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");
60 naddr myaddr
; /* system address */
61 char myname
[MAXHOSTNAMELEN
+1];
65 int supplier
; /* supply or broadcast updates */
67 int ipforwarding
= 1; /* kernel forwarding on */
69 int default_gateway
; /* 1=advertise default */
71 int ridhosts
; /* 1=reduce host routes */
72 int mhome
; /* 1=want multi-homed host route */
73 int advertise_mhome
; /* 1=must continue advertising it */
74 int auth_ok
= 1; /* 1=ignore auth if we do not care */
76 struct timeval epoch
; /* when started */
77 struct timeval clk
, prev_clk
;
78 static int usec_fudge
;
79 struct timeval now
; /* current idea of time */
84 struct timeval next_bcast
; /* next general broadcast */
85 struct timeval no_flash
= { /* inhibit flash update */
86 EPOCH
+SUPPLY_INTERVAL
, 0
89 struct timeval flush_kern_timer
;
93 int rip_sock
= -1; /* RIP socket */
94 struct interface
*rip_sock_mcast
; /* current multicast interface */
95 int rt_sock
; /* routing socket */
99 static int get_rip_sock(naddr
, int);
100 static void timevalsub(struct timeval
*, struct timeval
*, struct timeval
*);
110 struct timeval wtime
, t2
;
114 struct interface
*ifp
;
119 /* Some shells are badly broken and send SIGHUP to backgrounded
122 signal(SIGHUP
, SIG_IGN
);
124 openlog("routed", LOG_PID
| LOG_ODELAY
, LOG_DAEMON
);
127 gettimeofday(&clk
, 0);
130 epoch
.tv_sec
-= EPOCH
;
132 now_stale
= EPOCH
- STALE_TIME
;
133 now_expire
= EPOCH
- EXPIRE_TIME
;
134 now_garbage
= EPOCH
- GARBAGE_TIME
;
137 gethostname(myname
, sizeof(myname
)-1);
138 gethost(myname
, &myaddr
);
140 while ((n
= getopt(argc
, argv
, "sqdghmpAtvT:F:P:")) != -1) {
157 memset(&parm
, 0, sizeof(parm
));
158 parm
.parm_d_metric
= 1;
159 cp
= check_parms(&parm
);
161 msglog("bad -g: %s", cp
);
166 case 'h': /* suppress extra host routes */
170 case 'm': /* advertise host route */
171 mhome
= 1; /* on multi-homed hosts */
175 /* Ignore authentication if we do not care.
176 * Crazy as it is, that is what RFC 1723 requires.
189 case 'F': /* minimal routes for SLIP */
191 p
= strchr(optarg
,',');
192 if (p
&& *p
!= '\0') {
193 n
= (int)strtoul(p
+1, &q
, 0);
195 && n
<= HOPCNT_INFINITY
-1
199 if (!getnet(optarg
, &p_net
, &p_mask
)) {
200 msglog("bad network; \"-F %s\"",
204 memset(&parm
, 0, sizeof(parm
));
205 parm
.parm_net
= p_net
;
206 parm
.parm_mask
= p_mask
;
207 parm
.parm_d_metric
= n
;
208 cp
= check_parms(&parm
);
210 msglog("bad -F: %s", cp
);
214 /* handle arbitrary parameters.
217 cp
= parse_parms(q
, 0);
219 msglog("%s in \"-P %s\"", cp
, optarg
);
224 /* display version */
226 msglog("version 2.22");
236 if (tracename
== 0 && argc
>= 1) {
240 if (tracename
!= 0 && tracename
[0] == '\0')
244 logbad(0, "usage: routed [-sqdghmpAtv] [-T tracefile]"
245 " [-F net[,metric]] [-P parms]");
247 if (geteuid() != 0) {
250 logbad(0, "requires UID 0");
256 mib
[3] = IPCTL_FORWARDING
;
257 len
= sizeof(ipforwarding
);
258 if (sysctl(mib
, 4, &ipforwarding
, &len
, 0, 0) < 0)
259 LOGERR("sysctl(IPCTL_FORWARDING)");
263 msglog("-s incompatible with ipforwarding=0");
264 if (default_gateway
) {
265 msglog("-g incompatible with ipforwarding=0");
271 if (default_gateway
) {
272 if (supplier_set
&& !supplier
) {
273 msglog("-g and -q incompatible");
281 signal(SIGALRM
, sigalrm
);
283 signal(SIGHUP
, sigterm
); /* SIGHUP fatal during debugging */
284 signal(SIGTERM
, sigterm
);
285 signal(SIGINT
, sigterm
);
286 signal(SIGUSR1
, sigtrace_on
);
287 signal(SIGUSR2
, sigtrace_off
);
289 /* get into the background */
291 if (0 > _daemonize(background
? 0 : (_DF_NOCHDIR
|_DF_NOFORK
),
292 STDIN_FILENO
, STDOUT_FILENO
, STDERR_FILENO
))
293 BADERR(0, "_daemonize()");
295 if (background
&& daemon(0, 1) < 0)
296 BADERR(0,"daemon()");
300 srandom((int)(clk
.tv_sec
^ clk
.tv_usec
^ mypid
));
302 /* prepare socket connected to the kernel.
