[ovs-dev] [PATCH 2/4] dpif-netdev: Add SCTP support

Jesse Gross jesse at nicira.com
Fri Aug 31 15:45:08 UTC 2012


On Fri, Aug 31, 2012 at 3:44 AM, Joe Stringer <joe at wand.net.nz> wrote:
> Signed-off-by: Joe Stringer <joe at wand.net.nz>
> ---
> It seems that this commit breaks the following test, I'm not sure quite why:-
>
>     504: ofproto-dpif - NetFlow flow expiration

[...]

> diff --git a/include/linux/openvswitch.h b/include/linux/openvswitch.h
> index f5c9cca..63c83cb 100644
> --- a/include/linux/openvswitch.h
> +++ b/include/linux/openvswitch.h
> @@ -274,6 +274,7 @@ enum ovs_key_attr {
>         OVS_KEY_ATTR_IPV6,      /* struct ovs_key_ipv6 */
>         OVS_KEY_ATTR_TCP,       /* struct ovs_key_tcp */
>         OVS_KEY_ATTR_UDP,       /* struct ovs_key_udp */
> +       OVS_KEY_ATTR_SCTP,       /* struct ovs_key_sctp */
>         OVS_KEY_ATTR_ICMP,      /* struct ovs_key_icmp */
>         OVS_KEY_ATTR_ICMPV6,    /* struct ovs_key_icmpv6 */
>         OVS_KEY_ATTR_ARP,       /* struct ovs_key_arp */

You can't add new types in the middle of this enum because it will
break userspace/kernel compatibility for the existing fields.  New
types have to go at the end of the list (but before TUN_ID, which
isn't locked down yet).

This is actually why the unit test broke - it changed the value of
some of the fields and NetFlow outputs flows in hash order but the
unit test expects them to come in a particular order.



More information about the dev mailing list