2 * Copyright (c) 1983, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgment:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * $FreeBSD: src/sbin/routed/main.c,v 1.11.2.1 2000/08/14 17:00:03 sheldonh Exp $
34 * $DragonFly: src/sbin/routed/main.c,v 1.4 2005/03/16 21:21:34 cpressey Exp $
38 #include "pathnames.h"
46 #if !defined(sgi) && !defined(__NetBSD__)
48 "@(#) Copyright (c) 1983, 1988, 1993\n\
49 The Regents of the University of California. All rights reserved.\n";
50 static char sccsid[] __attribute__((unused)) = "@(#)main.c 8.1 (Berkeley) 6/5/93";
51 #elif defined(__NetBSD__)
53 __COPYRIGHT("@(#) Copyright (c) 1983, 1988, 1993\n\
54 The Regents of the University of California. All rights reserved.\n");
56 #ident "$FreeBSD: src/sbin/routed/main.c,v 1.11.2.1 2000/08/14 17:00:03 sheldonh Exp $"
61 naddr myaddr; /* system address */
62 char myname[MAXHOSTNAMELEN+1];
66 int supplier; /* supply or broadcast updates */
68 int ipforwarding = 1; /* kernel forwarding on */
70 int default_gateway; /* 1=advertise default */
72 int ridhosts; /* 1=reduce host routes */
73 int mhome; /* 1=want multi-homed host route */
74 int advertise_mhome; /* 1=must continue advertising it */
75 int auth_ok = 1; /* 1=ignore auth if we do not care */
77 struct timeval epoch; /* when started */
78 struct timeval clk, prev_clk;
79 static int usec_fudge;
80 struct timeval now; /* current idea of time */
85 struct timeval next_bcast; /* next general broadcast */
86 struct timeval no_flash = { /* inhibit flash update */
87 EPOCH+SUPPLY_INTERVAL, 0
90 struct timeval flush_kern_timer;
94 int rip_sock = -1; /* RIP socket */
95 struct interface *rip_sock_mcast; /* current multicast interface */
96 int rt_sock; /* routing socket */
100 static int get_rip_sock(naddr, int);
101 static void timevalsub(struct timeval *, struct timeval *, struct timeval *);
111 struct timeval wtime, t2;
115 struct interface *ifp;
120 /* Some shells are badly broken and send SIGHUP to backgrounded
123 signal(SIGHUP, SIG_IGN);
125 openlog("routed", LOG_PID | LOG_ODELAY, LOG_DAEMON);
128 gettimeofday(&clk, 0);
131 epoch.tv_sec -= EPOCH;
133 now_stale = EPOCH - STALE_TIME;
134 now_expire = EPOCH - EXPIRE_TIME;
135 now_garbage = EPOCH - GARBAGE_TIME;
138 gethostname(myname, sizeof(myname)-1);
139 gethost(myname, &myaddr);
141 while ((n = getopt(argc, argv, "sqdghmpAtvT:F:P:")) != -1) {
158 memset(&parm, 0, sizeof(parm));
159 parm.parm_d_metric = 1;
160 cp = check_parms(&parm);
162 msglog("bad -g: %s", cp);
167 case 'h': /* suppress extra host routes */
171 case 'm': /* advertise host route */
172 mhome = 1; /* on multi-homed hosts */
176 /* Ignore authentication if we do not care.
177 * Crazy as it is, that is what RFC 1723 requires.
190 case 'F': /* minimal routes for SLIP */
192 p = strchr(optarg,',');
193 if (p && *p != '\0') {
194 n = (int)strtoul(p+1, &q, 0);
196 && n <= HOPCNT_INFINITY-1
200 if (!getnet(optarg, &p_net, &p_mask)) {
201 msglog("bad network; \"-F %s\"",
205 memset(&parm, 0, sizeof(parm));
206 parm.parm_net = p_net;
207 parm.parm_mask = p_mask;
208 parm.parm_d_metric = n;
209 cp = check_parms(&parm);
211 msglog("bad -F: %s", cp);
215 /* handle arbitrary parameters.
