Home | History | Annotate | Line # | Download | only in net
rtsock.c revision 1.40.4.2
      1 /*	$NetBSD: rtsock.c,v 1.40.4.2 2003/06/25 09:26:12 msaitoh Exp $	*/
      2 
      3 /*
      4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
      5  * All rights reserved.
      6  *
      7  * Redistribution and use in source and binary forms, with or without
      8  * modification, are permitted provided that the following conditions
      9  * are met:
     10  * 1. Redistributions of source code must retain the above copyright
     11  *    notice, this list of conditions and the following disclaimer.
     12  * 2. Redistributions in binary form must reproduce the above copyright
     13  *    notice, this list of conditions and the following disclaimer in the
     14  *    documentation and/or other materials provided with the distribution.
     15  * 3. Neither the name of the project nor the names of its contributors
     16  *    may be used to endorse or promote products derived from this software
     17  *    without specific prior written permission.
     18  *
     19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
     20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
     23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     29  * SUCH DAMAGE.
     30  */
     31 
     32 /*
     33  * Copyright (c) 1988, 1991, 1993
     34  *	The Regents of the University of California.  All rights reserved.
     35  *
     36  * Redistribution and use in source and binary forms, with or without
     37  * modification, are permitted provided that the following conditions
     38  * are met:
     39  * 1. Redistributions of source code must retain the above copyright
     40  *    notice, this list of conditions and the following disclaimer.
     41  * 2. Redistributions in binary form must reproduce the above copyright
     42  *    notice, this list of conditions and the following disclaimer in the
     43  *    documentation and/or other materials provided with the distribution.
     44  * 3. All advertising materials mentioning features or use of this software
     45  *    must display the following acknowledgement:
     46  *	This product includes software developed by the University of
     47  *	California, Berkeley and its contributors.
     48  * 4. Neither the name of the University nor the names of its contributors
     49  *    may be used to endorse or promote products derived from this software
     50  *    without specific prior written permission.
     51  *
     52  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     53  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     54  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     55  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     56  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     57  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     58  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     59  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     60  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     61  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     62  * SUCH DAMAGE.
     63  *
     64  *	@(#)rtsock.c	8.7 (Berkeley) 10/12/95
     65  */
     66 
     67 #include "opt_inet.h"
     68 
     69 #include <sys/param.h>
     70 #include <sys/systm.h>
     71 #include <sys/proc.h>
     72 #include <sys/mbuf.h>
     73 #include <sys/socket.h>
     74 #include <sys/socketvar.h>
     75 #include <sys/domain.h>
     76 #include <sys/protosw.h>
     77 
     78 #include <vm/vm.h>
     79 #include <sys/sysctl.h>
     80 
     81 #include <net/if.h>
     82 #include <net/route.h>
     83 #include <net/raw_cb.h>
     84 
     85 #include <machine/stdarg.h>
     86 
     87 struct	sockaddr route_dst = { 2, PF_ROUTE, };
     88 struct	sockaddr route_src = { 2, PF_ROUTE, };
     89 struct	sockproto route_proto = { PF_ROUTE, };
     90 
     91 struct walkarg {
     92 	int	w_op;
     93 	int	w_arg;
     94 	int	w_given;
     95 	int	w_needed;
     96 	caddr_t	w_where;
     97 	int	w_tmemsize;
     98 	int	w_tmemneeded;
     99 	caddr_t	w_tmem;
    100 };
    101 
    102 static struct mbuf *rt_msg1 __P((int, struct rt_addrinfo *, caddr_t, int));
    103 static int rt_msg2 __P((int, struct rt_addrinfo *, caddr_t, struct walkarg *,
    104     int *));
    105 static void rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
    106 static int sysctl_dumpentry __P((struct radix_node *, void *));
    107 static int sysctl_iflist __P((int, struct walkarg *, int));
    108 static int sysctl_rtable __P((int *, u_int, void *, size_t *, void *, size_t));
    109 static __inline void rt_adjustcount __P((int, int));
    110 
    111 /* Sleazy use of local variables throughout file, warning!!!! */
    112 #define dst	info.rti_info[RTAX_DST]
    113 #define gate	info.rti_info[RTAX_GATEWAY]
    114 #define netmask	info.rti_info[RTAX_NETMASK]
    115 #define genmask	info.rti_info[RTAX_GENMASK]
    116 #define ifpaddr	info.rti_info[RTAX_IFP]
    117 #define ifaaddr	info.rti_info[RTAX_IFA]
    118 #define brdaddr	info.rti_info[RTAX_BRD]
    119 
    120 static __inline void
    121 rt_adjustcount(af, cnt)
    122 	int af, cnt;
    123 {
    124 	route_cb.any_count += cnt;
    125 	switch (af) {
    126 	case AF_INET:
    127 		route_cb.ip_count += cnt;
    128 		return;
    129 #ifdef INET6
    130 	case AF_INET6:
    131 		route_cb.ip6_count += cnt;
    132 		return;
    133 #endif
    134 	case AF_IPX:
    135 		route_cb.ipx_count += cnt;
    136 		return;
    137 	case AF_NS:
    138 		route_cb.ns_count += cnt;
    139 		return;
    140 	case AF_ISO:
    141 		route_cb.iso_count += cnt;
    142 		return;
    143 	}
    144 }
    145 
    146 /*ARGSUSED*/
    147 int
    148 route_usrreq(so, req, m, nam, control, p)
    149 	struct socket *so;
    150 	int req;
    151 	struct mbuf *m, *nam, *control;
    152 	struct proc *p;
    153 {
    154 	int error = 0;
    155 	struct rawcb *rp = sotorawcb(so);
    156 	int s;
    157 
    158 	if (req == PRU_ATTACH) {
    159 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
    160 		if ((so->so_pcb = rp) != NULL)
    161 			bzero(so->so_pcb, sizeof(*rp));
    162 
    163 	}
    164 	if (req == PRU_DETACH && rp)
    165 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
    166 	s = splsoftnet();
    167 
    168 	/*
    169 	 * Don't call raw_usrreq() in the attach case, because
    170 	 * we want to allow non-privileged processes to listen on
    171 	 * and send "safe" commands to the routing socket.
