Re: 2.5.74-mm1

[Anton Blanchard]

> Look at select_bad_process(), and the ->mm test in badness().  pdflush
> can never be chosen.
> 
> Nevertheless, there have been several report where kernel threads _are_ 
> being hit my the oom killer.  Any idea why that is?

Milton and I were just looking at this and it seems there is no locking
to prevent p->mm ending up NULL due to exit. And if p->mm does end up
NULL, you go off and kill all your kernel threads :)

Anton

	read_lock(&tasklist_lock);
	p = select_bad_process();

...

        oom_kill_task(p);
        /*
         * kill all processes that share the ->mm (i.e. all threads),
         * but are in a different thread group
         */
        do_each_thread(g, q)
                if (q->mm == p->mm && q->tgid != p->tgid)
                        oom_kill_task(q);
        while_each_thread(g, q);