On Wed 15-06-16 16:37:01, Oleg Nesterov wrote:

Michal,

I am going to ack the whole series, but send some nits/questions,

On 06/09, Michal Hocko wrote:

quoted

@@ -283,10 +283,22 @@ enum oom_scan_t oom_scan_process_thread(struct oom_control *oc,
 
 	/*
 	 * This task already has access to memory reserves and is being killed.
-	 * Don't allow any other task to have access to the reserves.
+	 * 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.
 	 */
-	if (!is_sysrq_oom(oc) && atomic_read(&task->signal->oom_victims))
-		return OOM_SCAN_ABORT;
+	if (!is_sysrq_oom(oc) && atomic_read(&task->signal->oom_victims)) {
+		struct task_struct *p = find_lock_task_mm(task);
+		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);

OK, but perhaps it would be beter to change oom_badness() to return zero if
MMF_OOM_REAPED is set?

We already do that:
	if (adj == OOM_SCORE_ADJ_MIN ||
			test_bit(MMF_OOM_REAPED, &p->mm->flags) ||
			in_vfork(p)) {
		task_unlock(p);
		return 0;
	}

It is kind of subtle that we have to check it 2 times but we would have
to rework this code much more because oom_badness only can tell to
ignore the task but not to abort scanning altogether currently. If we
should change this I would suggest a separate patch.
-- 
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>