Home | History | Annotate | Line # | Download | only in net
rtsock.c revision 1.35
      1 /*	$NetBSD: rtsock.c,v 1.35 2000/02/17 04:28:00 itojun 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 __inline void rt_adjustcount __P((int, int));
    107 
    108 /* Sleazy use of local variables throughout file, warning!!!! */
    109 #define dst	info.rti_info[RTAX_DST]
    110 #define gate	info.rti_info[RTAX_GATEWAY]
    111 #define netmask	info.rti_info[RTAX_NETMASK]
    112 #define genmask	info.rti_info[RTAX_GENMASK]
    113 #define ifpaddr	info.rti_info[RTAX_IFP]
    114 #define ifaaddr	info.rti_info[RTAX_IFA]
    115 #define brdaddr	info.rti_info[RTAX_BRD]
    116 
    117 static __inline void
    118 rt_adjustcount(af, cnt)
    119 	int af, cnt;
    120 {
    121 	route_cb.any_count += cnt;
    122 	switch (af) {
    123 	case AF_INET:
    124 		route_cb.ip_count += cnt;
    125 		return;
    126 #ifdef INET6
    127 	case AF_INET6:
    128 		route_cb.ip6_count += cnt;
    129 		return;
    130 #endif
    131 	case AF_IPX:
    132 		route_cb.ipx_count += cnt;
    133 		return;
    134 	case AF_NS:
    135 		route_cb.ns_count += cnt;
    136 		return;
    137 	case AF_ISO:
    138 		route_cb.iso_count += cnt;
    139 		return;
    140 	}
    141 }
    142 
    143 /*ARGSUSED*/
    144 int
    145 route_usrreq(so, req, m, nam, control, p)
    146 	register struct socket *so;
    147 	int req;
    148 	struct mbuf *m, *nam, *control;
    149 	struct proc *p;
    150 {
    151 	register int error = 0;
    152 	register struct rawcb *rp = sotorawcb(so);
    153 	int s;
    154 
    155 	if (req == PRU_ATTACH) {
    156 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
    157 		if ((so->so_pcb = rp) != NULL)
    158 			bzero(so->so_pcb, sizeof(*rp));
    159 
    160 	}
    161 	if (req == PRU_DETACH && rp)
    162 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
    163 	s = splsoftnet();
    164 
    165 	/*
    166 	 * Don't call raw_usrreq() in the attach case, because
    167 	 * we want to allow non-privileged processes to listen on
    168 	 * and send "safe" commands to the routing socket.
    169 	 */
    170 	if (req == PRU_ATTACH) {
    171 		if (p == 0)
    172 			error = EACCES;
    173 		else
    174 			error = raw_attach(so, (int)(long)nam);
    175 	} else
    176 		error = raw_usrreq(so, req, m, nam, control, p);
    177 
    178 	rp = sotorawcb(so);
    179 	if (req == PRU_ATTACH && rp) {
    180 		if (error) {
    181 			free((caddr_t)rp, M_PCB);
    182 			splx(s);
    183 			return (error);
    184 		}
    185 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
    186 		rp->rcb_laddr = &route_src;
    187 		rp->rcb_faddr = &route_dst;
    188 		soisconnected(so);
    189 		so->so_options |= SO_USELOOPBACK;
    190 	}
    191 	splx(s);
    192 	return (error);
    193 }
    194 
    195 /*ARGSUSED*/
    196 int
    197 #if __STDC__
    198 route_output(struct mbuf *m, ...)
