From mboxrd@z Thu Jan 1 00:00:00 1970 From: jamal Subject: Re: regression due to "flush SAD/SPD generate false events" Date: Thu, 18 Feb 2010 17:52:41 -0500 Message-ID: <1266533561.3879.132.camel@bigi> References: <20100218185510.GA18829@x200> <1266526416.3879.109.camel@bigi> <1266529558.3879.121.camel@bigi> <20100218.142735.230862605.davem@davemloft.net> Reply-To: hadi@cyberus.ca Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: adobriyan@gmail.com, netdev@vger.kernel.org To: David Miller Return-path: Received: from mail-qy0-f200.google.com ([209.85.221.200]:37115 "EHLO mail-qy0-f200.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751045Ab0BRWwq (ORCPT ); Thu, 18 Feb 2010 17:52:46 -0500 Received: by qyk38 with SMTP id 38so3284757qyk.1 for ; Thu, 18 Feb 2010 14:52:45 -0800 (PST) In-Reply-To: <20100218.142735.230862605.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, 2010-02-18 at 14:27 -0800, David Miller wrote: > Let's see how #2 works out, if we get any more breakage > cases reported then we'll have to revert again and go with > #1. Actually, i think i found the root cause. The sequence should be: 1) pfkey->kernel: flush kernel flushes SA/SPD 2)kernel->pfkey: respond to flush message 3)kernel->all user space listeners: i just flushed the SA/SPD #2 is always needed. #3 is wrong if SA/SPD is empty. kernel unfortunately sends a single broadcast which combines #2 and #3. So the proper fix is to break #2 and #3 into separate messages. And follow the rules above. I dont have time to fix it right now - but i will send a patch tommorow. cheers, jamal