Home | History | Annotate | Line # | Download | only in net
rtsock.c revision 1.44
      1 /*	$NetBSD: rtsock.c,v 1.44 2000/11/10 03:37:42 enami 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 <uvm/uvm_extern.h>
     79 
     80 #include <sys/sysctl.h>
     81 
     82 #include <net/if.h>
     83 #include <net/route.h>
     84 #include <net/raw_cb.h>
     85 
     86 #include <machine/stdarg.h>
     87 
     88 struct	sockaddr route_dst = { 2, PF_ROUTE, };
     89 struct	sockaddr route_src = { 2, PF_ROUTE, };
     90 struct	sockproto route_proto = { PF_ROUTE, };
     91 
     92 struct walkarg {
     93 	int	w_op;
     94 	int	w_arg;
     95 	int	w_given;
     96 	int	w_needed;
     97 	caddr_t	w_where;
     98 	int	w_tmemsize;
     99 	int	w_tmemneeded;
    100 	caddr_t	w_tmem;
    101 };
    102 
    103 static struct mbuf *rt_msg1 __P((int, struct rt_addrinfo *, caddr_t, int));
    104 static int rt_msg2 __P((int, struct rt_addrinfo *, caddr_t, struct walkarg *,
    105     int *));
    106 static int rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
    107 static int sysctl_dumpentry __P((struct radix_node *, void *));
    108 static int sysctl_iflist __P((int, struct walkarg *, int));
    109 static int sysctl_rtable __P((int *, u_int, void *, size_t *, void *, size_t));
    110 static __inline void rt_adjustcount __P((int, int));
    111 
    112 /* Sleazy use of local variables throughout file, warning!!!! */
    113 #define dst	info.rti_info[RTAX_DST]
    114 #define gate	info.rti_info[RTAX_GATEWAY]
    115 #define netmask	info.rti_info[RTAX_NETMASK]
    116 #define genmask	info.rti_info[RTAX_GENMASK]
    117 #define ifpaddr	info.rti_info[RTAX_IFP]
    118 #define ifaaddr	info.rti_info[RTAX_IFA]
    119 #define brdaddr	info.rti_info[RTAX_BRD]
    120 
    121 static __inline void
    122 rt_adjustcount(af, cnt)
    123 	int af, cnt;
    124 {
    125 	route_cb.any_count += cnt;
    126 	switch (af) {
    127 	case AF_INET:
    128 		route_cb.ip_count += cnt;
    129 		return;
    130 #ifdef INET6
    131 	case AF_INET6:
    132 		route_cb.ip6_count += cnt;
    133 		return;
    134 #endif
    135 	case AF_IPX:
    136 		route_cb.ipx_count += cnt;
    137 		return;
    138 	case AF_NS:
    139 		route_cb.ns_count += cnt;
    140 		return;
    141 	case AF_ISO:
    142 		route_cb.iso_count += cnt;
    143 		return;
    144 	}
    145 }
    146 
    147 /*ARGSUSED*/
    148 int
    149 route_usrreq(so, req, m, nam, control, p)
    150 	struct socket *so;
    151 	int req;
    152 	struct mbuf *m, *nam, *control;
    153 	struct proc *p;
    154 {
    155 	int error = 0;
    156 	struct rawcb *rp = sotorawcb(so);
    157 	int s;
    158 
    159 	if (req == PRU_ATTACH) {
    160 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
    161 		if ((so->so_pcb = rp) != NULL)
    162 			bzero(so->so_pcb, sizeof(*rp));
    163 
    164 	}
    165 	if (req == PRU_DETACH && rp)
    166 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
    167 	s = splsoftnet();
    168 
    169 	/*
    170 	 * Don't call raw_usrreq() in the attach case, because
    171 	 * we want to allow non-privileged processes to listen on
    172 	 * and send "safe" commands to the routing socket.
    173 	 */
    174 	if (req == PRU_ATTACH) {
    175 		if (p == 0)
    176 			error = EACCES;
    177 		else
    178 			error = raw_attach(so, (int)(long)nam);
    179 	} else
    180 		error = raw_usrreq(so, req, m, nam, control, p);
    181 
    182 	rp = sotorawcb(so);
    183 	if (req == PRU_ATTACH && rp) {
    184 		if (error) {
    185 			free((caddr_t)rp, M_PCB);
    186 			splx(s);
    187 			return (error);
    188 		}
    189 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
    190 		rp->rcb_laddr = &route_src;
    191 		rp->rcb_faddr = &route_dst;
    192 		soisconnected(so);
    193 		so->so_options |= SO_USELOOPBACK;
    194 	}
    195 	splx(s);
    196 	return (error);
    197 }
    198 
    199 /*ARGSUSED*/
    200 int
    201 #if __STDC__
    202 route_output(struct mbuf *m, ...)