304 rt_sock
= socket(AF_ROUTE
, SOCK_RAW
, 0);
306 BADERR(1,"rt_sock = socket()");
307 if (fcntl(rt_sock
, F_SETFL
, O_NONBLOCK
) == -1)
308 logbad(1, "fcntl(rt_sock) O_NONBLOCK: %s", strerror(errno
));
310 if (setsockopt(rt_sock
, SOL_SOCKET
,SO_USELOOPBACK
,
311 &off
,sizeof(off
)) < 0)
312 LOGERR("setsockopt(SO_USELOOPBACK,0)");
317 if (tracename
!= 0) {
318 strncpy(inittracename
, tracename
, sizeof(inittracename
)-1);
319 set_tracefile(inittracename
, "%s", -1);
321 tracelevel_msg("%s", -1); /* turn on tracing to stdio */
326 /* initialize radix tree */
329 /* Pick a random part of the second for our output to minimize
332 * Start broadcasting after hearing from other routers, and
333 * at a random time so a bunch of systems do not get synchronized
334 * after a power failure.
336 intvl_random(&next_bcast
, EPOCH
+MIN_WAITTIME
, EPOCH
+SUPPLY_INTERVAL
);
337 age_timer
.tv_usec
= next_bcast
.tv_usec
;
338 age_timer
.tv_sec
= EPOCH
+MIN_WAITTIME
;
339 rdisc_timer
= next_bcast
;
340 ifinit_timer
.tv_usec
= next_bcast
.tv_usec
;
342 /* Collect an initial view of the world by checking the interface
343 * configuration and the kludge file.
352 /* Now turn off stdio if not tracing */
353 if (new_tracelevel
== 0)
354 trace_close(background
);
356 /* Loop forever, listening and broadcasting.
360 gettimeofday(&clk
, 0);
361 if (prev_clk
.tv_sec
== clk
.tv_sec
362 && prev_clk
.tv_usec
== clk
.tv_usec
+usec_fudge
) {
363 /* Much of `routed` depends on time always advancing.
364 * On systems that do not guarantee that gettimeofday()
365 * produces unique timestamps even if called within
366 * a single tick, use trickery like that in classic
369 clk
.tv_usec
+= ++usec_fudge
;
374 timevalsub(&t2
, &clk
, &prev_clk
);
376 || t2
.tv_sec
> wtime
.tv_sec
+ 5) {
377 /* Deal with time changes before other
378 * housekeeping to keep everything straight.
383 trace_act("time changed by %d sec", (int)dt
);
387 timevalsub(&now
, &clk
, &epoch
);
388 now_stale
= now
.tv_sec
- STALE_TIME
;
389 now_expire
= now
.tv_sec
- EXPIRE_TIME
;
390 now_garbage
= now
.tv_sec
- GARBAGE_TIME
;
392 /* deal with signals that should affect tracing */
398 trace_off("exiting with signal %d", stopint
);
402 /* look for new or dead interfaces */
403 timevalsub(&wtime
, &ifinit_timer
, &now
);
404 if (wtime
.tv_sec
<= 0) {
411 /* Check the kernel table occassionally for mysteriously
414 timevalsub(&t2
, &flush_kern_timer
, &now
);
415 if (t2
.tv_sec
<= 0) {
417 flush_kern_timer
.tv_sec
= (now
.tv_sec
418 + CHECK_QUIET_INTERVAL
);
421 if (timercmp(&t2
, &wtime
, <))
424 /* If it is time, then broadcast our routes.
