[ovs-dev] [PATCH v3 0/2] Encap & Decap actions for MPLS packet type

Eelco Chaudron echaudro at redhat.com
Thu Mar 11 09:04:14 UTC 2021



On 11 Mar 2021, at 5:57, Varghese, Martin (Nokia - IN/Bangalore) wrote:

> Hi Eelco,
>
> As you suggested in a different email I have tried to use the new 
> ADD_MPLS  datapath action for the "push_mpls" userspace  action.
> But these tests fail.
> 487: mpls-xlate.at:3    MPLS xlate action
> 6230 488: mpls-xlate.at:147  MPLS xlate action - patch-port
> 6231 489: mpls-xlate.at:191  MPLS xlate action - group bucket
> 6232 1143: ofproto-dpif.at:7118 ofproto-dpif - sFlow packet sampling - 
> MPLS
> 6233 1165: ofproto-dpif.at:8204 ofproto-dpif - MPLS actions that 
> result in a userspace action
> 6234 1166: ofproto-dpif.at:8240 ofproto-dpif - MPLS actions that 
> result in a drop
> 6235 1177: ofproto-dpif.at:8534 ofproto-dpif megaflow - mpls
>
> I understand these tests fail as they don't see "PUSH_MPLS" datapath 
> action in datapath flows.
>
> Is there any framework to selectively use ADD_MPLS or PUSH_MPLS  in 
> test scripts  based on what is supported in datapath ?

With AT_SKIP_IF you can only skip the entire test, not do a conditional 
somewhere in the test.
So you could either write two test cases and execute them based on a 
specific condition, or you could pass the result through sed and replace 
the new keyword with the old one (if the parameters apply the same).

> Or, Should we continue to use PUSH_MPLS datapath action for userspace 
> push_mpls action ?

That I leave up to you, as long as it’s consistent in all cases.



More information about the dev mailing list