Hi Alexei,

On 02/08/2023 04:23, Alexei Starovoitov wrote:

On Tue, Aug 1, 2023 at 7:19 PM Geliang Tang [off-list ref] wrote:

quoted

On Mon, Jul 31, 2023 at 05:43:23PM -0700, Alexei Starovoitov wrote:

quoted

On Sat, Jul 29, 2023 at 05:57:21PM +0800, Geliang Tang wrote:

quoted

The main idea is to add a hook in sys_socket() to change the protocol id
from IPPROTO_TCP (or 0) to IPPROTO_MPTCP.

I still think it's a hack, but its blast radius is nicely contained.
And since I cannot propose any better I'm ok with it.

Patches 1-2 can be squashed into one.
Just like patches 3-6 as a single patch for selftests.

Thanks Alexei. I'll squash patch 1 and patch 2 into one, and squash patch 3
and patch 6 into one for selftests. But I prefer to keep patch 4 and patch 5
as is, since they were implemented in different times for different purposes.
They were merged into MPTCP repo on May 17 for "run MPTCP sched tests in a
dedicated ns" [1].

since they were sent to a different tree than don't send them here.
git will not like that during the merge window.

Thank you for the suggestion but that's OK to have these patches applied
in BPF tree because on MPTCP side, they have been applied in a "devel"
branch which is rebased on top of -net and net-next everyday. In other
words, we don't use it in pull requests and it is fine to apply these
patches in bpf-next or elsewhere.

Cheers,
Matt
-- 
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net