On 01/04/2025 15:49, Sabrina Dubroca wrote:
2025-03-18, 02:40:42 +0100, Antonio Quartulli wrote:+static int ovpn_udp_output(struct ovpn_peer *peer, struct dst_cache *cache, + struct sock *sk, struct sk_buff *skb) +{ + struct ovpn_bind *bind; + int ret; + + /* set sk to null if skb is already orphaned */ + if (!skb->destructor) + skb->sk = NULL; + + /* always permit openvpn-created packets to be (outside) fragmented */ + skb->ignore_df = 1;Have you tested this with IPv4 encap? AFAICT it doesn't have any effect because of the call chain: ovpn_udp4_output -> udp_tunnel_xmit_skb -> iptunnel_xmit -> skb_scrub_packet which does skb->ignore_df = 0; But since you pass df = 0 to udp_tunnel_xmit_skb, I suspect it works as intended despite skb_scrub_packet.
Yeah, seems so. Passing df = 0 basically does what we need. So you're right, that ignore_df = 1 is useless. Will drop it.
[note: that was the last comment I wanted to send, I have a few more suggestions that don't need to be addressed at this time]
Thanks! :-) I'll answer all other open comments in the meantime.. Regards,
+ + rcu_read_lock(); + bind = rcu_dereference(peer->bind); + if (unlikely(!bind)) { + net_warn_ratelimited("%s: no bind for remote peer %u\n", + netdev_name(peer->ovpn->dev), peer->id); + ret = -ENODEV; + goto out; + } + + switch (bind->remote.in4.sin_family) { + case AF_INET: + ret = ovpn_udp4_output(peer, bind, cache, sk, skb); + break; +#if IS_ENABLED(CONFIG_IPV6) + case AF_INET6: + ret = ovpn_udp6_output(peer, bind, cache, sk, skb); + break; +#endif + default: + ret = -EAFNOSUPPORT; + break; + } + +out: + rcu_read_unlock(); + return ret; +}
-- Antonio Quartulli OpenVPN Inc.
