ip_output.c revision 1542
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/malloc.h>
38#include <sys/mbuf.h>
39#include <sys/errno.h>
40#include <sys/protosw.h>
41#include <sys/socket.h>
42#include <sys/socketvar.h>
43
44#include <net/if.h>
45#include <net/route.h>
46
47#include <netinet/in.h>
48#include <netinet/in_systm.h>
49#include <netinet/ip.h>
50#include <netinet/in_pcb.h>
51#include <netinet/in_var.h>
52#include <netinet/ip_var.h>
53
54#ifdef vax
55#include <machine/mtpr.h>
56#endif
57
58static struct mbuf *ip_insertoptions __P((struct mbuf *, struct mbuf *, int *));
59static void ip_mloopback
60	__P((struct ifnet *, struct mbuf *, struct sockaddr_in *));
61
62/*
63 * IP output.  The packet in mbuf chain m contains a skeletal IP
64 * header (with len, off, ttl, proto, tos, src, dst).
65 * The mbuf chain containing the packet will be freed.
66 * The mbuf opt, if present, will not be freed.
67 */
68int
69ip_output(m0, opt, ro, flags, imo)
70	struct mbuf *m0;
71	struct mbuf *opt;
72	struct route *ro;
73	int flags;
74	struct ip_moptions *imo;
75{
76	register struct ip *ip, *mhip;
77	register struct ifnet *ifp;
78	register struct mbuf *m = m0;
79	register int hlen = sizeof (struct ip);
80	int len, off, error = 0;
81	struct route iproute;
82	struct sockaddr_in *dst;
83	struct in_ifaddr *ia;
84
85#ifdef	DIAGNOSTIC
86	if ((m->m_flags & M_PKTHDR) == 0)
87		panic("ip_output no HDR");
88#endif
89	if (opt) {
90		m = ip_insertoptions(m, opt, &len);
91		hlen = len;
92	}
93	ip = mtod(m, struct ip *);
94	/*
95	 * Fill in IP header.
96	 */
97	if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) {
98		ip->ip_v = IPVERSION;
99		ip->ip_off &= IP_DF;
100		ip->ip_id = htons(ip_id++);
101		ip->ip_hl = hlen >> 2;
102		ipstat.ips_localout++;
103	} else {
104		hlen = ip->ip_hl << 2;
105	}
106	/*
107	 * Route packet.
108	 */
109	if (ro == 0) {
110		ro = &iproute;
111		bzero((caddr_t)ro, sizeof (*ro));
112	}
113	dst = (struct sockaddr_in *)&ro->ro_dst;
114	/*
115	 * If there is a cached route,
116	 * check that it is to the same destination
117	 * and is still up.  If not, free it and try again.
118	 */
119	if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
120	   dst->sin_addr.s_addr != ip->ip_dst.s_addr)) {
121		RTFREE(ro->ro_rt);
122		ro->ro_rt = (struct rtentry *)0;
123	}
124	if (ro->ro_rt == 0) {
125		dst->sin_family = AF_INET;
126		dst->sin_len = sizeof(*dst);
127		dst->sin_addr = ip->ip_dst;
128	}
129	/*
130	 * If routing to interface only,
131	 * short circuit routing lookup.
132	 */
133#define ifatoia(ifa)	((struct in_ifaddr *)(ifa))
134#define sintosa(sin)	((struct sockaddr *)(sin))
135	if (flags & IP_ROUTETOIF) {
136		if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 &&
137		    (ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) {
138			ipstat.ips_noroute++;
139			error = ENETUNREACH;
140			goto bad;
141		}
142		ifp = ia->ia_ifp;
143		ip->ip_ttl = 1;
144	} else {
145		if (ro->ro_rt == 0)
146			rtalloc(ro);
147		if (ro->ro_rt == 0) {
148			ipstat.ips_noroute++;
149			error = EHOSTUNREACH;
150			goto bad;
151		}
152		ia = ifatoia(ro->ro_rt->rt_ifa);
153		ifp = ro->ro_rt->rt_ifp;
154		ro->ro_rt->rt_use++;
155		if (ro->ro_rt->rt_flags & RTF_GATEWAY)
156			dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway;
157	}
158	if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
159		struct in_multi *inm;
160		extern struct ifnet loif;
161
162		m->m_flags |= M_MCAST;
163		/*
164		 * IP destination address is multicast.  Make sure "dst"
165		 * still points to the address in "ro".  (It may have been
166		 * changed to point to a gateway address, above.)
