From: Edwin Török Date: Mon, 11 Mar 2019 15:06:32 +0000 Subject: [Cluster-devel] [PATCH] gfs2: Fix missed wakeups in find_insert_glock In-Reply-To: References: <20190306161459.23639-1-agruenba@redhat.com> <352d6ec4-2ee3-4348-ac13-ea3db6fc6702@citrix.com> Message-ID: <975e12ec-8f41-eaaf-a4e2-1cd99aa9a906@citrix.com> List-Id: To: cluster-devel.redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit On 07/03/2019 16:23, Andreas Gruenbacher wrote: > Hi Edwin, > > On Thu, 7 Mar 2019 at 16:55, Edwin T?r?k wrote: >> On 06/03/2019 16:14, Andreas Gruenbacher wrote: >>> Mark Syms has reported seeing tasks that are stuck waiting in >>> find_insert_glock. It turns out that struct lm_lockname contains four padding >>> bytes on 64-bit architectures that function glock_waitqueue doesn't skip when >>> hashing the glock name. As a result, we can end up waking up the wrong >>> waitqueue, and the waiting tasks will be stuck. >>> >>> Use offsetofend instead of sizeof to get the struct size without the padding. >>> >>> Reported-by: Mark Syms >>> Signed-off-by: Andreas Gruenbacher >> >> Thank you for your patch, initial testing looks promising when replacing the schedule_timeout patch with this one (no deadlocks/stuck tasks found so far). >> Will let you know when our testing completes. > > okay great, thanks. Hope we can still get this into the 5.1 kernel. Testing of this patch itself looked good (no new issues), however GFS2 still deadlocks on 4.19 even with this patch, see: https://www.redhat.com/archives/cluster-devel/2019-March/msg00000.html [the iomap deadlocks seems to be more rare now though] Thanks, --Edwin