From: David Ahern <redacted>
Date: Fri, 5 Oct 2018 15:18:11 -0600

One thing I missed in the rfc and v1 versions is strict attribute
parsing -- ie., there should be nothing remaining after nla_parse is
done. I have a new patch that adds an nlmsg_parse_strict and
nla_parse_strict that returns -EINVAL (with extack filled in) if that
happens. The new patch pushes the set over 20.

I can peel off the first 3 patches from this set which add extack to the
dumps and down to nlmsg_parse and send those separately if preferred.

Don't worry about it, just send the whole thing.

My rule with patch series sizes is "be reasonable", rather than a
strict number requirement.

Thanks.