Re: [PATCH] mm,oom: use per signal_struct flag rather than clear TIF_MEMDIE

[Oleg Nesterov <oleg@redhat.com>]

On 07/07, Michal Hocko wrote:
>
> On Sun 03-07-16 15:21:47, Oleg Nesterov wrote:
> > >
> > > I am not sure I can see security implications but I agree this is less
> > > than optimal,
> >
> > Well, just suppose that a memory hog execs a setuid application which does
> > something important, then we can kill it in some "inconsistent" state. Say,
> > after it created a file-lock which blocks other instances.
>
> How that would differ from selecting and killing the suid application
> right away?

in this case we at least check oom_score_adj/has_capability_noaudit(CAP_SYS_ADMIN)
before we decide to kill it.

> > And it is not clear to me why "child_points > victim_points" can be true if
> > the victim was chosen by select_bad_process() (to simplify the discussion,
> > lets ignore has_intersects_mems_allowed/etc).
>
> Because victim_points is a bit of misnomer. It doesn't have anything to
> do with selected victim's score. victim_points is 0 before the loop.

Ah, thanks. Yes I misread the code.

Oleg.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>