From mboxrd@z Thu Jan 1 00:00:00 1970 From: Pablo Neira Ayuso Subject: Re: [PATCH net-next 00/15] Simplify netfilter and network namespaces Date: Mon, 15 Jun 2015 17:20:24 +0200 Message-ID: <20150615152024.GA10973@salvia> References: <87616ppt3h.fsf@x220.int.ebiederm.org> <20150615070622.GA3315@salvia> <87twu9nh8s.fsf@x220.int.ebiederm.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: David Miller , netdev@vger.kernel.org, netfilter-devel@vger.kernel.org, Stephen Hemminger , Juanjo Ciarlante , Wensong Zhang , Simon Horman , Julian Anastasov , Patrick McHardy , Jozsef Kadlecsik , Jamal Hadi Salim , Steffen Klassert , Herbert Xu To: "Eric W. Biederman" Return-path: Content-Disposition: inline In-Reply-To: <87twu9nh8s.fsf@x220.int.ebiederm.org> Sender: netdev-owner@vger.kernel.org List-Id: netfilter-devel.vger.kernel.org On Mon, Jun 15, 2015 at 10:06:27AM -0500, Eric W. Biederman wrote: > Pablo Neira Ayuso writes: > > > Hi Eric, > > > > On Sun, Jun 14, 2015 at 10:07:30PM -0500, Eric W. Biederman wrote: > >> > >> While looking into what it would take to route packets out to network > >> devices in other network namespaces I started looking at the netfilter > >> hooks, and there is a lot of nasty code to figure out which network > >> namespace to filter the packets in. > >> > >> Just passing the network namespace into the netfilter hooks looks like a > >> significant simplication in the code, and worth it as the first thing > >> most netfilter hooks do is compute the network namespace. > >> > >> We probably even want per network namespace netfilter hooks as nftables > >> tries (and fails to use). nftables currently has a fairly serious bug > >> where the tables for one network namespace run for packets in all > >> network namespaces. I do not have a fix yet but I think this patchset > >> making it easy to get at the network namespace is a good starting > >> place for a fix. > > > > I have a patchset that is almost ready to convert netfilter to pernet > > hooks, it partially clashes with this. Will send it out today and will > > Cc you so you can review it. > > > > I think it should take me little time to rebase it on top of yours. > > It's rather large changeset, which quite a lot of boiler plate code > > though. > > Hmm. > > I will wait and take a look. I think going from where I am to pernet > netfilter hooks is about one simple patch, so I don't know where the > large about of boiler plate comes from. You have to register the struct nf_hook_ops for each nf_register_hooks() caller from the pernet .init. > I think I will code that up so I can verify that and we can compare > notes later today. This is basically done here, I'll post this asap.