Received-SPF: softfail (google.com: domain of transitioning linux-kernel-owner@vger.kernel.org does not designate 23.128.96.19 as permitted sender) client-ip=23.128.96.19;
MIME-Version: 1.0
References: <CAHmME9r_DbZWe4FsfebHSSf_iPctSe5S-w9bU3o8BN43raeURg@mail.gmail.com>
 <20151116203709.GA27178@oracle.com> <CAHmME9pNCqbcoqbOnx6p8poehAntyyy1jQhy=0_HjkJ8nvMQdw@mail.gmail.com>
 <1447712932.22599.77.camel@edumazet-glaptop2.roam.corp.google.com>
 <CAHmME9oTU7HwP5=qo=aFWe0YXv5EPGoREpF2k-QY7qTmkDeXEA@mail.gmail.com>
 <YmszSXueTxYOC41G@zx2c4.com> <04f72c85-557f-d67c-c751-85be65cb015a@gmail.com>
 <YmxTo2hVwcwhdvjO@zx2c4.com> <d9854c74-c209-9ea5-6c76-8390e867521b@gmail.com> <CAHmME9qXC-4OPc5xRbC6CQJcpzb96EXzNWAist5A8momYxvVUA@mail.gmail.com>
In-Reply-To: <CAHmME9qXC-4OPc5xRbC6CQJcpzb96EXzNWAist5A8momYxvVUA@mail.gmail.com>
From:   Eric Dumazet <edumazet@google.com>
Date:   Fri, 29 Apr 2022 15:05:23 -0700
Message-ID: <CANn89iLyNoCRrp6YYdy6kGhM7X2JQ9J4-LfEJCBvhYAv4N+FPA@mail.gmail.com>
Subject: Re: Routing loops & TTL tracking with tunnel devices
To:     "Jason A. Donenfeld" <Jason@zx2c4.com>
Cc:     Eric Dumazet <eric.dumazet@gmail.com>,
        Netdev <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Jakub Kicinski <kuba@kernel.org>,
        Hannes Frederic Sowa <hannes@stressinduktion.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Fri, Apr 29, 2022 at 2:54 PM Jason A. Donenfeld <Jason@zx2c4.com> wrote:
>
> Hi Eric,
>
> On Fri, Apr 29, 2022 at 11:14 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
> >
> >
> > On 4/29/22 14:07, Jason A. Donenfeld wrote:
> >
> > Hi Eric,
> >
> > On Fri, Apr 29, 2022 at 01:54:27PM -0700, Eric Dumazet wrote:
> >
> > Anyway, it'd be nice if there were a free u8 somewhere in sk_buff that I
> > could use for tracking times through the stack. Other kernels have this
> > but afaict Linux still does not. I looked into trying to overload some
> > existing fields -- tstamp/skb_mstamp_ns or queue_mapping -- which I was
> > thinking might be totally unused on TX?
> >
> > if skbs are stored in some internal wireguard queue, can not you use
> > skb->cb[],
> >
> > like many other layers do ?
> >
> > This isn't for some internal wireguard queue. The packets get sent out
> > of udp_tunnel_xmit_skb(), so they leave wireguard's queues.
> >
> >
> > OK, where is the queue then ?
> >
> > dev_xmit_recursion() is supposed to catch long chains of virtual devices.
>
> This is the long-chain-of-virtual-devices case indeed. But
> dev_xmit_recursion() does not help here, since that's just a per-cpu
> increment, assuming that the packet actually gets xmit'd in its
> ndo_start_xmit function. But in reality, wireguard queues up the
> packet, encrypts it in some worker later, and eventually transmits it
> with udp_tunnel_xmit_skb(). All the while ndo_start_xmit() has long
> returned. So no help from dev_xmit_recursion().
>

I assume you add encap headers to the skb ?

You could check if the wireguard header is there already, or if the
amount of headers is crazy.

net/core/flow_dissector.c might help.

You also can take a look at CONFIG_SKB_EXTENSIONS infrastructure.