426 if (supplier
|| advertise_mhome
) {
427 timevalsub(&t2
, &next_bcast
, &now
);
428 if (t2
.tv_sec
<= 0) {
429 /* Synchronize the aging and broadcast
430 * timers to minimize awakenings
436 /* It is desirable to send routing updates
437 * regularly. So schedule the next update
438 * 30 seconds after the previous one was
439 * scheduled, instead of 30 seconds after
440 * the previous update was finished.
441 * Even if we just started after discovering
442 * a 2nd interface or were otherwise delayed,
443 * pick a 30-second aniversary of the
444 * original broadcast time.
446 n
= 1 + (0-t2
.tv_sec
)/SUPPLY_INTERVAL
;
447 next_bcast
.tv_sec
+= n
*SUPPLY_INTERVAL
;
452 if (timercmp(&t2
, &wtime
, <))
456 /* If we need a flash update, either do it now or
457 * set the delay to end when it is time.
459 * If we are within MIN_WAITTIME seconds of a full update,
464 && no_flash
.tv_sec
+MIN_WAITTIME
< next_bcast
.tv_sec
) {
465 /* accurate to the millisecond */
466 if (!timercmp(&no_flash
, &now
, >))
468 timevalsub(&t2
, &no_flash
, &now
);
469 if (timercmp(&t2
, &wtime
, <))
473 /* trigger the main aging timer.
475 timevalsub(&t2
, &age_timer
, &now
);
476 if (t2
.tv_sec
<= 0) {
480 if (timercmp(&t2
, &wtime
, <))
483 /* update the kernel routing table
485 timevalsub(&t2
, &need_kern
, &now
);
486 if (t2
.tv_sec
<= 0) {
490 if (timercmp(&t2
, &wtime
, <))
493 /* take care of router discovery,
494 * but do it in the correct the millisecond
496 if (!timercmp(&rdisc_timer
, &now
, >)) {
500 timevalsub(&t2
, &rdisc_timer
, &now
);
501 if (timercmp(&t2
, &wtime
, <))
505 /* wait for input or a timer to expire.
509 n
= select(sock_max
, &ibits
, 0, 0, &wtime
);
511 if (n
< 0 && errno
!= EINTR
&& errno
!= EAGAIN
)
516 if (FD_ISSET(rt_sock
, &ibits
)) {
520 if (rdisc_sock
>= 0 && FD_ISSET(rdisc_sock
, &ibits
)) {
524 if (rip_sock
>= 0 && FD_ISSET(rip_sock
, &ibits
)) {
525 read_rip(rip_sock
, 0);
529 for (ifp
= ifnet
; n
> 0 && 0 != ifp
; ifp
= ifp
->int_next
) {
530 if (ifp
->int_rip_sock
>= 0
531 && FD_ISSET(ifp
->int_rip_sock
, &ibits
)) {
532 read_rip(ifp
->int_rip_sock
, ifp
);
542 sigalrm(int s UNUSED
)
544 /* Historically, SIGALRM would cause the daemon to check for
545 * new and broken interfaces.
547 ifinit_timer
.tv_sec
= now
.tv_sec
;
548 trace_act("SIGALRM");
552 /* watch for fatal signals */
557 signal(sig
, SIG_DFL
); /* catch it only once */
564 struct interface
*ifp
;
570 FD_SET(rt_sock
, &fdbits
);
571 if (sock_max
<= rt_sock
)
572 sock_max
= rt_sock
+1;
574 FD_SET(rip_sock
, &fdbits
);
575 if (sock_max
<= rip_sock
)
576 sock_max
= rip_sock
+1;
578 for (ifp
= ifnet
; 0 != ifp
; ifp
= ifp
->int_next
) {
579 if (ifp
->int_rip_sock
>= 0) {
580 FD_SET(ifp
->int_rip_sock
, &fdbits
);
581 if (sock_max
<= ifp
->int_rip_sock
)
582 sock_max
= ifp
->int_rip_sock
+1;
585 if (rdisc_sock
>= 0) {
586 FD_SET(rdisc_sock
, &fdbits
);
587 if (sock_max
<= rdisc_sock
)
588 sock_max
= rdisc_sock
+1;
598 #define MIN_SOCKBUF (4*1024)
