From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from verein.lst.de ([213.95.11.211]:60810 "EHLO newverein.lst.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751364AbdGSHgp (ORCPT ); Wed, 19 Jul 2017 03:36:45 -0400 Date: Wed, 19 Jul 2017 09:36:43 +0200 From: Christoph Hellwig Subject: Re: [PATCH, RFC] xfs: fix multi-AG deadlock in xfs_bunmapi Message-ID: <20170719073643.GA21259@lst.de> References: <20170718172545.18065-1-hch@lst.de> <20170718181535.GJ4224@magnolia> <20170718183351.GK4224@magnolia> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20170718183351.GK4224@magnolia> Sender: linux-xfs-owner@vger.kernel.org List-ID: List-Id: xfs To: "Darrick J. Wong" Cc: Nikolay Borisov , Christoph Hellwig , linux-xfs@vger.kernel.org On Tue, Jul 18, 2017 at 11:33:51AM -0700, Darrick J. Wong wrote: > On Tue, Jul 18, 2017 at 09:27:43PM +0300, Nikolay Borisov wrote: > > > > > > On 18.07.2017 21:15, Darrick J. Wong wrote: > > > On Tue, Jul 18, 2017 at 07:25:45PM +0200, Christoph Hellwig wrote: > > >> Just like in the allocator we must avoid touching multiple AGs out of > > >> order when freeing blocks, as freeing still locks the AGF and can cause > > >> the same AB-BA deadlocks as in the allocation path. > > >> > > >> Signed-off-by: Christoph Hellwig > > >> Reported-by: Nikolay Borisov > > > > > > Looks ok, though I wonder what the bug report looked like. :) > > > > https://www.spinics.net/lists/linux-xfs/msg05918.html > > Ah, ok. That's what I suspected, so I'm glad to have confirmation. Btw, I found a little bug when rebasing additional bunmapi work on top of the patch - we need to add a !isrt to the conditional for the AG check. But I'll wait for test results from Nikolay before resending it, as it should not affect normal non-RT setups.