On Wed, 2023-03-29 at 21:06 -0700, Jakub Kicinski wrote: > On Wed, 29 Mar 2023 23:56:31 +0200 Johannes Berg wrote: > > > + drop_reasons_unregister_subsys(SKB_DROP_REASON_SUBSYS_MAC80211_MONITOR); > > > + drop_reasons_unregister_subsys(SKB_DROP_REASON_SUBSYS_MAC80211_UNUSABLE); > > > + > > > rcu_barrier(); > > > > This is making me think that perhaps we don't want synchronize_rcu() > > inside drop_reasons_unregister_subsys(), since I have two now and also > > already have an rcu_barrier() ... so maybe just document that it's > > needed? > > premature optimization? some workload is reloading mac80211 in a loop? Yeah, true, nobody is going to do that :-) Let's leave it this way. johannes
