From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: oops in net/ipv4/icmp.c:icmp_send() with icmp_errors_use_inbound_ifaddr (fwd) Date: Mon, 21 May 2007 18:36:57 +0200 Message-ID: <4651CAA9.4050908@trash.net> References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-15 Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, jmorris@namei.org, Curtis@greenkey.net, davem@davemloft.net To: Herbert Xu Return-path: Received: from stinky.trash.net ([213.144.137.162]:59041 "EHLO stinky.trash.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755821AbXEUQjO (ORCPT ); Mon, 21 May 2007 12:39:14 -0400 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Herbert Xu wrote: > Patrick McHardy wrote: > >>This leaves the question what to do in the path after ip_output, >>when skb->dev points to the output device. We don't know the >>input device anymore, so there doesn't seem to be a way to make >>it do what the sysctl promises. > > > Perhaps we could change things so that the setting of skb->dev is > delayed until the packet has completely left the IP stack? It'd > require a massive audit though of the IP stack though. The IP stack shouldn't be too hard, we currently set skb->dev in ip_output, after that we only have the POST_ROUTING hook and ip_fragment which care. It wouldn't help with ipip/ip_gre though, at that point skb->dev must point to the tunnel device.