> > Grumbach, Emmanuel wrote: > > Why are mgmt frames different from data frames with respect to that? > > You have a separated queue for mgmt frames that is common to all the > contexts? > > Just like an "offchannel" queue? > > No, ath9k treats them the same, but if the current, active context is different > from the incoming frame's assigned context, then it is added to a software > queue for deferred transmission. But, before a switch to the next context > can happen, mac80211 proceeds to tear down the BSS/stations etc. and the > VIF/context is also removed. > I see - we also had this problem and because of this, we implemented the flush callback that allows to wait until all the queues are empty. So mac80211 will send the deauth, flush and only then it'd remove the VIF. This should allow the frame to make it to the air. Note that API explicitly states that mgd_prepare_tx is before assoc only. -- 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
