On Wed, Jun 12, 2013 at 6:19 PM, Simon Horman <[email protected]> wrote:
> On Wed, Jun 12, 2013 at 06:14:07PM -0700, Pravin Shelar wrote:
>> On Tue, Jun 11, 2013 at 2:26 PM, Jesse Gross <[email protected]> wrote:
>> > On Mon, Jun 10, 2013 at 1:07 AM, Simon Horman <[email protected]> wrote:
>> >> @@ -341,10 +376,14 @@ static int netdev_send(struct vport *vport, struct
>> >> sk_buff *skb)
>> >> nskb = skb->next;
>> >> skb->next = NULL;
>> >>
>> >> - skb = __vlan_put_tag(skb,
>> >> vlan_tx_tag_get(skb));
>> >> + if (vlan)
>> >> + skb = __vlan_put_tag(skb,
>> >> vlan_tx_tag_get(skb));
>> >> if (likely(skb)) {
>> >> len += skb->len;
>> >> - vlan_set_tci(skb, 0);
>> >> + if (mpls)
>> >> + skb->protocol =
>> >> mpls_protocol;
>> >> + if (vlan)
>> >> + vlan_set_tci(skb, 0);
>> >> dev_queue_xmit(skb);
>> >> }
>> >
>> > The order of setting the protocol field is different from the one
>> > below and I think this one isn't right because it means that the MPLS
>> > protocol can override the vlan.
>> >
>> > Pravin, can you take a look at this as well since you are working in
>> > the same area?
>>
>> I think this compatibility code can be moved to new compat code posted
>> on gre-restructuring code few changes.
>
> Sure that sounds reasonable but I am unsure of the status of those changes
> and I would rather not rely on something that isn't ready to merge yet.
Right. sorry abt delay.
But I think it is much closer now and it would be alot easy to handle
compatibility, so it is worth waiting for.
_______________________________________________
dev mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/dev
