1993-03-21 12:45:37 +03:00
|
|
|
/*
|
|
|
|
* Copyright (c) 1983, 1988 Regents of the University of California.
|
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* 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.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef lint
|
1993-08-01 21:54:45 +04:00
|
|
|
/*static char sccsid[] = "from: @(#)startup.c 5.19 (Berkeley) 2/28/91";*/
|
1993-12-11 04:47:28 +03:00
|
|
|
static char rcsid[] = "$Id: startup.c,v 1.5 1993/12/11 01:47:32 cgd Exp $";
|
1993-03-21 12:45:37 +03:00
|
|
|
#endif /* not lint */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Routing Table Management Daemon
|
|
|
|
*/
|
|
|
|
#include "defs.h"
|
|
|
|
#include <sys/ioctl.h>
|
|
|
|
#include <net/if.h>
|
|
|
|
#include <syslog.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include "pathnames.h"
|
|
|
|
|
|
|
|
struct interface *ifnet;
|
|
|
|
struct interface **ifnext = &ifnet;
|
|
|
|
int lookforinterfaces = 1;
|
|
|
|
int externalinterfaces = 0; /* # of remote and local interfaces */
|
|
|
|
int foundloopback; /* valid flag for loopaddr */
|
|
|
|
struct sockaddr loopaddr; /* our address on loopback */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Find the network interfaces which have configured themselves.
|
|
|
|
* If the interface is present but not yet up (for example an
|
|
|
|
* ARPANET IMP), set the lookforinterfaces flag so we'll
|
|
|
|
* come back later and look again.
|
|
|
|
*/
|
|
|
|
ifinit()
|
|
|
|
{
|
|
|
|
struct interface ifs, *ifp;
|
|
|
|
int s;
|
|
|
|
char buf[BUFSIZ], *cp, *cplim;
|
|
|
|
struct ifconf ifc;
|
|
|
|
struct ifreq ifreq, *ifr;
|
|
|
|
struct sockaddr_in *sin;
|
|
|
|
u_long i;
|
|
|
|
|
|
|
|
if ((s = socket(AF_INET, SOCK_DGRAM, 0)) < 0) {
|
|
|
|
syslog(LOG_ERR, "socket: %m");
|
|
|
|
close(s);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
ifc.ifc_len = sizeof (buf);
|
|
|
|
ifc.ifc_buf = buf;
|
|
|
|
if (ioctl(s, SIOCGIFCONF, (char *)&ifc) < 0) {
|
|
|
|
syslog(LOG_ERR, "ioctl (get interface configuration)");
|
|
|
|
close(s);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
ifr = ifc.ifc_req;
|
|
|
|
lookforinterfaces = 0;
|
|
|
|
#ifdef RTM_ADD
|
|
|
|
#define max(a, b) (a > b ? a : b)
|
|
|
|
#define size(p) max((p).sa_len, sizeof(p))
|
|
|
|
#else
|
|
|
|
#define size(p) (sizeof (p))
|
|
|
|
#endif
|
|
|
|
cplim = buf + ifc.ifc_len; /*skip over if's with big ifr_addr's */
|
|
|
|
for (cp = buf; cp < cplim;
|
|
|
|
cp += sizeof (ifr->ifr_name) + size(ifr->ifr_addr)) {
|
|
|
|
ifr = (struct ifreq *)cp;
|
|
|
|
bzero((char *)&ifs, sizeof(ifs));
|
|
|
|
ifs.int_addr = ifr->ifr_addr;
|
|
|
|
ifreq = *ifr;
|
|
|
|
if (ioctl(s, SIOCGIFFLAGS, (char *)&ifreq) < 0) {
|
|
|
|
syslog(LOG_ERR, "%s: ioctl (get interface flags)",
|
|
|
|
ifr->ifr_name);
|
|
|
|
continue;
|
|
|
|
}
|
1993-12-11 04:47:28 +03:00
|
|
|
ifs.int_flags =
|
|
|
|
(ifreq.ifr_flags & IFF_FLAGMASK) | IFF_INTERFACE;
|
1993-03-21 12:45:37 +03:00
|
|
|
if ((ifs.int_flags & IFF_UP) == 0 ||
|
|
|
|
ifr->ifr_addr.sa_family == AF_UNSPEC) {
|
|
|
|
lookforinterfaces = 1;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
/* argh, this'll have to change sometime */
|
|
|
|
if (ifs.int_addr.sa_family != AF_INET)
|
|
|
|
continue;
|
|
|
|
if (ifs.int_flags & IFF_POINTOPOINT) {
|
|
|
|
if (ioctl(s, SIOCGIFDSTADDR, (char *)&ifreq) < 0) {
|
|
|
|
syslog(LOG_ERR, "%s: ioctl (get dstaddr)",
|
|
|
|
ifr->ifr_name);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (ifr->ifr_addr.sa_family == AF_UNSPEC) {
|
|
|
|
lookforinterfaces = 1;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
ifs.int_dstaddr = ifreq.ifr_dstaddr;
|
|
|
|
}
|
|
|
|
/*
|
|
|
|
* already known to us?
