1995-10-26 21:28:30 +00:00
|
|
|
/*
|
|
|
|
* Copyright (c) 1985, 1993
|
|
|
|
* The Regents of the University of California. All rights reserved.
|
|
|
|
*
|
|
|
|
* Copyright (c) 1995 John Hay. All rights reserved.
|
|
|
|
*
|
|
|
|
* This file includes significant work done at Cornell University by
|
|
|
|
* Bill Nesheim. That work included by permission.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
* 3. All advertising materials mentioning features or use of this software
|
|
|
|
* must display the following acknowledgement:
|
|
|
|
* This product includes software developed by the University of
|
|
|
|
* California, Berkeley and its contributors.
|
|
|
|
* 4. Neither the name of the University nor the names of its contributors
|
|
|
|
* may be used to endorse or promote products derived from this software
|
|
|
|
* without specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
|
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
|
|
* SUCH DAMAGE.
|
|
|
|
*
|
1999-08-28 01:35:59 +00:00
|
|
|
* $FreeBSD$
|
1995-10-26 21:28:30 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef lint
|
2003-11-15 17:10:56 +00:00
|
|
|
static const char sccsid[] = "@(#)input.c 8.1 (Berkeley) 6/5/93";
|
1995-10-26 21:28:30 +00:00
|
|
|
#endif /* not lint */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* IPX Routing Table Management Daemon
|
|
|
|
*/
|
|
|
|
#include "defs.h"
|
|
|
|
|
|
|
|
struct sockaddr *
|
|
|
|
ipx_nettosa(net)
|
|
|
|
union ipx_net net;
|
|
|
|
{
|
|
|
|
static struct sockaddr_ipx sxn;
|
|
|
|
|
|
|
|
bzero(&sxn, sizeof (struct sockaddr_ipx));
|
|
|
|
sxn.sipx_family = AF_IPX;
|
|
|
|
sxn.sipx_len = sizeof (sxn);
|
|
|
|
sxn.sipx_addr.x_net = net;
|
|
|
|
return( (struct sockaddr *)&sxn);
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Process a newly received packet.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
rip_input(from, size)
|
|
|
|
struct sockaddr *from;
|
|
|
|
int size;
|
|
|
|
{
|
1997-07-06 07:38:36 +00:00
|
|
|
int newsize;
|
|
|
|
int rtchanged = 0;
|
1995-10-26 21:28:30 +00:00
|
|
|
struct rt_entry *rt;
|
|
|
|
struct netinfo *n;
|
|
|
|
struct interface *ifp = 0;
|
|
|
|
struct afswitch *afp;
|
|
|
|
struct sockaddr_ipx *ipxp;
|
|
|
|
|
|
|
|
ifp = if_ifwithnet(from);
|
|
|
|
ipxp = (struct sockaddr_ipx *)from;
|
|
|
|
if (ifp == 0) {
|
|
|
|
if(ftrace) {
|
|
|
|
fprintf(ftrace, "Received bogus packet from %s\n",
|
|
|
|
ipxdp_ntoa(&ipxp->sipx_addr));
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
TRACE_INPUT(ifp, from, size);
|
|
|
|
if (from->sa_family >= AF_MAX)
|
|
|
|
return;
|
|
|
|
afp = &afswitch[from->sa_family];
|
|
|
|
|
|
|
|
size -= sizeof (u_short) /* command */;
|
|
|
|
n = msg->rip_nets;
|
|
|
|
|
|
|
|
switch (ntohs(msg->rip_cmd)) {
|
|
|
|
|
|
|
|
case RIPCMD_REQUEST:
|
|
|
|
if (ipx_hosteq(satoipx_addr(ifp->int_addr), ipxp->sipx_addr))
|
|
|
|
return;
|
|
|
|
newsize = 0;
|
|
|
|
while (size > 0) {
|
|
|
|
if (size < sizeof (struct netinfo))
|
|
|
|
break;
|
|
|
|
size -= sizeof (struct netinfo);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A single entry with rip_dst == DSTNETS_ALL and
|
|
|
|
* metric ``infinity'' means ``all routes''.
|
|
|
|
*
|
|
|
|
* XXX According to the IPX RIP spec the metric
|
|
|
|
* and tick fields can be anything. So maybe we
|
|
|
|
* should not check the metric???
|
|
|
|
*/
|
|
|
|
if (ipx_neteqnn(n->rip_dst, ipx_anynet) &&
|
|
|
|
ntohs(n->rip_metric) == HOPCNT_INFINITY &&
|
|
|
|
size == 0) {
|
1997-07-06 07:38:36 +00:00
|
|
|
supply(from, 0, ifp, 0);
|
1995-10-26 21:28:30 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* request for specific nets
|
|
|
|
*/
|
|
|
|
rt = rtlookup(ipx_nettosa(n->rip_dst));
|
|
|
|
if (ftrace) {
|
|
|
|
fprintf(ftrace,
|
|
|
|
"specific request for %s",
|
|
|
|
ipxdp_nettoa(n->rip_dst));
|
|
|
|
fprintf(ftrace,
|
2003-10-26 05:33:59 +00:00
|
|
|
" yields route %lx\n",
|
|
|
|
(u_long)rt);
|
1995-10-26 21:28:30 +00:00
|
|
|
}
|
|
|
|
/*
|
|
|
|
* XXX We break out on the first net that isn't
|
|
|
|
* found. The specs is a bit vague here. I'm not
|
|
|
|
* sure what we should do.
