From mboxrd@z Thu Jan  1 00:00:00 1970
From: Eric Dumazet <eric.dumazet@gmail.com>
Subject: Re: Regression, bisected: reference leak with IPSec since ~2.6.31
Date: Tue, 21 Sep 2010 12:07:37 +0200
Message-ID: <1285063657.2617.226.camel@edumazet-laptop>
References: <20100921091248.GA8424@ff.dom.local>
	 <1285060860.2617.158.camel@edumazet-laptop>
	 <20100921093853.GA8664@ff.dom.local>
	 <1285062948.2617.204.camel@edumazet-laptop>
Mime-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: QUOTED-PRINTABLE
Cc: Nick Bowler <nbowler@elliptictech.com>,
	linux-kernel@vger.kernel.org, netdev@vger.kernel.org,
	"David S. Miller" <davem@davemloft.net>
To: Jarek Poplawski <jarkao2@gmail.com>
Return-path: <linux-kernel-owner@vger.kernel.org>
In-Reply-To: <1285062948.2617.204.camel@edumazet-laptop>
Sender: linux-kernel-owner@vger.kernel.org
List-Id: netdev.vger.kernel.org

Le mardi 21 septembre 2010 =C3=A0 11:55 +0200, Eric Dumazet a =C3=A9cri=
t :
> Le mardi 21 septembre 2010 =C3=A0 09:38 +0000, Jarek Poplawski a =C3=A9=
crit :
> > On Tue, Sep 21, 2010 at 11:21:00AM +0200, Eric Dumazet wrote:
>=20
> > I hope you're right with this.
> >=20
> > >=20
> > > I liked the :
> > >=20
> > > <check everything without changing state>
> > > if something wrong
> > > 	goto slow_path
> > > else
> > > 	<OK, lets do destructive things>
> > >=20
> >=20
> > But it's an optimization of the "unlikely" case btw. ;-)
> >=20
>=20
> This is a bug fix, in a complex function.
>=20
> in -next branch, we can try to be smart, adding more code in slow_pat=
h
> to revert what was done in the hope fast path would be taken.
> And pray we dont add another bug :)
>=20

My initial patch was somehing like :

<loop to check everything is ok>
  compute truesizes =3D sum (frag->truesize)
  not modifying frag->sk/frag->destructor

if problem -> goto slow_path
else {
	skb->truesize -=3D truesizes;
       send(skb)

	for each frag
		set frag->sk =3D original_sk
                set frag->destructor =3D sock_wfree
		send(frag)

Problem was if we got an error while sending skb or one frag :
It was leaking some truesize accounting.
(an extra loop was needed to set frag->sk/frag->destructor for remainin=
g
frags before freeing them)

So I added the extra loop to :

for each frag {
	set frag->sk =3D skb->sk
	set frag->destructor =3D sock_wfree
	skb->truesize -=3D frag->truesize
}

It removed the need to compute sum(frag->truesize) in a temp variable
(truesizes)