On Sat, Sep 20, 2014 at 01:27:49AM +0900, YAMAMOTO Takashi wrote: > it seems master produces table-features with NXM_0, NXM_1, > and even packet_regs for OF1.3. > > i don't think it's a good idea to include NXM stuff > because the OF standard doesn't seem to define how to parse them. > (the size of OXM ids are class-dependant)
I think that there is a bug here: we should be careful to include only NXM stuff that is not available via OXM. But I don't think it makes sense to include only standard fields. That would defeat much of the purpose of having an extensible match: it would make it more difficult for controllers to find out what fields are supported. The size of OXM ids is not class-dependent, except to the extent that experimenter OXM ids are 64 bits and the rest are 32 bits. > packet_regs seems like an OF1.5 stuff which should not be included > when speaking OF1.3. We can do that. I recently posted a series that completely redoes OXM and NXM handling. Among other changes, it adds support for experimenter OXM. I don't know whether it fixes this particular problem. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