    203 #else
    204 route_output(m, va_alist)
    205 	struct mbuf *m;
    206 	va_dcl
    207 #endif
    208 {
    209 	struct rt_msghdr *rtm = 0;
    210 	struct radix_node *rn = 0;
    211 	struct rtentry *rt = 0;
    212 	struct rtentry *saved_nrt = 0;
    213 	struct radix_node_head *rnh;
    214 	struct rt_addrinfo info;
    215 	int len, error = 0;
    216 	struct ifnet *ifp = 0;
    217 	struct ifaddr *ifa = 0;
    218 	struct socket *so;
    219 	va_list ap;
    220 
    221 	va_start(ap, m);
    222 	so = va_arg(ap, struct socket *);
    223 	va_end(ap);
    224 
    225 	bzero(&info, sizeof(info));
    226 #define senderr(e) do { error = e; goto flush;} while (0)
    227 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
    228 	   (m = m_pullup(m, sizeof(int32_t))) == 0))
    229 		return (ENOBUFS);
    230 	if ((m->m_flags & M_PKTHDR) == 0)
    231 		panic("route_output");
    232 	len = m->m_pkthdr.len;
    233 	if (len < sizeof(*rtm) ||
    234 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
    235 		dst = 0;
    236 		senderr(EINVAL);
    237 	}
    238 	R_Malloc(rtm, struct rt_msghdr *, len);
    239 	if (rtm == 0) {
    240 		dst = 0;
    241 		senderr(ENOBUFS);
    242 	}
    243 	m_copydata(m, 0, len, (caddr_t)rtm);
    244 	if (rtm->rtm_version != RTM_VERSION) {
    245 		dst = 0;
    246 		senderr(EPROTONOSUPPORT);
    247 	}
    248 	rtm->rtm_pid = curproc->p_pid;
    249 	info.rti_addrs = rtm->rtm_addrs;
    250 	if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info))
    251 		senderr(EINVAL);
    252 	if (dst == 0 || (dst->sa_family >= AF_MAX))
    253 		senderr(EINVAL);
    254 	if (gate != 0 && (gate->sa_family >= AF_MAX))
    255 		senderr(EINVAL);
    256 	if (genmask) {
    257 		struct radix_node *t;
    258 		t = rn_addmask((caddr_t)genmask, 0, 1);
    259 		if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0)
    260 			genmask = (struct sockaddr *)(t->rn_key);
    261 		else
    262 			senderr(ENOBUFS);
    263 	}
    264 
    265 	/*
    266 	 * Verify that the caller has the appropriate privilege; RTM_GET
    267 	 * is the only operation the non-superuser is allowed.
    268 	 */
    269 	if (rtm->rtm_type != RTM_GET &&
    270 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
    271 		senderr(EACCES);
    272 
    273 	switch (rtm->rtm_type) {
    274 
    275 	case RTM_ADD:
    276 		if (gate == 0)
    277 			senderr(EINVAL);
    278 		error = rtrequest(RTM_ADD, dst, gate, netmask,
    279 		    rtm->rtm_flags, &saved_nrt);
    280 		if (error == 0 && saved_nrt) {
    281 			rt_setmetrics(rtm->rtm_inits,
    282 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
    283 			saved_nrt->rt_refcnt--;
    284 			saved_nrt->rt_genmask = genmask;
    285 		}
    286 		break;
    287 
    288 	case RTM_DELETE:
    289 		error = rtrequest(RTM_DELETE, dst, gate, netmask,
    290 		    rtm->rtm_flags, &saved_nrt);
    291 		if (error == 0) {
    292 			(rt = saved_nrt)->rt_refcnt++;
    293 			goto report;
    294 		}
    295 		break;
    296 
    297 	case RTM_GET:
    298 	case RTM_CHANGE:
    299 	case RTM_LOCK:
    300 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
    301 			senderr(EAFNOSUPPORT);
    302 		}
    303 		rn = rnh->rnh_lookup(dst, netmask, rnh);
    304 		if (rn == NULL || (rn->rn_flags & RNF_ROOT) != 0) {
    305 			senderr(ESRCH);
    306 		}
    307 		rt = (struct rtentry *)rn;
    308 		rt->rt_refcnt++;
    309 
    310 		switch(rtm->rtm_type) {
    311 
    312 		case RTM_GET:
    313 		report:
    314 			dst = rt_key(rt);
    315 			gate = rt->rt_gateway;
    316 			netmask = rt_mask(rt);
    317 			genmask = rt->rt_genmask;
    318 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
    319 				if ((ifp = rt->rt_ifp) != NULL) {
    320 					ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
    321 					ifaaddr = rt->rt_ifa->ifa_addr;
    322 					if (ifp->if_flags & IFF_POINTOPOINT)
    323 						brdaddr = rt->rt_ifa->ifa_dstaddr;
    324 					else
    325 						brdaddr = 0;
    326 					rtm->rtm_index = ifp->if_index;
    327 				} else {
    328 					ifpaddr = 0;
    329 					ifaaddr = 0;
    330 				}
    331 			}
    332 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
    333 			    (struct walkarg *)0, &len);
    334 			if (len > rtm->rtm_msglen) {
    335 				struct rt_msghdr *new_rtm;
    336 				R_Malloc(new_rtm, struct rt_msghdr *, len);
    337 				if (new_rtm == 0)
    338 					senderr(ENOBUFS);
    339 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
    340 				Free(rtm); rtm = new_rtm;
    341 			}
