ip_output.c revision 1549
1/*
2 * Copyright (c) 1982, 1986, 1988, 1990, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 *    must display the following acknowledgement:
15 *	This product includes software developed by the University of
16 *	California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 *    may be used to endorse or promote products derived from this software
19 *    without specific prior written permission.
20 *
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
33 *	@(#)ip_output.c	8.3 (Berkeley) 1/21/94
34 */
35
36#include <sys/param.h>
37#include <sys/systm.h>
38#include <sys/malloc.h>
39#include <sys/mbuf.h>
40#include <sys/errno.h>
41#include <sys/protosw.h>
42#include <sys/socket.h>
43#include <sys/socketvar.h>
44
45#include <net/if.h>
46#include <net/route.h>
47
48#include <netinet/in.h>
49#include <netinet/in_systm.h>
50#include <netinet/ip.h>
51#include <netinet/in_pcb.h>
52#include <netinet/in_var.h>
53#include <netinet/ip_var.h>
54
55#ifdef vax
56#include <machine/mtpr.h>
57#endif
58
59static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *));
60static void ip_mloopback
61	__P((struct ifnet *, struct mbuf *, struct sockaddr_in *));
62
63/*
64 * IP output.  The packet in mbuf chain m contains a skeletal IP
65 * header (with len, off, ttl, proto, tos, src, dst).
66 * The mbuf chain containing the packet will be freed.
67 * The mbuf opt, if present, will not be freed.
68 */
69int
70ip_output(m0, opt, ro, flags, imo)
71	struct mbuf *m0;
72	struct mbuf *opt;
73	struct route *ro;
74	int flags;
75	struct ip_moptions *imo;
76{
77	register struct ip *ip, *mhip;
78	register struct ifnet *ifp;
79	register struct mbuf *m = m0;
80	register int hlen = sizeof (struct ip);
81	int len, off, error = 0;
82	struct route iproute;
83	struct sockaddr_in *dst;
84	struct in_ifaddr *ia;
85
86#ifdef	DIAGNOSTIC
87	if ((m->m_flags & M_PKTHDR) == 0)
88		panic("ip_output no HDR");
89#endif
90	if (opt) {
91		m = ip_insertoptions(m, opt, &len);
92		hlen = len;
93	}
94	ip = mtod(m, struct ip *);
95	/*
96	 * Fill in IP header.
97	 */
98	if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) {
99		ip->ip_v = IPVERSION;
100		ip->ip_off &= IP_DF;
101		ip->ip_id = htons(ip_id++);
102		ip->ip_hl = hlen >> 2;
103		ipstat.ips_localout++;
104	} else {
105		hlen = ip->ip_hl << 2;
106	}
107	/*
108	 * Route packet.
109	 */
110	if (ro == 0) {
111		ro = &iproute;
112		bzero((caddr_t)ro, sizeof (*ro));
113	}
114	dst = (struct sockaddr_in *)&ro->ro_dst;
115	/*
116	 * If there is a cached route,
117	 * check that it is to the same destination
118	 * and is still up.  If not, free it and try again.
119	 */
120	if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
121	   dst->sin_addr.s_addr != ip->ip_dst.s_addr)) {
122		RTFREE(ro->ro_rt);
123		ro->ro_rt = (struct rtentry *)0;
124	}
125	if (ro->ro_rt == 0) {
126		dst->sin_family = AF_INET;
127		dst->sin_len = sizeof(*dst);
128		dst->sin_addr = ip->ip_dst;
129	}
130	/*
131	 * If routing to interface only,
132	 * short circuit routing lookup.
133	 */
134#define ifatoia(ifa)	((struct in_ifaddr *)(ifa))
135#define sintosa(sin)	((struct sockaddr *)(sin))
136	if (flags & IP_ROUTETOIF) {
137		if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 &&
138		    (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) {
139			ipstat.ips_noroute++;
140			error = ENETUNREACH;
141			goto bad;
142		}
143		ifp = ia->ia_ifp;
144		ip->ip_ttl = 1;
145	} else {
146		if (ro->ro_rt == 0)
147			rtalloc(ro);
148		if (ro->ro_rt == 0) {
149			ipstat.ips_noroute++;
150			error = EHOSTUNREACH;
151			goto bad;
152		}
153		ia = ifatoia(ro->ro_rt->rt_ifa);
154		ifp = ro->ro_rt->rt_ifp;
155		ro->ro_rt->rt_use++;
156		if (ro->ro_rt->rt_flags & RTF_GATEWAY)
157			dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway;
158	}
159	if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
160		struct in_multi *inm;
161		extern struct ifnet loif;
162
163		m->m_flags |= M_MCAST;
164		/*
165		 * IP destination address is multicast.  Make sure "dst"
166		 * still points to the address in "ro".  (It may have been
167		 * changed to point to a gateway address, above.)
