From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [PATCH v3 04/12] l2tp: Add ppp device name to L2TP ppp session data Date: Thu, 01 Apr 2010 09:30:34 +0200 Message-ID: <1270107034.2229.5.camel@edumazet-laptop> References: <20100330161725.9628.69994.stgit@bert.katalix.com> <20100330161746.9628.40723.stgit@bert.katalix.com> <20100330092914.4f3e9ed3@nehalam> <4BB2FD0D.1080105@katalix.com> <20100331160811.16351eb7@s6510> <4BB4490F.8090406@katalix.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Stephen Hemminger , netdev@vger.kernel.org To: James Chapman Return-path: Received: from mail-bw0-f225.google.com ([209.85.218.225]:41451 "EHLO mail-bw0-f225.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750929Ab0DAHak (ORCPT ); Thu, 1 Apr 2010 03:30:40 -0400 Received: by bwz25 with SMTP id 25so616928bwz.28 for ; Thu, 01 Apr 2010 00:30:38 -0700 (PDT) In-Reply-To: <4BB4490F.8090406@katalix.com> Sender: netdev-owner@vger.kernel.org List-ID: Le jeudi 01 avril 2010 =C3=A0 08:19 +0100, James Chapman a =C3=A9crit : > There might be thousands of L2TP sessions in some setups. Populating > sysfs with a link for each of those sessions isn't practical. The > existing /proc file dumps its info as a single text file for this > reason. I'd also like to provide the device name in the session netli= nk > message, which is the interface used by l2tp userspace, so I need a > kernel API to retrieve the device name from ppp. >=20 > I like the suggestion of using debugfs for access to driver debug inf= o > though. I propose leaving the /proc file for L2TPv2 only, removing th= e > L2TPv3 data that I added to the proc file in this patch series, to > retain compatibility with the existing driver. This would show only > L2TPv2 sessions and tunnels. For new driver functionality (L2TPv3 etc= ), > use debugfs. The debugfs files would dump lists in a similar form to = the > current code, listing all tunnels (L2TPv2 and L2TPv3) in a single fil= e. > Using debugfs gives more flexibility for adding additional info later= , > as required. How does that sound? >=20 debugfs ? I dont get it, sorry. Why not using netlink, as most iproute2 utilities do ?