From mboxrd@z Thu Jan  1 00:00:00 1970
From: Pablo Neira Ayuso <pablo@netfilter.org>
Subject: Re: [PATCH RESEND] netfilter: bridge: unshare bridge info before
 change it
Date: Wed, 19 Nov 2014 12:49:09 +0100
Message-ID: <20141119114909.GA5014@salvia>
References: <1416203323-22745-1-git-send-email-gaofeng@cn.fujitsu.com>
 <20141118183419.GA5078@salvia>
 <546BFC4D.7070203@cn.fujitsu.com>
 <20141119111720.GA4537@salvia>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: netfilter-devel@vger.kernel.org
To: Gao feng <gaofeng@cn.fujitsu.com>
Return-path: <netfilter-devel-owner@vger.kernel.org>
Received: from mail.us.es ([193.147.175.20]:33763 "EHLO mail.us.es"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S1750728AbaKSLrQ (ORCPT <rfc822;netfilter-devel@vger.kernel.org>);
	Wed, 19 Nov 2014 06:47:16 -0500
Content-Disposition: inline
In-Reply-To: <20141119111720.GA4537@salvia>
Sender: netfilter-devel-owner@vger.kernel.org
List-ID: <netfilter-devel.vger.kernel.org>

On Wed, Nov 19, 2014 at 12:17:20PM +0100, Pablo Neira Ayuso wrote:
> On Wed, Nov 19, 2014 at 10:11:25AM +0800, Gao feng wrote:
> > On 11/19/2014 02:34 AM, Pablo Neira Ayuso wrote:
> > > On Mon, Nov 17, 2014 at 01:48:43PM +0800, Gao feng wrote:
> > >> diff --git a/net/bridge/br_netfilter.c b/net/bridge/br_netfilter.c
> > >> index 1a4f32c..b4612b9 100644
> > >> --- a/net/bridge/br_netfilter.c
> > >> +++ b/net/bridge/br_netfilter.c
> > >> @@ -653,7 +643,11 @@ static int br_nf_forward_finish(struct sk_buff *skb)
> > >>  		in = nf_bridge->physindev;
> > >>  		if (nf_bridge->mask & BRNF_PKT_TYPE) {
> > >>  			skb->pkt_type = PACKET_OTHERHOST;
> > >> -			nf_bridge->mask ^= BRNF_PKT_TYPE;
> > >> +
> > >> +			if (!nf_bridge_unset_mask(skb, BRNF_PKT_TYPE)) {
> > >> +				kfree_skb(skb);
> > >> +				return 0;
> > >> +			}
> > > 
> > > This can now release the packet and, thus, drop it.
> > > 
> > > However, br_nf_forward_ip() always returns NF_STOLEN.
> > > 
> > > Could you revisit the error paths and confirm they are correct?
> > 
> > Since br_nf_forward_ip() returns NF_STOLEN, br_nf_forward_finish should make
> > sure this skb will be released. if unser_mask for this skb is failed, we free
> > it. if unset successed, the hooks on NF_BR_FORWARD or br_forward_finish will
> > free this skb.
> 
> OK, but I think this should return NF_DROP instead of NF_STOLEN so you
> don't explicitly need to kfree_skb it.

After a further look, I noticed what I suggest needs to replace:

        NF_HOOK_THRESH(...);
        return 0;

by:
        return NF_HOOK_THRESH(...);

And review the return value of all the existing okfn() functions so
they return netfilter verdicts.

The existing code always returns NF_STOLEN from many other hooks, not
only when dropped but also when accepted to make this work...

I'm going to have a look at your v2. Thanks.