168		 */
169		dst = (struct sockaddr_in *)&ro->ro_dst;
170		/*
171		 * See if the caller provided any multicast options
172		 */
173		if (imo != NULL) {
174			ip->ip_ttl = imo->imo_multicast_ttl;
175			if (imo->imo_multicast_ifp != NULL)
176				ifp = imo->imo_multicast_ifp;
177		} else
178			ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL;
179		/*
180		 * Confirm that the outgoing interface supports multicast.
181		 */
182		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
183			ipstat.ips_noroute++;
184			error = ENETUNREACH;
185			goto bad;
186		}
187		/*
188		 * If source address not specified yet, use address
189		 * of outgoing interface.
190		 */
191		if (ip->ip_src.s_addr == INADDR_ANY) {
192			register struct in_ifaddr *ia;
193
194			for (ia = in_ifaddr; ia; ia = ia->ia_next)
195				if (ia->ia_ifp == ifp) {
196					ip->ip_src = IA_SIN(ia)->sin_addr;
197					break;
198				}
199		}
200
201		IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm);
202		if (inm != NULL &&
203		   (imo == NULL || imo->imo_multicast_loop)) {
204			/*
205			 * If we belong to the destination multicast group
206			 * on the outgoing interface, and the caller did not
207			 * forbid loopback, loop back a copy.
208			 */
209			ip_mloopback(ifp, m, dst);
210		}
211#ifdef MROUTING
212		else {
213			/*
214			 * If we are acting as a multicast router, perform
215			 * multicast forwarding as if the packet had just
216			 * arrived on the interface to which we are about
217			 * to send.  The multicast forwarding function
218			 * recursively calls this function, using the
219			 * IP_FORWARDING flag to prevent infinite recursion.
220			 *
221			 * Multicasts that are looped back by ip_mloopback(),
222			 * above, will be forwarded by the ip_input() routine,
223			 * if necessary.
224			 */
225			extern struct socket *ip_mrouter;
226			if (ip_mrouter && (flags & IP_FORWARDING) == 0) {
227				if (ip_mforward(m, ifp) != 0) {
228					m_freem(m);
229					goto done;
230				}
231			}
232		}
233#endif
234		/*
235		 * Multicasts with a time-to-live of zero may be looped-
236		 * back, above, but must not be transmitted on a network.
237		 * Also, multicasts addressed to the loopback interface
238		 * are not sent -- the above call to ip_mloopback() will
239		 * loop back a copy if this host actually belongs to the
240		 * destination group on the loopback interface.
241		 */
242		if (ip->ip_ttl == 0 || ifp == &loif) {
243			m_freem(m);
244			goto done;
245		}
246
247		goto sendit;
248	}
249#ifndef notdef
250	/*
251	 * If source address not specified yet, use address
252	 * of outgoing interface.
253	 */
254	if (ip->ip_src.s_addr == INADDR_ANY)
255		ip->ip_src = IA_SIN(ia)->sin_addr;
256#endif
257	/*
258	 * Look for broadcast address and
259	 * and verify user is allowed to send
260	 * such a packet.
261	 */
262	if (in_broadcast(dst->sin_addr, ifp)) {
263		if ((ifp->if_flags & IFF_BROADCAST) == 0) {
264			error = EADDRNOTAVAIL;
265			goto bad;
266		}
267		if ((flags & IP_ALLOWBROADCAST) == 0) {
268			error = EACCES;
269			goto bad;
270		}
271		/* don't allow broadcast messages to be fragmented */
272		if ((u_short)ip->ip_len > ifp->if_mtu) {
273			error = EMSGSIZE;
274			goto bad;
275		}
276		m->m_flags |= M_BCAST;
277	} else
278		m->m_flags &= ~M_BCAST;
279
280sendit:
281	/*
282	 * If small enough for interface, can just send directly.
