On Mon 29-08-22 12:40:05, Michal Hocko wrote:
> On Sun 28-08-22 13:50:09, Yu Zhao wrote:
> > On Tue, Aug 23, 2022 at 2:36 AM Michal Hocko <mhocko@xxxxxxxx> wrote:
> [...]
> > > You cannot really make any
> > > assumptions about oom_reaper and how quickly it is going to free the
> > > memory.
> >
> > Agreed. But here we are talking about heuristics, not dependencies on
> > certain behaviors. Assume we are playing a guessing game: there are
> > multiple mm_structs available for reclaim, would the oom-killed ones
> > be more profitable on average? I'd say no, because I assume it's more
> > likely than unlikely that the oom reaper is doing/to do its work. Note
> > that the assumption is about likelihood, hence arguably valid.
>
> Well, my main counter argument would be that we do not really want to
> carve last resort mechanism (which the oom reaper is) into any heuristic
> because any future changes into that mechanism will be much harder to
> justify and change. There is a cost of the maintenance that should be
> considered. While you might be right that this change would be
> beneficial, there is no actual proof of that. Historically we've had
> several examples of such a behavior which was really hard to change
> later on because the effect would be really hard to evaluate.
Forgot to mention the recent change as a clear example of the change
which would be have a higher burden to evaluate. e4a38402c36e
("oom_kill.c: futex: delay the OOM reaper to allow time for proper futex
cleanup") has changed the wake up logic to be triggered after a timeout.
This means that the task will be sitting there on the queue without any
actual reclaim done on it. The timeout itself can be changed in the
future and I would really hate to argue that changeing it from $FOO to
$FOO + epsilon breaks a very subtle dependency somewhere deep in the
reclaim path. From the oom reaper POV any timeout is reasonable becaude
this is the _last_ resort to resolve OOM stall/deadlock when the victim
cannot exit on its own for whatever reason. This is a considerably
different objective from "we want to optimize which taks to scan to
reclaim efficiently".
See my point?
--
Michal Hocko
SUSE Labs
