On Tue, Mar 18, 2014 at 05:16:03PM -0700, Andy Zhou wrote: > On Tue, Mar 18, 2014 at 11:10 AM, Ben Pfaff <b...@nicira.com> wrote: > > > On Tue, Mar 11, 2014 at 04:56:18PM -0700, Andy Zhou wrote: > > > Signed-off-by: Andy Zhou <az...@nicira.com> > > > > The new NXM fields are weird because OVS can report them (via > > nxm_put_32m() calls in nx-match.c) but it can't accept them at all > > (because the fields aren't listed in the table in meta-flow.c) or > > match on them. I don't know if that's intentional but it's > > inconsistent with the comments in nicira-ext.h that talk about masking > > (a field that can't be matched definitely can't be masked). > > In previous discussions, I think we concluded that we want > > recirculation to be transparent to userspace (and to OpenFlow), so I > > suspect that the recirculation ID, at least, should not be mentioned > > in nicira-ext.h or in nx-match.c (or meta-flow.[ch]). I'm not sure > > about dp_hash, but so far I think we only plan to use that internally > > as well, so I'd be inclined toward doing the same for it. > > > I would like those fields to be visible to the ovs-ofctl. I can add logic > to make sure > the controller can not set them.
That sounds ideal. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
