NAME
nsip —
software network interface
encapsulating NS packets in IP packets
SYNOPSIS
options NSIP
#include <netns/ns_if.h>
DESCRIPTION
Thensip interface is a software mechanism which may be
used to transmit Xerox NS(tm) packets through otherwise uncooperative
networks. It functions by prepending an IP header, and resubmitting the packet
through the UNIX IP machinery.
The super-user can advise the operating system of a willing partner by naming an IP address to be associated with an NS address. Presently, only specific hosts pairs are allowed, and for each host pair, an artificial point-to-point interface is constructed. At some future date, IP broadcast addresses or hosts may be paired with NS networks or hosts.
Specifically, a socket option of
SO_NSIP_ROUTE is set on a socket of family
AF_NS, type SOCK_DGRAM,
passing the following structure:
struct nsip_req {
struct sockaddr rq_ns; /* must be ns format destination */
struct sockaddr rq_ip; /* must be ip format gateway */
short rq_flags;
};
DIAGNOSTICS
- nsip%d: can't handle af%d.
- The interface was handed a message with addresses formatted in an unsuitable address family; the packet was dropped.
SEE ALSO
HISTORY
The nsip interface appeared in
4.3BSD.
BUGS
It is absurd to have a separate pseudo-device for each pt-to-pt
link. There is no way to change the IP address for an NS host once the
encapsulation interface is set up. The request should honor flags of
RTF_GATEWAY to indicate remote networks, and the
absence of RTF_UP should be a clue to remove that
partner. This was intended to postpone the necessity of rewriting reverse
ARP for the en(4) device, and to allow passing XNS packets through an
Arpanet-Milnet gateway, to facilitate testing between some co-operating
universities.