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: <CAGbU3_nRMbe8Syo3OHw6B4LXUheGiXXcLcaEQe0EAFTAB7xgng@mail.gmail.com>
 <CAGbU3_=ZywUOP1CKNQ6=P99SgX28_0iXSs81yP=vGFKv7JyMcQ@mail.gmail.com>
 <CAGbU3_krnjbeKnm6Zyn-tqYCHVZFBkB+oCP-UF_kVOGz=zkKFQ@mail.gmail.com> <20200822.140141.880909883327091452.davem@davemloft.net>
In-Reply-To: <20200822.140141.880909883327091452.davem@davemloft.net>
From:   Pascal Bouchareine <kalou@tfz.net>
Date:   Sun, 23 Aug 2020 15:28:23 -0700
Message-ID: <CAGbU3_m_2_6iq-ZCMJ0i7E=4nFi6o=akd9_QpeRSa26=U67yog@mail.gmail.com>
Subject: Re: [PATCH v2 2/2] net: socket: implement SO_DESCRIPTION
To:     David Miller <davem@davemloft.net>
Cc:     linux-kernel@vger.kernel.org, linux-api@vger.kernel.org,
        netdev@vger.kernel.org, Jakub Kicinski <kuba@kernel.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        Alexey Dobriyan <adobriyan@gmail.com>,
        Al Viro <viro@zeniv.linux.org.uk>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Sat, Aug 22, 2020 at 2:01 PM David Miller <davem@davemloft.net> wrote:
> > About the pid part -
> > On top of multiple pids to scan for a given socket, there's also the
> > security provided by /proc - I'm not sure what inet_diag does for that
> > So maybe users calling it will need to scan /proc for a long time anyway...
> >
> > Or is that doable?
>
> I'd like to kindly ask that you do more research into how this kind of
> information is advertised to the user using modern interfaces, and what
> kinds of permissions and checks are done for those.

If we wanted to get rid of having to scan /proc from userland when
using sock_diag to identify associated processes,
I suppose scanning for pids would be the most annoying part?

I understand sock_diag uses CAP_NET_ADMIN for some sensitive bits.

I thought it would require an additional bit of logic to let an
unprivileged user access its own socket "sensitive" data.

Your message makes me think I need to read a lot more about it, so
I'll try that - but more importantly
as you mention APIs and modern interfaces, I think eBPF is going to be
of great help to try and hack
around this data without disturbing existing APIs.

Thanks for taking the time to look into it