On Fri, 2024-08-23 at 18:26 +0200, Felix Fietkau wrote: > On 23.08.24 12:20, Johannes Berg wrote: > > On Mon, 2024-08-05 at 21:23 +0200, Felix Fietkau wrote: > > > This allows drivers to improve filtering of unwanted packets when using > > > different combinations of filter/monitor settings on radios. > > > > > > > Here, it would seem to make more sense to simply give the necessary per- > > interface information to the driver? Drivers likely have more > > restrictions and will need less complex data structures (or just > > iteration) to figure out what's eventually needed for each hw etc. > > > > Some drivers might anyway even have different filters per interface, > > even if they're operating on the same channel. > > I can do that as well. However, I do think it makes sense to have > per-radio tracking in mac80211 as well, because I'm pretty sure that not > just mt76, but ath12k would also be able to make use of this. But if you have per-link, deriving per-radio is effectively trivial, so I don't see why we should do the extra work to track it? johannes