    342 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
    343 			    (struct walkarg *)0, 0);
    344 			rtm->rtm_flags = rt->rt_flags;
    345 			rtm->rtm_rmx = rt->rt_rmx;
    346 			rtm->rtm_addrs = info.rti_addrs;
    347 			break;
    348 
    349 		case RTM_CHANGE:
    350 			if (gate && rt_setgate(rt, rt_key(rt), gate))
    351 				senderr(EDQUOT);
    352 			/* new gateway could require new ifaddr, ifp;
    353 			   flags may also be different; ifp may be specified
    354 			   by ll sockaddr when protocol address is ambiguous */
    355 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
    356 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
    357 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
    358 				    ifp);
    359 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
    360 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
    361 			    rt_key(rt), gate))))
    362 				ifp = ifa->ifa_ifp;
    363 			if (ifa) {
    364 				struct ifaddr *oifa = rt->rt_ifa;
    365 				if (oifa != ifa) {
    366 				    if (oifa && oifa->ifa_rtrequest)
    367 					oifa->ifa_rtrequest(RTM_DELETE,
    368 					rt, gate);
    369 				    IFAFREE(rt->rt_ifa);
    370 				    rt->rt_ifa = ifa;
    371 				    IFAREF(rt->rt_ifa);
    372 				    rt->rt_ifp = ifp;
    373 				}
    374 			}
    375 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
    376 			    &rt->rt_rmx);
    377 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
    378 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate);
    379 			if (genmask)
    380 				rt->rt_genmask = genmask;
    381 			/*
    382 			 * Fall into
    383 			 */
    384 		case RTM_LOCK:
    385 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
    386 			rt->rt_rmx.rmx_locks |=
    387 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
    388 			break;
    389 		}
    390 		break;
    391 
    392 	default:
    393 		senderr(EOPNOTSUPP);
    394 	}
    395 
    396 flush:
    397 	if (rtm) {
    398 		if (error)
    399 			rtm->rtm_errno = error;
    400 		else
    401 			rtm->rtm_flags |= RTF_DONE;
    402 	}
    403 	if (rt)
    404 		rtfree(rt);
    405     {
    406 	struct rawcb *rp = 0;
    407 	/*
    408 	 * Check to see if we don't want our own messages.
    409 	 */
    410 	if ((so->so_options & SO_USELOOPBACK) == 0) {
    411 		if (route_cb.any_count <= 1) {
    412 			if (rtm)
    413 				Free(rtm);
    414 			m_freem(m);
    415 			return (error);
    416 		}
    417 		/* There is another listener, so construct message */
    418 		rp = sotorawcb(so);
    419 	}
    420 	if (rtm) {
    421 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
    422 		Free(rtm);
    423 	}
    424 	if (rp)
    425 		rp->rcb_proto.sp_family = 0; /* Avoid us */
    426 	if (dst)
    427 		route_proto.sp_protocol = dst->sa_family;
    428 	raw_input(m, &route_proto, &route_src, &route_dst);
    429 	if (rp)
    430 		rp->rcb_proto.sp_family = PF_ROUTE;
    431     }
    432 	return (error);
    433 }
    434 
    435 void
    436 rt_setmetrics(which, in, out)
    437 	u_long which;
    438 	struct rt_metrics *in, *out;
    439 {
    440 #define metric(f, e) if (which & (f)) out->e = in->e;
    441 	metric(RTV_RPIPE, rmx_recvpipe);
    442 	metric(RTV_SPIPE, rmx_sendpipe);
    443 	metric(RTV_SSTHRESH, rmx_ssthresh);
    444 	metric(RTV_RTT, rmx_rtt);
    445 	metric(RTV_RTTVAR, rmx_rttvar);
    446 	metric(RTV_HOPCOUNT, rmx_hopcount);
    447 	metric(RTV_MTU, rmx_mtu);
    448 	metric(RTV_EXPIRE, rmx_expire);
    449 #undef metric
    450 }
    451 
    452 #define ROUNDUP(a) \
    453 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
    454 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
    455 
    456 static int
    457 rt_xaddrs(cp, cplim, rtinfo)
    458 	caddr_t cp, cplim;
    459 	struct rt_addrinfo *rtinfo;
    460 {
    461 	struct sockaddr *sa;
    462 	int i;
    463 
    464 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
    465 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
    466 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
    467 			continue;
    468 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
    469 		ADVANCE(cp, sa);
    470 	}
    471 
    472 	/* Check for extra addresses specified.  */
    473 	if ((rtinfo->rti_addrs & (~0 << i)) != 0)
    474 		return (1);
    475 	/* Check for bad data length.  */
    476 	if (cp != cplim) {
    477 		if (i == RTAX_NETMASK + 1 &&
    478 		    cp - ROUNDUP(sa->sa_len) + sa->sa_len == cplim)
    479 			/*
    480 			 * The last sockaddr was netmask.
