[ovs-dev] MPLS and VLAN QinQ patch

Jesse Gross jesse at nicira.com
Thu Jun 14 11:02:56 UTC 2012


On Thu, Jun 14, 2012 at 2:09 PM, ravi kerur <rkerur at gmail.com> wrote:
> On Wed, Jun 13, 2012 at 9:24 PM, ravi kerur <rkerur at gmail.com> wrote:
>> On Wed, Jun 13, 2012 at 7:58 PM, Jesse Gross <jesse at nicira.com> wrote:
>>> On Thu, Jun 14, 2012 at 4:51 AM, ravi kerur <rkerur at gmail.com> wrote:
>>>> There are additional things that needs to be addressed as well
>>>>
>>>> 1. offload code review, it's currently generic enough and getting near
>>>> line rate performance numbers.
>>>
>>> You can't just take what is done for vlans and copy that.  There is
>>> far too much code that you're adding to OVS.  Did you read what I
>>> wrote earlier about where to start?:
>>
>> <rk> How is adding far too much code to ovs related to offload? It is
>> handled similar to what vlan is doing for older kernel + in addition
>> it takes care of handling copy + restore skb->protocol since
>> skb_gso_segment relies on it and handle cases for non-gso packets to
>> calculate checksum. I don't understand your comments, have you looked
>> at latest patch?
>
> <rk> or are you saying that mpls label should be handled in skb as
> well esp. in ovs_skb_cb such that in future it will be added to
> upstream kernel skbs and can be used for mpls label offloading in
> NICs? I haven't taken this approach as its not sure whether it will
> ever happen? you can ignore it if this is not the case and let me know
> what you mean by "adding too much code in ovs and how it is related to
> offload?

No, I don't want to use out of band MPLS tags.  It's a real mess with
vlans.  I want you to update the upstream GSO code to handle MPLS and
as NICs expose MPLS offloading they can skip the software emulation.



More information about the dev mailing list