Tue, Jun 13, 2017 at 12:53:00PM CEST, j...@mojatatu.com wrote: >On 17-06-13 01:32 AM, Jiri Pirko wrote: >> Mon, Jun 12, 2017 at 09:58:33PM CEST, dsah...@gmail.com wrote: >> > On 6/12/17 1:22 PM, Jiri Pirko wrote: >> > > >> > > > 3. IMO since these are nla prefixes and new NLA type they should be in >> > > > uapi/linux/netlink.h >> > > Including NLA_* type enum? I think it is reasonable. >> > >> > well, maybe not the NLA_BITFIELD. That enum is for policy validation >> > kernel side so not really part of the API. >> >> Yeah, now I see it. Agreed. >> > >Jiri, you agreed to a name change? ;-> I want to have some of what >David A. ate yesterday.
I ok with either of the tree (1 Jamal's, 2 DaveA's) variants. > >I agree it is a good idea to have arbitrary size bitmask so we dont >run out of bit space but we need to restrict the max length possible. >I dont agree to using the netlink.h as the best location for this >but lets move on. Suggest a better place, I don't see one. >Do you or David A. want to take a crack at this? I am a little tied >up. No time right now. Maybe in July I could allocate some cycles for this.
