On Fri, Jul 1, 2016 at 5:59 PM, Pravin B Shelar <pshe...@ovn.org> wrote: > diff --git a/datapath/linux/compat/include/net/ip6_tunnel.h > b/datapath/linux/compat/include/net/ip6_tunnel.h > index ce65087..565b69b 100644 > --- a/datapath/linux/compat/include/net/ip6_tunnel.h > +++ b/datapath/linux/compat/include/net/ip6_tunnel.h > @@ -27,7 +27,7 @@ static inline void ip6tunnel_xmit(struct sock *sk, struct > sk_buff *skb, > else > pkt_len = err; > > - iptunnel_xmit_stats(pkt_len, &dev->stats, (struct pcpu_sw_netstats > __percpu *)dev->tstats); > + iptunnel_xmit_stats(dev, pkt_len); > }
The existing stats code vs. the upstream commit looks funny to me. The upstream version checks net_xmit_eval() and if it is an error, sets pkt_len to -1. However, the code here does something different and sets pkt_len to err in the non-error case. > diff --git a/datapath/linux/compat/ip_tunnels_core.c > b/datapath/linux/compat/ip_tunnels_core.c > index 31e4f3b..84684bf 100644 > --- a/datapath/linux/compat/ip_tunnels_core.c > +++ b/datapath/linux/compat/ip_tunnels_core.c > @@ -106,8 +107,9 @@ int ovs_iptunnel_handle_offloads(struct sk_buff *skb, > if (gso_type_mask) > fix_segment = NULL; > > +#if LINUX_VERSION_CODE < KERNEL_VERSION(3,18,0) > OVS_GSO_CB(skb)->fix_segment = fix_segment; > - > +#endif What is this change related to? _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
