On Mon 12-12-16 12:59:18, Michal Hocko wrote:
> On Mon 12-12-16 19:55:52, Tetsuo Handa wrote:
> > Since commit 862e3073b3eed13f
> > ("mm, oom: get rid of signal_struct::oom_victims")
> > changed to wait until MMF_OOM_SKIP is set rather than wait while
> > TIF_MEMDIE is set, rationale comment for commit e2fe14564d3316d1
> > ("oom_reaper: close race with exiting task") needs to be updated.
>
> True.
>
> > While holding oom_lock can make sure that other threads waiting for
> > oom_lock at __alloc_pages_may_oom() are given a chance to call
> > get_page_from_freelist() after the OOM reaper called unmap_page_range()
> > via __oom_reap_task_mm(), it can defer calling of __oom_reap_task_mm().
> >
> > Therefore, this patch moves oom_lock from __oom_reap_task_mm() to
> > oom_reap_task() (without any functional change). By doing so, the OOM
> > killer can call __oom_reap_task_mm() if we don't want to defer calling
> > of __oom_reap_task_mm() (e.g. when oom_evaluate_task() aborted by
> > finding existing OOM victim's mm without MMF_OOM_SKIP).
>
> But I fail to understand this part of the changelog. It sounds like a
> preparatory for other changes. There doesn't seem to be any other user
> of __oom_reap_task_mm in the current tree.
>
> Please send a patch which removes the comment which is no longer true
> on its own and feel free to add
>
> Acked-by: Michal Hocko <mhocko@xxxxxxxx>
>
> but do not make other changes if you do not have any follow up patch
> which would benefit from that.
Do you plan to pursue this?
--
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>