On Sun, Nov 30, 2014 at 2:32 PM, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote: > On Sun, 2014-11-30 at 14:26 +0200, Arik Nemtsov wrote: > >> I think Johannes' point was that it's easy to overrun the message size >> if there are a lot of wiphys. > > Yes. > >> So I'll simply add an iterator over all wiphys in "iw reg get" and >> kernel-mode will only return a single regdomain in each GET_REG >> invocation. > > Err, there's such a thing built into netlink already - just support > dumpit instead of doit :) That's cool. I'll use it then :) > iw will have to fall back to doit for older kernels though I guess. Well older kernels don't have this feature anyway, so it's fine to leave the doit as is. For backports, we are bringing this code with us anyway right? > >> About the "--all" suggestion - I think it's fine to not have backward >> compatibility in the output of "iw reg get"? So we can just output the >> global first, and then output private regdoms for all wiphys that have >> them. >> >> Does that sound ok? > > Yeah I wasn't taking about the iw display, and adding --all there > doesn't help for what I was concerned about. > >> Well you have to give a wiphy-idx in order to get a private regdom in >> the first place. And only new userspace will add a wiphy-idx in the >> first place.. > > Are you sure about the last part though? wpa_supplicant often passed a > netdev index instead of a wiphy index for example, so I could imagine it > passing a wiphy index here even though it was previously ignored? > > If it didn't though then I think there's no problem, there shouldn't > really be any userspace other than wpa_s and iw for this I guess/hope. It definitely doesn't pass it now. Otherwise we wouldn't have to change it :) Arik -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
