Patchwork sit tunnel slow since added GSO/TSO support

login
register
mail settings
Submitter Eric Dumazet
Date Dec. 15, 2013, 1:23 a.m.
Message ID <1387070629.19078.195.camel@edumazet-glaptop2.roam.corp.google.com>
Download mbox | patch
Permalink /patch/301291/
State RFC
Delegated to: David Miller
Headers show

Comments

Eric Dumazet - Dec. 15, 2013, 1:23 a.m.
On Sat, 2013-12-14 at 22:25 +0100, Michal Schmidt wrote:
> On 12/13/2013 11:54 PM, Eric Dumazet wrote:
> > I tried the same setup than before and got good behavior on current
> > net-next kernel
> >
> > modprobe sit
> > ip link set sit0 up
> > ip -6 addr add dev sit0 2002:af6:1153::/64
> > ip -6 ro add 2002:a00::/24 via fe80:: dev sit0
> 
> Eric,
> in my test I do not use sit0. I create a new tunnel interface:
> ip tun add sittun mode sit remote $REMOTE_IPV4 local $LOCAL_IPV4 \
>    dev $IFACE
> 
> Looking at "ip tun show" I see that sit0 has the nopmtudisc option by 
> default, while sittun does not.
> 
> As I wrote in my original message, using nopmtudisc when creating my 
> tunnel interface works around the problem.
> 
> So I think nopmtudisc is the key difference in our setups.

Oh right.

Iguess the following should help ?



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/net/ipv6/sit.c b/net/ipv6/sit.c
index 366fbba3359a..a710fdec42d3 100644
--- a/net/ipv6/sit.c
+++ b/net/ipv6/sit.c
@@ -924,7 +924,7 @@  static netdev_tx_t ipip6_tunnel_xmit(struct sk_buff *skb,
 		if (tunnel->parms.iph.daddr && skb_dst(skb))
 			skb_dst(skb)->ops->update_pmtu(skb_dst(skb), NULL, skb, mtu);
 
-		if (skb->len > mtu) {
+		if (skb->len > mtu && !skb_is_gso(skb)) {
 			icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
 			ip_rt_put(rt);
 			goto tx_error;