> > Maybe. I scanned it very quickly but I think there are more problems > > than this one. I will check it once again. Thanks! > ieee80211_add_station - LOCKED by rcu_read_lock That's actually also locked by rtnl, and shouldn't be locked by rcu, the two statements have to be reordered. > ieee80211_del_station - this one is NOT LOCKED by rcu (it's locked by rtnl) That's the way it should be as explained in "STA information lifetime rules" in sta_info.c > ieee80211_open - NOT LOCKED actually locked by rtnl as most netdev callbacks, so fine. > ieee80211_associated - NOT-LOCKED by rcu not supposed to be, is under rtnl > ieee80211_ibss_add_sta - NOT LOCKED (not locked at all) That's a bit of a corner case, maybe we should allow this, or maybe this is a bug in that it should lock the rtnl although the STA never went live. The rtnl locking requirement comes from key management so this can't actually cause problems (since mac80211's ibss code can't do encryption) but it should still be fixed. > sta_info_debugfs_add_work - NOT LOCKED that probably needs locking. I'll look at making patches, thanks for the review. johannes
Attachment:
signature.asc
Description: This is a digitally signed message part
