From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cong Wang Subject: Re: [RFC Patch 1/3] netpoll: add generic support for bridge and bonding devices Date: Tue, 23 Mar 2010 12:47:39 +0800 Message-ID: <4BA847EB.9040808@redhat.com> References: <20100322082059.4967.63492.sendpatchset@localhost.localdomain> <1269297081.3552.19.camel@calx> <4BA823D7.4010106@redhat.com> <20100322.204939.146100390.davem@davemloft.net> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: mpm@selenic.com, linux-kernel@vger.kernel.org, netdev@vger.kernel.org, bridge@lists.linux-foundation.org, gospo@redhat.com, nhorman@tuxdriver.com, shemminger@linux-foundation.org, bonding-devel@lists.sourceforge.net, fubar@us.ibm.com, jmoyer@redhat.com To: David Miller Return-path: In-Reply-To: <20100322.204939.146100390.davem@davemloft.net> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org David Miller wrote: > From: Cong Wang > Date: Tue, 23 Mar 2010 10:13:43 +0800 > >> Matt Mackall wrote: >>> Seems like a lot of interface for something to be used by only a >>> couple >>> core drivers. Hopefully Dave has an opinion here. >>> >> Yeah, I worry about this too, maybe we can group those methods >> for netpoll together into another struct, and just put a pointer >> here? > > This looks like it's tackled at the wrong layer, to be honest. > > Teaching all of these layers about eachother's states is > going to end up being a nightmare in the end. > > All of this "where is the npinfo" business can be handled > generically in net/core/dev.c I think, with none of these > callbacks. > > For example, something like "if dev lacks ->npinfo, check > it's master". This is a good point! I haven't tried but certainly this is worthy a try. Ideally those callbacks can be all removed, but I don't know if this is true practically. ;) I will try. > > Another thing, I wouldn't iterate over all devices, like I > see in the bonding poll controller method. Just whichever > one supports netpoll you see first, use it and exit > immediately. Don't send it to every single port, I can't > see how that might be desirable or useful. Yeah, for bonding case, probably. But for bridge case, I think we still need to check all, right? Thanks!