From mboxrd@z Thu Jan 1 00:00:00 1970 From: Erez Shitrit Subject: Re: [PATCH rdma-rc 3/7] IB/IPoIB: Fix race between ipoib_remove_one to sysfs functions Date: Wed, 8 Jun 2016 10:01:51 +0300 Message-ID: References: <1465042524-25852-1-git-send-email-leon@kernel.org> <1465042524-25852-4-git-send-email-leon@kernel.org> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Return-path: In-Reply-To: Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Or Gerlitz Cc: Erez Shitrit , Doug Ledford , "linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" List-Id: linux-rdma@vger.kernel.org On Tue, Jun 7, 2016 at 11:22 PM, Or Gerlitz wrote: > On Tue, Jun 7, 2016 at 1:45 PM, Erez Shitrit wrote: >> On Tue, Jun 7, 2016 at 10:31 AM, Or Gerlitz wrote: >>> On Sat, Jun 4, 2016 at 3:15 PM, Leon Romanovsky wrote: >>> >>>> Fixes: 862096a8bbf8 ("IB/ipoib: Add more rtnl_link_ops callbacks") >>>> Fixes: 9baa0b036410 ("IB/ipoib: Add rtnl_link_ops support") >>> >>> These two commits go back to 3.7-rc1 / Oct 2012 -- so I wonder how come we never >>> got reports on this ABBA deadlock happens in real life - any idea, did >>> you see the deadlock in >> >> Sure, it is from the real life, you can see the stack that i added to >> the commit message. > > yes, but the bug you're pointing on is out there for whole four years > and no one complained including you, can you explain that? Not really, it is a race, not easy to see it, but it is there, waiting.. > >> If you want to simulate that just add sleep in the remove_one while >> trying to change mode/add-child etc. > > so why need to simulate? is that impossible to get it on non simulated env? It happened in our regression system. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html