On Sun 03-07-16 11:45:34, Tetsuo Handa wrote:
> Michal Hocko wrote:
> > diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> > index 7d0a275df822..4ea4a649822d 100644
> > --- a/mm/oom_kill.c
> > +++ b/mm/oom_kill.c
> > @@ -286,16 +286,17 @@ enum oom_scan_t oom_scan_process_thread(struct oom_control *oc,
> > * Don't allow any other task to have access to the reserves unless
> > * the task has MMF_OOM_REAPED because chances that it would release
> > * any memory is quite low.
> > + * MMF_OOM_NOT_REAPABLE means that the oom_reaper backed off last time
> > + * so let it try again.
> > */
> > if (!is_sysrq_oom(oc) && atomic_read(&task->signal->oom_victims)) {
> > - struct task_struct *p = find_lock_task_mm(task);
> > + struct mm_struct *mm = task->signal->oom_mm;
> > enum oom_scan_t ret = OOM_SCAN_ABORT;
> >
> > - if (p) {
> > - if (test_bit(MMF_OOM_REAPED, &p->mm->flags))
> > - ret = OOM_SCAN_CONTINUE;
> > - task_unlock(p);
> > - }
> > + if (test_bit(MMF_OOM_REAPED, &mm->flags))
> > + ret = OOM_SCAN_CONTINUE;
> > + else if (test_bit(MMF_OOM_NOT_REAPABLE, &mm->flags))
> > + ret = OOM_SCAN_SELECT;
>
> I don't think this is useful.
Well, to be honest me neither but changing the retry logic is not in
scope of this patch. It just preserved the existing logic. I guess we
can get rid of it but that deserves a separate patch. The retry was
implemented to cover unlikely stalls when the lock is held but as this
hasn't ever been observed in the real life I would agree to remove it to
simplify the code (even though it is literally few lines of code). I was
probably overcautious when adding the flag.
--
Michal Hocko
SUSE Labs
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@xxxxxxxxx. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>