Hi David,

David Stevens wrote:
> OK, I see what you're trying to fix now.
> 
> I think the scope_id checks are not quite right-- they
> should be something like this:
> 
> if (addr_type&IPV6_ADDR_LINKLOCAL) {
>         if (addr_len >= sizeof(struct sockaddr_in6)) {
>                 if (sk->sk_bound_dev_if && usin->sin6_scope_id &&
>                     sk->sk_bound_dev_if != usin->sin6_scope_id) {
>                         err = -EINVAL;
>                         goto out;
>                 }
>                 if (usin->sin6_scope_id)
>                         sk->sk_bound_dev_if = usin->sin6_scope_id;
>                 if (!sk->sk_bound_dev_if &&
>                      (addr_type & IPV6_ADDR_MULTICAST))
>                         fl.oif = np->mcast_oif;

This assignment will not get us past the next check...

>         /* connect to the link-local addres requires an interface */
>         if (!sk->sk_bound_dev_if) {
>                 err = -EINVAL;
>                 goto out;
>         }

... and even if it did, fl.oif is over-written by sk_bound_dev_if just a 
few lines down.

>         If I did an SO_BINDTODEVICE and specified sin6_scope_id,
> then they better agree.
>         If I specified sin6_scope_id without SO_BINDTODEVICE, set
> the device to that.
>         If I get this far without a device and it's multicast, use 
> mcast_oif
>         If I get all through that and don't have a device, EINVAL.

You also need to check if mcast_oif matches sk_bind_dev_if here - it's 
actually done in the setsockopt() code already when we set it, 
duplicating it here isn't that big a deal.

How about the following patch?  It does not set sk_bound_dev_if to 
mcast_oif, but does allow the connect() to succeed.

-Brian


Signed-off-by: Brian Haley <brian.haley@hp.com>
---