283	 */
284	if ((u_short)ip->ip_len <= ifp->if_mtu) {
285		ip->ip_len = htons((u_short)ip->ip_len);
286		ip->ip_off = htons((u_short)ip->ip_off);
287		ip->ip_sum = 0;
288		ip->ip_sum = in_cksum(m, hlen);
289		error = (*ifp->if_output)(ifp, m,
290				(struct sockaddr *)dst, ro->ro_rt);
291		goto done;
292	}
293	/*
294	 * Too large for interface; fragment if possible.
295	 * Must be able to put at least 8 bytes per fragment.
296	 */
297	if (ip->ip_off & IP_DF) {
298		error = EMSGSIZE;
299		ipstat.ips_cantfrag++;
300		goto bad;
301	}
302	len = (ifp->if_mtu - hlen) &~ 7;
303	if (len < 8) {
304		error = EMSGSIZE;
305		goto bad;
306	}
307
308    {
309	int mhlen, firstlen = len;
310	struct mbuf **mnext = &m->m_nextpkt;
311
312	/*
313	 * Loop through length of segment after first fragment,
314	 * make new header and copy data of each part and link onto chain.
315	 */
316	m0 = m;
317	mhlen = sizeof (struct ip);
318	for (off = hlen + len; off < (u_short)ip->ip_len; off += len) {
319		MGETHDR(m, M_DONTWAIT, MT_HEADER);
320		if (m == 0) {
321			error = ENOBUFS;
322			ipstat.ips_odropped++;
323			goto sendorfree;
324		}
325		m->m_data += max_linkhdr;
326		mhip = mtod(m, struct ip *);
327		*mhip = *ip;
328		if (hlen > sizeof (struct ip)) {
329			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
330			mhip->ip_hl = mhlen >> 2;
331		}
332		m->m_len = mhlen;
333		mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF);
334		if (ip->ip_off & IP_MF)
335			mhip->ip_off |= IP_MF;
336		if (off + len >= (u_short)ip->ip_len)
337			len = (u_short)ip->ip_len - off;
338		else
339			mhip->ip_off |= IP_MF;
340		mhip->ip_len = htons((u_short)(len + mhlen));
341		m->m_next = m_copy(m0, off, len);
342		if (m->m_next == 0) {
343			(void) m_free(m);
344			error = ENOBUFS;	/* ??? */
345			ipstat.ips_odropped++;
346			goto sendorfree;
347		}
348		m->m_pkthdr.len = mhlen + len;
349		m->m_pkthdr.rcvif = (struct ifnet *)0;
350		mhip->ip_off = htons((u_short)mhip->ip_off);
351		mhip->ip_sum = 0;
352		mhip->ip_sum = in_cksum(m, mhlen);
353		*mnext = m;
354		mnext = &m->m_nextpkt;
355		ipstat.ips_ofragments++;
356	}
357	/*
358	 * Update first fragment by trimming what's been copied out
359	 * and updating header, then send each fragment (in order).
360	 */
361	m = m0;
362	m_adj(m, hlen + firstlen - (u_short)ip->ip_len);
363	m->m_pkthdr.len = hlen + firstlen;
364	ip->ip_len = htons((u_short)m->m_pkthdr.len);
365	ip->ip_off = htons((u_short)(ip->ip_off | IP_MF));
366	ip->ip_sum = 0;
367	ip->ip_sum = in_cksum(m, hlen);
368sendorfree:
369	for (m = m0; m; m = m0) {
370		m0 = m->m_nextpkt;
371		m->m_nextpkt = 0;
372		if (error == 0)
373			error = (*ifp->if_output)(ifp, m,
374			    (struct sockaddr *)dst, ro->ro_rt);
375		else
376			m_freem(m);
377	}
378
379	if (error == 0)
380		ipstat.ips_fragmented++;
381    }
382done:
383	if (ro == &iproute && (flags & IP_ROUTETOIF) == 0 && ro->ro_rt)
384		RTFREE(ro->ro_rt);
385	return (error);
386bad:
387	m_freem(m0);
388	goto done;
389}
390
391/*
392 * Insert IP options into preformed packet.