167		 */
168		dst = (struct sockaddr_in *)&ro->ro_dst;
169		/*
170		 * See if the caller provided any multicast options
171		 */
172		if (imo != NULL) {
173			ip->ip_ttl = imo->imo_multicast_ttl;
174			if (imo->imo_multicast_ifp != NULL)
175				ifp = imo->imo_multicast_ifp;
176		} else
177			ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL;
178		/*
179		 * Confirm that the outgoing interface supports multicast.
180		 */
181		if ((ifp->if_flags & IFF_MULTICAST) == 0) {
182			ipstat.ips_noroute++;
183			error = ENETUNREACH;
184			goto bad;
185		}
186		/*
187		 * If source address not specified yet, use address
188		 * of outgoing interface.
189		 */
190		if (ip->ip_src.s_addr == INADDR_ANY) {
191			register struct in_ifaddr *ia;
192
193			for (ia = in_ifaddr; ia; ia = ia->ia_next)
194				if (ia->ia_ifp == ifp) {
195					ip->ip_src = IA_SIN(ia)->sin_addr;
196					break;
197				}
198		}
199
200		IN_LOOKUP_MULTI(ip->ip_dst, ifp, inm);
201		if (inm != NULL &&
202		   (imo == NULL || imo->imo_multicast_loop)) {
203			/*
204			 * If we belong to the destination multicast group
205			 * on the outgoing interface, and the caller did not
206			 * forbid loopback, loop back a copy.
207			 */
208			ip_mloopback(ifp, m, dst);
209		}
210#ifdef MROUTING
211		else {
212			/*
213			 * If we are acting as a multicast router, perform
214			 * multicast forwarding as if the packet had just
215			 * arrived on the interface to which we are about
216			 * to send.  The multicast forwarding function
217			 * recursively calls this function, using the
218			 * IP_FORWARDING flag to prevent infinite recursion.
219			 *
220			 * Multicasts that are looped back by ip_mloopback(),
221			 * above, will be forwarded by the ip_input() routine,
222			 * if necessary.
223			 */
224			extern struct socket *ip_mrouter;
225			if (ip_mrouter && (flags & IP_FORWARDING) == 0) {
226				if (ip_mforward(m, ifp) != 0) {
227					m_freem(m);
228					goto done;
229				}
230			}
231		}
232#endif
233		/*
234		 * Multicasts with a time-to-live of zero may be looped-
235		 * back, above, but must not be transmitted on a network.
236		 * Also, multicasts addressed to the loopback interface
237		 * are not sent -- the above call to ip_mloopback() will
238		 * loop back a copy if this host actually belongs to the
239		 * destination group on the loopback interface.
240		 */
241		if (ip->ip_ttl == 0 || ifp == &loif) {
242			m_freem(m);
243			goto done;
244		}
245
246		goto sendit;
247	}
248#ifndef notdef
249	/*
250	 * If source address not specified yet, use address
251	 * of outgoing interface.
252	 */
253	if (ip->ip_src.s_addr == INADDR_ANY)
254		ip->ip_src = IA_SIN(ia)->sin_addr;
255#endif
256	/*
257	 * Look for broadcast address and
258	 * and verify user is allowed to send
259	 * such a packet.