    481 			 * We accept this for now for the sake of old
    482 			 * binaries or third party softwares.
    483 			 */
    484 			;
    485 		else
    486 			return (1);
    487 	}
    488 	return (0);
    489 }
    490 
    491 static struct mbuf *
    492 rt_msg1(type, rtinfo, data, datalen)
    493 	int type;
    494 	struct rt_addrinfo *rtinfo;
    495 	caddr_t data;
    496 	int datalen;
    497 {
    498 	struct rt_msghdr *rtm;
    499 	struct mbuf *m;
    500 	int i;
    501 	struct sockaddr *sa;
    502 	int len, dlen;
    503 
    504 	m = m_gethdr(M_DONTWAIT, MT_DATA);
    505 	if (m == 0)
    506 		return (m);
    507 	switch (type) {
    508 
    509 	case RTM_DELADDR:
    510 	case RTM_NEWADDR:
    511 		len = sizeof(struct ifa_msghdr);
    512 		break;
    513 
    514 #ifdef COMPAT_14
    515 	case RTM_OIFINFO:
    516 		len = sizeof(struct if_msghdr14);
    517 		break;
    518 #endif
    519 
    520 	case RTM_IFINFO:
    521 		len = sizeof(struct if_msghdr);
    522 		break;
    523 
    524 	case RTM_IFANNOUNCE:
    525 		len = sizeof(struct if_announcemsghdr);
    526 		break;
    527 
    528 	default:
    529 		len = sizeof(struct rt_msghdr);
    530 	}
    531 	if (len > MHLEN + MLEN)
    532 		panic("rt_msg1: message too long");
    533 	else if (len > MHLEN) {
    534 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
    535 		if (m->m_next == NULL) {
    536 			m_freem(m);
    537 			return (NULL);
    538 		}
    539 		m->m_pkthdr.len = len;
    540 		m->m_len = MHLEN;
    541 		m->m_next->m_len = len - MHLEN;
    542 	} else {
    543 		m->m_pkthdr.len = m->m_len = len;
    544 	}
    545 	m->m_pkthdr.rcvif = 0;
    546 	m_copyback(m, 0, datalen, data);
    547 	rtm = mtod(m, struct rt_msghdr *);
    548 	for (i = 0; i < RTAX_MAX; i++) {
    549 		if ((sa = rtinfo->rti_info[i]) == NULL)
    550 			continue;
    551 		rtinfo->rti_addrs |= (1 << i);
    552 		dlen = ROUNDUP(sa->sa_len);
    553 		m_copyback(m, len, dlen, (caddr_t)sa);
    554 		len += dlen;
    555 	}
    556 	rtm->rtm_msglen = len;
    557 	rtm->rtm_version = RTM_VERSION;
    558 	rtm->rtm_type = type;
    559 	return (m);
    560 }
    561 
    562 /*
    563  * rt_msg2
    564  *
    565  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
    566  *		returns the length of the message in 'lenp'.
    567  *
    568  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
    569  *	the message
    570  * otherwise walkarg's w_needed is updated and if the user buffer is
    571  *	specified and w_needed indicates space exists the information is copied
    572  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
    573  *	if the allocation fails ENOBUFS is returned.
