On Thu 26-09-19 09:26:13, David Hildenbrand wrote: [...] > I'd like to hear what Michal thinks. If we do want the cpu hotplug lock, > we can at least restrict it to the call paths (e.g., online_pages()) > where the lock is really needed and document that. Completely agreed. Conflating cpu and memory hotplug locks was a bad decision. If there are places which need both they should better use both lock explicitly. Now, the reality might turn out more complicated due to locks nesting but hiding the cpu lock into the mem hotplug is just not fixing that. -- Michal Hocko SUSE Labs
