Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > linux.kernel > #1689070
| From | Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> |
|---|---|
| Newsgroups | linux.kernel |
| Subject | Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. |
| Date | 2017-07-17 16:00 +0200 |
| Message-ID | <u4eJ4-5NI-7@gated-at.bofh.it> (permalink) |
| References | <u3PAZ-6Cr-9@gated-at.bofh.it> <u4a2L-2Ji-45@gated-at.bofh.it> |
| Organization | linux.* mail to news gateway |
Michal Hocko wrote: > On Sun 16-07-17 19:59:51, Tetsuo Handa wrote: > > Since the whole memory reclaim path has never been designed to handle the > > scheduling priority inversions, those locations which are assuming that > > execution of some code path shall eventually complete without using > > synchronization mechanisms can get stuck (livelock) due to scheduling > > priority inversions, for CPU time is not guaranteed to be yielded to some > > thread doing such code path. > > > > mutex_trylock() in __alloc_pages_may_oom() (waiting for oom_lock) and > > schedule_timeout_killable(1) in out_of_memory() (already held oom_lock) is > > one of such locations, and it was demonstrated using artificial stressing > > that the system gets stuck effectively forever because SCHED_IDLE priority > > thread is unable to resume execution at schedule_timeout_killable(1) if > > a lot of !SCHED_IDLE priority threads are wasting CPU time [1]. > > I do not understand this. All the contending tasks will go and sleep for > 1s. How can they preempt the lock holder? Not 1s. It sleeps for only 1 jiffies, which is 1ms if CONFIG_HZ=1000. And 1ms may not be long enough to allow the owner of oom_lock when there are many threads doing the same thing. I demonstrated that SCHED_IDLE oom_lock owner is completely defeated by a bunch of !SCHED_IDLE contending threads.
Back to linux.kernel | Previous | Next — Previous in thread | Next in thread | Find similar | Unroll thread
[PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2017-07-16 13:10 +0200
Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Michal Hocko <mhocko@kernel.org> - 2017-07-17 11:00 +0200
Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2017-07-17 16:00 +0200
Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Michal Hocko <mhocko@kernel.org> - 2017-07-17 17:20 +0200
Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Michal Hocko <mhocko@kernel.org> - 2017-07-17 17:30 +0200
Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> - 2017-07-17 23:50 +0200
Re: [PATCH v2] mm/page_alloc: Wait for oom_lock before retrying. Michal Hocko <mhocko@kernel.org> - 2017-07-18 11:10 +0200
csiph-web