393 * Adjust IP destination as required for IP source routing,
394 * as indicated by a non-zero in_addr at the start of the options.
395 */
396static struct mbuf *
397ip_insertoptions(m, opt, phlen)
398	register struct mbuf *m;
399	struct mbuf *opt;
400	int *phlen;
401{
402	register struct ipoption *p = mtod(opt, struct ipoption *);
403	struct mbuf *n;
404	register struct ip *ip = mtod(m, struct ip *);
405	unsigned optlen;
406
407	optlen = opt->m_len - sizeof(p->ipopt_dst);
408	if (optlen + (u_short)ip->ip_len > IP_MAXPACKET)
409		return (m);		/* XXX should fail */
410	if (p->ipopt_dst.s_addr)
411		ip->ip_dst = p->ipopt_dst;
412	if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
413		MGETHDR(n, M_DONTWAIT, MT_HEADER);
414		if (n == 0)
415			return (m);
416		n->m_pkthdr.len = m->m_pkthdr.len + optlen;
417		m->m_len -= sizeof(struct ip);
418		m->m_data += sizeof(struct ip);
419		n->m_next = m;
420		m = n;
421		m->m_len = optlen + sizeof(struct ip);
422		m->m_data += max_linkhdr;
423		bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
424	} else {
425		m->m_data -= optlen;
426		m->m_len += optlen;
427		m->m_pkthdr.len += optlen;
428		ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
429	}
430	ip = mtod(m, struct ip *);
431	bcopy((caddr_t)p->ipopt_list, (caddr_t)(ip + 1), (unsigned)optlen);
432	*phlen = sizeof(struct ip) + optlen;
433	ip->ip_len += optlen;
434	return (m);
435}
436
437/*
438 * Copy options from ip to jp,
439 * omitting those not copied during fragmentation.
440 */
441int
442ip_optcopy(ip, jp)
443	struct ip *ip, *jp;
444{
445	register u_char *cp, *dp;
446	int opt, optlen, cnt;
447
448	cp = (u_char *)(ip + 1);
449	dp = (u_char *)(jp + 1);
450	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
451	for (; cnt > 0; cnt -= optlen, cp += optlen) {
452		opt = cp[0];
453		if (opt == IPOPT_EOL)
454			break;
455		if (opt == IPOPT_NOP) {
456			/* Preserve for IP mcast tunnel's LSRR alignment. */
457			*dp++ = IPOPT_NOP;
458			optlen = 1;
459			continue;
460		} else
461			optlen = cp[IPOPT_OLEN];
462		/* bogus lengths should have been caught by ip_dooptions */
463		if (optlen > cnt)
464			optlen = cnt;
465		if (IPOPT_COPIED(opt)) {
466			bcopy((caddr_t)cp, (caddr_t)dp, (unsigned)optlen);
467			dp += optlen;
468		}
469	}
470	for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
471		*dp++ = IPOPT_EOL;
472	return (optlen);
473}
474
475/*
476 * IP socket option processing.
