From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Philip A. Prindeville" Subject: Re: [Linux-ATM-General] [PATCH] atm/br2684: netif_stop_queue() when atm device busy and netif_wake_queue() when we can send packets again. Date: Fri, 11 Sep 2009 12:56:59 -0700 Message-ID: <4AAAAB8B.1050304@redfish-solutions.com> References: <4A9901E7.9040804@hiramoto.org> <1251545092-18081-1-git-send-email-karl@hiramoto.org> <4AA95838.4010007@redfish-solutions.com> <4AA97004.2010904@hiramoto.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, linux-atm-general@lists.sourceforge.net To: Karl Hiramoto Return-path: Received: from mail.redfish-solutions.com ([66.232.79.143]:60566 "EHLO mail.redfish-solutions.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751923AbZIKT6U (ORCPT ); Fri, 11 Sep 2009 15:58:20 -0400 In-Reply-To: <4AA97004.2010904@hiramoto.org> Sender: netdev-owner@vger.kernel.org List-ID: Karl Hiramoto wrote: > Philip A. Prindeville wrote: >> I had noticed that with my Solos card, and using Qwest 7062/896kb/s >> service that I was typically only getting ~400kb/s upstream, so I >> thought that delayed transmitter restarts might be the culprit and >> decided to try out this patch. >> >> I'm running 2.6.27.26, and I modified the patch as below. >> >> Has anyone confirmed this patch (Karl's) against 2.6.27? >> >> Thanks, >> >> -Philip > I'd be interested in hearing comparisons with/without the patch. What > i have is no upstream packet loss with this patch, however slightly > lower total throughput. I think because the upper networking layers > take time to restart the packet flow. I'm not really sure if or how > many packets to upper layers buffer. I haven't had time to debug it > further. > > I don't have a solos card, but you may have to tweak the solos driver > sk_sndbuf value. > > -- > Karl I'm running with the patch right now... Can't tell if latency has been affected. Seems to be working a bit better, but I've not done any formalized testing. -Philip