On Wed 14-11-18 08:20:47, Wei Yang wrote: > On Wed, Nov 14, 2018 at 08:48:21AM +0100, Michal Hocko wrote: > >On Wed 14-11-18 07:43:41, Wei Yang wrote: > >> On Tue, Nov 13, 2018 at 10:07:58AM +0100, Michal Hocko wrote: > >> >On Tue 13-11-18 08:16:44, Wei Yang wrote: > >> > > >> >No, I believe we want all three of them. But reviewing > >> >for_each_populated_zone users and explicit checks for present/managed > >> >pages and unify them would be a step forward both a more optimal code > >> >and more maintainable code. I haven't checked but > >> >for_each_populated_zone would seem like a proper user for managed page > >> >counter. But that really requires to review all current users. > >> > > >> > >> To sync with your purpose, I searched the user of > >> for_each_populated_zone() and replace it with a new loop > >> for_each_managed_zone(). > > > >I do not think we really want a new iterator. Is there any users of > >for_each_populated_zone which would be interested in something else than > >managed pages? > > Your purpose is replace the populated_zone() in > for_each_populated_zone() with managed_zone()? Well, we might rename as well but I if we have only one or two users then an opencoded variant with populated_zone() check sounds better than a new iterator. > If this is the case, most of them is possible. Some places I am not sure > is: > > kernel/power/snapshot.c This one really looks like it wants the full pfn range whether it is managed or not. So changing this to opencoded for_each_zone + populated_zone check should be OK. > mm/huge_memory.c > mm/khugepaged.c These two are definitely page allocator related so they do care about managed. -- Michal Hocko SUSE Labs