    574  */
    575 static int
    576 rt_msg2(type, rtinfo, cp, w, lenp)
    577 	int type;
    578 	struct rt_addrinfo *rtinfo;
    579 	caddr_t cp;
    580 	struct walkarg *w;
    581 	int *lenp;
    582 {
    583 	int i;
    584 	int len, dlen, second_time = 0;
    585 	caddr_t cp0;
    586 
    587 	rtinfo->rti_addrs = 0;
    588 again:
    589 	switch (type) {
    590 
    591 	case RTM_DELADDR:
    592 	case RTM_NEWADDR:
    593 		len = sizeof(struct ifa_msghdr);
    594 		break;
    595 #ifdef COMPAT_14
    596 	case RTM_OIFINFO:
    597 		len = sizeof(struct if_msghdr14);
    598 		break;
    599 #endif
    600 
    601 	case RTM_IFINFO:
    602 		len = sizeof(struct if_msghdr);
    603 		break;
    604 
    605 	default:
    606 		len = sizeof(struct rt_msghdr);
    607 	}
    608 	if ((cp0 = cp) != NULL)
    609 		cp += len;
    610 	for (i = 0; i < RTAX_MAX; i++) {
    611 		struct sockaddr *sa;
    612 
    613 		if ((sa = rtinfo->rti_info[i]) == 0)
    614 			continue;
    615 		rtinfo->rti_addrs |= (1 << i);
    616 		dlen = ROUNDUP(sa->sa_len);
    617 		if (cp) {
    618 			bcopy(sa, cp, (unsigned)dlen);
    619 			cp += dlen;
    620 		}
    621 		len += dlen;
    622 	}
    623 	if (cp == 0 && w != NULL && !second_time) {
    624 		struct walkarg *rw = w;
    625 
    626 		rw->w_needed += len;
    627 		if (rw->w_needed <= 0 && rw->w_where) {
    628 			if (rw->w_tmemsize < len) {
    629 				if (rw->w_tmem)
    630 					free(rw->w_tmem, M_RTABLE);
    631 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
    632 				    M_NOWAIT);
    633 				if (rw->w_tmem)
    634 					rw->w_tmemsize = len;
    635 			}
    636 			if (rw->w_tmem) {
    637 				cp = rw->w_tmem;
    638 				second_time = 1;
    639 				goto again;
    640 			} else {
    641 				rw->w_tmemneeded = len;
    642 				return (ENOBUFS);
    643 			}
    644 		}
    645 	}
    646 	if (cp) {
    647 		struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
    648 
    649 		rtm->rtm_version = RTM_VERSION;
    650 		rtm->rtm_type = type;
    651 		rtm->rtm_msglen = len;
    652 	}
    653 	if (lenp)
    654 		*lenp = len;
    655 	return (0);
    656 }
    657 
    658 /*
    659  * This routine is called to generate a message from the routing
    660  * socket indicating that a redirect has occured, a routing lookup
    661  * has failed, or that a protocol has detected timeouts to a particular
    662  * destination.
    663  */
    664 void
    665 rt_missmsg(type, rtinfo, flags, error)
    666 	int type, flags, error;
    667 	struct rt_addrinfo *rtinfo;
    668 {
    669 	struct rt_msghdr rtm;
    670 	struct mbuf *m;
    671 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
    672 
    673 	if (route_cb.any_count == 0)
    674 		return;
    675 	bzero(&rtm, sizeof(rtm));
    676 	rtm.rtm_flags = RTF_DONE | flags;
    677 	rtm.rtm_errno = error;
    678 	m = rt_msg1(type, rtinfo, (caddr_t)&rtm, sizeof(rtm));
    679 	if (m == 0)
    680 		return;
    681 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
    682 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
    683 	raw_input(m, &route_proto, &route_src, &route_dst);
    684 }
    685 
    686 /*
    687  * This routine is called to generate a message from the routing
    688  * socket indicating that the status of a network interface has changed.
    689  */
    690 void
    691 rt_ifmsg(ifp)
    692 	struct ifnet *ifp;
    693 {
    694 	struct if_msghdr ifm;
    695 #ifdef COMPAT_14
    696 	struct if_msghdr14 oifm;
    697 #endif
    698 	struct mbuf *m;
    699 	struct rt_addrinfo info;
    700 
    701 	if (route_cb.any_count == 0)
    702 		return;
    703 	bzero(&info, sizeof(info));
    704 	bzero(&ifm, sizeof(ifm));
    705 	ifm.ifm_index = ifp->if_index;
    706 	ifm.ifm_flags = ifp->if_flags;
    707 	ifm.ifm_data = ifp->if_data;
    708 	ifm.ifm_addrs = 0;
    709 	m = rt_msg1(RTM_IFINFO, &info, (caddr_t)&ifm, sizeof(ifm));
    710 	if (m == 0)
    711 		return;
    712 	route_proto.sp_protocol = 0;
    713 	raw_input(m, &route_proto, &route_src, &route_dst);
    714 #ifdef COMPAT_14
    715 	bzero(&info, sizeof(info));
    716 	bzero(&oifm, sizeof(oifm));
    717 	oifm.ifm_index = ifp->if_index;
    718 	oifm.ifm_flags = ifp->if_flags;
    719 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
    720 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
    721 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
    722 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
    723 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
    724 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
    725 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
    726 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
    727 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
    728 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
    729 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
    730 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
    731 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
    732 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
    733 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
    734 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
    735 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
    736 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
    737 	oifm.ifm_addrs = 0;
    738 	m = rt_msg1(RTM_OIFINFO, &info, (caddr_t)&oifm, sizeof(oifm));
    739 	if (m == 0)
    740 		return;
    741 	route_proto.sp_protocol = 0;
    742 	raw_input(m, &route_proto, &route_src, &route_dst);
    743 #endif
    744 }
    745 
    746 /*
    747  * This is called to generate messages from the routing socket
    748  * indicating a network interface has had addresses associated with it.
