RE: IPv6 routing table implementation

RE: IPv6 routing table implementation

[Joerg Pommnitz]

 > The underlined lines will be creating a kernel panic ALWAYS.

What makes you think so? After all, the pointer is never dereferenced.
This is quite a common C trick (similiar to the offsetof macro).

Regards
  Joerg

=====
-- 
Regards
       Joerg


__________________________________________________________________

Gesendet von Yahoo! Mail - http://mail.yahoo.de
Yahoo! präsentiert als offizieller Sponsor das Fußball-Highlight des
Jahres: - http://www.FIFAworldcup.com

IPv6 routing table implementation

[Madhavi]

Hi

While going through the routing table implementation for linux-ipv6, I
found this piece of code.

struct fib6_node * fib6_lookup(struct fib6_node *root, struct in6_addr
			*daddr, struct in6_addr *saddr)
{
        struct lookup_args args[2];
        struct rt6_info *rt = NULL;
        struct fib6_node *fn;

        args[0].offset = (u8*) &rt->rt6i_dst - (u8*) rt;
			~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
        args[0].addr = daddr;

#ifdef CONFIG_IPV6_SUBTREES
        args[1].offset = (u8*) &rt->rt6i_src - (u8*) rt;
			~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
        args[1].addr = saddr;
#endif

.
.
.
.
}

I found out that this code is called from inet6_route_input() which will
always be called on the receiving end for IPv6 packets (ip6_rcv_finish()).

The underlined lines will be creating a kernel panic ALWAYS.

I am using 2.4.16 kernel. I have checked version 2.4.18 also and no
change in this part. Aren't these versions supposed to include a working
IPv6 implementation? Am I missing something?

regards
Madhavi.