ip_queue getting rtnetlink semaphore?

[Jason Lunz <j@trellisinc.com>]

In net/ipv4/netfilter/ip_queue.c, in netlink_receive_user_sk(), why does
it obtain the rtnl semaphore?

static void netlink_receive_user_sk(struct sock *sk, int len)
{
	do {
		struct sk_buff *skb;

		if (rtnl_shlock_nowait())
			return;
		while ((skb = skb_dequeue(&sk->receive_queue)) != NULL) {
			netlink_receive_user_skb(skb);
			kfree_skb(skb);
		}
		up(&rtnl_sem);
	} while (nfnl && nfnl->receive_queue.qlen);
}

What do routing netlink sockets have to do with the netlink sockets used
for userspace communication in ip_queue? If they're sharing that
semaphore, is it just named "rtnl" as a historical artifact?

thanks,

Jason
-
: send the line "unsubscribe linux-net" in
the body of a message to majordomo@vger.kernel.org