    749  * if we ever reverse the logic and replace messages TO the routing
    750  * socket indicate a request to configure interfaces, then it will
    751  * be unnecessary as the routing socket will automatically generate
    752  * copies of it.
    753  */
    754 void
    755 rt_newaddrmsg(cmd, ifa, error, rt)
    756 	int cmd, error;
    757 	struct ifaddr *ifa;
    758 	struct rtentry *rt;
    759 {
    760 	struct rt_addrinfo info;
    761 	struct sockaddr *sa = NULL;
    762 	int pass;
    763 	struct mbuf *m = NULL;
    764 	struct ifnet *ifp = ifa->ifa_ifp;
    765 
    766 	if (route_cb.any_count == 0)
    767 		return;
    768 	for (pass = 1; pass < 3; pass++) {
    769 		bzero(&info, sizeof(info));
    770 		if ((cmd == RTM_ADD && pass == 1) ||
    771 		    (cmd == RTM_DELETE && pass == 2)) {
    772 			struct ifa_msghdr ifam;
    773 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
    774 
    775 			ifaaddr = sa = ifa->ifa_addr;
    776 			ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
    777 			netmask = ifa->ifa_netmask;
    778 			brdaddr = ifa->ifa_dstaddr;
    779 			bzero(&ifam, sizeof(ifam));
    780 			ifam.ifam_index = ifp->if_index;
    781 			ifam.ifam_metric = ifa->ifa_metric;
    782 			ifam.ifam_flags = ifa->ifa_flags;
    783 			m = rt_msg1(ncmd, &info, (caddr_t)&ifam, sizeof(ifam));
    784 			if (m == NULL)
    785 				continue;
    786 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
    787 			    info.rti_addrs;
    788 		}
    789 		if ((cmd == RTM_ADD && pass == 2) ||
    790 		    (cmd == RTM_DELETE && pass == 1)) {
    791 			struct rt_msghdr rtm;
    792 
    793 			if (rt == 0)
    794 				continue;
    795 			netmask = rt_mask(rt);
    796 			dst = sa = rt_key(rt);
    797 			gate = rt->rt_gateway;
    798 			bzero(&rtm, sizeof(rtm));
    799 			rtm.rtm_index = ifp->if_index;
    800 			rtm.rtm_flags |= rt->rt_flags;
    801 			rtm.rtm_errno = error;
    802 			m = rt_msg1(cmd, &info, (caddr_t)&rtm, sizeof(rtm));
    803 			if (m == NULL)
    804 				continue;
    805 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
    806 		}
    807 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
    808 		raw_input(m, &route_proto, &route_src, &route_dst);
    809 	}
    810 }
    811 
    812 /*
    813  * This is called to generate routing socket messages indicating
    814  * network interface arrival and departure.
    815  */
    816 void
    817 rt_ifannouncemsg(ifp, what)
    818 	struct ifnet *ifp;
    819 	int what;
    820 {
    821 	struct if_announcemsghdr ifan;
    822 	struct mbuf *m;
    823 	struct rt_addrinfo info;
    824 
    825 	if (route_cb.any_count == 0)
    826 		return;
    827 	bzero(&info, sizeof(info));
    828 	bzero(&ifan, sizeof(ifan));
    829 	ifan.ifan_index = ifp->if_index;
    830 	strcpy(ifan.ifan_name, ifp->if_xname);
    831 	ifan.ifan_what = what;
    832 	m = rt_msg1(RTM_IFANNOUNCE, &info, (caddr_t)&ifan, sizeof(ifan));
    833 	if (m == 0)
    834 		return;
    835 	route_proto.sp_protocol = 0;
    836 	raw_input(m, &route_proto, &route_src, &route_dst);
    837 }
    838 
    839 /*
    840  * This is used in dumping the kernel table via sysctl().