218 cp = parse_parms(q, 0);
220 msglog("%s in \"-P %s\"", cp, optarg);
225 /* display version */
227 msglog("version 2.22");
237 if (tracename == 0 && argc >= 1) {
241 if (tracename != 0 && tracename[0] == '\0')
245 logbad(0, "usage: routed [-sqdghmpAtv] [-T tracefile]"
246 " [-F net[,metric]] [-P parms]");
248 if (geteuid() != 0) {
251 logbad(0, "requires UID 0");
257 mib[3] = IPCTL_FORWARDING;
258 len = sizeof(ipforwarding);
259 if (sysctl(mib, 4, &ipforwarding, &len, 0, 0) < 0)
260 LOGERR("sysctl(IPCTL_FORWARDING)");
264 msglog("-s incompatible with ipforwarding=0");
265 if (default_gateway) {
266 msglog("-g incompatible with ipforwarding=0");
272 if (default_gateway) {
273 if (supplier_set && !supplier) {
274 msglog("-g and -q incompatible");
282 signal(SIGALRM, sigalrm);
284 signal(SIGHUP, sigterm); /* SIGHUP fatal during debugging */
285 signal(SIGTERM, sigterm);
286 signal(SIGINT, sigterm);
287 signal(SIGUSR1, sigtrace_on);
288 signal(SIGUSR2, sigtrace_off);
290 /* get into the background */
292 if (0 > _daemonize(background ? 0 : (_DF_NOCHDIR|_DF_NOFORK),
293 STDIN_FILENO, STDOUT_FILENO, STDERR_FILENO))
294 BADERR(0, "_daemonize()");
296 if (background && daemon(0, 1) < 0)
297 BADERR(0,"daemon()");
301 srandom((int)(clk.tv_sec ^ clk.tv_usec ^ mypid));
303 /* prepare socket connected to the kernel.
305 rt_sock = socket(AF_ROUTE, SOCK_RAW, 0);
307 BADERR(1,"rt_sock = socket()");
308 if (fcntl(rt_sock, F_SETFL, O_NONBLOCK) == -1)
309 logbad(1, "fcntl(rt_sock) O_NONBLOCK: %s", strerror(errno));
311 if (setsockopt(rt_sock, SOL_SOCKET,SO_USELOOPBACK,
312 &off,sizeof(off)) < 0)
313 LOGERR("setsockopt(SO_USELOOPBACK,0)");
318 if (tracename != 0) {
319 strncpy(inittracename, tracename, sizeof(inittracename)-1);
320 set_tracefile(inittracename, "%s", -1);
322 tracelevel_msg("%s", -1); /* turn on tracing to stdio */
327 /* initialize radix tree */
330 /* Pick a random part of the second for our output to minimize
333 * Start broadcasting after hearing from other routers, and
334 * at a random time so a bunch of systems do not get synchronized
335 * after a power failure.
337 intvl_random(&next_bcast, EPOCH+MIN_WAITTIME, EPOCH+SUPPLY_INTERVAL);
338 age_timer.tv_usec = next_bcast.tv_usec;
339 age_timer.tv_sec = EPOCH+MIN_WAITTIME;
340 rdisc_timer = next_bcast;
341 ifinit_timer.tv_usec = next_bcast.tv_usec;
343 /* Collect an initial view of the world by checking the interface
344 * configuration and the kludge file.
353 /* Now turn off stdio if not tracing */
354 if (new_tracelevel == 0)
355 trace_close(background);
357 /* Loop forever, listening and broadcasting.
361 gettimeofday(&clk, 0);
362 if (prev_clk.tv_sec == clk.tv_sec
363 && prev_clk.tv_usec == clk.tv_usec+usec_fudge) {
364 /* Much of `routed` depends on time always advancing.
365 * On systems that do not guarantee that gettimeofday()
366 * produces unique timestamps even if called within
367 * a single tick, use trickery like that in classic
370 clk.tv_usec += ++usec_fudge;
375 timevalsub(&t2, &clk, &prev_clk);
377 || t2.tv_sec > wtime.tv_sec + 5) {
378 /* Deal with time changes before other
379 * housekeeping to keep everything straight.
