On Wed, Mar 25, 2015 at 06:01:48PM +0100, Vlastimil Babka wrote:

On 03/25/2015 03:15 PM, Tetsuo Handa wrote:

quoted

Johannes Weiner wrote:

quoted

diff --git a/mm/oom_kill.c b/mm/oom_kill.c
index 5cfda39b3268..e066ac7353a4 100644
--- a/mm/oom_kill.c
+++ b/mm/oom_kill.c

@@ -711,12 +711,15 @@ bool out_of_memory(struct zonelist *zonelist, gfp_t gfp_mask,
 		killed = 1;
 	}
 out:
+	if (test_thread_flag(TIF_MEMDIE))
+		return true;
 	/*
-	 * Give the killed threads a good chance of exiting before trying to
-	 * allocate memory again.
+	 * Wait for any outstanding OOM victims to die.  In rare cases
+	 * victims can get stuck behind the allocating tasks, so the
+	 * wait needs to be bounded.  It's crude alright, but cheaper
+	 * than keeping a global dependency tree between all tasks.
 	 */
-	if (killed)
-		schedule_timeout_killable(1);
+	wait_event_timeout(oom_victims_wait, !atomic_read(&oom_victims), HZ);

 	return true;
 }

out_of_memory() returning true with bounded wait effectively means that
wait forever without choosing subsequent OOM victims when first OOM victim
failed to die. The system will lock up, won't it?

And after patch 12, does this mean that you may not be waiting long enough
for the victim to die, before you fail the allocation, prematurely? I can
imagine there would be situations where the victim is not deadlocked, but
still take more than HZ to finish, no?

Arguably it should be reasonable to fail allocations once the OOM
victim is stuck for over a second and the OOM reserves have been
depleted.

On the other hand, we don't need to play it that tight, because that
timeout is only targetted for the victim-blocked-on-alloc situations
which aren't all that common.  Something like 5 seconds should still
be okay.

--
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>