477 */
478int
479ip_ctloutput(op, so, level, optname, mp)
480	int op;
481	struct socket *so;
482	int level, optname;
483	struct mbuf **mp;
484{
485	register struct inpcb *inp = sotoinpcb(so);
486	register struct mbuf *m = *mp;
487	register int optval = 0;
488	int error = 0;
489
490	if (level != IPPROTO_IP) {
491		error = EINVAL;
492		if (op == PRCO_SETOPT && *mp)
493			(void) m_free(*mp);
494	} else switch (op) {
495
496	case PRCO_SETOPT:
497		switch (optname) {
498		case IP_OPTIONS:
499#ifdef notyet
500		case IP_RETOPTS:
501			return (ip_pcbopts(optname, &inp->inp_options, m));
502#else
503			return (ip_pcbopts(&inp->inp_options, m));
504#endif
505
506		case IP_TOS:
507		case IP_TTL:
508		case IP_RECVOPTS:
509		case IP_RECVRETOPTS:
510		case IP_RECVDSTADDR:
511			if (m->m_len != sizeof(int))
512				error = EINVAL;
513			else {
514				optval = *mtod(m, int *);
515				switch (optname) {
516
517				case IP_TOS:
518					inp->inp_ip.ip_tos = optval;
519					break;
520
521				case IP_TTL:
522					inp->inp_ip.ip_ttl = optval;
523					break;
524#define	OPTSET(bit) \
525	if (optval) \
526		inp->inp_flags |= bit; \
527	else \
528		inp->inp_flags &= ~bit;
529
530				case IP_RECVOPTS:
531					OPTSET(INP_RECVOPTS);
532					break;
533
534				case IP_RECVRETOPTS:
535					OPTSET(INP_RECVRETOPTS);
536					break;
537
538				case IP_RECVDSTADDR:
539					OPTSET(INP_RECVDSTADDR);
540					break;
541				}
542			}
543			break;
544#undef OPTSET
545
546		case IP_MULTICAST_IF:
547		case IP_MULTICAST_TTL:
548		case IP_MULTICAST_LOOP:
549		case IP_ADD_MEMBERSHIP:
550		case IP_DROP_MEMBERSHIP:
551			error = ip_setmoptions(optname, &inp->inp_moptions, m);
552			break;
553
554		default:
555			error = ENOPROTOOPT;
556			break;
557		}
558		if (m)
559			(void)m_free(m);
560		break;
561
562	case PRCO_GETOPT:
563		switch (optname) {
564		case IP_OPTIONS:
565		case IP_RETOPTS:
566			*mp = m = m_get(M_WAIT, MT_SOOPTS);
567			if (inp->inp_options) {
568				m->m_len = inp->inp_options->m_len;
569				bcopy(mtod(inp->inp_options, caddr_t),
570				    mtod(m, caddr_t), (unsigned)m->m_len);
571			} else
572				m->m_len = 0;
573			break;
574
575		case IP_TOS:
576		case IP_TTL:
577		case IP_RECVOPTS:
578		case IP_RECVRETOPTS:
579		case IP_RECVDSTADDR:
580			*mp = m = m_get(M_WAIT, MT_SOOPTS);
581			m->m_len = sizeof(int);
582			switch (optname) {
583
584			case IP_TOS:
585				optval = inp->inp_ip.ip_tos;
586				break;
587
588			case IP_TTL:
589				optval = inp->inp_ip.ip_ttl;
590				break;
591
592#define	OPTBIT(bit)	(inp->inp_flags & bit ? 1 : 0)
593
594			case IP_RECVOPTS:
595				optval = OPTBIT(INP_RECVOPTS);
596				break;
597
598			case IP_RECVRETOPTS:
599				optval = OPTBIT(INP_RECVRETOPTS);
600				break;
601
602			case IP_RECVDSTADDR:
603				optval = OPTBIT(INP_RECVDSTADDR);
604				break;
605			}
606			*mtod(m, int *) = optval;
607			break;
608
609		case IP_MULTICAST_IF:
610		case IP_MULTICAST_TTL:
611		case IP_MULTICAST_LOOP:
612		case IP_ADD_MEMBERSHIP:
613		case IP_DROP_MEMBERSHIP:
614			error = ip_getmoptions(optname, inp->inp_moptions, mp);
615			break;
616
617		default:
618			error = ENOPROTOOPT;
619			break;
620		}
621		break;
622	}
623	return (error);
624}
625
626/*
627 * Set up IP options in pcb for insertion in output packets.
628 * Store in mbuf with pointer in pcbopt, adding pseudo-option
629 * with destination address if source routed.
630 */
631int
632#ifdef notyet
633ip_pcbopts(optname, pcbopt, m)
634	int optname;
635#else
636ip_pcbopts(pcbopt, m)
637#endif
638	struct mbuf **pcbopt;
639	register struct mbuf *m;
640{
641	register cnt, optlen;
642	register u_char *cp;
643	u_char opt;
644
645	/* turn off any old options */
646	if (*pcbopt)
647		(void)m_free(*pcbopt);
648	*pcbopt = 0;
649	if (m == (struct mbuf *)0 || m->m_len == 0) {
650		/*
651		 * Only turning off any previous options.
