Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <20220118070029.1157324-1-liu3101@purdue.edu>
In-Reply-To: <20220118070029.1157324-1-liu3101@purdue.edu>
From:   Eric Dumazet <edumazet@google.com>
Date:   Tue, 18 Jan 2022 00:40:56 -0800
Message-ID: <CANn89iJevGzP5r6sPXpX=pSxPJWZQHjKKYekZpFTG9xEq50pMg@mail.gmail.com>
Subject: Re: [PATCH net] net: fix information leakage in /proc/net/ptype
To:     Congyu Liu <liu3101@purdue.edu>
Cc:     David Miller <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>,
        Yajun Deng <yajun.deng@linux.dev>,
        Willem de Bruijn <willemb@google.com>,
        Marc Kleine-Budde <mkl@pengutronix.de>, rsanger@wand.net.nz,
        Wang Hai <wanghai38@huawei.com>,
        Pablo Neira Ayuso <pablo@netfilter.org>,
        jiapeng.chong@linux.alibaba.com, netdev <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Jan 17, 2022 at 11:01 PM Congyu Liu <liu3101@purdue.edu> wrote:
>
> In one net namespace, after creating a packet socket without binding
> it to a device, users in other net namespaces can observe the new
> `packet_type` added by this packet socket by reading `/proc/net/ptype`
> file. I believe this is minor information leakage as packet socket is
> namespace aware.
>
> Add a function pointer in `packet_type` to retrieve the net namespace
> of corresponding packet socket. In `ptype_seq_show`, if this
> function pointer is not NULL, use it to determine if certain ptype
> should be shown.
>
> Signed-off-by: Congyu Liu <liu3101@purdue.edu>
> ---
>  include/linux/netdevice.h |  1 +
>  net/core/net-procfs.c     |  3 ++-
>  net/packet/af_packet.c    | 18 ++++++++++++++++++
>  3 files changed, 21 insertions(+), 1 deletion(-)
>
> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
> index 3213c7227b59..72d3601850c5 100644
> --- a/include/linux/netdevice.h
> +++ b/include/linux/netdevice.h
> @@ -2548,6 +2548,7 @@ struct packet_type {
>                                               struct net_device *);
>         bool                    (*id_match)(struct packet_type *ptype,
>                                             struct sock *sk);
> +       struct net              *(*get_net) (struct packet_type *ptype);
>         void                    *af_packet_priv;
>         struct list_head        list;
>  };

Patch looks fine, but the question is:

Can an af_packet socket created in netns A can be moved to netns B later ?

As the answer is probably no, it seems we could simply add a 'struct
net'  pointer
 in 'struct packet type', no need for a function.

Thanks.