260	 */
261	if (in_broadcast(dst->sin_addr, ifp)) {
262		if ((ifp->if_flags & IFF_BROADCAST) == 0) {
263			error = EADDRNOTAVAIL;
264			goto bad;
265		}
266		if ((flags & IP_ALLOWBROADCAST) == 0) {
267			error = EACCES;
268			goto bad;
269		}
270		/* don't allow broadcast messages to be fragmented */
271		if ((u_short)ip->ip_len > ifp->if_mtu) {
272			error = EMSGSIZE;
273			goto bad;
274		}
275		m->m_flags |= M_BCAST;
276	} else
277		m->m_flags &= ~M_BCAST;
278
279sendit:
280	/*
281	 * If small enough for interface, can just send directly.
282	 */
283	if ((u_short)ip->ip_len <= ifp->if_mtu) {
284		ip->ip_len = htons((u_short)ip->ip_len);
285		ip->ip_off = htons((u_short)ip->ip_off);
286		ip->ip_sum = 0;
287		ip->ip_sum = in_cksum(m, hlen);
288		error = (*ifp->if_output)(ifp, m,
289				(struct sockaddr *)dst, ro->ro_rt);
290		goto done;
291	}
292	/*
293	 * Too large for interface; fragment if possible.
294	 * Must be able to put at least 8 bytes per fragment.
295	 */
296	if (ip->ip_off & IP_DF) {
297		error = EMSGSIZE;
298		ipstat.ips_cantfrag++;
299		goto bad;
300	}
301	len = (ifp->if_mtu - hlen) &~ 7;
302	if (len < 8) {
303		error = EMSGSIZE;
304		goto bad;
305	}
306
307    {
308	int mhlen, firstlen = len;
309	struct mbuf **mnext = &m->m_nextpkt;
310
311	/*
312	 * Loop through length of segment after first fragment,
313	 * make new header and copy data of each part and link onto chain.
314	 */
315	m0 = m;
316	mhlen = sizeof (struct ip);
317	for (off = hlen + len; off < (u_short)ip->ip_len; off += len) {
318		MGETHDR(m, M_DONTWAIT, MT_HEADER);
319		if (m == 0) {
320			error = ENOBUFS;
321			ipstat.ips_odropped++;
322			goto sendorfree;
323		}
324		m->m_data += max_linkhdr;
325		mhip = mtod(m, struct ip *);
326		*mhip = *ip;
327		if (hlen > sizeof (struct ip)) {
328			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
329			mhip->ip_hl = mhlen >> 2;
330		}
331		m->m_len = mhlen;
332		mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF);
333		if (ip->ip_off & IP_MF)
334			mhip->ip_off |= IP_MF;
335		if (off + len >= (u_short)ip->ip_len)
336			len = (u_short)ip->ip_len - off;
337		else
338			mhip->ip_off |= IP_MF;
339		mhip->ip_len = htons((u_short)(len + mhlen));
340		m->m_next = m_copy(m0, off, len);
341		if (m->m_next == 0) {
342			(void) m_free(m);
343			error = ENOBUFS;	/* ??? */
344			ipstat.ips_odropped++;
345			goto sendorfree;
346		}
347		m->m_pkthdr.len = mhlen + len;
348		m->m_pkthdr.rcvif = (struct ifnet *)0;
349		mhip->ip_off = htons((u_short)mhip->ip_off);
350		mhip->ip_sum = 0;
351		mhip->ip_sum = in_cksum(m, mhlen);
352		*mnext = m;
353		mnext = &m->m_nextpkt;
354		ipstat.ips_ofragments++;
355	}
356	/*
357	 * Update first fragment by trimming what's been copied out
358	 * and updating header, then send each fragment (in order).
