From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: PROBLEM: BUG (NULL ptr dereference in ipv4_dst_check) Date: Fri, 29 Jul 2011 08:19:02 -0700 (PDT) Message-ID: <20110729.081902.300678107767426313.davem@davemloft.net> References: <1311950488.2843.23.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> <4E32C76B.5010700@hippy.csoma.elte.hu> <1311952306.2843.27.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: synapse@hippy.csoma.elte.hu, netdev@vger.kernel.org To: eric.dumazet@gmail.com Return-path: Received: from shards.monkeyblade.net ([198.137.202.13]:42804 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751137Ab1G2PTL (ORCPT ); Fri, 29 Jul 2011 11:19:11 -0400 In-Reply-To: <1311952306.2843.27.camel@edumazet-HP-Compaq-6005-Pro-SFF-PC> Sender: netdev-owner@vger.kernel.org List-ID: From: Eric Dumazet Date: Fri, 29 Jul 2011 17:11:46 +0200 > Thats tricky, because I am not sure we dont need RCU protection since we > can now exchange dst neighbour on the fly. > > Following patch would only reduce the window of bug, not a complete > fix... > > David, any opinion on this ? Indeed, old code worked because we invalidated entire route cache entry, and we never before ran arp_bind_neighbour() except on new route cache entires before they become globally visible. I think when we change an existing neigh we will need to release old neigh via RCU, at a minimum.