From: Johannes Berg <johannes@xxxxxxxxxxxxxxxx> Date: Wed, 31 Jul 2019 17:50:56 +0200 > There's a fair number of changes here, so I thought I'd get them out. > I've included two Intel driver cleanups because Luca is on vacation, > I'm covering for him, and doing it all in one tree let me merge all > of the patches at once (including mac80211 that depends on that); > Kalle is aware. > > Also, though this isn't very interesting yet, I've started looking at > weaning the wireless subsystem off the RTNL for all operations, as it > can cause significant lock contention, especially with slow USB devices. > The real patches for that are some way off, but one preparation here is > to use generic netlink's parallel_ops=true, to avoid trading one place > with contention for another in the future, and to avoid adding more > genl_family_attrbuf() usage (since that's no longer possible with the > parallel_ops setting). > > Please pull and let me know if there's any problem. Pulled, thanks Johannes. I'm sure people like Florian Westphal will also appreciate your RTNL elimination work...
