Re: [PATCH 1/1] mm, oom_adj: don't loop through tasks in __set_oom_adj when not necessary

[Suren Baghdasaryan]

On Thu, Aug 20, 2020 at 9:29 AM Christian Brauner
<christian.brauner@xxxxxxxxxx> wrote:
>
> On Thu, Aug 20, 2020 at 06:26:45PM +0200, Michal Hocko wrote:
> > On Thu 20-08-20 08:56:53, Suren Baghdasaryan wrote:
> > [...]
> > > Catching up on the discussion which was going on while I was asleep...
> > > So it sounds like there is a consensus that oom_adj should be moved to
> > > mm_struct rather than trying to synchronize it among tasks sharing mm.
> > > That sounds reasonable to me too. Michal answered all the earlier
> > > questions about this patch, so I won't be reiterating them, thanks
> > > Michal. If any questions are still lingering about the original patch
> > > I'll be glad to answer them.
> >
> > I think it still makes some sense to go with a simpler (aka less tricky)
> > solution which would be your original patch with an incremental fix for
> > vfork and the proper ordering (http://lkml.kernel.org/r/20200820124109.GI5033@xxxxxxxxxxxxxx)
> > and then make a more complex shift to mm struct on top of that. The
> > former will be less tricky to backport to stable IMHO.
>
> /me nods

Ah, ok. Then I'll incorporate these changes, re-test and re-post as v2. Thanks!

>
> Christian