    841  */
    842 static int
    843 sysctl_dumpentry(rn, v)
    844 	struct radix_node *rn;
    845 	void *v;
    846 {
    847 	struct walkarg *w = v;
    848 	struct rtentry *rt = (struct rtentry *)rn;
    849 	int error = 0, size;
    850 	struct rt_addrinfo info;
    851 
    852 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
    853 		return 0;
    854 	bzero(&info, sizeof(info));
    855 	dst = rt_key(rt);
    856 	gate = rt->rt_gateway;
    857 	netmask = rt_mask(rt);
    858 	genmask = rt->rt_genmask;
    859 	if (rt->rt_ifp) {
    860 		ifpaddr = rt->rt_ifp->if_addrlist.tqh_first->ifa_addr;
    861 		ifaaddr = rt->rt_ifa->ifa_addr;
    862 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
    863 			brdaddr = rt->rt_ifa->ifa_dstaddr;
    864 	}
    865 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
    866 		return (error);
    867 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    868 		struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
    869 
    870 		rtm->rtm_flags = rt->rt_flags;
    871 		rtm->rtm_use = rt->rt_use;
    872 		rtm->rtm_rmx = rt->rt_rmx;
    873 		rtm->rtm_index = rt->rt_ifp->if_index;
    874 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
    875 		rtm->rtm_addrs = info.rti_addrs;
    876 		if ((error = copyout(rtm, w->w_where, size)) != 0)
    877 			w->w_where = NULL;
    878 		else
    879 			w->w_where += size;
    880 	}
    881 	return (error);
    882 }
    883 
    884 static int
    885 sysctl_iflist(af, w, type)
    886 	int	af;
    887 	struct	walkarg *w;
    888 	int type;
    889 {
    890 	struct ifnet *ifp;
    891 	struct ifaddr *ifa;
    892 	struct	rt_addrinfo info;
    893 	int	len, error = 0;
    894 
    895 	bzero(&info, sizeof(info));
    896 	for (ifp = ifnet.tqh_first; ifp != 0; ifp = ifp->if_list.tqe_next) {
    897 		if (w->w_arg && w->w_arg != ifp->if_index)
    898 			continue;
    899 		ifa = ifp->if_addrlist.tqh_first;
    900 		ifpaddr = ifa->ifa_addr;
    901 		switch(type) {
    902 		case NET_RT_IFLIST:
    903 			error =
    904 			    rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w, &len);
    905 			break;
    906 #ifdef COMPAT_14
    907 		case NET_RT_OIFLIST:
    908 			error =
    909 			    rt_msg2(RTM_OIFINFO, &info, (caddr_t)0, w, &len);
    910 			break;
    911 #endif
    912 		default:
    913 			panic("sysctl_iflist(1)");
    914 		}
    915 		if (error)
    916 			return (error);
    917 		ifpaddr = 0;
    918 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    919 			switch(type) {
    920 			case NET_RT_IFLIST: {
    921 				struct if_msghdr *ifm;
    922 
    923 				ifm = (struct if_msghdr *)w->w_tmem;
    924 				ifm->ifm_index = ifp->if_index;
    925 				ifm->ifm_flags = ifp->if_flags;
    926 				ifm->ifm_data = ifp->if_data;
    927 				ifm->ifm_addrs = info.rti_addrs;
    928 				error = copyout(ifm, w->w_where, len);
    929 				if (error)
    930 					return (error);
    931 				w->w_where += len;
    932 				break;
    933 			}
    934 
    935 #ifdef COMPAT_14
    936 			case NET_RT_OIFLIST: {
    937 				struct if_msghdr14 *ifm;
    938 
    939 				ifm = (struct if_msghdr14 *)w->w_tmem;
    940 				ifm->ifm_index = ifp->if_index;
    941 				ifm->ifm_flags = ifp->if_flags;
    942 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
    943 				ifm->ifm_data.ifi_addrlen =
    944 				    ifp->if_data.ifi_addrlen;
    945 				ifm->ifm_data.ifi_hdrlen =
    946 				    ifp->if_data.ifi_hdrlen;
    947 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
    948 				ifm->ifm_data.ifi_metric =
    949 				    ifp->if_data.ifi_metric;
    950 				ifm->ifm_data.ifi_baudrate =
    951 				    ifp->if_data.ifi_baudrate;
    952 				ifm->ifm_data.ifi_ipackets =
    953 				    ifp->if_data.ifi_ipackets;
    954 				ifm->ifm_data.ifi_ierrors =
    955 				    ifp->if_data.ifi_ierrors;
    956 				ifm->ifm_data.ifi_opackets =
    957 				    ifp->if_data.ifi_opackets;
    958 				ifm->ifm_data.ifi_oerrors =
    959 				    ifp->if_data.ifi_oerrors;
    960 				ifm->ifm_data.ifi_collisions =
    961 				    ifp->if_data.ifi_collisions;
    962 				ifm->ifm_data.ifi_ibytes =
    963 				    ifp->if_data.ifi_ibytes;
    964 				ifm->ifm_data.ifi_obytes =
    965 				    ifp->if_data.ifi_obytes;
    966 				ifm->ifm_data.ifi_imcasts =
