On Fri, Aug 12, 2011 at 11:48:42AM -0700, Ethan Jackson wrote: > Thanks for the review. I've supplied an incremental below. The series has > changed enough that I'm going to resend the entire thing for completeness.
Thanks, I would appreciate that. > > Let's add a requirement that pad[] in struct nx_action_output_reg be > > all-bytes-zero, in case we want to extend it later. > > I was wondering about this. Historically we haven't taken this approach with > padding fields, but I did notice we did so for the resubmit table action. I > don't expect we will need extensions for this particular action, but I'm fine > with enforcing it anyways. Thanks. I've just started coming around to the idea that we should enforce all-zeros in padding fields. It make extensions easier later, it could also help detect bugs in which malloc()'d memory isn't being properly cleared, and it doesn't really cost much on either end. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
