From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: dst->obsolete has become pointless Date: Tue, 08 Nov 2011 12:20:20 -0500 (EST) Message-ID: <20111108.122020.1080743546477280623.davem@davemloft.net> References: <20111104.230910.520924516201406800.davem@davemloft.net> <20111108093424.GG22141@secunet.com> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, timo.teras@iki.fi To: steffen.klassert@secunet.com Return-path: Received: from shards.monkeyblade.net ([198.137.202.13]:35416 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754448Ab1KHRUg (ORCPT ); Tue, 8 Nov 2011 12:20:36 -0500 In-Reply-To: <20111108093424.GG22141@secunet.com> Sender: netdev-owner@vger.kernel.org List-ID: From: Steffen Klassert Date: Tue, 8 Nov 2011 10:34:24 +0100 > I don't know what to do with DecNET, but we probaply need to decide > about the future of dst->obsolete before we can fix the ipv4 PMTU > problems. Possible fixes might depend on whether ->dst_check() is > always invoked or not. Simplest thing to do is to move dst->obsolete check into DecNET's ->dst_check() handler, then call ->dst_check() unconditionally. Then we can just set dst->obsolete to zero for all route types, and kill the "initial_obsolete" argument to dst_alloc() and associated logic. As things are currently implemented, because of how we elide the full table scan and flush, ipv4 and ipv6 really need to do the serial number check every time so we have to keep things as they are. Things get more interesting with the routing cache removed, of course, but that is still a long ways off. :-)