From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bart Van Assche Subject: Re: 3.15-rc4: circular locking dependency triggered by dm-multipath Date: Tue, 06 May 2014 15:46:03 +0200 Message-ID: <5368E79B.8000208@acm.org> References: <5368ABE8.9050909@acm.org> <20140506132211.GA22945@redhat.com> Reply-To: device-mapper development Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20140506132211.GA22945@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: Mike Snitzer Cc: device-mapper development List-Id: dm-devel.ids On 05/06/14 15:22, Mike Snitzer wrote: > On Tue, May 06 2014 at 5:31am -0400, > Bart Van Assche wrote: >> Has anyone else perhaps already run into this ? > > Not seen this, how easily is it reproduced? > > These recent commits are clearly the issue: > > e809917735e dm mpath: push back requests instead of queueing > 3e9f1be1b40 dm mpath: remove process_queued_ios() > > Need to take a closer look at the lockdep splat you provided. All I did was to simulate a cable pull while running a fio test on top of a dm-multipath device on top of a few SCSI devices. This lockdep report was generated the first time I ran the cable pull simulation. I have not yet tried to reproduce this but it's not impossible that this is 100% reproducible. The same test runs fine with kernel 3.14. Bart.