652		 */
653		if (m)
654			(void)m_free(m);
655		return (0);
656	}
657
658#ifndef	vax
659	if (m->m_len % sizeof(long))
660		goto bad;
661#endif
662	/*
663	 * IP first-hop destination address will be stored before
664	 * actual options; move other options back
665	 * and clear it when none present.
666	 */
667	if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
668		goto bad;
669	cnt = m->m_len;
670	m->m_len += sizeof(struct in_addr);
671	cp = mtod(m, u_char *) + sizeof(struct in_addr);
672	ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt);
673	bzero(mtod(m, caddr_t), sizeof(struct in_addr));
674
675	for (; cnt > 0; cnt -= optlen, cp += optlen) {
676		opt = cp[IPOPT_OPTVAL];
677		if (opt == IPOPT_EOL)
678			break;
679		if (opt == IPOPT_NOP)
680			optlen = 1;
681		else {
682			optlen = cp[IPOPT_OLEN];
683			if (optlen <= IPOPT_OLEN || optlen > cnt)
684				goto bad;
685		}
686		switch (opt) {
687
688		default:
689			break;
690
691		case IPOPT_LSRR:
692		case IPOPT_SSRR:
693			/*
694			 * user process specifies route as:
695			 *	->A->B->C->D
696			 * D must be our final destination (but we can't
697			 * check that since we may not have connected yet).
698			 * A is first hop destination, which doesn't appear in
699			 * actual IP option, but is stored before the options.
700			 */
701			if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
702				goto bad;
703			m->m_len -= sizeof(struct in_addr);
704			cnt -= sizeof(struct in_addr);
705			optlen -= sizeof(struct in_addr);
706			cp[IPOPT_OLEN] = optlen;
707			/*
708			 * Move first hop before start of options.
709			 */
710			bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
711			    sizeof(struct in_addr));
712			/*
713			 * Then copy rest of options back
714			 * to close up the deleted entry.
715			 */
716			ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] +
717			    sizeof(struct in_addr)),
718			    (caddr_t)&cp[IPOPT_OFFSET+1],
719			    (unsigned)cnt + sizeof(struct in_addr));
720			break;
721		}
722	}
723	if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
724		goto bad;
725	*pcbopt = m;
726	return (0);
727
728bad:
729	(void)m_free(m);
730	return (EINVAL);
731}
732
733/*
734 * Set the IP multicast options in response to user setsockopt().
735 */
736int
737ip_setmoptions(optname, imop, m)
738	int optname;
739	struct ip_moptions **imop;
740	struct mbuf *m;
741{
742	register int error = 0;
743	u_char loop;
744	register int i;
745	struct in_addr addr;
746	register struct ip_mreq *mreq;
747	register struct ifnet *ifp;
748	register struct ip_moptions *imo = *imop;
749	struct route ro;
750	register struct sockaddr_in *dst;
751
752	if (imo == NULL) {
753		/*
754		 * No multicast option buffer attached to the pcb;
755		 * allocate one and initialize to default values.
756		 */
757		imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS,
758		    M_WAITOK);
759
760		if (imo == NULL)
761			return (ENOBUFS);
762		*imop = imo;
763		imo->imo_multicast_ifp = NULL;
764		imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL;
765		imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP;
766		imo->imo_num_memberships = 0;
767	}
768
769	switch (optname) {
770
771	case IP_MULTICAST_IF:
772		/*
773		 * Select the interface for outgoing multicast packets.
774		 */
775		if (m == NULL || m->m_len != sizeof(struct in_addr)) {
776			error = EINVAL;
777			break;
778		}
779		addr = *(mtod(m, struct in_addr *));
780		/*
781		 * INADDR_ANY is used to remove a previous selection.
782		 * When no interface is selected, a default one is
783		 * chosen every time a multicast packet is sent.
784		 */
785		if (addr.s_addr == INADDR_ANY) {
786			imo->imo_multicast_ifp = NULL;
787			break;
788		}
789		/*
790		 * The selected interface is identified by its local
791		 * IP address.  Find the interface and confirm that
792		 * it supports multicasting.
793		 */
794		INADDR_TO_IFP(addr, ifp);
795		if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
796			error = EADDRNOTAVAIL;
797			break;
798		}
799		imo->imo_multicast_ifp = ifp;
800		break;
801
802	case IP_MULTICAST_TTL:
803		/*
804		 * Set the IP time-to-live for outgoing multicast packets.