    967 				    ifp->if_data.ifi_imcasts;
    968 				ifm->ifm_data.ifi_omcasts =
    969 				    ifp->if_data.ifi_omcasts;
    970 				ifm->ifm_data.ifi_iqdrops =
    971 				    ifp->if_data.ifi_iqdrops;
    972 				ifm->ifm_data.ifi_noproto =
    973 				    ifp->if_data.ifi_noproto;
    974 				ifm->ifm_data.ifi_lastchange =
    975 				    ifp->if_data.ifi_lastchange;
    976 				ifm->ifm_addrs = info.rti_addrs;
    977 				error = copyout(ifm, w->w_where, len);
    978 				if (error)
    979 					return (error);
    980 				w->w_where += len;
    981 				break;
    982 			}
    983 #endif
    984 			default:
    985 				panic("sysctl_iflist(2)");
    986 			}
    987 		}
    988 		while ((ifa = ifa->ifa_list.tqe_next) != NULL) {
    989 			if (af && af != ifa->ifa_addr->sa_family)
    990 				continue;
    991 			ifaaddr = ifa->ifa_addr;
    992 			netmask = ifa->ifa_netmask;
    993 			brdaddr = ifa->ifa_dstaddr;
    994 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
    995 				return (error);
    996 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
    997 				struct ifa_msghdr *ifam;
    998 
    999 				ifam = (struct ifa_msghdr *)w->w_tmem;
   1000 				ifam->ifam_index = ifa->ifa_ifp->if_index;
   1001 				ifam->ifam_flags = ifa->ifa_flags;
   1002 				ifam->ifam_metric = ifa->ifa_metric;
   1003 				ifam->ifam_addrs = info.rti_addrs;
   1004 				error = copyout(w->w_tmem, w->w_where, len);
   1005 				if (error)
   1006 					return (error);
   1007 				w->w_where += len;
   1008 			}
   1009 		}
   1010 		ifaaddr = netmask = brdaddr = 0;
   1011 	}
   1012 	return (0);
   1013 }
   1014 
   1015 static int
   1016 sysctl_rtable(name, namelen, where, given, new, newlen)
   1017 	int	*name;
   1018 	u_int	namelen;
   1019 	void 	*where;
   1020 	size_t	*given;
   1021 	void	*new;
   1022 	size_t	newlen;
   1023 {
   1024 	struct radix_node_head *rnh;
   1025 	int	i, s, error = EINVAL;
   1026 	u_char  af;
   1027 	struct	walkarg w;
   1028 
   1029 	if (new)
   1030 		return (EPERM);
   1031 	if (namelen != 3)
   1032 		return (EINVAL);
   1033 	af = name[0];
   1034 	w.w_tmemneeded = 0;
   1035 	w.w_tmemsize = 0;
   1036 	w.w_tmem = NULL;
   1037 again:
   1038 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
   1039 	if (w.w_tmemneeded) {
   1040 		w.w_tmem = (caddr_t) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
   1041 		w.w_tmemsize = w.w_tmemneeded;
   1042 		w.w_tmemneeded = 0;
   1043 	}
   1044 	w.w_op = name[1];
   1045 	w.w_arg = name[2];
   1046 	w.w_given = *given;
   1047 	w.w_needed = 0 - w.w_given;
   1048 	w.w_where = where;
   1049 
   1050 	s = splsoftnet();
   1051 	switch (w.w_op) {
   1052 
   1053 	case NET_RT_DUMP:
   1054 	case NET_RT_FLAGS:
   1055 		for (i = 1; i <= AF_MAX; i++)
   1056 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
   1057 			    (error = (*rnh->rnh_walktree)(rnh,
   1058 			    sysctl_dumpentry, &w)))
   1059 				break;
   1060 		break;
   1061 
   1062 #ifdef COMPAT_14
   1063 	case NET_RT_OIFLIST:
   1064 		error = sysctl_iflist(af, &w, w.w_op);
   1065 		break;
   1066 #endif
   1067 
   1068 	case NET_RT_IFLIST:
   1069 		error = sysctl_iflist(af, &w, w.w_op);
   1070 	}
   1071 	splx(s);
   1072 
   1073 	/* check to see if we couldn't allocate memory with NOWAIT */
   1074 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
   1075 		goto again;
   1076 
   1077 	if (w.w_tmem)
   1078 		free(w.w_tmem, M_RTABLE);
   1079 	w.w_needed += w.w_given;
   1080 	if (where) {
   1081 		*given = w.w_where - (caddr_t) where;
   1082 		if (*given < w.w_needed)
   1083 			return (ENOMEM);
   1084 	} else {
   1085 		*given = (11 * w.w_needed) / 10;
   1086 	}
   1087 	return (error);
   1088 }
   1089 
   1090 /*
   1091  * Definitions of protocols supported in the ROUTE domain.
   1092  */
   1093 
   1094 extern	struct domain routedomain;		/* or at least forward */
   1095 
   1096 struct protosw routesw[] = {
   1097 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
   1098   raw_input,	route_output,	raw_ctlinput,	0,
   1099   route_usrreq,
   1100   raw_init,	0,		0,		0,
   1101   sysctl_rtable,
   1102 }
   1103 };
   1104 
   1105 struct domain routedomain =
   1106     { PF_ROUTE, "route", route_init, 0, 0,
   1107       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
   1108