    199 #else
    200 route_output(m, va_alist)
    201 	struct mbuf *m;
    202 	va_dcl
    203 #endif
    204 {
    205 	register struct rt_msghdr *rtm = 0;
    206 	register struct rtentry *rt = 0;
    207 	struct rtentry *saved_nrt = 0;
    208 	struct radix_node_head *rnh;
    209 	struct rt_addrinfo info;
    210 	int len, error = 0;
    211 	struct ifnet *ifp = 0;
    212 	struct ifaddr *ifa = 0;
    213 	struct socket *so;
    214 	va_list ap;
    215 
    216 	va_start(ap, m);
    217 	so = va_arg(ap, struct socket *);
    218 	va_end(ap);
    219 
    220 
    221 #define senderr(e) { error = e; goto flush;}
    222 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
    223 	   (m = m_pullup(m, sizeof(int32_t))) == 0))
    224 		return (ENOBUFS);
    225 	if ((m->m_flags & M_PKTHDR) == 0)
    226 		panic("route_output");
    227 	len = m->m_pkthdr.len;
    228 	if (len < sizeof(*rtm) ||
    229 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
    230 		dst = 0;
    231 		senderr(EINVAL);
    232 	}
    233 	R_Malloc(rtm, struct rt_msghdr *, len);
    234 	if (rtm == 0) {
    235 		dst = 0;
    236 		senderr(ENOBUFS);
    237 	}
    238 	m_copydata(m, 0, len, (caddr_t)rtm);
    239 	if (rtm->rtm_version != RTM_VERSION) {
    240 		dst = 0;
    241 		senderr(EPROTONOSUPPORT);
    242 	}
    243 	rtm->rtm_pid = curproc->p_pid;
    244 	info.rti_addrs = rtm->rtm_addrs;
    245 	rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info);
    246 	if (dst == 0 || (dst->sa_family >= AF_MAX))
    247 		senderr(EINVAL);
    248 	if (gate != 0 && (gate->sa_family >= AF_MAX))
    249 		senderr(EINVAL);
    250 	if (genmask) {
    251 		struct radix_node *t;
    252 		t = rn_addmask((caddr_t)genmask, 0, 1);
    253 		if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0)
    254 			genmask = (struct sockaddr *)(t->rn_key);
    255 		else
    256 			senderr(ENOBUFS);
    257 	}
    258 
    259 	/*
    260 	 * Verify that the caller has the appropriate privilege; RTM_GET
    261 	 * is the only operation the non-superuser is allowed.
    262 	 */
    263 	if (rtm->rtm_type != RTM_GET &&
    264 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
    265 		senderr(EACCES);
    266 
    267 	switch (rtm->rtm_type) {
    268 
    269 	case RTM_ADD:
    270 		if (gate == 0)
    271 			senderr(EINVAL);
    272 		error = rtrequest(RTM_ADD, dst, gate, netmask,
    273 		    rtm->rtm_flags, &saved_nrt);
    274 		if (error == 0 && saved_nrt) {
    275 			rt_setmetrics(rtm->rtm_inits,
    276 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
    277 			saved_nrt->rt_refcnt--;
    278 			saved_nrt->rt_genmask = genmask;
    279 		}
    280 		break;
    281 
    282 	case RTM_DELETE:
    283 		error = rtrequest(RTM_DELETE, dst, gate, netmask,
    284 		    rtm->rtm_flags, &saved_nrt);
    285 		if (error == 0) {
    286 			(rt = saved_nrt)->rt_refcnt++;
    287 			goto report;
    288 		}
    289 		break;
    290 
    291 	case RTM_GET:
    292 	case RTM_CHANGE:
    293 	case RTM_LOCK:
    294 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
    295 			senderr(EAFNOSUPPORT);
    296 		} else if ((rt = (struct rtentry *)
    297 		    rnh->rnh_lookup(dst, netmask, rnh)) != NULL)
    298 			rt->rt_refcnt++;
    299 		else
    300 			senderr(ESRCH);
    301 		switch(rtm->rtm_type) {
    302 
    303 		case RTM_GET:
    304 		report:
    305 			dst = rt_key(rt);
    306 			gate = rt->rt_gateway;
    307 			netmask = rt_mask(rt);
    308 			genmask = rt->rt_genmask;
    309 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
    310 				if ((ifp = rt->rt_ifp) != NULL) {
    311 					ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
    312 					ifaaddr = rt->rt_ifa->ifa_addr;
    313 					if (ifp->if_flags & IFF_POINTOPOINT)
    314 						brdaddr = rt->rt_ifa->ifa_dstaddr;
    315 					else
    316 						brdaddr = 0;
    317 					rtm->rtm_index = ifp->if_index;
    318 				} else {
    319 					ifpaddr = 0;
    320 					ifaaddr = 0;
    321 			    }
    322 			}
    323 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
    324 			    (struct walkarg *)0, &len);
    325 			if (len > rtm->rtm_msglen) {
    326 				struct rt_msghdr *new_rtm;
    327 				R_Malloc(new_rtm, struct rt_msghdr *, len);