805		 */
806		if (m == NULL || m->m_len != 1) {
807			error = EINVAL;
808			break;
809		}
810		imo->imo_multicast_ttl = *(mtod(m, u_char *));
811		break;
812
813	case IP_MULTICAST_LOOP:
814		/*
815		 * Set the loopback flag for outgoing multicast packets.
816		 * Must be zero or one.
817		 */
818		if (m == NULL || m->m_len != 1 ||
819		   (loop = *(mtod(m, u_char *))) > 1) {
820			error = EINVAL;
821			break;
822		}
823		imo->imo_multicast_loop = loop;
824		break;
825
826	case IP_ADD_MEMBERSHIP:
827		/*
828		 * Add a multicast group membership.
829		 * Group must be a valid IP multicast address.
830		 */
831		if (m == NULL || m->m_len != sizeof(struct ip_mreq)) {
832			error = EINVAL;
833			break;
834		}
835		mreq = mtod(m, struct ip_mreq *);
836		if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) {
837			error = EINVAL;
838			break;
839		}
840		/*
841		 * If no interface address was provided, use the interface of
842		 * the route to the given multicast address.
843		 */
844		if (mreq->imr_interface.s_addr == INADDR_ANY) {
845			ro.ro_rt = NULL;
846			dst = (struct sockaddr_in *)&ro.ro_dst;
847			dst->sin_len = sizeof(*dst);
848			dst->sin_family = AF_INET;
849			dst->sin_addr = mreq->imr_multiaddr;
850			rtalloc(&ro);
851			if (ro.ro_rt == NULL) {
852				error = EADDRNOTAVAIL;
853				break;
854			}
855			ifp = ro.ro_rt->rt_ifp;
856			rtfree(ro.ro_rt);
857		}
858		else {
859			INADDR_TO_IFP(mreq->imr_interface, ifp);
860		}
861		/*
862		 * See if we found an interface, and confirm that it
863		 * supports multicast.
864		 */
865		if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
866			error = EADDRNOTAVAIL;
867			break;
868		}
869		/*
870		 * See if the membership already exists or if all the
871		 * membership slots are full.
872		 */
873		for (i = 0; i < imo->imo_num_memberships; ++i) {
874			if (imo->imo_membership[i]->inm_ifp == ifp &&
875			    imo->imo_membership[i]->inm_addr.s_addr
876						== mreq->imr_multiaddr.s_addr)
877				break;
878		}
879		if (i < imo->imo_num_memberships) {
880			error = EADDRINUSE;
881			break;
882		}
883		if (i == IP_MAX_MEMBERSHIPS) {
884			error = ETOOMANYREFS;
885			break;
886		}
887		/*
888		 * Everything looks good; add a new record to the multicast
889		 * address list for the given interface.
890		 */
891		if ((imo->imo_membership[i] =
892		    in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) {
893			error = ENOBUFS;
894			break;
895		}
896		++imo->imo_num_memberships;
897		break;
898
899	case IP_DROP_MEMBERSHIP:
900		/*
901		 * Drop a multicast group membership.
902		 * Group must be a valid IP multicast address.
903		 */
904		if (m == NULL || m->m_len != sizeof(struct ip_mreq)) {
905			error = EINVAL;
906			break;
907		}
908		mreq = mtod(m, struct ip_mreq *);
909		if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) {
910			error = EINVAL;
911			break;
912		}
913		/*
914		 * If an interface address was specified, get a pointer
915		 * to its ifnet structure.
916		 */
917		if (mreq->imr_interface.s_addr == INADDR_ANY)
918			ifp = NULL;
919		else {
920			INADDR_TO_IFP(mreq->imr_interface, ifp);
921			if (ifp == NULL) {
922				error = EADDRNOTAVAIL;
923				break;
924			}
925		}
926		/*
927		 * Find the membership in the membership array.
928		 */
929		for (i = 0; i < imo->imo_num_memberships; ++i) {
930			if ((ifp == NULL ||
931			     imo->imo_membership[i]->inm_ifp == ifp) &&
932			     imo->imo_membership[i]->inm_addr.s_addr ==
933			     mreq->imr_multiaddr.s_addr)
934				break;
935		}
936		if (i == imo->imo_num_memberships) {
937			error = EADDRNOTAVAIL;
938			break;
939		}
940		/*
941		 * Give up the multicast address record to which the
942		 * membership points.