|
|
|
|
*/
|
|
|
|
if (rt == 0)
|
|
|
|
return;
|
|
|
|
/* XXX
|
|
|
|
* According to the spec we should not include
|
|
|
|
* information about networks for which the number
|
|
|
|
* of hops is 16.
|
|
|
|
*/
|
|
|
|
if (rt->rt_metric == (HOPCNT_INFINITY-1))
|
|
|
|
return;
|
|
|
|
n->rip_metric = htons( rt == 0 ? HOPCNT_INFINITY :
|
|
|
|
min(rt->rt_metric+1, HOPCNT_INFINITY));
|
|
|
|
n->rip_ticks = htons(rt->rt_ticks+1);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We use split horizon with a twist. If the requested
|
|
|
|
* net is the directly connected net we supply an
|
|
|
|
* answer. This is so that the host can learn about
|
|
|
|
* the routers on its net.
|
|
|
|
*/
|
|
|
|
{
|
|
|
|
register struct rt_entry *trt = rt;
|
|
|
|
|
|
|
|
while (trt) {
|
|
|
|
if ((trt->rt_ifp == ifp) &&
|
|
|
|
!ipx_neteqnn(n->rip_dst,
|
|
|
|
satoipx_addr(ifp->int_addr).x_net))
|
|
|
|
return;
|
|
|
|
trt = trt->rt_clone;
|
|
|
|
}
|
|
|
|
n++;
|
|
|
|
newsize += sizeof (struct netinfo);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (newsize > 0) {
|
|
|
|
msg->rip_cmd = htons(RIPCMD_RESPONSE);
|
|
|
|
newsize += sizeof (u_short);
|
|
|
|
/* should check for if with dstaddr(from) first */
|
|
|
|
(*afp->af_output)(ripsock, 0, from, newsize);
|
|
|
|
TRACE_OUTPUT(ifp, from, newsize);
|
|
|
|
if (ftrace) {
|
|
|
|
/* XXX This should not happen anymore. */
|
|
|
|
if(ifp == 0)
|
|
|
|
fprintf(ftrace, "--- ifp = 0\n");
|
|
|
|
else
|
|
|
|
fprintf(ftrace,
|
|
|
|
"request arrived on interface %s\n",
|
|
|
|
ifp->int_name);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
|
|
|
|
case RIPCMD_RESPONSE:
|
|
|
|
/* verify message came from a router */
|
|
|
|
if ((*afp->af_portmatch)(from) == 0)
|
|
|
|
return;
|
|
|
|
(*afp->af_canon)(from);
|
|
|
|
/* are we talking to ourselves? */
|
|
|
|
if ((ifp = if_ifwithaddr(from)) != 0) {
|
|
|
|
rt = rtfind(from);
|
1997-07-06 07:38:36 +00:00
|
|
|
if (rt == 0 || (rt->rt_state & RTS_INTERFACE) == 0) {
|
1995-10-26 21:28:30 +00:00
|
|
|
addrouteforif(ifp);
|
1997-07-06 07:38:36 +00:00
|
|
|
rtchanged = 1;
|
|
|
|
} else
|
1995-10-26 21:28:30 +00:00
|
|
|
rt->rt_timer = 0;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
/* Update timer for interface on which the packet arrived.
|
|
|
|
* If from other end of a point-to-point link that isn't
|
|
|
|
* in the routing tables, (re-)add the route.
|
|
|
|
*/
|
|
|
|
if ((rt = rtfind(from)) && (rt->rt_state & RTS_INTERFACE)) {
|
|
|
|
if(ftrace) fprintf(ftrace, "Got route\n");
|
|
|
|
rt->rt_timer = 0;
|
|
|
|
} else if ((ifp = if_ifwithdstaddr(from)) != 0) {
|
|
|
|
if(ftrace) fprintf(ftrace, "Got partner\n");
|
|
|
|
addrouteforif(ifp);
|
1997-07-06 07:38:36 +00:00
|
|
|
rtchanged = 1;
|
1995-10-26 21:28:30 +00:00
|
|
|
}
|
|
|
|
for (; size > 0; size -= sizeof (struct netinfo), n++) {
|
|
|
|
struct sockaddr *sa;
|
|
|
|
if (size < sizeof (struct netinfo))
|
|
|
|
break;
|
1997-07-06 07:38:36 +00:00
|
|
|
if ((unsigned) ntohs(n->rip_metric) > HOPCNT_INFINITY)
|
1995-10-26 21:28:30 +00:00
|
|
|
continue;
|
|
|
|
rt = rtfind(sa = ipx_nettosa(n->rip_dst));
|
|
|
|
if (rt == 0) {
|
1997-07-06 07:38:36 +00:00
|
|
|
if (ntohs(n->rip_metric) == HOPCNT_INFINITY)
|
|
|
|
continue;
|
1995-10-26 21:28:30 +00:00
|
|
|
rtadd(sa, from, ntohs(n->rip_metric),
|
|
|
|
ntohs(n->rip_ticks), 0);
|
1997-07-06 07:38:36 +00:00
|
|
|
rtchanged = 1;
|
1995-10-26 21:28:30 +00:00
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* A clone is a different route to the same net
|
|
|
|
* with exactly the same cost (ticks and metric).