384 trace_act("time changed by %d sec", (int)dt);
388 timevalsub(&now, &clk, &epoch);
389 now_stale = now.tv_sec - STALE_TIME;
390 now_expire = now.tv_sec - EXPIRE_TIME;
391 now_garbage = now.tv_sec - GARBAGE_TIME;
393 /* deal with signals that should affect tracing */
399 trace_off("exiting with signal %d", stopint);
403 /* look for new or dead interfaces */
404 timevalsub(&wtime, &ifinit_timer, &now);
405 if (wtime.tv_sec <= 0) {
412 /* Check the kernel table occassionally for mysteriously
415 timevalsub(&t2, &flush_kern_timer, &now);
416 if (t2.tv_sec <= 0) {
418 flush_kern_timer.tv_sec = (now.tv_sec
419 + CHECK_QUIET_INTERVAL);
422 if (timercmp(&t2, &wtime, <))
425 /* If it is time, then broadcast our routes.
427 if (supplier || advertise_mhome) {
428 timevalsub(&t2, &next_bcast, &now);
429 if (t2.tv_sec <= 0) {
430 /* Synchronize the aging and broadcast
431 * timers to minimize awakenings
437 /* It is desirable to send routing updates
438 * regularly. So schedule the next update
439 * 30 seconds after the previous one was
440 * scheduled, instead of 30 seconds after
441 * the previous update was finished.
442 * Even if we just started after discovering
443 * a 2nd interface or were otherwise delayed,
444 * pick a 30-second aniversary of the
445 * original broadcast time.
447 n = 1 + (0-t2.tv_sec)/SUPPLY_INTERVAL;
448 next_bcast.tv_sec += n*SUPPLY_INTERVAL;
453 if (timercmp(&t2, &wtime, <))
457 /* If we need a flash update, either do it now or
458 * set the delay to end when it is time.
460 * If we are within MIN_WAITTIME seconds of a full update,
465 && no_flash.tv_sec+MIN_WAITTIME < next_bcast.tv_sec) {
466 /* accurate to the millisecond */
467 if (!timercmp(&no_flash, &now, >))
469 timevalsub(&t2, &no_flash, &now);
470 if (timercmp(&t2, &wtime, <))
474 /* trigger the main aging timer.
476 timevalsub(&t2, &age_timer, &now);
477 if (t2.tv_sec <= 0) {
481 if (timercmp(&t2, &wtime, <))
484 /* update the kernel routing table
486 timevalsub(&t2, &need_kern, &now);
487 if (t2.tv_sec <= 0) {
491 if (timercmp(&t2, &wtime, <))
494 /* take care of router discovery,
495 * but do it in the correct the millisecond
497 if (!timercmp(&rdisc_timer, &now, >)) {
501 timevalsub(&t2, &rdisc_timer, &now);
502 if (timercmp(&t2, &wtime, <))
506 /* wait for input or a timer to expire.
510 n = select(sock_max, &ibits, 0, 0, &wtime);
512 if (n < 0 && errno != EINTR && errno != EAGAIN)
517 if (FD_ISSET(rt_sock, &ibits)) {
521 if (rdisc_sock >= 0 && FD_ISSET(rdisc_sock, &ibits)) {
525 if (rip_sock >= 0 && FD_ISSET(rip_sock, &ibits)) {
526 read_rip(rip_sock, 0);
530 for (ifp = ifnet; n > 0 && 0 != ifp; ifp = ifp->int_next) {
531 if (ifp->int_rip_sock >= 0
532 && FD_ISSET(ifp->int_rip_sock, &ibits)) {
533 read_rip(ifp->int_rip_sock, ifp);
543 sigalrm(int s UNUSED)
545 /* Historically, SIGALRM would cause the daemon to check for
546 * new and broken interfaces.