    172 	 */
    173 	if (req == PRU_ATTACH) {
    174 		if (p == 0)
    175 			error = EACCES;
    176 		else
    177 			error = raw_attach(so, (int)(long)nam);
    178 	} else
    179 		error = raw_usrreq(so, req, m, nam, control, p);
    180 
    181 	rp = sotorawcb(so);
    182 	if (req == PRU_ATTACH && rp) {
    183 		if (error) {
    184 			free((caddr_t)rp, M_PCB);
    185 			splx(s);
    186 			return (error);
    187 		}
    188 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
    189 		rp->rcb_laddr = &route_src;
    190 		rp->rcb_faddr = &route_dst;
    191 		soisconnected(so);
    192 		so->so_options |= SO_USELOOPBACK;
    193 	}
    194 	splx(s);
    195 	return (error);
    196 }
    197 
    198 /*ARGSUSED*/
    199 int
    200 #if __STDC__
    201 route_output(struct mbuf *m, ...)
    202 #else
    203 route_output(m, va_alist)
    204 	struct mbuf *m;
    205 	va_dcl
    206 #endif
    207 {
    208 	struct rt_msghdr *rtm = 0;
    209 	struct radix_node *rn = 0;
    210 	struct rtentry *rt = 0;
    211 	struct rtentry *saved_nrt = 0;
    212 	struct radix_node_head *rnh;
    213 	struct rt_addrinfo info;
    214 	int len, error = 0;
    215 	struct ifnet *ifp = 0;
    216 	struct ifaddr *ifa = 0;
    217 	struct socket *so;
    218 	va_list ap;
    219 
    220 	va_start(ap, m);
    221 	so = va_arg(ap, struct socket *);
    222 	va_end(ap);
    223 
    224 	bzero(&info, sizeof(info));
    225 #define senderr(e) do { error = e; goto flush;} while (0)
    226 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
    227 	   (m = m_pullup(m, sizeof(int32_t))) == 0))
    228 		return (ENOBUFS);
    229 	if ((m->m_flags & M_PKTHDR) == 0)
    230 		panic("route_output");
    231 	len = m->m_pkthdr.len;
    232 	if (len < sizeof(*rtm) ||
    233 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
    234 		dst = 0;
    235 		senderr(EINVAL);
    236 	}
    237 	R_Malloc(rtm, struct rt_msghdr *, len);
    238 	if (rtm == 0) {
    239 		dst = 0;
    240 		senderr(ENOBUFS);
    241 	}
    242 	m_copydata(m, 0, len, (caddr_t)rtm);
    243 	if (rtm->rtm_version != RTM_VERSION) {
    244 		dst = 0;
    245 		senderr(EPROTONOSUPPORT);
    246 	}
    247 	rtm->rtm_pid = curproc->p_pid;
    248 	info.rti_addrs = rtm->rtm_addrs;
    249 	rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info);
    250 	if (dst == 0 || (dst->sa_family >= AF_MAX))
    251 		senderr(EINVAL);
    252 	if (gate != 0 && (gate->sa_family >= AF_MAX))
    253 		senderr(EINVAL);
    254 	if (genmask) {
    255 		struct radix_node *t;
    256 		t = rn_addmask((caddr_t)genmask, 0, 1);
    257 		if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0)
    258 			genmask = (struct sockaddr *)(t->rn_key);
    259 		else
    260 			senderr(ENOBUFS);
    261 	}
    262 
    263 	/*
    264 	 * Verify that the caller has the appropriate privilege; RTM_GET
    265 	 * is the only operation the non-superuser is allowed.
