Re: [PATCHv2 net-next 02/17] udp6: move the mss check after udp gso tunnel processing

From: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
To: Xin Long <lucien.xin@gmail.com>
Cc: Willem de Bruijn <willemdebruijn.kernel@gmail.com>,
	network dev <netdev@vger.kernel.org>,
	linux-sctp@vger.kernel.org,
	Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>,
	Neil Horman <nhorman@tuxdriver.com>,
	Michael Tuexen <tuexen@fh-muenster.de>,
	David Miller <davem@davemloft.net>
Subject: Re: [PATCHv2 net-next 02/17] udp6: move the mss check after udp gso tunnel processing
Date: Fri, 09 Oct 2020 13:59:26 +0000	[thread overview]
Message-ID: <CA+FuTSd1MZusO_9W8tkn99jGCu94o3K+6e8zROz_FNEMWt6Crg@mail.gmail.com> (raw)
In-Reply-To: <CADvbK_en7mePKdmMaLr9V8hTdmjf2bSVpSrid2CjharJtvD6YQ@mail.gmail.com>

On Thu, Oct 8, 2020 at 9:49 PM Xin Long <lucien.xin@gmail.com> wrote:
>
> On Thu, Oct 8, 2020 at 8:45 PM Willem de Bruijn
> <willemdebruijn.kernel@gmail.com> wrote:
> >
> > On Thu, Oct 8, 2020 at 5:48 AM Xin Long <lucien.xin@gmail.com> wrote:
> > >
> > > For some protocol's gso, like SCTP, it's using GSO_BY_FRAGS for
> > > gso_size. When using UDP to encapsulate its packet, it will
> > > return error in udp6_ufo_fragment() as skb->len < gso_size,
> > > and it will never go to the gso tunnel processing.
> > >
> > > So we should move this check after udp gso tunnel processing,
> > > the same as udp4_ufo_fragment() does. While at it, also tidy
> > > the variables up.
> >
> > Please don't mix a new feature and code cleanup.
> Hi, Willem,
>
> Tidying up variables are not worth a single patch, that's what I was
> thinking. I can leave the variables as it is if you wish in this patch.
>
> >
> > This patch changes almost every line of the function due to
> > indentation changes. But the only relevant part is
> >
> > "
> >         mss = skb_shinfo(skb)->gso_size;
> >         if (unlikely(skb->len <= mss))
> >                 goto out;
> >
> >         if (skb->encapsulation && skb_shinfo(skb)->gso_type &
> >             (SKB_GSO_UDP_TUNNEL|SKB_GSO_UDP_TUNNEL_CSUM))
> >                 segs = skb_udp_tunnel_segment(skb, features, true);
> >         else {
> >                 /* irrelevant here */
> >         }
> >
> > out:
> >         return segs;
> > }
> > "
> >
> > Is it a sufficient change to just skip the mss check if mss = GSO_BY_FRAGS?
> It is sufficient.
>
> But I think we'd better keep the code here consistent with ipv4's if
> there's no other reason to do 'skb->len <= mss' check at the first.
>
> We can go with if-else as you showed above now, then do a cleanup in
> the future. What do you think?

I do think that's a better approach, thanks.

Mixing in feature changes with 150+ lines of code refactoring makes it
hard to spot the behavioral change.

Too often the additional purportedly noop refactoring proves to have
subtle behavioral changes after all. It's easier to review for this if
that patch is separate. Not saying it's the case here, of course.

Refactoring also adds a layer of indirection when using git blame to
understand code or for people who have the existing code memorized, so
has a fairly high bar imho.