Re: [PATCH] Compilation fixes for 2.1.127
Andi Kleen (ak@muc.de)
Tue, 10 Nov 1998 17:05:53 +0100
On Tue, Nov 10, 1998 at 04:57:25PM +0100, Henner Eisen wrote:
> >>>>> "Andi" == Andi Kleen <ak@muc.de> writes:
> >> Is dst_link_failure() intended to be a public interface?
>
> Andi> Yes. icmp_send on the other hand isn't: it is only supposed
> Andi> to be called from IPv4, it doesn't work with IPv6, it is a
> Andi> layer violation.
>
> Yes. I was just wondering whether this is the proper way for net device
> drivers to signal link failures. As the link failure detected by the device
> affects all attached network protocol stacks, would it make sense to
> handle this event by means of a dedicated netdevice notifier?
No. The proper way to communicate such things to higher layers is over the
destination entry (e.g. that is how path mtu discovery is handled). The higher
layers pick the condition up as soon as they access the dst_entry again -
usually with the next packet which is the first time such a link failure is
interesting anyways.
-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/