Re: BPF redirect API design issue for BPF-prog MTU feedback?

From: Jesper Dangaard Brouer <brouer@redhat.com>
To: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Cc: "Marek Zavodsky" <marek.zavodsky@gmail.com>,
	"Daniel Borkmann" <daniel@iogearbox.net>,
	"Lorenz Bauer" <lmb@cloudflare.com>,
	"Maciej Żenczykowski" <maze@google.com>,
	"Saeed Mahameed" <saeed@kernel.org>,
	"Daniel Borkmann" <borkmann@iogearbox.net>,
	"Alexei Starovoitov" <alexei.starovoitov@gmail.com>,
	BPF-dev-list <bpf@vger.kernel.org>,
	"netdev@vger.kernel.org" <netdev@vger.kernel.org>,
	"Lorenzo Bianconi" <lorenzo.bianconi@redhat.com>,
	"John Fastabend" <john.fastabend@gmail.com>,
	"Jakub Kicinski" <kuba@kernel.org>,
	"Shaun Crampton" <shaun@tigera.io>,
	"David Miller" <davem@davemloft.net>,
	"Marek Majkowski" <marek@cloudflare.com>,
	brouer@redhat.com
Subject: Re: BPF redirect API design issue for BPF-prog MTU feedback?
Date: Tue, 22 Sep 2020 11:15:43 +0200	[thread overview]
Message-ID: <20200922111543.7cefb3b1@carbon> (raw)
In-Reply-To: <CA+FuTSdx_a+DGG5dSFRB3wkowwNb1ZXHFed=qA3sj5y6U3VtiA@mail.gmail.com>

On Mon, 21 Sep 2020 23:17:16 +0200
Willem de Bruijn <willemdebruijn.kernel@gmail.com> wrote:

> On Mon, Sep 21, 2020 at 6:22 PM Marek Zavodsky <marek.zavodsky@gmail.com> wrote:
> >
> > Hi guys,
> >
> > My kernel knowledge is small, but I experienced this (similar) issue
> > with packet encapsulation (not a redirect), therefore modifying the
> > redirect branch would not help in my case.
> >
> > I'm working on a TC program to do GUE encap/decap (IP + UDP + GUE,
> > outer header has extra 52B).
> > There are no issues with small packets. But when I use curl to
> > download big file HTTP server chunks data randomly. Some packets have
> > MTU size, others are even bigger. Big packets are not an issue,
> > however MTU sized packets fail on bpf_skb_adjust_room with -524
> > (ENOTSUPP).  
> 
> This is a related, but different, unresolved issue at the boundary of
> GSO packets. Packets that are not GSO, but would exceed MTU once
> encapsulated, will cause adjust room to fail:
> 
>             (!shrink && (skb->len + len_diff_abs > len_max &&
>                          !skb_is_gso(skb))))
>                 return -ENOTSUPP;
> 
> As admin, this can be addressed by setting a lower route MTU on routes
> that may be encapsulated. But that is not very obvious or transparent.

Your issue is very much related, and even-though it is not related to
redirect, I also want to address and allow your use-case (in the
patchset that I'm collecting input for now).

I do think this patch[1][2] will actually solve your problem.  Could
you please try to apply and test this to make sure?  (as we have
discussed on this list, that patch is not a 100% solution, and I will
work on a better solution).

[1] https://lore.kernel.org/bpf/159921182827.1260200.9699352760916903781.stgit@firesoul
[2] http://patchwork.ozlabs.org/project/netdev/patch/159921182827.1260200.9699352760916903781.stgit@firesoul/
-- 
Best regards,
  Jesper Dangaard Brouer
  MSc.CS, Principal Kernel Engineer at Red Hat
  LinkedIn: http://www.linkedin.com/in/brouer