On Thu 22-03-18 09:06:14, Yang Shi wrote: > > > On 3/22/18 2:10 AM, Michal Hocko wrote: > > On Wed 21-03-18 15:36:12, Yang Shi wrote: > > > > > > On 3/21/18 2:23 PM, Michal Hocko wrote: [...] > > > > pages and that is quite easy to move out of the write lock. That would > > > > be an improvement already and it should be risk safe. If even that is > > > > not sufficient then using range locking should help a lot. There > > > > shouldn't be really any other address space operations within the range > > > > most of the time so this would be basically non-contended access. > > > It might depend on how the range is defined. Too big range may lead to > > > surprisingly more contention, but too small range may bring in too much > > > lock/unlock operations. > > The full vma will have to be range locked. So there is nothing small or large. > > It sounds not helpful to a single large vma case since just one range lock > for the vma, it sounds equal to mmap_sem. This is not how the range locking works. If we have a range lock per mm then exclusive ranges are not contending. So if you are unmapping one vma and want to create a new mapping or fault into a different range then you are basically lockless. -- Michal Hocko SUSE Labs