548 ifinit_timer.tv_sec = now.tv_sec;
549 trace_act("SIGALRM");
553 /* watch for fatal signals */
558 signal(sig, SIG_DFL); /* catch it only once */
565 struct interface *ifp;
571 FD_SET(rt_sock, &fdbits);
572 if (sock_max <= rt_sock)
573 sock_max = rt_sock+1;
575 FD_SET(rip_sock, &fdbits);
576 if (sock_max <= rip_sock)
577 sock_max = rip_sock+1;
579 for (ifp = ifnet; 0 != ifp; ifp = ifp->int_next) {
580 if (ifp->int_rip_sock >= 0) {
581 FD_SET(ifp->int_rip_sock, &fdbits);
582 if (sock_max <= ifp->int_rip_sock)
583 sock_max = ifp->int_rip_sock+1;
586 if (rdisc_sock >= 0) {
587 FD_SET(rdisc_sock, &fdbits);
588 if (sock_max <= rdisc_sock)
589 sock_max = rdisc_sock+1;
599 #define MIN_SOCKBUF (4*1024)
602 if (fcntl(sock, F_SETFL, O_NONBLOCK) == -1)
603 logbad(1, "fcntl(%s) O_NONBLOCK: %s",
604 name, strerror(errno));
606 if (setsockopt(sock, SOL_SOCKET,SO_BROADCAST, &on,sizeof(on)) < 0)
607 msglog("setsockopt(%s,SO_BROADCAST): %s",
608 name, strerror(errno));
609 #ifdef USE_PASSIFNAME
611 if (setsockopt(sock, SOL_SOCKET, SO_PASSIFNAME, &on,sizeof(on)) < 0)
612 msglog("setsockopt(%s,SO_PASSIFNAME): %s",
613 name, strerror(errno));
616 if (rbuf >= MIN_SOCKBUF) {
617 if (setsockopt(sock, SOL_SOCKET, SO_RCVBUF,
618 &rbuf, sizeof(rbuf)) < 0)
619 msglog("setsockopt(%s,SO_RCVBUF=%d): %s",
620 name, rbuf, strerror(errno));
622 for (rbuf = 60*1024; ; rbuf -= 4096) {
623 if (setsockopt(sock, SOL_SOCKET, SO_RCVBUF,
624 &rbuf, sizeof(rbuf)) == 0) {
625 trace_act("RCVBUF=%d", rbuf);
628 if (rbuf < MIN_SOCKBUF) {
629 msglog("setsockopt(%s,SO_RCVBUF = %d): %s",
630 name, rbuf, strerror(errno));
640 static int /* <0 or file descriptor */
641 get_rip_sock(naddr addr,
642 int serious) /* 1=failure to bind is serious */
644 struct sockaddr_in in;
649 if ((s = socket(AF_INET, SOCK_DGRAM, 0)) < 0)
650 BADERR(1,"rip_sock = socket()");
652 memset(&in, 0, sizeof(in));
654 in.sin_len = sizeof(in);
656 in.sin_family = AF_INET;
657 in.sin_port = htons(RIP_PORT);
658 in.sin_addr.s_addr = addr;
659 if (bind(s, (struct sockaddr *)&in, sizeof(in)) < 0) {
661 BADERR(errno != EADDRINUSE, "bind(rip_sock)");
664 fix_sock(s,"rip_sock");
667 if (setsockopt(s, IPPROTO_IP, IP_MULTICAST_TTL,
668 &ttl, sizeof(ttl)) < 0)
669 DBGERR(1,"rip_sock setsockopt(IP_MULTICAST_TTL)");
675 /* turn off main RIP socket */
679 struct interface *ifp;
683 if (rip_sock >= 0 && !mhome) {
684 trace_act("turn off RIP");
689 /* get non-broadcast sockets to listen to queries.
691 for (ifp = ifnet; ifp != 0; ifp = ifp->int_next) {
692 if (ifp->int_state & IS_REMOTE)
694 if (ifp->int_rip_sock < 0) {
695 addr = ((ifp->int_if_flags & IFF_POINTOPOINT)
698 ifp->int_rip_sock = get_rip_sock(addr, 0);
709 /* turn on RIP multicast input via an interface
712 rip_mcast_on(struct interface *ifp)
716 if (!IS_RIP_IN_OFF(ifp->int_state)
717 && (ifp->int_if_flags & IFF_MULTICAST)
719 && !(ifp->int_if_flags & IFF_POINTOPOINT)
721 && !(ifp->int_state & IS_ALIAS)) {
722 m.imr_multiaddr.s_addr = htonl(INADDR_RIP_GROUP);
723 m.imr_interface.s_addr = ((ifp->int_if_flags & IFF_POINTOPOINT)
726 if (setsockopt(rip_sock,IPPROTO_IP, IP_ADD_MEMBERSHIP,
728 LOGERR("setsockopt(IP_ADD_MEMBERSHIP RIP)");
733 /* Prepare socket used for RIP.
736 rip_on(struct interface *ifp)
738 /* If the main RIP socket is already alive, only start receiving
739 * multicasts for this interface.
