From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jarek Poplawski Subject: Re: [RFC PATCH] Regression in linux 2.6.32 virtio_net seen with vhost-net Date: Thu, 17 Dec 2009 12:19:11 +0000 Message-ID: <20091217121911.GA8654@ff.dom.local> References: <20091217112754.GA7755@ff.dom.local> <20091217114535.GA30429@gondor.apana.org.au> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Krishna Kumar2 , Sridhar Samudrala , mst@redhat.com, netdev@vger.kernel.org, Rusty Russell , "David S. Miller" To: Herbert Xu Return-path: Received: from mail-fx0-f221.google.com ([209.85.220.221]:62244 "EHLO mail-fx0-f221.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S934016AbZLQMTQ (ORCPT ); Thu, 17 Dec 2009 07:19:16 -0500 Received: by fxm21 with SMTP id 21so1824565fxm.21 for ; Thu, 17 Dec 2009 04:19:14 -0800 (PST) Content-Disposition: inline In-Reply-To: <20091217114535.GA30429@gondor.apana.org.au> Sender: netdev-owner@vger.kernel.org List-ID: On Thu, Dec 17, 2009 at 07:45:35PM +0800, Herbert Xu wrote: > On Thu, Dec 17, 2009 at 11:27:55AM +0000, Jarek Poplawski wrote: > > > > So we _should_ get into sch_direct_xmit when the queue was stopped... > > I guess Herbert might forget the multiqueue change, and Sridhar isn't > > missing much. ;-) > > Hmm, if that's the case then this new multiqueue code is seriously > broken. It means that once the queue is stopped, every new tx packet > will cause an unnecessary dequeue/requeue, up to 1000. Hmm, I can even remember who inspired this change ;-) But I'm not sure this really broken: it seems it's "optimized" for non-stopped case. Btw, IIRC, Krishna worked on changing this for uniqueues - I wonder if there could be much difference in testing? Cheers, Jarek P.