On Fri, Dec 13, 2013 at 10:25 AM, Pravin B Shelar <pshe...@nicira.com> wrote: > In case of flow-miss kernel sends skb to userspace and then > userspace sends execute netlink msg to switch the packet. But > in this process skb rxhash is not restored. skb rxhash is used > for calculating tunnel source port for vxlan. If rxhash is not > restored packet which is sent to miss call handler can get > different source port compared to rest of packet of the flow. > > Following patch adds support for packet metadata which is part > of packet-attributes. Kernel can send packet metadata and > userspace echoes it back to kernel. > Currently packet metadata consist of skb rxhash. > > Signed-off-by: Pravin B Shelar <pshe...@nicira.com>
I've been thinking about this and I keep wondering whether it is really the right solution. It solves one level of the problem but if you have a controller that is receiving packet ins and spitting them back out (the classic OpenFlow model) then you have exactly the same issue again. From a reordering perspective, it probably doesn't matter because the time difference between packets would be very large but it still doesn't seem right. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
