Re: [patch 3.5-rc3] mm, oom: fix potential killing of thread that is disabled from oom killing

[KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>]

On Fri, Jun 22, 2012 at 5:44 PM, David Rientjes <rientjes@google.com> wrote:
> /proc/sys/vm/oom_kill_allocating_task will immediately kill current when
> the oom killer is called to avoid a potentially expensive tasklist scan
> for large systems.
>
> Currently, however, it is not checking current's oom_score_adj value
> which may be OOM_SCORE_ADJ_MIN, meaning that it has been disabled from
> oom killing.
>
> This patch avoids killing current in such a condition and simply falls
> back to the tasklist scan since memory still needs to be freed.
>
> Signed-off-by: David Rientjes <rientjes@google.com>
> ---
>  mm/oom_kill.c |    4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mm/oom_kill.c b/mm/oom_kill.c
> --- a/mm/oom_kill.c
> +++ b/mm/oom_kill.c
> @@ -720,9 +720,9 @@ void out_of_memory(struct zonelist *zonelist, gfp_t gfp_mask,
>        check_panic_on_oom(constraint, gfp_mask, order, mpol_mask);
>
>        read_lock(&tasklist_lock);
> -       if (sysctl_oom_kill_allocating_task &&
> +       if (sysctl_oom_kill_allocating_task && current->mm &&
>            !oom_unkillable_task(current, NULL, nodemask) &&
> -           current->mm) {
> +           current->signal->oom_score_adj != OOM_SCORE_ADJ_MIN) {
>                oom_kill_process(current, gfp_mask, order, 0, totalpages, NULL,
>                                 nodemask,
>                                 "Out of memory (oom_kill_allocating_task)");

Seems straight forward and reasonable.

Acked-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>

--
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>