Message-ID: <1523373245.2654.182.camel@codethink.co.uk>
Subject: Re: [PATCH 4.4 38/97] netfilter: xt_CT: fix refcnt leak on error
 path
From: Ben Hutchings <ben.hutchings@codethink.co.uk>
To: Gao Feng <fgao@ikuai8.com>, Liping Zhang <zlpnobody@gmail.com>,
        Pablo Neira Ayuso <pablo@netfilter.org>
Cc: stable@vger.kernel.org,
        Sasha Levin <alexander.levin@microsoft.com>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        LKML <linux-kernel@vger.kernel.org>
Date: Tue, 10 Apr 2018 16:14:05 +0100
In-Reply-To: <1522777564.2654.115.camel@codethink.co.uk>
References: <20180323094157.535925724@linuxfoundation.org>
         <20180323094159.781131756@linuxfoundation.org>
         <1522777564.2654.115.camel@codethink.co.uk>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 8bit
Sender: stable-owner@vger.kernel.org

On Tue, 2018-04-03 at 18:46 +0100, Ben Hutchings wrote:
> On Fri, 2018-03-23 at 10:54 +0100, Greg Kroah-Hartman wrote:
> > 4.4-stable review patch.  If anyone has any objections, please let me know.
> > 
> > ------------------
> > 
> > From: Gao Feng <fgao@ikuai8.com>
> > 
> > 
> > [ Upstream commit 470acf55a021713869b9bcc967268ac90c8a0fac ]
[...] 
> > @@ -249,7 +252,7 @@ static int xt_ct_tg_check(const struct x
> >  	if (info->timeout[0]) {
> >  		ret = xt_ct_set_timeout(ct, par, info->timeout);
> >  		if (ret < 0)
> > -			goto err3;
> > +			goto err4;
> >  	}
> >  	__set_bit(IPS_CONFIRMED_BIT, &ct->status);
> >  	nf_conntrack_get(&ct->ct_general);
> > @@ -257,6 +260,10 @@ out:
> >  	info->ct = ct;
> >  	return 0;
> >  
> > +err4:
> > +	help = nfct_help(ct);
> > +	if (help)
> > +		module_put(help->helper->me);
> >  err3:
> >  	nf_ct_tmpl_free(ct);
> >  err2:
> 
> This does not.  nf_ct_tmpl_free() calls nf_ct_ext_destroy() which I
> think will call back into xt_ct_tg_destroy().  So I think the module
> reference is already dropped here and we mustn't do it twice.  Am I
> missing something?

I still don't understand this code, but I have verified that the
module_put() is needed on this error path to balance the module
reference count.  I.e. this fix is good.

Ben.

-- 
Ben Hutchings
Software Developer, Codethink Ltd.