|
|
|
|
* This allows multiple point-to-point links
|
|
|
|
* to share a source address (possibly with one
|
|
|
|
* other link), but assumes that there will not be
|
|
|
|
* multiple links with the same destination address.
|
|
|
|
*/
|
|
|
|
if (ifs.int_flags & IFF_POINTOPOINT) {
|
|
|
|
if (if_ifwithdstaddr(&ifs.int_dstaddr))
|
|
|
|
continue;
|
|
|
|
} else if (if_ifwithaddr(&ifs.int_addr))
|
|
|
|
continue;
|
|
|
|
if (ifs.int_flags & IFF_LOOPBACK) {
|
|
|
|
ifs.int_flags |= IFF_PASSIVE;
|
|
|
|
foundloopback = 1;
|
|
|
|
loopaddr = ifs.int_addr;
|
|
|
|
for (ifp = ifnet; ifp; ifp = ifp->int_next)
|
|
|
|
if (ifp->int_flags & IFF_POINTOPOINT)
|
|
|
|
add_ptopt_localrt(ifp);
|
|
|
|
}
|
|
|
|
if (ifs.int_flags & IFF_BROADCAST) {
|
|
|
|
if (ioctl(s, SIOCGIFBRDADDR, (char *)&ifreq) < 0) {
|
|
|
|
syslog(LOG_ERR, "%s: ioctl (get broadaddr)",
|
|
|
|
ifr->ifr_name);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
#ifndef sun
|
|
|
|
ifs.int_broadaddr = ifreq.ifr_broadaddr;
|
|
|
|
#else
|
|
|
|
ifs.int_broadaddr = ifreq.ifr_addr;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
#ifdef SIOCGIFMETRIC
|
|
|
|
if (ioctl(s, SIOCGIFMETRIC, (char *)&ifreq) < 0) {
|
|
|
|
syslog(LOG_ERR, "%s: ioctl (get metric)",
|
|
|
|
ifr->ifr_name);
|
|
|
|
ifs.int_metric = 0;
|
|
|
|
} else
|
|
|
|
ifs.int_metric = ifreq.ifr_metric;
|
|
|
|
#else
|
|
|
|
ifs.int_metric = 0;
|
|
|
|
#endif
|
|
|
|
/*
|
|
|
|
* Use a minimum metric of one;
|
|
|
|
* treat the interface metric (default 0)
|
|
|
|
* as an increment to the hop count of one.
