On Wed, 2024-02-21 at 08:57 +0000, Ping-Ke Shih wrote: > > > > > > No, firmware doesn't modify content of probe response frame. > > > > > > > Can you get that fixed? ;-) > > > > With differing beacon intervals etc., I don't know there's a good way to > > keep the counters even with a semblance of correctness, especially if we > > don't know when the beacon was transmitted? > > > > Or maybe just fill it in the driver since you probably have some beacon > > timing data more easily accessible? > > If driver can get CSA or ECSA offset simply, I probably can fill a reasonable > CSA counter (not sure if I can get 100% accurate counter for now), but > seemingly neither ieee80211_tx_control nor ieee80211_tx_info (SKB_CB) doesn't > have these offsets. Any suggestions? We're not there yet! This thread is debating how we want to handle it. Although .. you have a point, we have that issue now already, and we don't pass the CSA offsets in probe responses if hostapd is filling them in. I guess we also have work to do on this. > I wonder our out-of-tree driver generates probe response itself. Let's me > check how it does. 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. johannes
