Date: Fri, 11 Mar 2016 18:28:08 +0100
From: Peter Zijlstra <peterz@infradead.org>
To: Niklas Cassel <niklas.cassel@axis.com>
Cc: tj@kernel.org,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>
Subject: Re: [BUG] sched: leaf_cfs_rq_list use after free
Message-ID: <20160311172808.GF6344@twins.programming.kicks-ass.net>
References: <56D9664D.8080503@axis.com>
 <20160310125417.GW6344@twins.programming.kicks-ass.net>
 <56E2FA0F.4000803@axis.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <56E2FA0F.4000803@axis.com>
User-Agent: Mutt/1.5.21 (2012-12-30)
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 837
Lines: 24

On Fri, Mar 11, 2016 at 06:02:07PM +0100, Niklas Cassel wrote:
> > Can you describe a reliable reproducer?
> 
> Right now I just do a reboot loop, systemd creates cgroups during boot
> this dump took 84 reboots. Harder to reproduce with more trace_printks.

Bugger, I don't have a single system with systemd on, so that'll be hard
to reproduce.

> 
> CONFIG_SCHED_AUTOGROUP looks interesting though, might try that next week.
> 
> Appears that the call comes from idle_balance or rebalance_domains,
> since they are the only ones that call load_balance.

Weird, balancing should very much be inside the cgroup, so if balancing
moves anything, there must be tasks to begin with, which still should
not be the case if we've called offline.

FWIW:

echo 1 > /debug/tracing/options/stacktrace

will add a stack trace to each trace_printk().