Peter Memishian wrote:
> > There appears to be a fetal implementation or ND Proxy in ip_ndp.c (see
> > references to NCE_F_PROXY) whose origins are a bit of a mystery. The
> > flag and the code surrounding it (currently the dangling NDF_PROXY_ON and
> > NDF_PROXY_OFF flags for the SIOCLIFSETND ioctl, and twiddling the
> > override bit in Neighbor Advertisements if NCE_F_PROXY is set) appear to
> > be dead.
> >
> > Much more infrastructure needs to be in place for ND Proxy to work
> > (including forwarding ND messages, and joining solicited-node multicast
> > groups for nodes being proxied, etc.), and this isolated flag does not
> > figure in this solution. An existing ND Proxy RFC exists (RFC 4389), but
> > this flag is not part of an implementation of this experimental protocol.
> >
> > My opinion is that this flag and the code surrounding it is in fact dead
> > and does not do anything. Does anyone have any additional insight into
> > this? I think it could safely be removed.
>
> A related bit of seemingly-dead code is the NDF_ANYCAST_ON/NDF_ANYCAST_OFF
> logic. Seems like all of this stuff should go.
While I don't think the proxy flags are used, the NCE_F_ANYCAST is set
based on IFF_ANYCAST, and NDF_ANYCAST_ON is set based on NCE_F_ANYCAST.
Erik
_______________________________________________
networking-discuss mailing list
[email protected]
