Hi, On Tue 19-05-20 22:28:56, Michel Lespinasse wrote: > Reposting this patch series on top of v5.7-rc6. I think this is ready > for inclusion into the -mm tree; however there were some minor points > of feedback to address and also it was easier to regenerate a full > version after the v5.5 (only updating patches 09/10 and 10/10) caused > some confusion. > > > This patch series adds a new mmap locking API replacing the existing > mmap_sem lock and unlocks. Initially the API is just implemente in terms > of inlined rwsem calls, so it doesn't provide any new functionality. > > There are two justifications for the new API: > > - At first, it provides an easy hooking point to instrument mmap_sem > locking latencies independently of any other rwsems. > > - In the future, it may be a starting point for replacing the rwsem > implementation with a different one, such as range locks. This is > something that is being explored, even though there is no wide concensus > about this possible direction yet. > (see https://patchwork.kernel.org/cover/11401483/) Sorry that I am jumping late to the game. Even though I hoped that the new API would somehow start using a concept of address ranges, because I believe this is a fundamental lock granularity for this mmap_sem, this is definitely a step in the right direction. Getting rid of all direct usage of mmap_sem is an improvement in itself. Feel free to add Acked-by: Michal Hocko <mhocko@xxxxxxxx> Thanks! -- Michal Hocko SUSE Labs