    266 	 */
    267 	if (rtm->rtm_type != RTM_GET &&
    268 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
    269 		senderr(EACCES);
    270 
    271 	switch (rtm->rtm_type) {
    272 
    273 	case RTM_ADD:
    274 		if (gate == 0)
    275 			senderr(EINVAL);
    276 		error = rtrequest(RTM_ADD, dst, gate, netmask,
    277 		    rtm->rtm_flags, &saved_nrt);
    278 		if (error == 0 && saved_nrt) {
    279 			rt_setmetrics(rtm->rtm_inits,
    280 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
    281 			saved_nrt->rt_refcnt--;
    282 			saved_nrt->rt_genmask = genmask;
    283 		}
    284 		break;
    285 
    286 	case RTM_DELETE:
    287 		error = rtrequest(RTM_DELETE, dst, gate, netmask,
    288 		    rtm->rtm_flags, &saved_nrt);
    289 		if (error == 0) {
    290 			(rt = saved_nrt)->rt_refcnt++;
    291 			goto report;
    292 		}
    293 		break;
    294 
    295 	case RTM_GET:
    296 	case RTM_CHANGE:
    297 	case RTM_LOCK:
    298 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
    299 			senderr(EAFNOSUPPORT);
    300 		}
    301 		rn = rnh->rnh_lookup(dst, netmask, rnh);
    302 		if (rn == NULL || (rn->rn_flags & RNF_ROOT) != 0) {
    303 			senderr(ESRCH);
    304 		}
    305 		rt = (struct rtentry *)rn;
    306 		rt->rt_refcnt++;
    307 		if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
    308 			struct radix_node *rn;
    309 			extern struct radix_node_head *mask_rnhead;
    310 
    311 			if (Bcmp(dst, rt_key(rt), dst->sa_len) != 0)
    312 				senderr(ESRCH);
    313 			if (netmask && (rn = rn_search(netmask,
    314 					    mask_rnhead->rnh_treetop)))
    315 				netmask = (struct sockaddr *)rn->rn_key;
    316 			for (rn = rt->rt_nodes; rn; rn = rn->rn_dupedkey)
    317 				if (netmask == (struct sockaddr *)rn->rn_mask)
    318 					break;
    319 			if (rn == 0)
    320 				senderr(ETOOMANYREFS);
    321 			rt = (struct rtentry *)rn;
    322 		}
    323 
    324 		switch(rtm->rtm_type) {
    325 
    326 		case RTM_GET:
    327 		report:
    328 			dst = rt_key(rt);
    329 			gate = rt->rt_gateway;
    330 			netmask = rt_mask(rt);
    331 			genmask = rt->rt_genmask;
    332 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
    333 				if ((ifp = rt->rt_ifp) != NULL) {
    334 					ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
    335 					ifaaddr = rt->rt_ifa->ifa_addr;
    336 					if (ifp->if_flags & IFF_POINTOPOINT)
    337 						brdaddr = rt->rt_ifa->ifa_dstaddr;
    338 					else
    339 						brdaddr = 0;
    340 					rtm->rtm_index = ifp->if_index;
    341 				} else {
    342 					ifpaddr = 0;
    343 					ifaaddr = 0;
    344 				}
    345 			}
    346 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
    347 			    (struct walkarg *)0, &len);
    348 			if (len > rtm->rtm_msglen) {
    349 				struct rt_msghdr *new_rtm;
    350 				R_Malloc(new_rtm, struct rt_msghdr *, len);
    351 				if (new_rtm == 0)
    352 					senderr(ENOBUFS);
    353 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
    354 				Free(rtm); rtm = new_rtm;
    355 			}
    356 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
    357 			    (struct walkarg *)0, 0);
    358 			rtm->rtm_flags = rt->rt_flags;
    359 			rtm->rtm_rmx = rt->rt_rmx;
    360 			rtm->rtm_addrs = info.rti_addrs;
    361 			break;
    362 
    363 		case RTM_CHANGE:
    364 			if (gate && rt_setgate(rt, rt_key(rt), gate))
    365 				senderr(EDQUOT);
    366 			/* new gateway could require new ifaddr, ifp;
    367 			   flags may also be different; ifp may be specified
    368 			   by ll sockaddr when protocol address is ambiguous */
    369 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
    370 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
    371 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
    372 				    ifp);
    373 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
    374 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
    375 			    rt_key(rt), gate))))
    376 				ifp = ifa->ifa_ifp;
    377 			if (ifa) {
    378 				struct ifaddr *oifa = rt->rt_ifa;
    379 				if (oifa != ifa) {
    380 				    if (oifa && oifa->ifa_rtrequest)
    381 					oifa->ifa_rtrequest(RTM_DELETE,
    382 					rt, gate);
    383 				    IFAFREE(rt->rt_ifa);
    384 				    rt->rt_ifa = ifa;
    385 				    IFAREF(rt->rt_ifa);
    386 				    rt->rt_ifp = ifp;
    387 				}
    388 			}
    389 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
    390 			    &rt->rt_rmx);
    391 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
    392 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate);
    393 			if (genmask)
    394 				rt->rt_genmask = genmask;
    395 			/*
    396 			 * Fall into
    397 			 */
    398 		case RTM_LOCK:
    399 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
    400 			rt->rt_rmx.rmx_locks |=
    401 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
    402 			break;
    403 		}
    404 		break;
    405 
    406 	default:
    407 		senderr(EOPNOTSUPP);
    408 	}
    409 
    410 flush:
    411 	if (rtm) {
    412 		if (error)
    413 			rtm->rtm_errno = error;
    414 		else
    415 			rtm->rtm_flags |= RTF_DONE;
    416 	}
    417 	if (rt)
    418 		rtfree(rt);
    419     {
    420 	struct rawcb *rp = 0;
    421 	/*
    422 	 * Check to see if we don't want our own messages.
