From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753301AbaAOXnR (ORCPT ); Wed, 15 Jan 2014 18:43:17 -0500 Received: from zene.cmpxchg.org ([85.214.230.12]:46852 "EHLO zene.cmpxchg.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752195AbaAOXnO (ORCPT ); Wed, 15 Jan 2014 18:43:14 -0500 Date: Wed, 15 Jan 2014 18:43:08 -0500 From: Johannes Weiner To: Andrew Morton Cc: David Rientjes , Michal Hocko , linux-mm@kvack.org, linux-kernel@vger.kernel.org Subject: [patch] mm: oom_kill: revert 3% system memory bonus for privileged tasks Message-ID: <20140115234308.GB4407@cmpxchg.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org With a63d83f427fb ("oom: badness heuristic rewrite"), the OOM killer tries to avoid killing privileged tasks by subtracting 3% of overall memory (system or cgroup) from their per-task consumption. But as a result, all root tasks that consume less than 3% of overall memory are considered equal, and so it only takes 33+ privileged tasks pushing the system out of memory for the OOM killer to do something stupid and kill sshd or dhclient. For example, on a 32G machine it can't tell the difference between the 1M agetty and the 10G fork bomb member. The changelog describes this 3% boost as the equivalent to the global overcommit limit being 3% higher for privileged tasks, but this is not the same as discounting 3% of overall memory from _every privileged task individually_ during OOM selection. Revert back to the old priority boost of pretending root tasks are only a quarter of their actual size. Signed-off-by: Johannes Weiner --- mm/oom_kill.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/mm/oom_kill.c b/mm/oom_kill.c index 1e4a600a6163..1b0011c3d9e2 100644 --- a/mm/oom_kill.c +++ b/mm/oom_kill.c @@ -166,11 +166,11 @@ unsigned long oom_badness(struct task_struct *p, struct mem_cgroup *memcg, task_unlock(p); /* - * Root processes get 3% bonus, just like the __vm_enough_memory() - * implementation used by LSMs. + * Memory consumption being equal, prefer killing an + * unprivileged task over a root task. */ if (has_capability_noaudit(p, CAP_SYS_ADMIN)) - adj -= 30; + points /= 4; /* Normalize to oom_score_adj units */ adj *= totalpages / 1000; -- 1.8.4.2