|
|
|
|
* They must all be recorded because those interfaces
|
|
|
|
* must be handled in the same way as the first route
|
|
|
|
* to that net. ie When using the split horizon
|
|
|
|
* algorithm we must look at these interfaces also.
|
|
|
|
*
|
|
|
|
* Update if from gateway and different,
|
|
|
|
* from anywhere and less ticks or
|
|
|
|
* if same ticks and shorter,
|
|
|
|
* or getting stale and equivalent.
|
|
|
|
*/
|
|
|
|
if (!equal(from, &rt->rt_router) &&
|
1997-07-06 07:38:36 +00:00
|
|
|
ntohs(n->rip_ticks) == rt->rt_ticks &&
|
|
|
|
ntohs(n->rip_metric) == rt->rt_metric &&
|
|
|
|
ntohs(n->rip_metric) != HOPCNT_INFINITY) {
|
1995-10-26 21:28:30 +00:00
|
|
|
register struct rt_entry *trt = rt->rt_clone;
|
|
|
|
|
|
|
|
while (trt) {
|
|
|
|
if (equal(from, &trt->rt_router)) {
|
|
|
|
trt->rt_timer = 0;
|
|
|
|
break;
|
|
|
|
}
|
1995-12-04 10:36:02 +00:00
|
|
|
trt = trt->rt_clone;
|
1995-10-26 21:28:30 +00:00
|
|
|
}
|
|
|
|
if (trt == NULL) {
|
|
|
|
rtadd_clone(rt, sa, from,
|
|
|
|
ntohs(n->rip_metric),
|
|
|
|
ntohs(n->rip_ticks), 0);
|
|
|
|
}
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if ((equal(from, &rt->rt_router) &&
|
|
|
|
((ntohs(n->rip_ticks) != rt->rt_ticks) ||
|
|
|
|
(ntohs(n->rip_metric) != rt->rt_metric))) ||
|
|
|
|
(ntohs(n->rip_ticks) < rt->rt_ticks) ||
|
|
|
|
((ntohs(n->rip_ticks) == rt->rt_ticks) &&
|
|
|
|
(ntohs(n->rip_metric) < rt->rt_metric)) ||
|
|
|
|
(rt->rt_timer > (EXPIRE_TIME*2/3) &&
|
1997-07-06 07:38:36 +00:00
|
|
|
rt->rt_metric == ntohs(n->rip_metric) &&
|
|
|
|
ntohs(n->rip_metric) != HOPCNT_INFINITY)) {
|
1995-10-26 21:28:30 +00:00
|
|
|
rtchange(rt, from, ntohs(n->rip_metric),
|
|
|
|
ntohs(n->rip_ticks));
|
1997-07-06 07:38:36 +00:00
|
|
|
if (ntohs(n->rip_metric) == HOPCNT_INFINITY)
|
|
|
|
rt->rt_timer = EXPIRE_TIME;
|
|
|
|
else
|
|
|
|
rt->rt_timer = 0;
|
|
|
|
rtchanged = 1;
|
1995-10-26 21:28:30 +00:00
|
|
|
} else if (equal(from, &rt->rt_router) &&
|
|
|
|
(ntohs(n->rip_ticks) == rt->rt_ticks) &&
|
1997-07-06 07:38:36 +00:00
|
|
|
(ntohs(n->rip_metric) == rt->rt_metric) &&
|
|
|
|
(ntohs(n->rip_metric) != HOPCNT_INFINITY)) {
|
1995-10-26 21:28:30 +00:00
|
|
|
rt->rt_timer = 0;
|
|
|
|
}
|
|
|
|
}
|
1997-07-06 07:38:36 +00:00
|
|
|
if (rtchanged) {
|
|
|
|
register struct rthash *rh;
|
|
|
|
register struct rt_entry *rt;
|
|
|
|
|
|
|
|
toall(supply, NULL, 1);
|
|
|
|
for (rh = nethash; rh < &nethash[ROUTEHASHSIZ]; rh++)
|
|
|
|
for (rt = rh->rt_forw;
|
|
|
|
rt != (struct rt_entry *)rh;
|
|
|
|
rt = rt->rt_forw)
|
|
|
|
rt->rt_state &= ~RTS_CHANGED;
|
|
|
|
}
|
|
|
|
|
1995-10-26 21:28:30 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|