On Fri, 2019-10-25 at 17:20 +0200, Matthieu Baerts wrote:
Hi Paolo,
On 25/10/2019 09:53, Matthieu Baerts wrote:
> - MPTCP v1 support → part 2:
> - Paolo started to look at the RFC to support Christoph
> - Christoph will be the submitter of the patch
> - *Matth*: publish a branch with
mptcp.org + MPTCPv1
I didn't have a lot of time to look at it but you can find
mptcp.org
kernel with Christoph patches (and a small fix) to support MPTCPv1 on my
Github account:
https://github.com/matttbe/mptcp_net-next/tree/mptcp_v0.95_mptcp_v1
https://github.com/matttbe/mptcp_net-next/tree/mptcp_trunk_mptcp_v1
I would recommend to use the branch based on mptcp_v0.95 (last upstream
LTS, v4.19.55).
Before testing, don't forget to set:
sysctl net.mptcp.mptcp_version=1
sysctl net.mptcp.mptcp_enabled=1
Thank you very much, the above will help!
So far, I *think* the establishment of the connection is OK (SYN,
SYN+ACK, ACK) but not after: it seems the first data packet has bad
options. I would need to look more to understand why.
Do you see connection/test failures, or is just the wireshark dissector
output? Note that wireshark can't currently correctly dissect mptcp v1
MP_CAPABLE ACK option with data_len and/or checksum - it's a wireshark
bug/missing feature
Cheers,
Paolo