Messages in this thread | | | Subject | Re: [PATCH] mm, vmstat: Allow WQ concurrency to discover memory reclaim doesn't make any progress | From | Tetsuo Handa <> | Date | Wed, 25 Nov 2015 20:54:13 +0900 |
| |
Michal Hocko wrote: > Anyway I think that the issue is not solely theoretical. WQ_MEM_RECLAIM > is simply not working if the allocation path doesn't sleep currently and > my understanding of what Tejun claims [2] is that that reimplementing WQ > concurrency would be too intrusive and lacks sufficient justification > because other kernel paths do sleep. This patch tries to reduce the > sleep only to worker threads which should not cause any problems to > regular tasks.
I received many unexplained hangup/reboot reports from customers when I was working at support center. But we can't answer whether real people ever hit this problem because we have no watchdog for memory allocation stalls. I want one like http://lkml.kernel.org/r/201511250024.AAE78692.QVOtFFOSFOMLJH@I-love.SAKURA.ne.jp as I wrote off-list ( "mm,oom: The reason why I continue proposing timeout based approach." ). It will help with judging when we tackle TIF_MEMDIE livelock problem.
What I can say is that RHEL6 (a 2.6.32-based distro) backported the wait_iff_congested() changes and therefore people might really hit this problem.
| |