747 /* If the main RIP socket is off and it makes sense to turn it on,
748 * then turn it on for all of the interfaces.
749 * It makes sense if either router discovery is off, or if
750 * router discover is on and at most one interface is doing RIP.
752 if (rip_interfaces > 0 && (!rdisc_ok || rip_interfaces > 1)) {
753 trace_act("turn on RIP");
755 /* Close all of the query sockets so that we can open
756 * the main socket. SO_REUSEPORT is not a solution,
757 * since that would let two daemons bind to the broadcast
760 for (ifp = ifnet; ifp != 0; ifp = ifp->int_next) {
761 if (ifp->int_rip_sock >= 0) {
762 close(ifp->int_rip_sock);
763 ifp->int_rip_sock = -1;
767 rip_sock = get_rip_sock(INADDR_ANY, 1);
770 /* Do not advertise anything until we have heard something
772 if (next_bcast.tv_sec < now.tv_sec+MIN_WAITTIME)
773 next_bcast.tv_sec = now.tv_sec+MIN_WAITTIME;
775 for (ifp = ifnet; ifp != 0; ifp = ifp->int_next) {
776 ifp->int_query_time = NEVER;
779 ifinit_timer.tv_sec = now.tv_sec;
782 && !(ifp->int_state & IS_REMOTE)
783 && ifp->int_rip_sock < 0) {
784 /* RIP is off, so ensure there are sockets on which
785 * to listen for queries.
787 ifp->int_rip_sock = get_rip_sock(ifp->int_addr, 0);
794 /* die if malloc(3) fails
797 rtmalloc(size_t size,
800 void *p = malloc(size);
802 logbad(1,"malloc(%lu) failed in %s", (u_long)size, msg);
807 /* get a random instant in an interval
810 intvl_random(struct timeval *tp, /* put value here */
811 u_long lo, /* value is after this second */
812 u_long hi) /* and before this */
814 tp->tv_sec = (time_t)(hi == lo
816 : (lo + random() % ((hi - lo))));
817 tp->tv_usec = random() % 1000000;
822 timevaladd(struct timeval *t1,
826 t1->tv_sec += t2->tv_sec;
827 if ((t1->tv_usec += t2->tv_usec) >= 1000000) {
829 t1->tv_usec -= 1000000;
837 timevalsub(struct timeval *t1,
841 t1->tv_sec = t2->tv_sec - t3->tv_sec;
842 if ((t1->tv_usec = t2->tv_usec - t3->tv_usec) < 0) {
844 t1->tv_usec += 1000000;
849 /* put a message into the system log
852 msglog(const char *p, ...)
859 vsyslog(LOG_ERR, p, args);
862 if (ftrace == stdout)
863 fputs("routed: ", ftrace);
864 vfprintf(ftrace, p, args);
870 /* Put a message about a bad system into the system log if
871 * we have not complained about it recently.
873 * It is desirable to complain about all bad systems, but not too often.
874 * In the worst case, it is not practical to keep track of all bad systems.
875 * For example, there can be many systems with the wrong password.
878 msglim(struct msg_limit *lim, naddr addr, const char *p, ...)
882 struct msg_sub *ms1, *ms;
887 /* look for the oldest slot in the table
888 * or the slot for the bad router.
890 ms = ms1 = lim->subs;
891 for (i = MSG_SUBJECT_N; ; i--, ms1++) {
893 /* Reuse a slot at most once every 10 minutes.
895 if (lim->reuse > now.tv_sec) {
899 lim->reuse = now.tv_sec + 10*60;
903 if (ms->addr == addr) {
904 /* Repeat a complaint about a given system at
907 if (ms->until > now.tv_sec)
911 if (ms->until < ms1->until)
916 ms->until = now.tv_sec + 60*60; /* 60 minutes */
919 for (p1 = p; *p1 == ' '; p1++)
921 vsyslog(LOG_ERR, p1, args);
924 /* always display the message if tracing */
926 vfprintf(ftrace, p, args);
933 logbad(int dump, const char *p, ...)
940 vsyslog(LOG_ERR, p, args);
942 fputs("routed: ", stderr);
943 vfprintf(stderr, p, args);
944 fputs("; giving up\n",stderr);