601 if (fcntl(sock
, F_SETFL
, O_NONBLOCK
) == -1)
602 logbad(1, "fcntl(%s) O_NONBLOCK: %s",
603 name
, strerror(errno
));
605 if (setsockopt(sock
, SOL_SOCKET
,SO_BROADCAST
, &on
,sizeof(on
)) < 0)
606 msglog("setsockopt(%s,SO_BROADCAST): %s",
607 name
, strerror(errno
));
608 #ifdef USE_PASSIFNAME
610 if (setsockopt(sock
, SOL_SOCKET
, SO_PASSIFNAME
, &on
,sizeof(on
)) < 0)
611 msglog("setsockopt(%s,SO_PASSIFNAME): %s",
612 name
, strerror(errno
));
615 if (rbuf
>= MIN_SOCKBUF
) {
616 if (setsockopt(sock
, SOL_SOCKET
, SO_RCVBUF
,
617 &rbuf
, sizeof(rbuf
)) < 0)
618 msglog("setsockopt(%s,SO_RCVBUF=%d): %s",
619 name
, rbuf
, strerror(errno
));
621 for (rbuf
= 60*1024; ; rbuf
-= 4096) {
622 if (setsockopt(sock
, SOL_SOCKET
, SO_RCVBUF
,
623 &rbuf
, sizeof(rbuf
)) == 0) {
624 trace_act("RCVBUF=%d", rbuf
);
627 if (rbuf
< MIN_SOCKBUF
) {
628 msglog("setsockopt(%s,SO_RCVBUF = %d): %s",
629 name
, rbuf
, strerror(errno
));
639 static int /* <0 or file descriptor */
640 get_rip_sock(naddr addr
,
641 int serious
) /* 1=failure to bind is serious */
643 struct sockaddr_in in
;
648 if ((s
= socket(AF_INET
, SOCK_DGRAM
, 0)) < 0)
649 BADERR(1,"rip_sock = socket()");
651 memset(&in
, 0, sizeof(in
));
653 in
.sin_len
= sizeof(in
);
655 in
.sin_family
= AF_INET
;
656 in
.sin_port
= htons(RIP_PORT
);
657 in
.sin_addr
.s_addr
= addr
;
658 if (bind(s
, (struct sockaddr
*)&in
, sizeof(in
)) < 0) {
660 BADERR(errno
!= EADDRINUSE
, "bind(rip_sock)");
663 fix_sock(s
,"rip_sock");
666 if (setsockopt(s
, IPPROTO_IP
, IP_MULTICAST_TTL
,
667 &ttl
, sizeof(ttl
)) < 0)
668 DBGERR(1,"rip_sock setsockopt(IP_MULTICAST_TTL)");
674 /* turn off main RIP socket */
678 struct interface
*ifp
;
682 if (rip_sock
>= 0 && !mhome
) {
683 trace_act("turn off RIP");
688 /* get non-broadcast sockets to listen to queries.
690 for (ifp
= ifnet
; ifp
!= 0; ifp
= ifp
->int_next
) {
691 if (ifp
->int_state
& IS_REMOTE
)
693 if (ifp
->int_rip_sock
< 0) {
694 addr
= ((ifp
->int_if_flags
& IFF_POINTOPOINT
)
697 ifp
->int_rip_sock
= get_rip_sock(addr
, 0);
708 /* turn on RIP multicast input via an interface
711 rip_mcast_on(struct interface
*ifp
)
715 if (!IS_RIP_IN_OFF(ifp
->int_state
)
716 && (ifp
->int_if_flags
& IFF_MULTICAST
)
718 && !(ifp
->int_if_flags
& IFF_POINTOPOINT
)
720 && !(ifp
->int_state
& IS_ALIAS
)) {
721 m
.imr_multiaddr
.s_addr
= htonl(INADDR_RIP_GROUP
);
722 m
.imr_interface
.s_addr
= ((ifp
->int_if_flags
& IFF_POINTOPOINT
)
725 if (setsockopt(rip_sock
,IPPROTO_IP
, IP_ADD_MEMBERSHIP
,
727 LOGERR("setsockopt(IP_ADD_MEMBERSHIP RIP)");
732 /* Prepare socket used for RIP.
735 rip_on(struct interface
*ifp
)
737 /* If the main RIP socket is already alive, only start receiving
738 * multicasts for this interface.
746 /* If the main RIP socket is off and it makes sense to turn it on,
747 * then turn it on for all of the interfaces.
748 * It makes sense if either router discovery is off, or if
749 * router discover is on and at most one interface is doing RIP.
