Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Message-ID: <cdba4a3b7f31ae8ece81be270233032fe774bd86.camel@samsung.com>
Subject: Re: [PATCH v2] netfilter: xt_owner: Add supplementary groups option
From:   Lukasz Pawelczyk <l.pawelczyk@samsung.com>
To:     Eric Dumazet <eric.dumazet@gmail.com>,
        Pablo Neira Ayuso <pablo@netfilter.org>,
        Jozsef Kadlecsik <kadlec@blackhole.kfki.hu>,
        Florian Westphal <fw@strlen.de>,
        "David S. Miller" <davem@davemloft.net>,
        netfilter-devel@vger.kernel.org, coreteam@netfilter.org,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Cc:     Lukasz Pawelczyk <havner@gmail.com>
Date:   Wed, 08 May 2019 17:25:35 +0200
In-Reply-To: <98f71c64-3887-b715-effb-894224a71ef9@gmail.com>
User-Agent: Evolution 3.30.5 (3.30.5-1.fc29) 
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset="utf-8"
CMS-TYPE: 201P
References: <CGME20190508141219eucas1p1e5a899714747b497499976113ea9681f@eucas1p1.samsung.com>
        <20190508141211.4191-1-l.pawelczyk@samsung.com>
        <98f71c64-3887-b715-effb-894224a71ef9@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, 2019-05-08 at 07:58 -0700, Eric Dumazet wrote:
> 
> On 5/8/19 10:12 AM, Lukasz Pawelczyk wrote:
> > The XT_SUPPL_GROUPS flag causes GIDs specified with XT_OWNER_GID to
> > be also checked in the supplementary groups of a process.
> > 
> > Signed-off-by: Lukasz Pawelczyk <l.pawelczyk@samsung.com>
> > ---
> >  include/uapi/linux/netfilter/xt_owner.h |  1 +
> >  net/netfilter/xt_owner.c                | 23 ++++++++++++++++++++-
> > --
> >  2 files changed, 21 insertions(+), 3 deletions(-)
> > 
> > diff --git a/include/uapi/linux/netfilter/xt_owner.h
> > b/include/uapi/linux/netfilter/xt_owner.h
> > index fa3ad84957d5..d646f0dc3466 100644
> > --- a/include/uapi/linux/netfilter/xt_owner.h
> > +++ b/include/uapi/linux/netfilter/xt_owner.h
> > @@ -8,6 +8,7 @@ enum {
> >  	XT_OWNER_UID    = 1 << 0,
> >  	XT_OWNER_GID    = 1 << 1,
> >  	XT_OWNER_SOCKET = 1 << 2,
> > +	XT_SUPPL_GROUPS = 1 << 3,
> >  };
> >  
> >  struct xt_owner_match_info {
> > diff --git a/net/netfilter/xt_owner.c b/net/netfilter/xt_owner.c
> > index 46686fb73784..283a1fb5cc52 100644
> > --- a/net/netfilter/xt_owner.c
> > +++ b/net/netfilter/xt_owner.c
> > @@ -91,11 +91,28 @@ owner_mt(const struct sk_buff *skb, struct
> > xt_action_param *par)
> >  	}
> >  
> >  	if (info->match & XT_OWNER_GID) {
> > +		unsigned int i, match = false;
> >  		kgid_t gid_min = make_kgid(net->user_ns, info-
> > >gid_min);
> >  		kgid_t gid_max = make_kgid(net->user_ns, info-
> > >gid_max);
> > -		if ((gid_gte(filp->f_cred->fsgid, gid_min) &&
> > -		     gid_lte(filp->f_cred->fsgid, gid_max)) ^
> > -		    !(info->invert & XT_OWNER_GID))
> > +		struct group_info *gi = filp->f_cred->group_info;
> > +
> > +		if (gid_gte(filp->f_cred->fsgid, gid_min) &&
> > +		    gid_lte(filp->f_cred->fsgid, gid_max))
> > +			match = true;
> > +
> > +		if (!match && (info->match & XT_SUPPL_GROUPS) && gi) {
> > +			for (i = 0; i < gi->ngroups; ++i) {
> > +				kgid_t group = gi->gid[i];
> > +
> > +				if (gid_gte(group, gid_min) &&
> > +				    gid_lte(group, gid_max)) {
> > +					match = true;
> > +					break;
> > +				}
> > +			}
> > +		}
> > +
> > +		if (match ^ !(info->invert & XT_OWNER_GID))
> >  			return false;
> >  	}
> >  
> > 
> 
> How can this be safe on SMP ?
> 

From what I see after the group_info rework some time ago this struct
is never modified. It's replaced. Would get_group_info/put_group_info
around the code be enough?


-- 
Lukasz Pawelczyk
Samsung R&D Institute Poland
Samsung Electronics