    328 				if (new_rtm == 0)
    329 					senderr(ENOBUFS);
    330 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
    331 				Free(rtm); rtm = new_rtm;
    332 			}
    333 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
    334 			    (struct walkarg *)0, 0);
    335 			rtm->rtm_flags = rt->rt_flags;
    336 			rtm->rtm_rmx = rt->rt_rmx;
    337 			rtm->rtm_addrs = info.rti_addrs;
    338 			break;
    339 
    340 		case RTM_CHANGE:
    341 			if (gate && rt_setgate(rt, rt_key(rt), gate))
    342 				senderr(EDQUOT);
    343 			/* new gateway could require new ifaddr, ifp;
    344 			   flags may also be different; ifp may be specified
    345 			   by ll sockaddr when protocol address is ambiguous */
    346 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
    347 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
    348 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
    349 				    ifp);
    350 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
    351 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
    352 			    rt_key(rt), gate))))
    353 				ifp = ifa->ifa_ifp;
    354 			if (ifa) {
    355 				register struct ifaddr *oifa = rt->rt_ifa;
    356 				if (oifa != ifa) {
    357 				    if (oifa && oifa->ifa_rtrequest)
    358 					oifa->ifa_rtrequest(RTM_DELETE,
    359 					rt, gate);
    360 				    IFAFREE(rt->rt_ifa);
    361 				    rt->rt_ifa = ifa;
    362 				    IFAREF(rt->rt_ifa);
    363 				    rt->rt_ifp = ifp;
    364 				}
    365 			}
    366 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
    367 			    &rt->rt_rmx);
    368 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
    369 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate);
    370 			if (genmask)
    371 				rt->rt_genmask = genmask;
    372 			/*
    373 			 * Fall into
    374 			 */
    375 		case RTM_LOCK:
    376 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
    377 			rt->rt_rmx.rmx_locks |=
    378 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
    379 			break;
    380 		}
    381 		break;
    382 
    383 	default:
    384 		senderr(EOPNOTSUPP);
    385 	}
    386 
    387 flush:
    388 	if (rtm) {
    389 		if (error)
    390 			rtm->rtm_errno = error;
    391 		else
    392 			rtm->rtm_flags |= RTF_DONE;
    393 	}
    394 	if (rt)
    395 		rtfree(rt);
    396     {
    397 	register struct rawcb *rp = 0;
    398 	/*
    399 	 * Check to see if we don't want our own messages.
    400 	 */
    401 	if ((so->so_options & SO_USELOOPBACK) == 0) {
    402 		if (route_cb.any_count <= 1) {
    403 			if (rtm)
    404 				Free(rtm);
    405 			m_freem(m);
    406 			return (error);
    407 		}
    408 		/* There is another listener, so construct message */
    409 		rp = sotorawcb(so);
    410 	}
    411 	if (rtm) {
    412 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
    413 		Free(rtm);
    414 	}
    415 	if (rp)
    416 		rp->rcb_proto.sp_family = 0; /* Avoid us */
    417 	if (dst)
    418 		route_proto.sp_protocol = dst->sa_family;
    419 	raw_input(m, &route_proto, &route_src, &route_dst);
    420 	if (rp)
    421 		rp->rcb_proto.sp_family = PF_ROUTE;
    422     }
    423 	return (error);
    424 }
    425 
    426 void
    427 rt_setmetrics(which, in, out)
    428 	u_long which;
    429 	register struct rt_metrics *in, *out;
    430 {
    431 #define metric(f, e) if (which & (f)) out->e = in->e;
    432 	metric(RTV_RPIPE, rmx_recvpipe);
    433 	metric(RTV_SPIPE, rmx_sendpipe);
    434 	metric(RTV_SSTHRESH, rmx_ssthresh);
    435 	metric(RTV_RTT, rmx_rtt);
    436 	metric(RTV_RTTVAR, rmx_rttvar);
    437 	metric(RTV_HOPCOUNT, rmx_hopcount);
    438 	metric(RTV_MTU, rmx_mtu);
    439 	metric(RTV_EXPIRE, rmx_expire);
    440 #undef metric
    441 }
    442 
    443 #define ROUNDUP(a) \
    444 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
    445 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
    446 
    447 static void
    448 rt_xaddrs(cp, cplim, rtinfo)
    449 	register caddr_t cp, cplim;
    450 	register struct rt_addrinfo *rtinfo;
    451 {
    452 	register struct sockaddr *sa;
    453 	register int i;
    454 
    455 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