|
|
|
|
*/
|
|
|
|
ifs.int_metric++;
|
|
|
|
if (ioctl(s, SIOCGIFNETMASK, (char *)&ifreq) < 0) {
|
|
|
|
syslog(LOG_ERR, "%s: ioctl (get netmask)",
|
|
|
|
ifr->ifr_name);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
sin = (struct sockaddr_in *)&ifreq.ifr_addr;
|
|
|
|
ifs.int_subnetmask = ntohl(sin->sin_addr.s_addr);
|
|
|
|
sin = (struct sockaddr_in *)&ifs.int_addr;
|
|
|
|
i = ntohl(sin->sin_addr.s_addr);
|
|
|
|
if (IN_CLASSA(i))
|
|
|
|
ifs.int_netmask = IN_CLASSA_NET;
|
|
|
|
else if (IN_CLASSB(i))
|
|
|
|
ifs.int_netmask = IN_CLASSB_NET;
|
|
|
|
else
|
|
|
|
ifs.int_netmask = IN_CLASSC_NET;
|
|
|
|
ifs.int_net = i & ifs.int_netmask;
|
|
|
|
ifs.int_subnet = i & ifs.int_subnetmask;
|
|
|
|
if (ifs.int_subnetmask != ifs.int_netmask)
|
|
|
|
ifs.int_flags |= IFF_SUBNET;
|
|
|
|
ifp = (struct interface *)malloc(sizeof (struct interface));
|
|
|
|
if (ifp == 0) {
|
|
|
|
printf("routed: out of memory\n");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
*ifp = ifs;
|
|
|
|
/*
|
|
|
|
* Count the # of directly connected networks
|
|
|
|
* and point to point links which aren't looped
|
|
|
|
* back to ourself. This is used below to
|
|
|
|
* decide if we should be a routing ``supplier''.
|
|
|
|
*/
|
|
|
|
if ((ifs.int_flags & IFF_LOOPBACK) == 0 &&
|
|
|
|
((ifs.int_flags & IFF_POINTOPOINT) == 0 ||
|
|
|
|
if_ifwithaddr(&ifs.int_dstaddr) == 0))
|
|
|
|
externalinterfaces++;
|
|
|
|
/*
|
|
|
|
* If we have a point-to-point link, we want to act
|
|
|
|
* as a supplier even if it's our only interface,
|
|
|
|
* as that's the only way our peer on the other end
|
|
|
|
* can tell that the link is up.
|
|
|
|
*/
|
|
|
|
if ((ifs.int_flags & IFF_POINTOPOINT) && supplier < 0)
|
|
|
|
supplier = 1;
|
|
|
|
ifp->int_name = malloc(strlen(ifr->ifr_name) + 1);
|
|
|
|
if (ifp->int_name == 0) {
|
|
|
|
fprintf(stderr, "routed: ifinit: out of memory\n");
|
|
|
|
syslog(LOG_ERR, "routed: ifinit: out of memory\n");
|
|
|
|
close(s);
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
strcpy(ifp->int_name, ifr->ifr_name);
|
|
|
|
*ifnext = ifp;
|
|
|
|
ifnext = &ifp->int_next;
|
|
|
|
traceinit(ifp);
|
|
|
|
addrouteforif(ifp);
|
|
|
|
}
|
|
|
|
if (externalinterfaces > 1 && supplier < 0)
|
|
|
|
supplier = 1;
|
|
|
|
close(s);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add route for interface if not currently installed.
|
|
|
|
* Create route to other end if a point-to-point link,
|
|
|
|
* otherwise a route to this (sub)network.
|
|
|
|
* INTERNET SPECIFIC.
|
|
|
|
*/
|
|
|
|
addrouteforif(ifp)
|
|
|
|
register struct interface *ifp;
|
|
|
|
{
|
|
|
|
struct sockaddr_in net;
|
|
|
|
struct sockaddr *dst;
|
|
|
|
int state;
|
|
|
|
register struct rt_entry *rt;
|
|
|
|
|
|
|
|
if (ifp->int_flags & IFF_POINTOPOINT)
|
|
|
|
dst = &ifp->int_dstaddr;
|
|
|
|
else {
|
|
|
|
bzero((char *)&net, sizeof (net));
|
|
|
|
net.sin_family = AF_INET;
|
|
|
|
net.sin_addr = inet_makeaddr(ifp->int_subnet, INADDR_ANY);
|
|
|
|
dst = (struct sockaddr *)&net;
|
|
|
|
}
|
|
|
|
rt = rtfind(dst);
|
|
|
|
if (rt &&
|
|
|
|
(rt->rt_state & (RTS_INTERFACE | RTS_INTERNAL)) == RTS_INTERFACE)
|
|
|
|
return;
|
|
|
|
if (rt)
|
|
|
|
rtdelete(rt);
|
|
|
|
/*
|
|
|
|
* If interface on subnetted network,
|
|
|
|
* install route to network as well.