    423 	 */
    424 	if ((so->so_options & SO_USELOOPBACK) == 0) {
    425 		if (route_cb.any_count <= 1) {
    426 			if (rtm)
    427 				Free(rtm);
    428 			m_freem(m);
    429 			return (error);
    430 		}
    431 		/* There is another listener, so construct message */
    432 		rp = sotorawcb(so);
    433 	}
    434 	if (rtm) {
    435 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
    436 		Free(rtm);
    437 	}
    438 	if (rp)
    439 		rp->rcb_proto.sp_family = 0; /* Avoid us */
    440 	if (dst)
    441 		route_proto.sp_protocol = dst->sa_family;
    442 	raw_input(m, &route_proto, &route_src, &route_dst);
    443 	if (rp)
    444 		rp->rcb_proto.sp_family = PF_ROUTE;
    445     }
    446 	return (error);
    447 }
    448 
    449 void
    450 rt_setmetrics(which, in, out)
    451 	u_long which;
    452 	struct rt_metrics *in, *out;
    453 {
    454 #define metric(f, e) if (which & (f)) out->e = in->e;
    455 	metric(RTV_RPIPE, rmx_recvpipe);
    456 	metric(RTV_SPIPE, rmx_sendpipe);
    457 	metric(RTV_SSTHRESH, rmx_ssthresh);
    458 	metric(RTV_RTT, rmx_rtt);
    459 	metric(RTV_RTTVAR, rmx_rttvar);
    460 	metric(RTV_HOPCOUNT, rmx_hopcount);
    461 	metric(RTV_MTU, rmx_mtu);
    462 	metric(RTV_EXPIRE, rmx_expire);
    463 #undef metric
    464 }
    465 
    466 #define ROUNDUP(a) \
    467 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
    468 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
    469 
    470 static void
    471 rt_xaddrs(cp, cplim, rtinfo)
    472 	caddr_t cp, cplim;
    473 	struct rt_addrinfo *rtinfo;
    474 {
    475 	struct sockaddr *sa;
    476 	int i;
    477 
    478 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
    479 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
    480 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
    481 			continue;
    482 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
    483 		ADVANCE(cp, sa);
    484 	}
    485 }
    486 
    487 static struct mbuf *
    488 rt_msg1(type, rtinfo, data, datalen)
    489 	int type;
    490 	struct rt_addrinfo *rtinfo;
    491 	caddr_t data;
    492 	int datalen;
    493 {
    494 	struct rt_msghdr *rtm;
    495 	struct mbuf *m;
    496 	int i;
    497 	struct sockaddr *sa;
    498 	int len, dlen;
    499 
    500 	m = m_gethdr(M_DONTWAIT, MT_DATA);
    501 	if (m == 0)
    502 		return (m);
    503 	switch (type) {
    504 
    505 	case RTM_DELADDR:
    506 	case RTM_NEWADDR:
    507 		len = sizeof(struct ifa_msghdr);
    508 		break;
    509 
    510 #ifdef COMPAT_14
    511 	case RTM_OIFINFO:
    512 		len = sizeof(struct if_msghdr14);
    513 		break;
    514 #endif
    515 
    516 	case RTM_IFINFO:
    517 		len = sizeof(struct if_msghdr);
    518 		break;
    519 
    520 	case RTM_IFANNOUNCE:
    521 		len = sizeof(struct if_announcemsghdr);
    522 		break;
    523 
    524 	default:
    525 		len = sizeof(struct rt_msghdr);
    526 	}
    527 	if (len > MHLEN + MLEN)
    528 		panic("rt_msg1: message too long");
    529 	else if (len > MHLEN) {
    530 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
    531 		if (m->m_next == NULL) {
    532 			m_freem(m);
    533 			return (NULL);
    534 		}
    535 		m->m_pkthdr.len = len;
    536 		m->m_len = MHLEN;
    537 		m->m_next->m_len = len - MHLEN;
    538 	} else {
    539 		m->m_pkthdr.len = m->m_len = len;
    540 	}
    541 	m->m_pkthdr.rcvif = 0;
    542 	m_copyback(m, 0, datalen, data);
    543 	rtm = mtod(m, struct rt_msghdr *);
    544 	for (i = 0; i < RTAX_MAX; i++) {
    545 		if ((sa = rtinfo->rti_info[i]) == NULL)
    546 			continue;
    547 		rtinfo->rti_addrs |= (1 << i);
    548 		dlen = ROUNDUP(sa->sa_len);
    549 		m_copyback(m, len, dlen, (caddr_t)sa);
    550 		len += dlen;
    551 	}
    552 	rtm->rtm_msglen = len;
    553 	rtm->rtm_version = RTM_VERSION;
    554 	rtm->rtm_type = type;
    555 	return (m);
    556 }
    557 
    558 /*
    559  * rt_msg2
    560  *
    561  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
    562  *		returns the length of the message in 'lenp'.