    456 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
    457 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
    458 			continue;
    459 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
    460 		ADVANCE(cp, sa);
    461 	}
    462 }
    463 
    464 static struct mbuf *
    465 rt_msg1(type, rtinfo, data, datalen)
    466 	int type;
    467 	register struct rt_addrinfo *rtinfo;
    468 	caddr_t data;
    469 	int datalen;
    470 {
    471 	register struct rt_msghdr *rtm;
    472 	register struct mbuf *m;
    473 	register int i;
    474 	register struct sockaddr *sa;
    475 	int len, dlen;
    476 
    477 	m = m_gethdr(M_DONTWAIT, MT_DATA);
    478 	if (m == 0)
    479 		return (m);
    480 	switch (type) {
    481 
    482 	case RTM_DELADDR:
    483 	case RTM_NEWADDR:
    484 		len = sizeof(struct ifa_msghdr);
    485 		break;
    486 
    487 #ifdef COMPAT_14
    488 	case RTM_OIFINFO:
    489 		len = sizeof(struct if_msghdr14);
    490 		break;
    491 #endif
    492 
    493 	case RTM_IFINFO:
    494 		len = sizeof(struct if_msghdr);
    495 		break;
    496 
    497 	default:
    498 		len = sizeof(struct rt_msghdr);
    499 	}
    500 	if (len > MHLEN + MLEN)
    501 		panic("rt_msg1: message too long");
    502 	else if (len > MHLEN) {
    503 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
    504 		if (m->m_next == NULL) {
    505 			m_freem(m);
    506 			return (NULL);
    507 		}
    508 		m->m_pkthdr.len = len;
    509 		m->m_len = MHLEN;
    510 		m->m_next->m_len = len - MHLEN;
    511 	} else {
    512 		m->m_pkthdr.len = m->m_len = len;
    513 	}
    514 	m->m_pkthdr.rcvif = 0;
    515 	m_copyback(m, 0, datalen, data);
    516 	rtm = mtod(m, struct rt_msghdr *);
    517 	for (i = 0; i < RTAX_MAX; i++) {
    518 		if ((sa = rtinfo->rti_info[i]) == NULL)
    519 			continue;
    520 		rtinfo->rti_addrs |= (1 << i);
    521 		dlen = ROUNDUP(sa->sa_len);
    522 		m_copyback(m, len, dlen, (caddr_t)sa);
    523 		len += dlen;
    524 	}
    525 	rtm->rtm_msglen = len;
    526 	rtm->rtm_version = RTM_VERSION;
    527 	rtm->rtm_type = type;
    528 	return (m);
    529 }
    530 
    531 /*
    532  * rt_msg2
    533  *
    534  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
    535  *		returns the length of the message in 'lenp'.
    536  *
    537  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
    538  *	the message
    539  * otherwise walkarg's w_needed is updated and if the user buffer is
    540  *	specified and w_needed indicates space exists the information is copied
    541  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
    542  *	if the allocation fails ENOBUFS is returned.
    543  */
    544 static int
    545 rt_msg2(type, rtinfo, cp, w, lenp)
    546 	int type;
    547 	register struct rt_addrinfo *rtinfo;
    548 	caddr_t cp;
    549 	struct walkarg *w;
    550 	int *lenp;
    551 {
    552 	register int i;
    553 	int len, dlen, second_time = 0;
    554 	caddr_t cp0;
    555 
    556 	rtinfo->rti_addrs = 0;
    557 again:
    558 	switch (type) {
    559 
    560 	case RTM_DELADDR:
    561 	case RTM_NEWADDR:
    562 		len = sizeof(struct ifa_msghdr);
    563 		break;
    564 #ifdef COMPAT_14
    565 	case RTM_OIFINFO:
    566 		len = sizeof(struct if_msghdr14);
    567 		break;
    568 #endif
    569 
    570 	case RTM_IFINFO:
    571 		len = sizeof(struct if_msghdr);
    572 		break;
    573 
    574 	default:
    575 		len = sizeof(struct rt_msghdr);
    576 	}
    577 	if ((cp0 = cp) != NULL)
    578 		cp += len;
    579 	for (i = 0; i < RTAX_MAX; i++) {
    580 		register struct sockaddr *sa;
    581 
    582 		if ((sa = rtinfo->rti_info[i]) == 0)
    583 			continue;
    584 		rtinfo->rti_addrs |= (1 << i);
    585 		dlen = ROUNDUP(sa->sa_len);
    586 		if (cp) {
    587 			bcopy(sa, cp, (unsigned)dlen);
    588 			cp += dlen;
    589 		}
    590 		len += dlen;
    591 	}
    592 	if (cp == 0 && w != NULL && !second_time) {
    593 		register struct walkarg *rw = w;
    594 
    595 		rw->w_needed += len;
    596 		if (rw->w_needed <= 0 && rw->w_where) {
    597 			if (rw->w_tmemsize < len) {
    598 				if (rw->w_tmem)
    599 					free(rw->w_tmem, M_RTABLE);
    600 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
    601 				    M_NOWAIT);
