On Thu 07-01-16 22:31:32, Tetsuo Handa wrote:
[...]

I think we need to filter at select_bad_process() and oom_kill_process().

When P has no children, P is chosen and TIF_MEMDIE is set on P. But P can
be chosen forever due to P->signal->oom_score_adj == OOM_SCORE_ADJ_MAX
even if the OOM reaper reclaimed P's mm. We need to ensure that
oom_kill_process() is not called with P if P already has TIF_MEMDIE.

Hmm. Any task is allowed to set its oom_score_adj that way and I
guess we should really make sure that at least sysrq+f will make some
progress. This is what I would do. Again I think this is worth a
separate patch. Unless there are any objections I will roll out what I
have and post 3 separate patches.
---

diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 45e51ad2f7cf..ee34a51bd65a 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c

@@ -333,6 +333,14 @@ static struct task_struct *select_bad_process(struct oom_control *oc,
 		if (points == chosen_points && thread_group_leader(chosen))
 			continue;
 
+		/*
+		 * If the current major task is already ooom killed and this
+		 * is sysrq+f request then we rather choose somebody else
+		 * because the current oom victim might be stuck.
+		 */
+		if (is_sysrq_oom(sc) && test_tsk_thread_flag(p, TIF_MEMDIE))
+			continue;
+
 		chosen = p;
 		chosen_points = points;
 	}

-- 
Michal Hocko
SUSE Labs

--
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>