    563  *
    564  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
    565  *	the message
    566  * otherwise walkarg's w_needed is updated and if the user buffer is
    567  *	specified and w_needed indicates space exists the information is copied
    568  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
    569  *	if the allocation fails ENOBUFS is returned.
    570  */
    571 static int
    572 rt_msg2(type, rtinfo, cp, w, lenp)
    573 	int type;
    574 	struct rt_addrinfo *rtinfo;
    575 	caddr_t cp;
    576 	struct walkarg *w;
    577 	int *lenp;
    578 {
    579 	int i;
    580 	int len, dlen, second_time = 0;
    581 	caddr_t cp0;
    582 
    583 	rtinfo->rti_addrs = 0;
    584 again:
    585 	switch (type) {
    586 
    587 	case RTM_DELADDR:
    588 	case RTM_NEWADDR:
    589 		len = sizeof(struct ifa_msghdr);
    590 		break;
    591 #ifdef COMPAT_14
    592 	case RTM_OIFINFO:
    593 		len = sizeof(struct if_msghdr14);
    594 		break;
    595 #endif
    596 
    597 	case RTM_IFINFO:
    598 		len = sizeof(struct if_msghdr);
    599 		break;
    600 
    601 	default:
    602 		len = sizeof(struct rt_msghdr);
    603 	}
    604 	if ((cp0 = cp) != NULL)
    605 		cp += len;
    606 	for (i = 0; i < RTAX_MAX; i++) {
    607 		struct sockaddr *sa;
    608 
    609 		if ((sa = rtinfo->rti_info[i]) == 0)
    610 			continue;
    611 		rtinfo->rti_addrs |= (1 << i);
    612 		dlen = ROUNDUP(sa->sa_len);
    613 		if (cp) {
    614 			bcopy(sa, cp, (unsigned)dlen);
    615 			cp += dlen;
    616 		}
    617 		len += dlen;
    618 	}
    619 	if (cp == 0 && w != NULL && !second_time) {
    620 		struct walkarg *rw = w;
    621 
    622 		rw->w_needed += len;
    623 		if (rw->w_needed <= 0 && rw->w_where) {
    624 			if (rw->w_tmemsize < len) {
    625 				if (rw->w_tmem)
    626 					free(rw->w_tmem, M_RTABLE);
    627 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
    628 				    M_NOWAIT);
    629 				if (rw->w_tmem)
    630 					rw->w_tmemsize = len;
    631 			}
    632 			if (rw->w_tmem) {
    633 				cp = rw->w_tmem;
    634 				second_time = 1;
    635 				goto again;
    636 			} else {
    637 				rw->w_tmemneeded = len;
    638 				return (ENOBUFS);
    639 			}
    640 		}
    641 	}
    642 	if (cp) {
    643 		struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
    644 
    645 		rtm->rtm_version = RTM_VERSION;
    646 		rtm->rtm_type = type;
    647 		rtm->rtm_msglen = len;
    648 	}
    649 	if (lenp)
    650 		*lenp = len;
    651 	return (0);
    652 }
    653 
    654 /*
    655  * This routine is called to generate a message from the routing
    656  * socket indicating that a redirect has occured, a routing lookup
    657  * has failed, or that a protocol has detected timeouts to a particular
    658  * destination.
    659  */
    660 void
    661 rt_missmsg(type, rtinfo, flags, error)
    662 	int type, flags, error;
    663 	struct rt_addrinfo *rtinfo;
    664 {
    665 	struct rt_msghdr rtm;
    666 	struct mbuf *m;
    667 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
    668 
    669 	if (route_cb.any_count == 0)
    670 		return;
    671 	bzero(&rtm, sizeof(rtm));
    672 	rtm.rtm_flags = RTF_DONE | flags;
    673 	rtm.rtm_errno = error;
    674 	m = rt_msg1(type, rtinfo, (caddr_t)&rtm, sizeof(rtm));
    675 	if (m == 0)
    676 		return;
    677 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
    678 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
    679 	raw_input(m, &route_proto, &route_src, &route_dst);
    680 }
    681 
    682 /*
    683  * This routine is called to generate a message from the routing
    684  * socket indicating that the status of a network interface has changed.