    602 				if (rw->w_tmem)
    603 					rw->w_tmemsize = len;
    604 			}
    605 			if (rw->w_tmem) {
    606 				cp = rw->w_tmem;
    607 				second_time = 1;
    608 				goto again;
    609 			} else {
    610 				rw->w_tmemneeded = len;
    611 				return (ENOBUFS);
    612 			}
    613 		}
    614 	}
    615 	if (cp) {
    616 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
    617 
    618 		rtm->rtm_version = RTM_VERSION;
    619 		rtm->rtm_type = type;
    620 		rtm->rtm_msglen = len;
    621 	}
    622 	if (lenp)
    623 		*lenp = len;
    624 	return (0);
    625 }
    626 
    627 /*
    628  * This routine is called to generate a message from the routing
    629  * socket indicating that a redirect has occured, a routing lookup
    630  * has failed, or that a protocol has detected timeouts to a particular
    631  * destination.
    632  */
    633 void
    634 rt_missmsg(type, rtinfo, flags, error)
    635 	int type, flags, error;
    636 	register struct rt_addrinfo *rtinfo;
    637 {
    638 	struct rt_msghdr rtm;
    639 	register struct mbuf *m;
    640 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
    641 
    642 	if (route_cb.any_count == 0)
    643 		return;
    644 	bzero(&rtm, sizeof(rtm));
    645 	rtm.rtm_flags = RTF_DONE | flags;
    646 	rtm.rtm_errno = error;
    647 	m = rt_msg1(type, rtinfo, (caddr_t)&rtm, sizeof(rtm));
    648 	if (m == 0)
    649 		return;
    650 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
    651 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
    652 	raw_input(m, &route_proto, &route_src, &route_dst);
    653 }
    654 
    655 /*
    656  * This routine is called to generate a message from the routing
    657  * socket indicating that the status of a network interface has changed.
    658  */
    659 void
    660 rt_ifmsg(ifp)
    661 	register struct ifnet *ifp;
    662 {
    663 	struct if_msghdr ifm;
    664 #ifdef COMPAT_14
    665 	struct if_msghdr14 oifm;
    666 #endif
    667 	struct mbuf *m;
    668 	struct rt_addrinfo info;
    669 
    670 	if (route_cb.any_count == 0)
    671 		return;
    672 	bzero(&info, sizeof(info));
    673 	bzero(&ifm, sizeof(ifm));
    674 	ifm.ifm_index = ifp->if_index;
    675 	ifm.ifm_flags = ifp->if_flags;
    676 	ifm.ifm_data = ifp->if_data;
    677 	ifm.ifm_addrs = 0;
    678 	m = rt_msg1(RTM_IFINFO, &info, (caddr_t)&ifm, sizeof(ifm));
    679 	if (m == 0)
    680 		return;
    681 	route_proto.sp_protocol = 0;
    682 	raw_input(m, &route_proto, &route_src, &route_dst);
    683 #ifdef COMPAT_14
    684 	bzero(&info, sizeof(info));
    685 	bzero(&oifm, sizeof(ifm));
    686 	oifm.ifm_index = ifp->if_index;
    687 	oifm.ifm_flags = ifp->if_flags;
    688 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
    689 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
    690 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
    691 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
    692 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
    693 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
    694 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
    695 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
    696 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
    697 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
    698 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
    699 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
    700 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
    701 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
    702 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
    703 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
    704 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
    705 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
    706 	oifm.ifm_addrs = 0;
    707 	m = rt_msg1(RTM_OIFINFO, &info, (caddr_t)&oifm, sizeof(oifm));
    708 	if (m == 0)
    709 		return;
    710 	route_proto.sp_protocol = 0;
    711 	raw_input(m, &route_proto, &route_src, &route_dst);
    712 #endif
    713 }
    714 
    715 /*
    716  * This is called to generate messages from the routing socket
    717  * indicating a network interface has had addresses associated with it.
