On 11 August 2014 17:05, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote: > On Mon, 2014-07-28 at 15:16 +0200, Michal Kazior wrote: >> When a device driver is unloaded local->interfaces >> list is cleared. If there was more than 1 >> interface running and connected (bound to a >> chanctx) then chantype recalc was called and it >> ended up with compat being NULL causing a call >> trace warning. >> >> Warn if compat becomes NULL as a result of >> incompatible bss_conf.chandef of interfaces bound >> to a given channel context only. > > Err, this slipped my radar - where does this need to go? No big deal. The warning is just a small nuisance with multi-vif. You won't see this on a typical system. > > Also - is this really the right thing? Why does this happen in this > situation? Shouldn't they all be not connected or not have a chanctx or > be compatible? If you eject, e.g. plug a usb wifi device out, its driver will call ieee80211_unregister_hw() which calls ieee80211_remove_interfaces(). If there are still running interfaces that have a netdev (probably to avoid a deadlock on iflist_mtx?) each interface is *moved* to a temporary list for a later call to unregister_netdevice_many(). While unregister_netdevice_many() executes the local->interfaces is empty. ieee80211_recalc_chanctx_chantype() is called whenever a chanctx is unassigned from a vif as long as there are still references to that chanctx. If there is just 1 interface when device is ejected, then ieee80211_recalc_chanctx_chantype() is not called. If there is more than 1 running interfaces then the removal of each but the last one calls to ieee80211_recalc_chanctx_chantype(). Since local->interfaces is empty compat pointer stays NULL and this triggers WARN_ON_ONCE (but only the first one effectively dumps a trace). Another way to fix this would probably be to change ieee80211_remove_interfaces() to remove all interfaces one-by-one (instead of in bulk). It seems iflist_mtx can be temporarily released while iterating over interfaces for the sake of calling unregister_netdevice() for each one because RTNL is held all the time guaranteeing no new interfaces are added in the meantime. I'm not sure if it's perfectly safe to replace unregister_netdevice_many() just like that though. I can look more into it. Michał -- 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
