From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752342AbcADTP7 (ORCPT ); Mon, 4 Jan 2016 14:15:59 -0500 Received: from h2.hallyn.com ([78.46.35.8]:33736 "EHLO h2.hallyn.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751590AbcADTPz (ORCPT ); Mon, 4 Jan 2016 14:15:55 -0500 Date: Mon, 4 Jan 2016 13:15:53 -0600 From: "Serge E. Hallyn" To: Tejun Heo Cc: Jeremiah Mahler , cgroups@vger.kernel.org, Serge Hallyn , Stephen Rothwell , linux-kernel@vger.kernel.org Subject: Re: cgroup: BUG: unable to handle kernel NULL pointer dereference Message-ID: <20160104191553.GA21384@mail.hallyn.com> References: <20160101234028.GA1750@hudson.localdomain> <20160102115437.GE3660@htj.duckdns.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20160102115437.GE3660@htj.duckdns.org> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sat, Jan 02, 2016 at 06:54:37AM -0500, Tejun Heo wrote: > On Fri, Jan 01, 2016 at 03:40:28PM -0800, Jeremiah Mahler wrote: > > all, > > > > When running the latest linux-next (20151231) two of my machines > > hang early in the boot sequence. The initial message is for a > > NULL pointer dereference. > > > > BUG: unable to handle kernel NULL pointer dereference at 0000000000000030 > > > > And the RIP line refers to cgroup_path. > > > > RIP [] cgroup_path+0x30/0x80 > > > > Attached are pictures of the back trace. > > > > Let me know if I can do anything else to help. I will investigate the > > problem further if I get a chance. > > This is most likely from the recent changes from cgroup ns support. > Reverted the patchset for now as it also introduced a bisectability > issue. Serge, can you please look into this one? Hi Tejun, an updated set is at https://git.kernel.org/cgit/linux/kernel/git/sergeh/linux-security.git/commit/?h=cgroupns.v9 Would you like a new patchset send to the list, or is the github tree ok? This branch: 1. puts lookup_one_len() under inode lock 2. makes cgroup_path non-namespaced, so only calls to cgroup_path_ns() are namespaced. 3. makes cgroup_path{,_ns} take the needed locks, since external callers cannot do so. 4. Fixes the bisectability problem of to_cg_ns() being defined after use (patch 4 -> patch 3) thanks, -serge