 +669,10 @@ static void ieee80211_ibss_disconnect(struct ieee80211_sub_if_data *sdata)
>       ifibss->state = IEEE80211_IBSS_MLME_SEARCH;
>
>       netif_carrier_off(sdata->dev);
> +
> +     sdata->vif.bss_conf.csa_active = false;
> +     ieee80211_vif_unblock_queues_csa(sdata);
> +

As far as I can see, the driver never finds out that the switch was
dropped here. ath9k_htc only clears priv->csa_vif once the countdown
completes with csa_active still set, or when the interface goes away.

So if I read it right, after leaving mid countdown and joining again, the
next switch with a countdown now gets past mac80211 and runs into this in
ath9k_htc_channel_switch_beacon():

        if (WARN_ON(priv->csa_vif))
                return;

Before this patch mac80211 stopped it earlier, so this couldn't happen in
IBSS. AP mode seems to have the same problem already after
ieee80211_stop_ap(), so I think it is better fixed in ath9k_htc, maybe by
dropping csa_vif when beaconing stops.

Since you have worked on ath9k already, would you be willing to send a
fix for that too?

Best regards,
Luka Gejak