From mboxrd@z Thu Jan 1 00:00:00 1970 From: Patrick McHardy Subject: Re: [PATCH 2.6 0/12]: netfilter update Date: Wed, 22 Sep 2004 03:42:24 +0200 Sender: netfilter-devel-bounces@lists.netfilter.org Message-ID: <4150D880.70108@trash.net> References: <414F9DFC.701@trash.net> <20040921143611.6a6e2f60.davem@redhat.com> <4150BB62.2060904@trash.net> <20040921171857.2c720c8e.davem@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Cc: netfilter-devel@lists.netfilter.org Return-path: To: "David S. Miller" In-Reply-To: <20040921171857.2c720c8e.davem@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: netfilter-devel-bounces@lists.netfilter.org List-Id: netfilter-devel.vger.kernel.org David S. Miller wrote: >Thanks, two problems: > >1) Please layout the new skbuff.h structure changes more > intelligently. Currently the netfilter stuff is: > > pointer > u32 > pointer > u32 > > and this wastes a lot of space on 64-bitters. > Oops. That wasn't very smart of me. >2) There is no way I'm amicable to exporting __tcp_v4_lookup() > and friends to modules. > > Are you absolutely sure that the existing scheme cannot > work properly? I'm talking about relying upon skb->sk > being sane. > > The existing scheme has one problem, it takes files->file_lock in a BH, but the patch doesn't mean to fix this. Besides this, it works fine. The patch extends the owner match so it can also be used in the input path, where skb->sk isn't set. This is quite useful for filtering input traffic based on user etc., unfortunately it can't be done without exporting {tcp,udp}_v4_lookup. Do you see an alternative to dropping the patch ? >So what I think I'll do is merge in all the changes other >than the ipt_owner.c change by hand into a tree I'll push >to Linus, I'll also fix the layout ordering problem which >I mentioned above in #1 > > Thanks. Regards Patrick