    685  */
    686 void
    687 rt_ifmsg(ifp)
    688 	struct ifnet *ifp;
    689 {
    690 	struct if_msghdr ifm;
    691 #ifdef COMPAT_14
    692 	struct if_msghdr14 oifm;
    693 #endif
    694 	struct mbuf *m;
    695 	struct rt_addrinfo info;
    696 
    697 	if (route_cb.any_count == 0)
    698 		return;
    699 	bzero(&info, sizeof(info));
    700 	bzero(&ifm, sizeof(ifm));
    701 	ifm.ifm_index = ifp->if_index;
    702 	ifm.ifm_flags = ifp->if_flags;
    703 	ifm.ifm_data = ifp->if_data;
    704 	ifm.ifm_addrs = 0;
    705 	m = rt_msg1(RTM_IFINFO, &info, (caddr_t)&ifm, sizeof(ifm));
    706 	if (m == 0)
    707 		return;
    708 	route_proto.sp_protocol = 0;
    709 	raw_input(m, &route_proto, &route_src, &route_dst);
    710 #ifdef COMPAT_14
    711 	bzero(&info, sizeof(info));
    712 	bzero(&oifm, sizeof(oifm));
    713 	oifm.ifm_index = ifp->if_index;
    714 	oifm.ifm_flags = ifp->if_flags;
    715 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
    716 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
    717 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
    718 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
    719 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
    720 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
    721 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
    722 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
    723 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
    724 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
    725 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
    726 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
    727 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
    728 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
    729 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
    730 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
    731 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
    732 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
    733 	oifm.ifm_addrs = 0;
    734 	m = rt_msg1(RTM_OIFINFO, &info, (caddr_t)&oifm, sizeof(oifm));
    735 	if (m == 0)
    736 		return;
    737 	route_proto.sp_protocol = 0;
    738 	raw_input(m, &route_proto, &route_src, &route_dst);
    739 #endif
    740 }
    741 
    742 /*
    743  * This is called to generate messages from the routing socket
    744  * indicating a network interface has had addresses associated with it.
    745  * if we ever reverse the logic and replace messages TO the routing
    746  * socket indicate a request to configure interfaces, then it will
    747  * be unnecessary as the routing socket will automatically generate
    748  * copies of it.
    749  */
    750 void
    751 rt_newaddrmsg(cmd, ifa, error, rt)
    752 	int cmd, error;
    753 	struct ifaddr *ifa;
    754 	struct rtentry *rt;
    755 {
    756 	struct rt_addrinfo info;
    757 	struct sockaddr *sa = NULL;
    758 	int pass;
    759 	struct mbuf *m = NULL;
    760 	struct ifnet *ifp = ifa->ifa_ifp;
    761 
    762 	if (route_cb.any_count == 0)
    763 		return;
    764 	for (pass = 1; pass < 3; pass++) {
    765 		bzero(&info, sizeof(info));
    766 		if ((cmd == RTM_ADD && pass == 1) ||
    767 		    (cmd == RTM_DELETE && pass == 2)) {
    768 			struct ifa_msghdr ifam;
    769 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
    770 
    771 			ifaaddr = sa = ifa->ifa_addr;
    772 			ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
    773 			netmask = ifa->ifa_netmask;
    774 			brdaddr = ifa->ifa_dstaddr;
    775 			bzero(&ifam, sizeof(ifam));
    776 			ifam.ifam_index = ifp->if_index;
    777 			ifam.ifam_metric = ifa->ifa_metric;
    778 			ifam.ifam_flags = ifa->ifa_flags;
    779 			m = rt_msg1(ncmd, &info, (caddr_t)&ifam, sizeof(ifam));
    780 			if (m == NULL)
    781 				continue;
    782 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
    783 			    info.rti_addrs;
    784 		}
    785 		if ((cmd == RTM_ADD && pass == 2) ||
    786 		    (cmd == RTM_DELETE && pass == 1)) {
    787 			struct rt_msghdr rtm;
    788 
    789 			if (rt == 0)
    790 				continue;
    791 			netmask = rt_mask(rt);
    792 			dst = sa = rt_key(rt);
    793 			gate = rt->rt_gateway;
    794 			bzero(&rtm, sizeof(rtm));
    795 			rtm.rtm_index = ifp->if_index;
    796 			rtm.rtm_flags |= rt->rt_flags;
    797 			rtm.rtm_errno = error;
    798 			m = rt_msg1(cmd, &info, (caddr_t)&rtm, sizeof(rtm));
    799 			if (m == NULL)
    800 				continue;
    801 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
    802 		}
    803 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
    804 		raw_input(m, &route_proto, &route_src, &route_dst);
    805 	}
    806 }
    807 
    808 /*
    809  * This is called to generate routing socket messages indicating
    810  * network interface arrival and departure.
    811  */
    812 void
    813 rt_ifannouncemsg(ifp, what)
    814 	struct ifnet *ifp;
    815 	int what;
    816 {
    817 	struct if_announcemsghdr ifan;
    818 	struct mbuf *m;
    819 	struct rt_addrinfo info;
    820 
    821 	if (route_cb.any_count == 0)
    822 		return;
    823 	bzero(&info, sizeof(info));
    824 	bzero(&ifan, sizeof(ifan));
    825 	ifan.ifan_index = ifp->if_index;
    826 	strcpy(ifan.ifan_name, ifp->if_xname);
    827 	ifan.ifan_what = what;
    828 	m = rt_msg1(RTM_IFANNOUNCE, &info, (caddr_t)&ifan, sizeof(ifan));
    829 	if (m == 0)
    830 		return;
    831 	route_proto.sp_protocol = 0;
    832 	raw_input(m, &route_proto, &route_src, &route_dst);
    833 }
    834 
    835 /*
    836  * This is used in dumping the kernel table via sysctl().
