Re: [PATCH] mm,oom: Don't call schedule_timeout_killable() with oom_lock held.

[Michal Hocko <mhocko@kernel.org>]

On Fri 01-06-18 14:11:10, Andrew Morton wrote:
> On Fri, 1 Jun 2018 17:28:01 +0200 Michal Hocko <mhocko@kernel.org> wrote:
> 
> > On Tue 29-05-18 16:07:00, Andrew Morton wrote:
> > > On Tue, 29 May 2018 09:17:41 +0200 Michal Hocko <mhocko@kernel.org> wrote:
> > > 
> > > > > I suggest applying
> > > > > this patch first, and then fix "mm, oom: cgroup-aware OOM killer" patch.
> > > > 
> > > > Well, I hope the whole pile gets merged in the upcoming merge window
> > > > rather than stall even more.
> > > 
> > > I'm more inclined to drop it all.  David has identified significant
> > > shortcomings and I'm not seeing a way of addressing those shortcomings
> > > in a backward-compatible fashion.  Therefore there is no way forward
> > > at present.
> > 
> > Well, I thought we have argued about those "shortcomings" back and forth
> > and expressed that they are not really a problem for workloads which are
> > going to use the feature. The backward compatibility has been explained
> > as well AFAICT.
> 
> Feel free to re-explain.  It's the only way we'll get there.

OK, I will go and my points to the last version of the patchset.

> David has proposed an alternative patchset.  IIRC Roman gave that a
> one-line positive response but I don't think it has seen a lot of
> attention?

I plan to go and revisit that. My preliminary feedback is that a more
generic policy API is really tricky and the patchset has many holes
there. But I will come with a more specific feedback in the respective
thread.

-- 
Michal Hocko
SUSE Labs