Hi Florian,
Thank you for your answer!
On 01/10/2019 17:00, Florian Westphal wrote:
Matthieu Baerts <matthieu.baerts(a)tessares.net> wrote:
> At the last meeting, we said it would be good to send a new RFCv2 to
> netdev before the next meeting while windows are closed. I guess today
> would be good, no?
Thanks for bringing this up.
> What do we still need to do?
> - the cover letter: Mat is it still OK to do it on your side?
> - any other patches to apply before this?
I think current export branch is fine for RFCv2.
> Here are the pending patches:
[..]
> - mptcp: Remove all traces of checksum support: do we want it?
Do you mean in general or for RFC?
For RFC I think it depends on how much work you want to
get yourself into right now, as it would need to be squashed, sending
it as extra patch is strange.
In general. It is linked to my message I sent on the thread linked to
the patch:
====
On my side, I was looking at applying this patch. And after having
split it in 5 smaller patches and starting applying half of them, I
don't know if it is a good idea to fully drop the checksum (partial)
support.
The main part of this patch is removing the parsing of the checksum
option. But in the commits where this is introduced, we also parse
other possible options we could have and we act differently if one
option is not supported. And in these commits, it makes sense to do
that.
We could see that as "we are parsing stuff we don't need". But we are
already doing that for the "backup" bit for example. Or for the
"MP_JOIN" while we don't support it when subflow_request_sock
structure is introduced, etc.
In other words, do we really need to remove this code linked to the
checksum? If we plan never to support it, it makes sense. But I guess
in the near future, we will want to support it, no?
So should I continue applying this patch?
====
But I can continue to apply it, I already did the splitting work. But I
preferred to raise the question now to avoid a revert of all of this :)
> - mptcp: add MIB counter infrastructure: Waiting for accept (last review)
> - mptcp: allow MPTCP sockets by default: can be applied
> - mptcp: prefix mptcp_ to exposed pm_ routines.
These three would be nice to have for sure.
I can work on that! (Once the first one is accepted :) )
> For the moment, I am blocked with "mptcp: Remove all traces
of checksum
> support" but I can drop this rebase to work on other patches if others
> are required for the RFCv2.
>
> What's your point of view on this?
I think merging in the checksum support remove before RFCv2 and the
other three (MIB, mptcp-enable-by-default, mptcp_pm_ prefix) would be
great. I think its even ok to delay the series for a few days to get
those in.
If we have an agreement on the patch removing the checksum support, I
can apply everything needed tomorrow and the whole patch-set can be sent
during the evening/night for us if it is OK for Mat!
Let me know if i can help.
Thank you!
It should be fine for me to apply the different patches :)
Cheers,
Matt
--
Matthieu Baerts | R&D Engineer
matthieu.baerts(a)tessares.net
Tessares SA | Hybrid Access Solutions
www.tessares.net
1 Avenue Jean Monnet, 1348 Louvain-la-Neuve, Belgium