On Mon, 2023-06-12 at 10:26 +0000, Greg Kroah-Hartman wrote: > From: Johannes Berg <johannes.berg@xxxxxxxxx> > > [ Upstream commit f7e60032c6618dfd643c7210d5cba2789e2de2e2 ] > > This should use wiphy_lock() now instead of requiring the > RTNL, since __cfg80211_leave() via cfg80211_leave() is now > requiring that lock to be held. You should perhaps hold off on this. While all this is correct, I missed something that Dan found later: https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless.git/commit/?id=996c3117dae4c02b38a3cb68e5c2aec9d907ec15 I'll have this in the next pull request. I suppose _both_ should go to stable, and nobody ever seems to run into this patch (at least lockdep would loudly complain), but stills seems better in the short term to have missing locking than a deadlock. johannes
