From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andi Kleen Subject: Re: IPv6 tunnel scalability problem Date: Mon, 01 Sep 2008 14:04:58 +0200 Message-ID: <8763pgxead.fsf@basil.nowhere.org> References: <200808311958.51510.rdenis@simphalempin.com> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: netdev@vger.kernel.org, Bernhard Schmidt To: =?utf-8?Q?R=C3=A9mi?= Denis-Courmont Return-path: Received: from one.firstfloor.org ([213.235.205.2]:45088 "EHLO one.firstfloor.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751886AbYIAMFA convert rfc822-to-8bit (ORCPT ); Mon, 1 Sep 2008 08:05:00 -0400 In-Reply-To: <200808311958.51510.rdenis@simphalempin.com> (=?utf-8?Q?R?= =?utf-8?Q?=C3=A9mi?= Denis-Courmont's message of "Sun, 31 Aug 2008 19:58:51 +0300") Sender: netdev-owner@vger.kernel.org List-ID: R=C3=A9mi Denis-Courmont writes: > I have been maintaining a TUN-based Linux implementation of Teredo/RF= C4380. On=20 > a busy node, this can trigger quite many peers on the virtual point-t= o-point=20 > tunnel interface. I have received complaints that the whole thing see= ms to=20 > hit some severe performance bottleneck when this happens. It is not c= lear to=20 > me at this point whether it's a kernel or a user problem. So I have b= een=20 > writing a stress test. You could use oprofile to find out where the CPU time is going. That would work even during normal operation. Post the opreport -l outp= ut -Andi --=20 ak@linux.intel.com