From mboxrd@z Thu Jan 1 00:00:00 1970 From: Bill Fink Subject: Re: [PATCH 3/3]: tg3: Manage TX backlog in-driver. Date: Fri, 20 Jun 2008 15:41:24 -0400 Message-ID: <20080620154124.eb91fd5e.billfink@mindspring.com> References: <20080619.041024.116139711.davem@davemloft.net> <20080620145233.3e11b6fe.billfink@mindspring.com> <20080620.120422.193703152.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Cc: krkumar2@in.ibm.com, mchan@broadcom.com, netdev@vger.kernel.org, vinay@linux.vnet.ibm.com To: David Miller Return-path: Received: from elasmtp-scoter.atl.sa.earthlink.net ([209.86.89.67]:47368 "EHLO elasmtp-scoter.atl.sa.earthlink.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751277AbYFTTl2 (ORCPT ); Fri, 20 Jun 2008 15:41:28 -0400 In-Reply-To: <20080620.120422.193703152.davem@davemloft.net> Sender: netdev-owner@vger.kernel.org List-ID: On Fri, 20 Jun 2008, David Miller wrote: > From: Bill Fink > Date: Fri, 20 Jun 2008 14:52:33 -0400 > > > I have a general question about this new tx queueing model, which > > I haven't seen discussed to this point. > > > > Although hopefully not frequent events, if the tx queue is kept in > > the driver rather than the network midlayer, what are the ramifications > > of a routing change which requires changing the output to a new interface, > > considering for example that on our 10-GigE interfaces we typically set > > txqueuelen to 10000. > > All of the packets would have been in the existing mid-layer generic > backlog anyways, way past the routing decisions. Nothing about > behavior in this area would be changing. Great. I wasn't aware the mid-layer generic backlog was after the routing decisions. Thanks for educating me (and others). I learn new things about the Linux network stack all the time, plus it's always changing, so can be a challenge to keep up with it if you're not a full-time developer. > > Similarly, what are the ramifications of such a change to the bonding > > driver (either in a load balancing or active/backup scenario) when one > > of the interfaces fails (again hopefully a rare event). > > Since the bonding driver acts like a pass-thru, and because of the > above, I expect no real ramifications in this area as well. Also good to know. -Thanks -Bill