    718  * if we ever reverse the logic and replace messages TO the routing
    719  * socket indicate a request to configure interfaces, then it will
    720  * be unnecessary as the routing socket will automatically generate
    721  * copies of it.
    722  */
    723 void
    724 rt_newaddrmsg(cmd, ifa, error, rt)
    725 	int cmd, error;
    726 	register struct ifaddr *ifa;
    727 	register struct rtentry *rt;
    728 {
    729 	struct rt_addrinfo info;
    730 	struct sockaddr *sa = NULL;
    731 	int pass;
    732 	struct mbuf *m = NULL;
    733 	struct ifnet *ifp = ifa->ifa_ifp;
    734 
    735 	if (route_cb.any_count == 0)
    736 		return;
    737 	for (pass = 1; pass < 3; pass++) {
    738 		bzero(&info, sizeof(info));
    739 		if ((cmd == RTM_ADD && pass == 1) ||
    740 		    (cmd == RTM_DELETE && pass == 2)) {
    741 			struct ifa_msghdr ifam;
    742 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
    743 
    744 			ifaaddr = sa = ifa->ifa_addr;
    745 			ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
    746 			netmask = ifa->ifa_netmask;
    747 			brdaddr = ifa->ifa_dstaddr;
    748 			bzero(&ifam, sizeof(ifam));
    749 			ifam.ifam_index = ifp->if_index;
    750 			ifam.ifam_metric = ifa->ifa_metric;
    751 			ifam.ifam_flags = ifa->ifa_flags;
    752 			m = rt_msg1(ncmd, &info, (caddr_t)&ifam, sizeof(ifam));
    753 			if (m == NULL)
    754 				continue;
    755 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
    756 			    info.rti_addrs;
    757 		}
    758 		if ((cmd == RTM_ADD && pass == 2) ||
    759 		    (cmd == RTM_DELETE && pass == 1)) {
    760 			struct rt_msghdr rtm;
    761 
    762 			if (rt == 0)
    763 				continue;
    764 			netmask = rt_mask(rt);
    765 			dst = sa = rt_key(rt);
    766 			gate = rt->rt_gateway;
    767 			bzero(&rtm, sizeof(rtm));
    768 			rtm.rtm_index = ifp->if_index;
    769 			rtm.rtm_flags |= rt->rt_flags;
    770 			rtm.rtm_errno = error;
    771 			m = rt_msg1(cmd, &info, (caddr_t)&rtm, sizeof(rtm));
    772 			if (m == NULL)
    773 				continue;
    774 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
    775 		}
    776 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
    777 		raw_input(m, &route_proto, &route_src, &route_dst);
    778 	}
    779 }
    780 
    781 /*
    782  * This is used in dumping the kernel table via sysctl().
    783  */
    784 int
    785 sysctl_dumpentry(rn, v)
    786 	struct radix_node *rn;
    787 	register void *v;
    788 {
    789 	register struct walkarg *w = v;
    790 	register struct rtentry *rt = (struct rtentry *)rn;
    791 	int error = 0, size;
    792 	struct rt_addrinfo info;
    793 
    794 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
    795 		return 0;
    796 	bzero(&info, sizeof(info));
    797 	dst = rt_key(rt);
    798 	gate = rt->rt_gateway;
    799 	netmask = rt_mask(rt);
    800 	genmask = rt->rt_genmask;
    801 	if (rt->rt_ifp) {
    802 		ifpaddr = rt->rt_ifp->if_addrlist.tqh_first->ifa_addr;
    803 		ifaaddr = rt->rt_ifa->ifa_addr;
    804 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
    805 			brdaddr = rt->rt_ifa->ifa_dstaddr;
    806 	}
    807 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
    808 		return (error);
    809 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    810 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
    811 
    812 		rtm->rtm_flags = rt->rt_flags;
    813 		rtm->rtm_use = rt->rt_use;
    814 		rtm->rtm_rmx = rt->rt_rmx;
    815 		rtm->rtm_index = rt->rt_ifp->if_index;
    816 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
    817 		rtm->rtm_addrs = info.rti_addrs;
    818 		if ((error = copyout(rtm, w->w_where, size)) != 0)