|
|
|
|
* This is meant for external viewers.
|
|
|
|
*/
|
|
|
|
if ((ifp->int_flags & (IFF_SUBNET|IFF_POINTOPOINT)) == IFF_SUBNET) {
|
|
|
|
struct in_addr subnet;
|
|
|
|
|
|
|
|
subnet = net.sin_addr;
|
|
|
|
net.sin_addr = inet_makeaddr(ifp->int_net, INADDR_ANY);
|
|
|
|
rt = rtfind(dst);
|
|
|
|
if (rt == 0)
|
|
|
|
rtadd(dst, &ifp->int_addr, ifp->int_metric,
|
|
|
|
((ifp->int_flags & (IFF_INTERFACE|IFF_REMOTE)) |
|
|
|
|
RTS_PASSIVE | RTS_INTERNAL | RTS_SUBNET));
|
|
|
|
else if ((rt->rt_state & (RTS_INTERNAL|RTS_SUBNET)) ==
|
|
|
|
(RTS_INTERNAL|RTS_SUBNET) &&
|
|
|
|
ifp->int_metric < rt->rt_metric)
|
|
|
|
rtchange(rt, &rt->rt_router, ifp->int_metric);
|
|
|
|
net.sin_addr = subnet;
|
|
|
|
}
|
|
|
|
if (ifp->int_transitions++ > 0)
|
|
|
|
syslog(LOG_ERR, "re-installing interface %s", ifp->int_name);
|
|
|
|
state = ifp->int_flags &
|
|
|
|
(IFF_INTERFACE | IFF_PASSIVE | IFF_REMOTE | IFF_SUBNET);
|
|
|
|
if (ifp->int_flags & IFF_POINTOPOINT &&
|
|
|
|
(ntohl(((struct sockaddr_in *)&ifp->int_dstaddr)->sin_addr.s_addr) &
|
|
|
|
ifp->int_netmask) != ifp->int_net)
|
|
|
|
state &= ~RTS_SUBNET;
|
|
|
|
if (ifp->int_flags & IFF_LOOPBACK)
|
|
|
|
state |= RTS_EXTERNAL;
|
|
|
|
rtadd(dst, &ifp->int_addr, ifp->int_metric, state);
|
|
|
|
if (ifp->int_flags & IFF_POINTOPOINT && foundloopback)
|
|
|
|
add_ptopt_localrt(ifp);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add route to local end of point-to-point using loopback.
|
|
|
|
* If a route to this network is being sent to neighbors on other nets,
|
|
|
|
* mark this route as subnet so we don't have to propagate it too.
|
|
|
|
*/
|
|
|
|
add_ptopt_localrt(ifp)
|
|
|
|
register struct interface *ifp;
|
|
|
|
{
|
|
|
|
struct rt_entry *rt;
|
|
|
|
struct sockaddr *dst;
|
|
|
|
struct sockaddr_in net;
|
|
|
|
int state;
|
|
|
|
|
|
|
|
state = RTS_INTERFACE | RTS_PASSIVE;
|
|
|
|
|
|
|
|
/* look for route to logical network */
|
|
|
|
bzero((char *)&net, sizeof (net));
|
|
|
|
net.sin_family = AF_INET;
|
|
|
|
net.sin_addr = inet_makeaddr(ifp->int_net, INADDR_ANY);
|
|
|
|
dst = (struct sockaddr *)&net;
|
|
|
|
rt = rtfind(dst);
|
|
|
|
if (rt && rt->rt_state & RTS_INTERNAL)
|
|
|
|
state |= RTS_SUBNET;
|
|
|
|
|
|
|
|
dst = &ifp->int_addr;
|
|
|
|
if (rt = rtfind(dst)) {
|
|
|
|
if (rt && rt->rt_state & RTS_INTERFACE)
|
|
|
|
return;
|
|
|
|
rtdelete(rt);
|
|
|
|
}
|
|
|
|
rtadd(dst, &loopaddr, 1, state);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* As a concession to the ARPANET we read a list of gateways
|
|
|
|
* from /etc/gateways and add them to our tables. This file
|
|
|
|
* exists at each ARPANET gateway and indicates a set of ``remote''
|
|
|
|
* gateways (i.e. a gateway which we can't immediately determine
|
|
|
|
* if it's present or not as we can do for those directly connected
|
|
|
|
* at the hardware level). If a gateway is marked ``passive''
|
|
|
|
* in the file, then we assume it doesn't have a routing process
|
|
|
|
* of our design and simply assume it's always present. Those
|
|
|
|
* not marked passive are treated as if they were directly
|
|
|
|
* connected -- they're added into the interface list so we'll
|
|
|
|
* send them routing updates.