    837  */
    838 static int
    839 sysctl_dumpentry(rn, v)
    840 	struct radix_node *rn;
    841 	void *v;
    842 {
    843 	struct walkarg *w = v;
    844 	struct rtentry *rt = (struct rtentry *)rn;
    845 	int error = 0, size;
    846 	struct rt_addrinfo info;
    847 
    848 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
    849 		return 0;
    850 	bzero(&info, sizeof(info));
    851 	dst = rt_key(rt);
    852 	gate = rt->rt_gateway;
    853 	netmask = rt_mask(rt);
    854 	genmask = rt->rt_genmask;
    855 	if (rt->rt_ifp) {
    856 		ifpaddr = rt->rt_ifp->if_addrlist.tqh_first->ifa_addr;
    857 		ifaaddr = rt->rt_ifa->ifa_addr;
    858 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
    859 			brdaddr = rt->rt_ifa->ifa_dstaddr;
    860 	}
    861 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
    862 		return (error);
    863 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    864 		struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
    865 
    866 		rtm->rtm_flags = rt->rt_flags;
    867 		rtm->rtm_use = rt->rt_use;
    868 		rtm->rtm_rmx = rt->rt_rmx;
    869 		rtm->rtm_index = rt->rt_ifp->if_index;
    870 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
    871 		rtm->rtm_addrs = info.rti_addrs;
    872 		if ((error = copyout(rtm, w->w_where, size)) != 0)
    873 			w->w_where = NULL;
    874 		else
    875 			w->w_where += size;
    876 	}
    877 	return (error);
    878 }
    879 
    880 static int
    881 sysctl_iflist(af, w, type)
    882 	int	af;
    883 	struct	walkarg *w;
    884 	int type;
    885 {
    886 	struct ifnet *ifp;
    887 	struct ifaddr *ifa;
    888 	struct	rt_addrinfo info;
    889 	int	len, error = 0;
    890 
    891 	bzero(&info, sizeof(info));
    892 	for (ifp = ifnet.tqh_first; ifp != 0; ifp = ifp->if_list.tqe_next) {
    893 		if (w->w_arg && w->w_arg != ifp->if_index)
    894 			continue;
    895 		ifa = ifp->if_addrlist.tqh_first;
    896 		ifpaddr = ifa->ifa_addr;
    897 		switch(type) {
    898 		case NET_RT_IFLIST:
    899 			error =
    900 			    rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w, &len);
    901 			break;
    902 #ifdef COMPAT_14
    903 		case NET_RT_OIFLIST:
    904 			error =
    905 			    rt_msg2(RTM_OIFINFO, &info, (caddr_t)0, w, &len);
    906 			break;
    907 #endif
    908 		default:
    909 			panic("sysctl_iflist(1)");
    910 		}
    911 		if (error)
    912 			return (error);
    913 		ifpaddr = 0;
    914 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    915 			switch(type) {
    916 			case NET_RT_IFLIST: {
    917 				struct if_msghdr *ifm;
    918 
    919 				ifm = (struct if_msghdr *)w->w_tmem;
    920 				ifm->ifm_index = ifp->if_index;
    921 				ifm->ifm_flags = ifp->if_flags;
    922 				ifm->ifm_data = ifp->if_data;
    923 				ifm->ifm_addrs = info.rti_addrs;
    924 				error = copyout(ifm, w->w_where, len);
    925 				if (error)
    926 					return (error);
    927 				w->w_where += len;
    928 				break;
    929 			}
    930 
    931 #ifdef COMPAT_14
    932 			case NET_RT_OIFLIST: {
    933 				struct if_msghdr14 *ifm;
    934 
    935 				ifm = (struct if_msghdr14 *)w->w_tmem;
    936 				ifm->ifm_index = ifp->if_index;
    937 				ifm->ifm_flags = ifp->if_flags;
    938 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
    939 				ifm->ifm_data.ifi_addrlen =
    940 				    ifp->if_data.ifi_addrlen;
    941 				ifm->ifm_data.ifi_hdrlen =
    942 				    ifp->if_data.ifi_hdrlen;
    943 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
    944 				ifm->ifm_data.ifi_metric =
    945 				    ifp->if_data.ifi_metric;
    946 				ifm->ifm_data.ifi_baudrate =
    947 				    ifp->if_data.ifi_baudrate;
    948 				ifm->ifm_data.ifi_ipackets =
    949 				    ifp->if_data.ifi_ipackets;
    950 				ifm->ifm_data.ifi_ierrors =
    951 				    ifp->if_data.ifi_ierrors;
    952 				ifm->ifm_data.ifi_opackets =
    953 				    ifp->if_data.ifi_opackets;
    954 				ifm->ifm_data.ifi_oerrors =
    955 				    ifp->if_data.ifi_oerrors;
    956 				ifm->ifm_data.ifi_collisions =
    957 				    ifp->if_data.ifi_collisions;
    958 				ifm->ifm_data.ifi_ibytes =
    959 				    ifp->if_data.ifi_ibytes;
    960 				ifm->ifm_data.ifi_obytes =
    961 				    ifp->if_data.ifi_obytes;
    962 				ifm->ifm_data.ifi_imcasts =