751 if (rip_interfaces
> 0 && (!rdisc_ok
|| rip_interfaces
> 1)) {
752 trace_act("turn on RIP");
754 /* Close all of the query sockets so that we can open
755 * the main socket. SO_REUSEPORT is not a solution,
756 * since that would let two daemons bind to the broadcast
759 for (ifp
= ifnet
; ifp
!= 0; ifp
= ifp
->int_next
) {
760 if (ifp
->int_rip_sock
>= 0) {
761 close(ifp
->int_rip_sock
);
762 ifp
->int_rip_sock
= -1;
766 rip_sock
= get_rip_sock(INADDR_ANY
, 1);
769 /* Do not advertise anything until we have heard something
771 if (next_bcast
.tv_sec
< now
.tv_sec
+MIN_WAITTIME
)
772 next_bcast
.tv_sec
= now
.tv_sec
+MIN_WAITTIME
;
774 for (ifp
= ifnet
; ifp
!= 0; ifp
= ifp
->int_next
) {
775 ifp
->int_query_time
= NEVER
;
778 ifinit_timer
.tv_sec
= now
.tv_sec
;
781 && !(ifp
->int_state
& IS_REMOTE
)
782 && ifp
->int_rip_sock
< 0) {
783 /* RIP is off, so ensure there are sockets on which
784 * to listen for queries.
786 ifp
->int_rip_sock
= get_rip_sock(ifp
->int_addr
, 0);
793 /* die if malloc(3) fails
796 rtmalloc(size_t size
,
799 void *p
= malloc(size
);
801 logbad(1,"malloc(%lu) failed in %s", (u_long
)size
, msg
);
806 /* get a random instant in an interval
809 intvl_random(struct timeval
*tp
, /* put value here */
810 u_long lo
, /* value is after this second */
811 u_long hi
) /* and before this */
813 tp
->tv_sec
= (time_t)(hi
== lo
815 : (lo
+ random() % ((hi
- lo
))));
816 tp
->tv_usec
= random() % 1000000;
821 timevaladd(struct timeval
*t1
,
825 t1
->tv_sec
+= t2
->tv_sec
;
826 if ((t1
->tv_usec
+= t2
->tv_usec
) >= 1000000) {
828 t1
->tv_usec
-= 1000000;
836 timevalsub(struct timeval
*t1
,
840 t1
->tv_sec
= t2
->tv_sec
- t3
->tv_sec
;
841 if ((t1
->tv_usec
= t2
->tv_usec
- t3
->tv_usec
) < 0) {
843 t1
->tv_usec
+= 1000000;
848 /* put a message into the system log
851 msglog(const char *p
, ...)
858 vsyslog(LOG_ERR
, p
, args
);
861 if (ftrace
== stdout
)
862 fputs("routed: ", ftrace
);
863 vfprintf(ftrace
, p
, args
);
869 /* Put a message about a bad system into the system log if
870 * we have not complained about it recently.
872 * It is desirable to complain about all bad systems, but not too often.
873 * In the worst case, it is not practical to keep track of all bad systems.
874 * For example, there can be many systems with the wrong password.
877 msglim(struct msg_limit
*lim
, naddr addr
, const char *p
, ...)
881 struct msg_sub
*ms1
, *ms
;
886 /* look for the oldest slot in the table
887 * or the slot for the bad router.
889 ms
= ms1
= lim
->subs
;
890 for (i
= MSG_SUBJECT_N
; ; i
--, ms1
++) {
892 /* Reuse a slot at most once every 10 minutes.
894 if (lim
->reuse
> now
.tv_sec
) {
898 lim
->reuse
= now
.tv_sec
+ 10*60;
902 if (ms
->addr
== addr
) {
903 /* Repeat a complaint about a given system at
906 if (ms
->until
> now
.tv_sec
)
910 if (ms
->until
< ms1
->until
)
915 ms
->until
= now
.tv_sec
+ 60*60; /* 60 minutes */
918 for (p1
= p
; *p1
== ' '; p1
++)
920 vsyslog(LOG_ERR
, p1
, args
);
923 /* always display the message if tracing */
925 vfprintf(ftrace
, p
, args
);
932 logbad(int dump
, const char *p
, ...)
939 vsyslog(LOG_ERR
, p
, args
);
941 fputs("routed: ", stderr
);
942 vfprintf(stderr
, p
, args
);
943 fputs("; giving up\n",stderr
);