|
|
|
|
*
|
|
|
|
* PASSIVE ENTRIES AREN'T NEEDED OR USED ON GATEWAYS RUNNING EGP.
|
|
|
|
*/
|
|
|
|
gwkludge()
|
|
|
|
{
|
|
|
|
struct sockaddr_in dst, gate;
|
|
|
|
FILE *fp;
|
|
|
|
char *type, *dname, *gname, *qual, buf[BUFSIZ];
|
|
|
|
struct interface *ifp;
|
|
|
|
int metric, n;
|
|
|
|
struct rt_entry route;
|
|
|
|
|
|
|
|
fp = fopen(_PATH_GATEWAYS, "r");
|
|
|
|
if (fp == NULL)
|
|
|
|
return;
|
|
|
|
qual = buf;
|
|
|
|
dname = buf + 64;
|
|
|
|
gname = buf + ((BUFSIZ - 64) / 3);
|
|
|
|
type = buf + (((BUFSIZ - 64) * 2) / 3);
|
|
|
|
bzero((char *)&dst, sizeof (dst));
|
|
|
|
bzero((char *)&gate, sizeof (gate));
|
|
|
|
bzero((char *)&route, sizeof(route));
|
|
|
|
/* format: {net | host} XX gateway XX metric DD [passive | external]\n */
|
|
|
|
#define readentry(fp) \
|
|
|
|
fscanf((fp), "%s %s gateway %s metric %d %s\n", \
|
|
|
|
type, dname, gname, &metric, qual)
|
|
|
|
for (;;) {
|
|
|
|
if ((n = readentry(fp)) == EOF)
|
|
|
|
break;
|
|
|
|
if (!getnetorhostname(type, dname, &dst))
|
|
|
|
continue;
|
|
|
|
if (!gethostnameornumber(gname, &gate))
|
|
|
|
continue;
|
|
|
|
if (metric == 0) /* XXX */
|
|
|
|
metric = 1;
|
|
|
|
if (strcmp(qual, "passive") == 0) {
|
|
|
|
/*
|
|
|
|
* Passive entries aren't placed in our tables,
|
|
|
|
* only the kernel's, so we don't copy all of the
|
|
|
|
* external routing information within a net.
|
|
|
|
* Internal machines should use the default
|
|
|
|
* route to a suitable gateway (like us).
|
|
|
|
*/
|
|
|
|
route.rt_dst = *(struct sockaddr *) &dst;
|
|
|
|
route.rt_router = *(struct sockaddr *) &gate;
|
|
|
|
route.rt_flags = RTF_UP;
|
|
|
|
if (strcmp(type, "host") == 0)
|
|
|
|
route.rt_flags |= RTF_HOST;
|
|
|
|
if (metric)
|
|
|
|
route.rt_flags |= RTF_GATEWAY;
|
|
|
|
(void) ioctl(s, SIOCADDRT, (char *)&route.rt_rt);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
if (strcmp(qual, "external") == 0) {
|
|
|
|
/*
|
|
|
|
* Entries marked external are handled
|
|
|
|
* by other means, e.g. EGP,
|
|
|
|
* and are placed in our tables only
|
|
|
|
* to prevent overriding them
|
|
|
|
* with something else.