359	 */
360	m = m0;
361	m_adj(m, hlen + firstlen - (u_short)ip->ip_len);
362	m->m_pkthdr.len = hlen + firstlen;
363	ip->ip_len = htons((u_short)m->m_pkthdr.len);
364	ip->ip_off = htons((u_short)(ip->ip_off | IP_MF));
365	ip->ip_sum = 0;
366	ip->ip_sum = in_cksum(m, hlen);
367sendorfree:
368	for (m = m0; m; m = m0) {
369		m0 = m->m_nextpkt;
370		m->m_nextpkt = 0;
371		if (error == 0)
372			error = (*ifp->if_output)(ifp, m,
373			    (struct sockaddr *)dst, ro->ro_rt);
374		else
375			m_freem(m);
376	}
377
378	if (error == 0)
379		ipstat.ips_fragmented++;
380    }
381done:
382	if (ro == &iproute && (flags & IP_ROUTETOIF) == 0 && ro->ro_rt)
383		RTFREE(ro->ro_rt);
384	return (error);
385bad:
386	m_freem(m0);
387	goto done;
388}
389
390/*
391 * Insert IP options into preformed packet.
392 * Adjust IP destination as required for IP source routing,
393 * as indicated by a non-zero in_addr at the start of the options.
394 */
395static struct mbuf *
396ip_insertoptions(m, opt, phlen)
397	register struct mbuf *m;
398	struct mbuf *opt;
399	int *phlen;
400{
401	register struct ipoption *p = mtod(opt, struct ipoption *);
402	struct mbuf *n;
403	register struct ip *ip = mtod(m, struct ip *);
404	unsigned optlen;
405
406	optlen = opt->m_len - sizeof(p->ipopt_dst);
407	if (optlen + (u_short)ip->ip_len > IP_MAXPACKET)
408		return (m);		/* XXX should fail */
409	if (p->ipopt_dst.s_addr)
410		ip->ip_dst = p->ipopt_dst;
411	if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
412		MGETHDR(n, M_DONTWAIT, MT_HEADER);
413		if (n == 0)
414			return (m);
415		n->m_pkthdr.len = m->m_pkthdr.len + optlen;
416		m->m_len -= sizeof(struct ip);
417		m->m_data += sizeof(struct ip);
418		n->m_next = m;
419		m = n;
420		m->m_len = optlen + sizeof(struct ip);
421		m->m_data += max_linkhdr;
422		bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
423	} else {
424		m->m_data -= optlen;
425		m->m_len += optlen;
426		m->m_pkthdr.len += optlen;
427		ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
428	}
429	ip = mtod(m, struct ip *);
430	bcopy((caddr_t)p->ipopt_list, (caddr_t)(ip + 1), (unsigned)optlen);
431	*phlen = sizeof(struct ip) + optlen;
432	ip->ip_len += optlen;
433	return (m);
434}
435
436/*
437 * Copy options from ip to jp,
438 * omitting those not copied during fragmentation.
439 */
440int
441ip_optcopy(ip, jp)
442	struct ip *ip, *jp;
443{
444	register u_char *cp, *dp;
445	int opt, optlen, cnt;
446
447	cp = (u_char *)(ip + 1);
448	dp = (u_char *)(jp + 1);
449	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
450	for (; cnt > 0; cnt -= optlen, cp += optlen) {
451		opt = cp[0];
452		if (opt == IPOPT_EOL)
453			break;
454		if (opt == IPOPT_NOP) {
455			/* Preserve for IP mcast tunnel's LSRR alignment. */
456			*dp++ = IPOPT_NOP;
457			optlen = 1;
458			continue;
459		} else
460			optlen = cp[IPOPT_OLEN];
461		/* bogus lengths should have been caught by ip_dooptions */
462		if (optlen > cnt)
463			optlen = cnt;
464		if (IPOPT_COPIED(opt)) {
465			bcopy((caddr_t)cp, (caddr_t)dp, (unsigned)optlen);
466			dp += optlen;
467		}
468	}
469	for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
470		*dp++ = IPOPT_EOL;
471	return (optlen);
472}
473
474/*
475 * IP socket option processing.
