From mboxrd@z Thu Jan 1 00:00:00 1970 From: Nikolay Aleksandrov Subject: Re: [PATCH net-next 0/7] net: ipmr: more cleanups and mfc netlink support Date: Thu, 26 Nov 2015 17:48:47 +0100 Message-ID: <565737EF.9030504@cumulusnetworks.com> References: <1448547830-21716-1-git-send-email-razor@blackwall.org> <20151126163502.GB4094@lunn.ch> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, davem@davemloft.net To: Andrew Lunn , Nikolay Aleksandrov Return-path: Received: from mail-wm0-f44.google.com ([74.125.82.44]:34942 "EHLO mail-wm0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752859AbbKZQst (ORCPT ); Thu, 26 Nov 2015 11:48:49 -0500 Received: by wmuu63 with SMTP id u63so28186923wmu.0 for ; Thu, 26 Nov 2015 08:48:48 -0800 (PST) In-Reply-To: <20151126163502.GB4094@lunn.ch> Sender: netdev-owner@vger.kernel.org List-ID: On 11/26/2015 05:35 PM, Andrew Lunn wrote: > On Thu, Nov 26, 2015 at 03:23:43PM +0100, Nikolay Aleksandrov wrote: >> From: Nikolay Aleksandrov >> >> Hi, >> This set continues with the minor cleanups in the first 6 patches and >> patch 7 adds the first new feature - MFC manipulation via netlink. > > Hi Nikolay > > Great to see somebody working on multicast. > > Do you have any plans to remove the 32 VIF limit? > > I've run out if VIFs on a VPN server with lots of clients. I took the > "quick fix" of recompiling the kernel and userspace with 256 VIFs. But > it would be nice if you could define a netlink API which does not have > this limit and modify the core code to be more dynamic. > > Andrew > Hi Andrew, I'm glad people are still using it. :-) To answer your question - yes, I plan to remove that limit, in fact I have a private version where it's done and this API is the first step, you can see it uses the generic rtnh which supports ifindex passing, so it'll be a matter of a few small changes to remove it. The hardest part is keeping current user-space compatible with everything, but I have the necessary workarounds for that. In the end it should be much more usable and to some extent without mrtsock application running as well. Thanks, Nik