From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1LXLp6-0003Zt-Cx for qemu-devel@nongnu.org; Wed, 11 Feb 2009 15:36:44 -0500 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1LXLp6-0003Zg-0s for qemu-devel@nongnu.org; Wed, 11 Feb 2009 15:36:44 -0500 Received: from [199.232.76.173] (port=49812 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1LXLp5-0003Zd-Pc for qemu-devel@nongnu.org; Wed, 11 Feb 2009 15:36:43 -0500 Received: from g5t0006.atlanta.hp.com ([15.192.0.43]:2246) by monty-python.gnu.org with esmtps (TLS-1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.60) (envelope-from ) id 1LXLp5-0001xY-Hw for qemu-devel@nongnu.org; Wed, 11 Feb 2009 15:36:43 -0500 From: Alex Williamson In-Reply-To: <1234383541.14052.247.camel@blaa> References: <20090210212841.9760.96780.stgit@kvm.aw> <1234380678.14052.238.camel@blaa> <1234381866.7026.1524.camel@lappy> <1234383541.14052.247.camel@blaa> Content-Type: text/plain Date: Wed, 11 Feb 2009 13:37:19 -0700 Message-Id: <1234384639.7026.1533.camel@lappy> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Subject: [Qemu-devel] Re: [PATCH 0/4] qemu: TAP filtering support Reply-To: qemu-devel@nongnu.org List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Mark McLoughlin Cc: qemu-devel@nongnu.org, kvm@vger.kernel.org On Wed, 2009-02-11 at 20:19 +0000, Mark McLoughlin wrote: > On Wed, 2009-02-11 at 12:51 -0700, Alex Williamson wrote: > > > - What do we need rxfilter=on|off on the command line for? > > > > Primarily because the current tun driver in Linux has a bug that it can > > drop unicast packets requested to be included in the filter if it > > overflows the exact match table. > > > > http://www.spinics.net/lists/netdev/msg88451.html > > Oh dear. Does this suggest we should make rxfilter=off the default? > > Is there any way we can automatically detect we're running on a broken > kernel? TUNSETTXFITLER has only existed since 2.6.26, so the ioctl will fail on anything older and it will be disabled anyway. The patch will fix .29 and should get rolled into .28 stable, so we're looking at an exposure of 2 kernel releases. Unfortunately a few community distros went out on those kernels, so perhaps the prudent approach would be to make the default disabled until we're a few releases beyond. I don't know any way you could detect it outside of ugly parsing of uname -r. Thanks, Alex