Re: [PATCH net-next 10/10] ovpn: ensure sk is still valid during cleanup

On 13/05/2025 11:19, Antonio Quartulli wrote:
[...]

quoted

Side note: the ovpn_socket refcount release/detach path looks wrong, at
least in case of an UDP socket, as ovpn_udp_socket_detach() calls
setup_udp_tunnel_sock() which in turns will try to _increment_ various
core counters, instead of decreasing them (i.e. udp_encap_enable should
be wrongly accounted after that call).

You're right.
I had the impression I needed to "undo" the setup.
I see now that the encap key is decremented in the UDP sock destroy, 
right after having called my implementation of .destroy().

I'll drop the call to setup_udp_tunnel_sock() with empty config then.

Paolo, the reason for calling setup_udp_tunnel_sock() with an empty 
config was to "reset" its encap state (for udp_tunnel).

Technically a UDP socket could go back to being a pure userspace socket 
only (i.e. when all peers using that socket have been deleted), 
therefore I wanted to make sure that the kernel would not try to 
intercept its packets anymore.

I understand setup_udp_tunnel_sock(cfg={}) is not correct, but what are 
my options?

1) I could nullify manually sk_user_data and udp_sk(sk)->encap_type;

2) I could implement something like "stop_udp_tunnel_sock()" that 
reverts what setup_udp_tunnel_sock() had done and call it.

Opinions?
I'd go with 2.

Thanks a lot!


-- 
Antonio Quartulli
OpenVPN Inc.