Stared at the code for a bit, and fed this to the Linux DCO server
testbed (excercising multisocket only lightly).
Some of the changes are really straightforward (like not calling
socket_set_listen_persistent() for dco_win, where the socket lives
in kernel space), or just not calling (windows-only) wait_signal()
for servers where win-DCO needs different handling.
The change to multi_tcp_process_outgoing_link() feels wrong, though
- yes, "what we do there" should obviously not been done for a DGRAM
packet, but why is the function then called in the first place? So
I guess there are a few more acts coming in this multisocket/event
drama...
Anyway, this does not really affect Linux/FreeBSD DCO, and testing
says it's not breaking anything - and it fixes Win/Server DCO, so
one step forward at a time.
Your patch has been applied to the master branch.
commit 831c585e96587e28d1390d165d89b4152ba03bd3
Author: Gianmarco De Gregori
Date: Thu Apr 10 23:50:29 2025 +0200
Fix dco_win and multisocket interaction
Signed-off-by: Gianmarco De Gregori <[email protected]>
Acked-by: Lev Stipakov <[email protected]>
Message-Id: <[email protected]>
URL:
https://www.mail-archive.com/[email protected]/msg31419.html
Signed-off-by: Gert Doering <[email protected]>
--
kind regards,
Gert Doering
_______________________________________________
Openvpn-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openvpn-devel
