(usagi-users 03390) Re: MLDv2 linux implementation questions

Hi David,

This patch doesn't fix the problem. Taking my test program, not it returns
EADDRINUSE after the second call when it should say OK, because I am
using the same channel number for every group, but I join different
sources: I join (S,G), then (S1,G), then (S2,G).

This patch works for ASM groups, because I am not supposed to join the
same group twice on a socket, but not for SSM where I can join several
channels on the same sockets with the same channel identifier.

Also, the EADDRINUSE is not documented in RFC3678.

Thanks,

Hoerdt Mickaël

David Stevens wrote:

>Hoerdt,
>        Here's a fix. I'm going to see if I can do the further
>optimization of leaving the group when the filter is (INCLUDE,empty)
>for this case, since that would use less socket buffer space, but
>this alone corrects the problem.
>
>                                        +-DLS
>
>(in-line for viewing, attached for applying)
>--- linux-2.6.11.12/net/ipv6/mcast.c    2005-03-01 23:38:25.000000000 
>-0800
>+++ linux-2.6.11.12MCT/net/ipv6/mcast.c 2005-06-14 10:31:24.000000000 
>-0700
>@@ -188,6 +188,16 @@ int ipv6_sock_mc_join(struct sock *sk, i
>        if (!ipv6_addr_is_multicast(addr))
>                return -EINVAL;
> 
>+       read_lock_bh(&ipv6_sk_mc_lock);
>+       for (mc_lst=np->ipv6_mc_list; mc_lst; mc_lst=mc_lst->next) {
>+               if ((ifindex == 0 || mc_lst->ifindex == ifindex) &&
>+                   ipv6_addr_equal(&mc_lst->addr, addr)) {
>+                       read_unlock_bh(&ipv6_sk_mc_lock);
>+                       return -EADDRINUSE;
>+               }
>+       }
>+       read_unlock_bh(&ipv6_sk_mc_lock);
>+
>        mc_lst = sock_kmalloc(sk, sizeof(struct ipv6_mc_socklist), 
>GFP_KERNEL);
> 
>        if (mc_lst == NULL)
>
>
>  
>