943		 */
944		in_delmulti(imo->imo_membership[i]);
945		/*
946		 * Remove the gap in the membership array.
947		 */
948		for (++i; i < imo->imo_num_memberships; ++i)
949			imo->imo_membership[i-1] = imo->imo_membership[i];
950		--imo->imo_num_memberships;
951		break;
952
953	default:
954		error = EOPNOTSUPP;
955		break;
956	}
957
958	/*
959	 * If all options have default values, no need to keep the mbuf.
960	 */
961	if (imo->imo_multicast_ifp == NULL &&
962	    imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL &&
963	    imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP &&
964	    imo->imo_num_memberships == 0) {
965		free(*imop, M_IPMOPTS);
966		*imop = NULL;
967	}
968
969	return (error);
970}
971
972/*
973 * Return the IP multicast options in response to user getsockopt().
974 */
975int
976ip_getmoptions(optname, imo, mp)
977	int optname;
978	register struct ip_moptions *imo;
979	register struct mbuf **mp;
980{
981	u_char *ttl;
982	u_char *loop;
983	struct in_addr *addr;
984	struct in_ifaddr *ia;
985
986	*mp = m_get(M_WAIT, MT_SOOPTS);
987
988	switch (optname) {
989
990	case IP_MULTICAST_IF:
991		addr = mtod(*mp, struct in_addr *);
992		(*mp)->m_len = sizeof(struct in_addr);
993		if (imo == NULL || imo->imo_multicast_ifp == NULL)
994			addr->s_addr = INADDR_ANY;
995		else {
996			IFP_TO_IA(imo->imo_multicast_ifp, ia);
997			addr->s_addr = (ia == NULL) ? INADDR_ANY
998					: IA_SIN(ia)->sin_addr.s_addr;
999		}
1000		return (0);
1001
1002	case IP_MULTICAST_TTL:
1003		ttl = mtod(*mp, u_char *);
1004		(*mp)->m_len = 1;
1005		*ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL
1006				     : imo->imo_multicast_ttl;
1007		return (0);
1008
1009	case IP_MULTICAST_LOOP:
1010		loop = mtod(*mp, u_char *);
1011		(*mp)->m_len = 1;
1012		*loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP
1013				      : imo->imo_multicast_loop;
1014		return (0);
1015
1016	default:
1017		return (EOPNOTSUPP);
1018	}
1019}
1020
1021/*
1022 * Discard the IP multicast options.
1023 */
1024void
1025ip_freemoptions(imo)
1026	register struct ip_moptions *imo;
1027{
1028	register int i;
1029
1030	if (imo != NULL) {
1031		for (i = 0; i < imo->imo_num_memberships; ++i)
1032			in_delmulti(imo->imo_membership[i]);
1033		free(imo, M_IPMOPTS);
1034	}
1035}
1036
1037/*
1038 * Routine called from ip_output() to loop back a copy of an IP multicast
1039 * packet to the input queue of a specified interface.  Note that this
1040 * calls the output routine of the loopback "driver", but with an interface
1041 * pointer that might NOT be &loif -- easier than replicating that code here.
1042 */
1043static void
1044ip_mloopback(ifp, m, dst)
1045	struct ifnet *ifp;
1046	register struct mbuf *m;
1047	register struct sockaddr_in *dst;
1048{
1049	register struct ip *ip;
1050	struct mbuf *copym;
1051
1052	copym = m_copy(m, 0, M_COPYALL);
1053	if (copym != NULL) {
1054		/*
1055		 * We don't bother to fragment if the IP length is greater
1056		 * than the interface's MTU.  Can this possibly matter?
1057		 */
1058		ip = mtod(copym, struct ip *);
1059		ip->ip_len = htons((u_short)ip->ip_len);
1060		ip->ip_off = htons((u_short)ip->ip_off);
1061		ip->ip_sum = 0;
1062		ip->ip_sum = in_cksum(copym, ip->ip_hl << 2);
1063		(void) looutput(ifp, copym, (struct sockaddr *)dst, NULL);
1064	}
1065}
1066