On Wed, 2024-02-21 at 09:17 +0000, Ping-Ke Shih wrote: > > > -----Original Message----- > > From: Johannes Berg <johannes@xxxxxxxxxxxxxxxx> > > Sent: Wednesday, February 21, 2024 5:00 PM > > To: Ping-Ke Shih <pkshih@xxxxxxxxxxx>; Aditya Kumar Singh <quic_adisi@xxxxxxxxxxx> > > Cc: linux-wireless@xxxxxxxxxxxxxxx; Jeff Johnson <quic_jjohnson@xxxxxxxxxxx>; Ilan Peer > > <ilan.peer@xxxxxxxxx>; Jouni Malinen <j@xxxxx>; Ryder Lee <ryder.lee@xxxxxxxxxxxx>; Arend van Spriel > > <arend.vanspriel@xxxxxxxxxxxx>; Felix Fietkau <nbd@xxxxxxxx> > > Subject: Re: [PATCH v8 4/5] wifi: mac80211: start and finalize channel switch on link basis > > > > You _could_ also actually just implement probe response "offload" in the > > driver, then you get the template from hostapd/mac80211, and that should > > come with the offsets to fill in. Might be easier, overall. > > Agree. That will be easier. What I only concern is hostapd can still work well > without probe request, which is dropped by driver implementing offload. That's what happens today, so yeah, that should work? You need to not answer&drop certain P2P probe requests though. But since we're talking about this in WiFi7 AP context, you could just make it different depending on that, I think we get the templates all the time in the kernel and just ignore them if not handled. johannes
