From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andrew Lunn Subject: Re: [PATCH net-next 1/2] dsa: bonding: implement HW bonding Date: Sat, 21 Feb 2015 21:43:17 +0100 Message-ID: <20150221204317.GA29884@lunn.ch> References: <1424429473-4601-1-git-send-email-jonasj76@gmail.com> <1424429473-4601-2-git-send-email-jonasj76@gmail.com> <54E763A6.4020505@gmail.com> <20150221165715.GE2092@nanopsycho.orion> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Jonas Johansson , Florian Fainelli , netdev@vger.kernel.org, Jonas Johansson , Scott Feldman To: Jiri Pirko Return-path: Received: from vps0.lunn.ch ([178.209.37.122]:47202 "EHLO vps0.lunn.ch" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750746AbbBUUqN (ORCPT ); Sat, 21 Feb 2015 15:46:13 -0500 Content-Disposition: inline In-Reply-To: <20150221165715.GE2092@nanopsycho.orion> Sender: netdev-owner@vger.kernel.org List-ID: > Hmm, I'm not sure I understand correctly what the marvel hw is capable of. > Would you care to explain it a bit ? (or maybe I should read the > datasheet) The mv88e6060 data sheet is publicly available. It is a rather old chip, but i guess trunking has not changed too much since then. All the newer chips require an NDA to get the datahseet. What i've done before is get the switch to combine two ports into a trunk. It will then load balance packets over the two ports. If one of the ports fails, i.e. link down, it will then concentrate all the traffic over the one remaining port. i.e some basic form of redundancy/failover. Not covered by this patchset, but what would be interesting, is getting trunking working towards the host. Many of the recent wireless APs have two ethernet interfaces connected to the switch. The current DSA architecture only allows one of them to be used for traffic to/from the host from/to the switch. Andrew