    963 				    ifp->if_data.ifi_imcasts;
    964 				ifm->ifm_data.ifi_omcasts =
    965 				    ifp->if_data.ifi_omcasts;
    966 				ifm->ifm_data.ifi_iqdrops =
    967 				    ifp->if_data.ifi_iqdrops;
    968 				ifm->ifm_data.ifi_noproto =
    969 				    ifp->if_data.ifi_noproto;
    970 				ifm->ifm_data.ifi_lastchange =
    971 				    ifp->if_data.ifi_lastchange;
    972 				ifm->ifm_addrs = info.rti_addrs;
    973 				error = copyout(ifm, w->w_where, len);
    974 				if (error)
    975 					return (error);
    976 				w->w_where += len;
    977 				break;
    978 			}
    979 #endif
    980 			default:
    981 				panic("sysctl_iflist(2)");
    982 			}
    983 		}
    984 		while ((ifa = ifa->ifa_list.tqe_next) != NULL) {
    985 			if (af && af != ifa->ifa_addr->sa_family)
    986 				continue;
    987 			ifaaddr = ifa->ifa_addr;
    988 			netmask = ifa->ifa_netmask;
    989 			brdaddr = ifa->ifa_dstaddr;
    990 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
    991 				return (error);
    992 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    993 				struct ifa_msghdr *ifam;
    994 
    995 				ifam = (struct ifa_msghdr *)w->w_tmem;
    996 				ifam->ifam_index = ifa->ifa_ifp->if_index;
    997 				ifam->ifam_flags = ifa->ifa_flags;
    998 				ifam->ifam_metric = ifa->ifa_metric;
    999 				ifam->ifam_addrs = info.rti_addrs;
   1000 				error = copyout(w->w_tmem, w->w_where, len);
   1001 				if (error)
   1002 					return (error);
   1003 				w->w_where += len;
   1004 			}
   1005 		}
   1006 		ifaaddr = netmask = brdaddr = 0;
   1007 	}
   1008 	return (0);
   1009 }
   1010 
   1011 static int
   1012 sysctl_rtable(name, namelen, where, given, new, newlen)
   1013 	int	*name;
   1014 	u_int	namelen;
   1015 	void 	*where;
   1016 	size_t	*given;
   1017 	void	*new;
   1018 	size_t	newlen;
   1019 {
   1020 	struct radix_node_head *rnh;
   1021 	int	i, s, error = EINVAL;
   1022 	u_char  af;
   1023 	struct	walkarg w;
   1024 
   1025 	if (new)
   1026 		return (EPERM);
   1027 	if (namelen != 3)
   1028 		return (EINVAL);
   1029 	af = name[0];
   1030 	w.w_tmemneeded = 0;
   1031 	w.w_tmemsize = 0;
   1032 	w.w_tmem = NULL;
   1033 again:
   1034 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
   1035 	if (w.w_tmemneeded) {
   1036 		w.w_tmem = (caddr_t) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
   1037 		w.w_tmemsize = w.w_tmemneeded;
   1038 		w.w_tmemneeded = 0;
   1039 	}
   1040 	w.w_op = name[1];
   1041 	w.w_arg = name[2];
   1042 	w.w_given = *given;
   1043 	w.w_needed = 0 - w.w_given;
   1044 	w.w_where = where;
   1045 
   1046 	s = splsoftnet();
   1047 	switch (w.w_op) {
   1048 
   1049 	case NET_RT_DUMP:
   1050 	case NET_RT_FLAGS:
   1051 		for (i = 1; i <= AF_MAX; i++)
   1052 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
   1053 			    (error = (*rnh->rnh_walktree)(rnh,
   1054 			    sysctl_dumpentry, &w)))
   1055 				break;
   1056 		break;
   1057 
   1058 #ifdef COMPAT_14
   1059 	case NET_RT_OIFLIST:
   1060 		error = sysctl_iflist(af, &w, w.w_op);
   1061 		break;
   1062 #endif
   1063 
   1064 	case NET_RT_IFLIST:
   1065 		error = sysctl_iflist(af, &w, w.w_op);
   1066 	}
   1067 	splx(s);
   1068 
   1069 	/* check to see if we couldn't allocate memory with NOWAIT */
   1070 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
   1071 		goto again;
   1072 
   1073 	if (w.w_tmem)
   1074 		free(w.w_tmem, M_RTABLE);
   1075 	w.w_needed += w.w_given;
   1076 	if (where) {
   1077 		*given = w.w_where - (caddr_t) where;
   1078 		if (*given < w.w_needed)
   1079 			return (ENOMEM);
   1080 	} else {
   1081 		*given = (11 * w.w_needed) / 10;
   1082 	}
   1083 	return (error);
   1084 }
   1085 
   1086 /*
   1087  * Definitions of protocols supported in the ROUTE domain.
   1088  */
   1089 
   1090 extern	struct domain routedomain;		/* or at least forward */
   1091 
   1092 struct protosw routesw[] = {
   1093 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
   1094   raw_input,	route_output,	raw_ctlinput,	0,
   1095   route_usrreq,
   1096   raw_init,	0,		0,		0,
   1097   sysctl_rtable,
   1098 }
   1099 };
   1100 
   1101 struct domain routedomain =
   1102     { PF_ROUTE, "route", route_init, 0, 0,
   1103       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
   1104