On Tue, Nov 4, 2008 at 4:37 AM, Bob Copeland <me@xxxxxxxxxxxxxxx> wrote: > On Mon, Nov 3, 2008 at 10:17 AM, Bob Copeland <me@xxxxxxxxxxxxxxx> wrote: >> On Sun, Nov 2, 2008 at 11:52 PM, Luis R. Rodriguez <mcgrof@xxxxxxxxx> wrote: >>>>>> Sorry, my fault, this was a direct result of >>>>>> 60c7e22196fb4230b76db1f5fb283e811b8f3fb3 "ath5k: honor >> [...] >>> OK Vasanth confirmed this is correct, and ath9k actually has incorrect >>> behavior so we have to fix that there as well. If stable got the ath5k >>> changes to the filter then yes we should push it there too. > > BTW perhaps we should still look at the filter for ath5k since, once > associated, we still get beacons from every AP in the world instead of > just those with our BSSID. > > FIF_BCN_PRBRESP_PROMISC > > This flag is set during scanning to indicate to the hardware > that it should not filter beacons or probe responses by BSSID. > Filtering them can greatly reduce the amount of processing > mac80211 needs to do and the amount of CPU wakeups, so > you should honour this flag if possible. > > Right now with ath5k there's basically no difference between having or > not having that flag in station mode, even though we do set the bssid > in ath5k_hw_set_associd AFAICT. Agreed, what we would need is as Johannes has been suggesting a way to implement this properly in mac80211. Come to think of it mac80211 also isn't aware of all the other filter flags ath5k/ath9k devices are capable of, I'd be inclined to move all that to mac80211 that way mac80211 *always* tells us the exact filter flags needed and we don't get some strange internal mode checks as we do now. I posted a strategy we can take in the other thread to handle the beacon filter in mac80211, let me know what you think. Luis -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
