From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: generic/04[89] fail on XFS due to change in writeback code [4.2-rc1 regression] Date: Thu, 13 Aug 2015 15:16:02 -0400 Message-ID: <20150813191602.GF4496@mtj.duckdns.org> References: <20150812101204.GE17933@dhcp-13-216.nay.redhat.com> <20150813004435.GN3902@dastard> <20150813153442.GE4496@mtj.duckdns.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Cc: axboe@fb.com, linux-fsdevel@vger.kernel.org, jack@suse.cz, Eryu Guan , xfs@oss.sgi.com To: Dave Chinner Return-path: Content-Disposition: inline In-Reply-To: <20150813153442.GE4496@mtj.duckdns.org> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: xfs-bounces@oss.sgi.com Sender: xfs-bounces@oss.sgi.com List-Id: linux-fsdevel.vger.kernel.org On Thu, Aug 13, 2015 at 11:34:42AM -0400, Tejun Heo wrote: > On Thu, Aug 13, 2015 at 10:44:35AM +1000, Dave Chinner wrote: > > It might be a couple of days before I really get a chance to dig > > into this, so it might be best if Tejun can look into it first. > > Yeap, have been looking into it since yesterday. I have some > suspicions. I'll write once I know more. So, here are what I've found out till now. * I can't reproduce it for some reason. * There's a bug in b_dirty_time handling. sync_inodes_sb() should schedule writebacks regardless of b_dirty_time but it currently isn't. I'm working on a patch to fix it. * But I can't see how the above bug would lead to failure of size sync. One possibility is that wb_has_dirty_io() and/or bdi_has_dirty_io() is getting out of sync for some reason. I'll write up a debug patch for this. Thanks. -- tejun _______________________________________________ xfs mailing list xfs@oss.sgi.com http://oss.sgi.com/mailman/listinfo/xfs