> I believe
http://bugzilla.kernel.org/show_bug.cgi?id=10318 is a false
> positive. There's no way in which networking will be using highmem pages
> here, so it won't be taking the KM_USER0 kmap slot, so there's no point in
> performing these checks.