So there was this function that looked slightly odd after the multisocket
event changes, and in the process from v1 to v7 in this patch, the offending
code ("udp_flags = ES_ERROR") got thrown out completely ;-)
I won't claim I understand most of the event / flags magic here, but this
part seems far more understandable than before now, and it seems to do
the right thing dealing with a socket array, correlating "we have a
pending multi instance" with "what socket to send the packet out?".
Some further surgery to p2mp_iow_flags() and multi_io_wait() might
make this even better :-) (include the first in the caller, and get
rid of the has_udp_in_local_list() outer check) - but this is future
work.
*This* has been tested across multiple version on the t_server testbed
(and everything works), and I understand Gianmarco has fed it to the
gremlins and nothing crashed either. This is about as good as it gets,
I think...
Your patch has been applied to the master and release/2.7 branch.
commit 1281d264a259bf8e95695bbb831737a4778d4afa (master)
commit ae5083017d70c695813cd7974d1fabbf0eb7307a (release/2.7)
Author: Gianmarco De Gregori
Date: Tue Jun 16 18:21:23 2026 +0200
Multisocket: use event engine rwflags for UDP I/O
Signed-off-by: Gianmarco De Gregori <[email protected]>
Acked-by: Gert Doering <[email protected]>
Gerrit URL: https://gerrit.openvpn.net/c/openvpn/+/1635
Message-Id: <[email protected]>
URL:
https://www.mail-archive.com/[email protected]/msg37198.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
