[PATCH] Revert "wpa_supplicant: Send EAPOL frames over nl80211 where available"

[Krishna Chaitanya]

On Thu, Jul 20, 2023 at 12:08 AM Fabio Estevam <festevam at gmail.com> wrote:
>
> On Wed, Jul 19, 2023 at 3:20 PM Krishna Chaitanya
> <chaitanya.mgit at gmail.com> wrote:
>
> > Almost all searches for that message point to network manager:
> > https://wiki.archlinux.org/title/NetworkManager#Secrets_were_required,_but_not_provided
>
> Again, not relevant to my use case.
I am confused, are you saying that the log from the network manager is
not applicable for all scenarios?
>
> Please accept the commit from you and Brendan caused a regression on
> my ath10k system.
>
> It is 100% reproducible.
>
> > > "I have tested this on DMG (11ad/ay) devices with an out-of-tree Linux
> > >  driver that does not use mac80211."
> > Yes, it's tested using a cfg80211-based FullMAC driver.
> > > So it looks like this commit tried to fix an out-of-tree Linux driver,
> > > but broke ath10k.
> > mac80211 sets this flag by default and I just tested with
> > mac80211_hwsim and was successfully
> > able to establish the secure connection (WPA2 and WPA3).
>
> Good, but how do we go about solving the regression?
To confirm the issue, can you try with mac80211_hwsim? You can
run hostapd + wpa_supplicant and see if the issue is reproducible?