    819 			w->w_where = NULL;
    820 		else
    821 			w->w_where += size;
    822 	}
    823 	return (error);
    824 }
    825 
    826 int
    827 sysctl_iflist(af, w, type)
    828 	int	af;
    829 	register struct	walkarg *w;
    830 	int type;
    831 {
    832 	register struct ifnet *ifp;
    833 	register struct ifaddr *ifa;
    834 	struct	rt_addrinfo info;
    835 	int	len, error = 0;
    836 
    837 	bzero(&info, sizeof(info));
    838 	for (ifp = ifnet.tqh_first; ifp != 0; ifp = ifp->if_list.tqe_next) {
    839 		if (w->w_arg && w->w_arg != ifp->if_index)
    840 			continue;
    841 		ifa = ifp->if_addrlist.tqh_first;
    842 		ifpaddr = ifa->ifa_addr;
    843 		switch(type) {
    844 		case NET_RT_IFLIST:
    845 			error =
    846 			    rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w, &len);
    847 			break;
    848 #ifdef COMPAT_14
    849 		case NET_RT_OIFLIST:
    850 			error =
    851 			    rt_msg2(RTM_OIFINFO, &info, (caddr_t)0, w, &len);
    852 			break;
    853 #endif
    854 		default:
    855 			panic("sysctl_iflist(1)");
    856 		}
    857 		if (error)
    858 			return (error);
    859 		ifpaddr = 0;
    860 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    861 			switch(type) {
    862 			case NET_RT_IFLIST: {
    863 				register struct if_msghdr *ifm;
    864 
    865 				ifm = (struct if_msghdr *)w->w_tmem;
    866 				ifm->ifm_index = ifp->if_index;
    867 				ifm->ifm_flags = ifp->if_flags;
    868 				ifm->ifm_data = ifp->if_data;
    869 				ifm->ifm_addrs = info.rti_addrs;
    870 				error = copyout(ifm, w->w_where, len);
    871 				if (error)
    872 					return (error);
    873 				w->w_where += len;
    874 				break;
    875 			}
    876 
    877 #ifdef COMPAT_14
    878 			case NET_RT_OIFLIST: {
    879 				register struct if_msghdr14 *ifm;
    880 
    881 				ifm = (struct if_msghdr14 *)w->w_tmem;
    882 				ifm->ifm_index = ifp->if_index;
    883 				ifm->ifm_flags = ifp->if_flags;
    884 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
    885 				ifm->ifm_data.ifi_addrlen =
    886 				    ifp->if_data.ifi_addrlen;
    887 				ifm->ifm_data.ifi_hdrlen =
    888 				    ifp->if_data.ifi_hdrlen;
    889 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
    890 				ifm->ifm_data.ifi_metric =
    891 				    ifp->if_data.ifi_metric;
    892 				ifm->ifm_data.ifi_baudrate =
    893 				    ifp->if_data.ifi_baudrate;
    894 				ifm->ifm_data.ifi_ipackets =
    895 				    ifp->if_data.ifi_ipackets;
    896 				ifm->ifm_data.ifi_ierrors =
    897 				    ifp->if_data.ifi_ierrors;
    898 				ifm->ifm_data.ifi_opackets =
    899 				    ifp->if_data.ifi_opackets;
    900 				ifm->ifm_data.ifi_oerrors =
    901 				    ifp->if_data.ifi_oerrors;
    902 				ifm->ifm_data.ifi_collisions =
    903 				    ifp->if_data.ifi_collisions;
    904 				ifm->ifm_data.ifi_ibytes =
    905 				    ifp->if_data.ifi_ibytes;
    906 				ifm->ifm_data.ifi_obytes =
    907 				    ifp->if_data.ifi_obytes;
    908 				ifm->ifm_data.ifi_imcasts =
    909 				    ifp->if_data.ifi_imcasts;
    910 				ifm->ifm_data.ifi_omcasts =
    911 				    ifp->if_data.ifi_omcasts;
    912 				ifm->ifm_data.ifi_iqdrops =
    913 				    ifp->if_data.ifi_iqdrops;
    914 				ifm->ifm_data.ifi_noproto =
    915 				    ifp->if_data.ifi_noproto;
    916 				ifm->ifm_data.ifi_lastchange =
    917 				    ifp->if_data.ifi_lastchange;
    918 				ifm->ifm_addrs = info.rti_addrs;
    919 				error = copyout(ifm, w->w_where, len);
    920 				if (error)
    921 					return (error);
    922 				w->w_where += len;
    923 				break;
    924 			}
    925 #endif
    926 			default:
    927 				panic("sysctl_iflist(2)");
