Messages in this thread | | | Date | Thu, 26 Nov 2015 16:03:30 +0100 | From | Michal Hocko <> | Subject | Re: [PATCH] mm/vmstat: retrieve more accurate vmstat value |
| |
On Thu 26-11-15 10:56:12, Joonsoo Kim wrote: > On Wed, Nov 25, 2015 at 01:00:22PM +0100, Michal Hocko wrote: > > On Tue 24-11-15 15:22:03, Joonsoo Kim wrote: > > > When I tested compaction in low memory condition, I found that > > > my benchmark is stuck in congestion_wait() at shrink_inactive_list(). > > > This stuck last for 1 sec and after then it can escape. More investigation > > > shows that it is due to stale vmstat value. vmstat is updated every 1 sec > > > so it is stuck for 1 sec. > > > > Wouldn't it be sufficient to use zone_page_state_snapshot in > > too_many_isolated? > > Yes, it would work in this case. But, I prefer this patch because > all zone_page_state() users get this benefit.
Just to make it clear, I am not against your patch in general. I am just not sure it would help for too_many_isolated case where a significant drift might occur on remote cpus as well so I am not really sure that is appropriate for the issue you are seeing. -- Michal Hocko SUSE Labs
| |