Q about ip_local_deliver_finish

[Alexey Vyskubov]

Hello.

I have a question (maybe stupid but I spent a lot of time trying to find an
answer)about ip_local_deliver_finish() in 2.4.4. kernel.
I'll be very grateful if someone explains me what's happening.

There is the following piece of code in ip_local_deliver_finish()
(in net/ipv4/ip_input.c):

if(ipprot != NULL) {
	if(raw_sk == NULL &&
	   ipprot->next == NULL &&
	   ipprot->protocol == protocol) {
		int ret;
		/* Fast path... */
		ret = ipprot->handler(skb);
		return ret;
	} else {
		flag = ip_run_ipprot(skb, skb->nh.iph, ipprot, (raw_sk != NULL));
	}

[some comments skipped]

if(raw_sk != NULL) {	/* Shift to last raw user */
	raw_rcv(raw_sk, skb);
	sock_put(raw_sk);
} else if (!flag) {	/* Free and report errors */
	icmp_send(skb, ICMP_DEST_UNREACH, ICMP_PROT_UNREACH, 0);	
out:
	kfree_skb(skb);
}
}

return 0;

But in ip_run_ipprot the return value of ipprot->handler is just ignored,
code below:

if(skb2 != NULL) {
	ret = 1;
	ipprot->handler(skb2);
}

If I understand correctly it means that if we have two different protocols
with the same (modulo MAX_INET_PROTOS) protocol number then
ip_local_deliver will return the return value of ipprot->handler for the
first protocol in the chain
inet_protos[protocol number modulo MAX_INET_PROTOS] and *always zero* for the
second. Why? Is it a bug or I just do not understand something?

-- 
Alexey