|
|
|
|
*/
|
|
|
|
rtadd(&dst, &gate, metric, RTS_EXTERNAL|RTS_PASSIVE);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
/* assume no duplicate entries */
|
|
|
|
externalinterfaces++;
|
|
|
|
ifp = (struct interface *)malloc(sizeof (*ifp));
|
|
|
|
bzero((char *)ifp, sizeof (*ifp));
|
|
|
|
ifp->int_flags = IFF_REMOTE;
|
|
|
|
/* can't identify broadcast capability */
|
|
|
|
ifp->int_net = inet_netof(dst.sin_addr);
|
|
|
|
if (strcmp(type, "host") == 0) {
|
|
|
|
ifp->int_flags |= IFF_POINTOPOINT;
|
|
|
|
ifp->int_dstaddr = *((struct sockaddr *)&dst);
|
|
|
|
}
|
|
|
|
ifp->int_addr = *((struct sockaddr *)&gate);
|
|
|
|
ifp->int_metric = metric;
|
|
|
|
ifp->int_next = ifnet;
|
|
|
|
ifnet = ifp;
|
|
|
|
addrouteforif(ifp);
|
|
|
|
}
|
|
|
|
fclose(fp);
|
|
|
|
}
|
|
|
|
|
|
|
|
getnetorhostname(type, name, sin)
|
|
|
|
char *type, *name;
|
|
|
|
struct sockaddr_in *sin;
|
|
|
|
{
|
|
|
|
|
|
|
|
if (strcmp(type, "net") == 0) {
|
|
|
|
struct netent *np = getnetbyname(name);
|
|
|
|
int n;
|
|
|
|
|
|
|
|
if (np == 0)
|
|
|
|
n = inet_network(name);
|
|
|
|
else {
|
|
|
|
if (np->n_addrtype != AF_INET)
|
|
|
|
return (0);
|
|
|
|
n = np->n_net;
|
|
|
|
/*
|
|
|
|
* getnetbyname returns right-adjusted value.
|
|
|
|
*/
|
|
|
|
if (n < 128)
|
|
|
|
n <<= IN_CLASSA_NSHIFT;
|
|
|
|
else if (n < 65536)
|
|
|
|
n <<= IN_CLASSB_NSHIFT;
|
|
|
|
else
|
|
|
|
n <<= IN_CLASSC_NSHIFT;
|
|
|
|
}
|
|
|
|
sin->sin_family = AF_INET;
|
|
|
|
sin->sin_addr = inet_makeaddr(n, INADDR_ANY);
|
|
|
|
return (1);
|
|
|
|
}
|
|
|
|
if (strcmp(type, "host") == 0) {
|
|
|
|
struct hostent *hp = gethostbyname(name);
|
|
|
|
|
|
|
|
if (hp == 0)
|
|
|
|
sin->sin_addr.s_addr = inet_addr(name);
|
|
|
|
else {
|
|
|
|
if (hp->h_addrtype != AF_INET)
|
|
|
|
return (0);
|
|
|
|
bcopy(hp->h_addr, &sin->sin_addr, hp->h_length);
|
|
|
|
}
|
|
|
|
sin->sin_family = AF_INET;
|
|
|
|
return (1);
|
|
|
|
}
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
gethostnameornumber(name, sin)
|
|
|
|
char *name;
|
|
|
|
struct sockaddr_in *sin;
|
|
|
|
{
|
|
|
|
struct hostent *hp;
|
|
|
|
|
|
|
|
hp = gethostbyname(name);
|
|
|
|
if (hp) {
|
|
|
|
bcopy(hp->h_addr, &sin->sin_addr, hp->h_length);
|
|
|
|
sin->sin_family = hp->h_addrtype;
|
|
|
|
return (1);
|
|
|
|
}
|
|
|
|
sin->sin_addr.s_addr = inet_addr(name);
|
|
|
|
sin->sin_family = AF_INET;
|
|
|
|
return (sin->sin_addr.s_addr != -1);
|
|
|
|
}
|