Re: [PATCH 0/10 -v4] Handle oom bypass more gracefully

[Michal Hocko]

On Tue 14-06-16 22:17:40, Oleg Nesterov wrote:
> On 06/13, Michal Hocko wrote:
> >
> > On Mon 13-06-16 13:23:48, Michal Hocko wrote:
> > > On Thu 09-06-16 13:52:07, Michal Hocko wrote:
> > > > I would like to explore ways how to remove kthreads (use_mm) special
> > > > case. It shouldn't be that hard, we just have to teach the page fault
> > > > handler to recognize oom victim mm and enforce EFAULT for kthreads
> > > > which have borrowed that mm.
> > >
> > > So I was trying to come up with solution for this which would require to
> > > hook into the pagefault an enforce EFAULT when the mm is being reaped
> > > by the oom_repaer. Not hard but then I have checked the current users
> > > and none of them is really needing to read from the userspace (aka
> > > copy_from_user/get_user). So we actually do not need to do anything
> > > special.
> >
> > As pointed out by Tetsuo [1] vhost does realy on copy_from_user.
> 
> Tetsuo, Michal, but do we really care?
> 
> I have no idea what vhost does, but obviously this should not lead to kernel
> crash or something like this, otherwise it should be fixed. If we are going
> to kill the owner of dev->mm anyway, why should we worry about vhost_worker()
> which can fail to access this ->mm after that?

This needs a deeper investigation. It relies on some state flags copied
from the userspace. I suspect it might misbehave but let's leave this
alone for a while. It is more complicated than I expected.

-- 
Michal Hocko
SUSE Labs