From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753827AbXCSMMj (ORCPT ); Mon, 19 Mar 2007 08:12:39 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1753834AbXCSMMj (ORCPT ); Mon, 19 Mar 2007 08:12:39 -0400 Received: from ug-out-1314.google.com ([66.249.92.169]:42926 "EHLO ug-out-1314.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753827AbXCSMM3 (ORCPT ); Mon, 19 Mar 2007 08:12:29 -0400 Date: Mon, 19 Mar 2007 14:13:10 +0200 From: "Michael S. Tsirkin" To: Alexey Kuznetsov Cc: "Michael S. Tsirkin" , Linux Kernel Mailing List , netdev@vger.kernel.org, general@lists.openfabrics.org, Roland Dreier , David Miller Subject: Re: dst_ifdown breaks infiniband? Message-ID: <20070319121310.GE18497@mellanox.co.il> Reply-To: "Michael S. Tsirkin" References: <20070318155532.GG7958@mellanox.co.il> <20070318191238.GA20518@ms2.inr.ac.ru> <20070318195355.GB11078@mellanox.co.il> <20070318201826.GB27004@ms2.inr.ac.ru> <20070319093632.GB8386@mellanox.co.il> <20070319120534.GA28187@ms2.inr.ac.ru> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20070319120534.GA28187@ms2.inr.ac.ru> User-Agent: Mutt/1.5.11 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org > Quoting Alexey Kuznetsov : > Subject: Re: dst_ifdown breaks infiniband? > > Hello! > > > If a device driver sets neigh_destructor in neigh_params, this could > > get called after the device has been unregistered and the driver module > > removed. > > It is the same problem: if dst->neighbour holds neighbour, it should > not hold device. parms->dev is not supposed to be used after > neigh_parms_release(). F.e. set parms->dev to NULL to catch bad references. Yes. I fixed that - simply checking that neighbour->dev is a loopback device is sufficient to detect the fact that the device is being unregistered. > Do you search for a way to find real inifiniband device in > ipoib_neigh_destructor()? No, not anymore. > I guess you will not be able. I agree it's not possible. > The problem is logical: if destructor needs device, neighbour entry > _somehow_ have to hold reference to the device (via neigh->dev, neigh->parms, > whatever). Hence, if we hold neighbour entry, unregister cannot be completed. > Therefore, destructor cannot refer to device. Q.E.D. :-) > > Seems, releasing dst->neighbour is inevitable. infiniband sets parm->neigh_destructor, and I search for a way to prevent this destructor from being called after the module has been unloaded. Ideas? -- MST