Re: [PATCH net-next v2 1/5] rtnetlink: provide permanent hardware address in RTM_NEWLINK

On Tue, 2019-12-10 at 13:27 -0700, David Ahern wrote:

On 12/10/19 1:23 PM, Johannes Berg wrote:

quoted

On Tue, 2019-12-10 at 21:22 +0100, Johannes Berg wrote:

quoted

On Tue, 2019-12-10 at 09:51 -0800, Jakub Kicinski wrote:

quoted

On Tue, 10 Dec 2019 14:07:53 +0100 (CET), Michal Kubecek wrote:

quoted

@@ -1822,6 +1826,7 @@ static const struct nla_policy ifla_policy[IFLA_MAX+1] = {
 	[IFLA_PROP_LIST]	= { .type = NLA_NESTED },
 	[IFLA_ALT_IFNAME]	= { .type = NLA_STRING,
 				    .len = ALTIFNAMSIZ - 1 },
+	[IFLA_PERM_ADDRESS]	= { .type = NLA_REJECT },
 };
 
 static const struct nla_policy ifla_info_policy[IFLA_INFO_MAX+1] = {

Jiri, I just noticed ifla_policy didn't get strict_start_type set when
ALT_IFNAME was added, should we add it in net? 🤔

Does it need one? It shouldn't be used with
nla_parse_nested_deprecated(), and if it's used with nla_parse_nested()
then it doesn't matter?

No, wait. I misread, you said "when ALT_IFNAME was added" but somehow I
managed to read "when it was added"...

So yeah, it should have one. Dunno about net, your call. I'd probably
not bother for an NLA_REJECT attribute, there's little use including it
anyway.

It's new in net, so it has to be there not net-next.

Oh, ok. Well, I was actually thinking to just add it on the next
attribute or so, but I guess now that we're discussing it there's a
higher chance of it actually happening :)

johannes