    928 			}
    929 		}
    930 		while ((ifa = ifa->ifa_list.tqe_next) != NULL) {
    931 			if (af && af != ifa->ifa_addr->sa_family)
    932 				continue;
    933 			ifaaddr = ifa->ifa_addr;
    934 			netmask = ifa->ifa_netmask;
    935 			brdaddr = ifa->ifa_dstaddr;
    936 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
    937 				return (error);
    938 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    939 				register struct ifa_msghdr *ifam;
    940 
    941 				ifam = (struct ifa_msghdr *)w->w_tmem;
    942 				ifam->ifam_index = ifa->ifa_ifp->if_index;
    943 				ifam->ifam_flags = ifa->ifa_flags;
    944 				ifam->ifam_metric = ifa->ifa_metric;
    945 				ifam->ifam_addrs = info.rti_addrs;
    946 				error = copyout(w->w_tmem, w->w_where, len);
    947 				if (error)
    948 					return (error);
    949 				w->w_where += len;
    950 			}
    951 		}
    952 		ifaaddr = netmask = brdaddr = 0;
    953 	}
    954 	return (0);
    955 }
    956 
    957 int
    958 sysctl_rtable(name, namelen, where, given, new, newlen)
    959 	int	*name;
    960 	u_int	namelen;
    961 	void 	*where;
    962 	size_t	*given;
    963 	void	*new;
    964 	size_t	newlen;
    965 {
    966 	register struct radix_node_head *rnh;
    967 	int	i, s, error = EINVAL;
    968 	u_char  af;
    969 	struct	walkarg w;
    970 
    971 	if (new)
    972 		return (EPERM);
    973 	if (namelen != 3)
    974 		return (EINVAL);
    975 	af = name[0];
    976 	w.w_tmemneeded = 0;
    977 	w.w_tmemsize = 0;
    978 	w.w_tmem = NULL;
    979 again:
    980 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
    981 	if (w.w_tmemneeded) {
    982 		w.w_tmem = (caddr_t) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
    983 		w.w_tmemsize = w.w_tmemneeded;
    984 		w.w_tmemneeded = 0;
    985 	}
    986 	w.w_op = name[1];
    987 	w.w_arg = name[2];
    988 	w.w_given = *given;
    989 	w.w_needed = 0 - w.w_given;
    990 	w.w_where = where;
    991 
    992 	s = splsoftnet();
    993 	switch (w.w_op) {
    994 
    995 	case NET_RT_DUMP:
    996 	case NET_RT_FLAGS:
    997 		for (i = 1; i <= AF_MAX; i++)
    998 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
    999 			    (error = (*rnh->rnh_walktree)(rnh,
   1000 			    sysctl_dumpentry, &w)))
   1001 				break;
   1002 		break;
   1003 
   1004 #ifdef COMPAT_14
   1005 	case NET_RT_OIFLIST:
   1006 		error = sysctl_iflist(af, &w, w.w_op);
   1007 		break;
   1008 #endif
   1009 
   1010 	case NET_RT_IFLIST:
   1011 		error = sysctl_iflist(af, &w, w.w_op);
   1012 	}
   1013 	splx(s);
   1014 
   1015 	/* check to see if we couldn't allocate memory with NOWAIT */
   1016 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
   1017 		goto again;
   1018 
   1019 	if (w.w_tmem)
   1020 		free(w.w_tmem, M_RTABLE);
   1021 	w.w_needed += w.w_given;
   1022 	if (where) {
   1023 		*given = w.w_where - (caddr_t) where;
   1024 		if (*given < w.w_needed)
   1025 			return (ENOMEM);
   1026 	} else {
   1027 		*given = (11 * w.w_needed) / 10;
   1028 	}
   1029 	return (error);
   1030 }
   1031 
   1032 /*
   1033  * Definitions of protocols supported in the ROUTE domain.
   1034  */
   1035 
   1036 extern	struct domain routedomain;		/* or at least forward */
   1037 
   1038 struct protosw routesw[] = {
   1039 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
   1040   raw_input,	route_output,	raw_ctlinput,	0,
   1041   route_usrreq,
   1042   raw_init,	0,		0,		0,
   1043   sysctl_rtable,
   1044 }
   1045 };
   1046 
   1047 struct domain routedomain =
   1048     { PF_ROUTE, "route", route_init, 0, 0,
   1049       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
   1050