476 */
477int
478ip_ctloutput(op, so, level, optname, mp)
479	int op;
480	struct socket *so;
481	int level, optname;
482	struct mbuf **mp;
483{
484	register struct inpcb *inp = sotoinpcb(so);
485	register struct mbuf *m = *mp;
486	register int optval;
487	int error = 0;
488
489	if (level != IPPROTO_IP) {
490		error = EINVAL;
491		if (op == PRCO_SETOPT && *mp)
492			(void) m_free(*mp);
493	} else switch (op) {
494
495	case PRCO_SETOPT:
496		switch (optname) {
497		case IP_OPTIONS:
498#ifdef notyet
499		case IP_RETOPTS:
500			return (ip_pcbopts(optname, &inp->inp_options, m));
501#else
502			return (ip_pcbopts(&inp->inp_options, m));
503#endif
504
505		case IP_TOS:
506		case IP_TTL:
507		case IP_RECVOPTS:
508		case IP_RECVRETOPTS:
509		case IP_RECVDSTADDR:
510			if (m->m_len != sizeof(int))
511				error = EINVAL;
512			else {
513				optval = *mtod(m, int *);
514				switch (optname) {
515
516				case IP_TOS:
517					inp->inp_ip.ip_tos = optval;
518					break;
519
520				case IP_TTL:
521					inp->inp_ip.ip_ttl = optval;
522					break;
523#define	OPTSET(bit) \
524	if (optval) \
525		inp->inp_flags |= bit; \
526	else \
527		inp->inp_flags &= ~bit;
528
529				case IP_RECVOPTS:
530					OPTSET(INP_RECVOPTS);
531					break;
532
533				case IP_RECVRETOPTS:
534					OPTSET(INP_RECVRETOPTS);
535					break;
536
537				case IP_RECVDSTADDR:
538					OPTSET(INP_RECVDSTADDR);
539					break;
540				}
541			}
542			break;
543#undef OPTSET
544
545		case IP_MULTICAST_IF:
546		case IP_MULTICAST_TTL:
547		case IP_MULTICAST_LOOP:
548		case IP_ADD_MEMBERSHIP:
549		case IP_DROP_MEMBERSHIP:
550			error = ip_setmoptions(optname, &inp->inp_moptions, m);
551			break;
552
553		default:
554			error = ENOPROTOOPT;
555			break;
556		}
557		if (m)
558			(void)m_free(m);
559		break;
560
561	case PRCO_GETOPT:
562		switch (optname) {
563		case IP_OPTIONS:
564		case IP_RETOPTS:
565			*mp = m = m_get(M_WAIT, MT_SOOPTS);
566			if (inp->inp_options) {
567				m->m_len = inp->inp_options->m_len;
568				bcopy(mtod(inp->inp_options, caddr_t),
569				    mtod(m, caddr_t), (unsigned)m->m_len);
570			} else
571				m->m_len = 0;
572			break;
573
574		case IP_TOS:
575		case IP_TTL:
576		case IP_RECVOPTS:
577		case IP_RECVRETOPTS:
578		case IP_RECVDSTADDR:
579			*mp = m = m_get(M_WAIT, MT_SOOPTS);
580			m->m_len = sizeof(int);
581			switch (optname) {
582
583			case IP_TOS:
584				optval = inp->inp_ip.ip_tos;
585				break;
586
587			case IP_TTL:
588				optval = inp->inp_ip.ip_ttl;
589				break;
590
591#define	OPTBIT(bit)	(inp->inp_flags & bit ? 1 : 0)
592
593			case IP_RECVOPTS:
594				optval = OPTBIT(INP_RECVOPTS);
595				break;
596
597			case IP_RECVRETOPTS:
598				optval = OPTBIT(INP_RECVRETOPTS);
599				break;
600
601			case IP_RECVDSTADDR:
602				optval = OPTBIT(INP_RECVDSTADDR);
603				break;
604			}
605			*mtod(m, int *) = optval;
606			break;
607
608		case IP_MULTICAST_IF:
609		case IP_MULTICAST_TTL:
610		case IP_MULTICAST_LOOP:
611		case IP_ADD_MEMBERSHIP:
612		case IP_DROP_MEMBERSHIP:
613			error = ip_getmoptions(optname, inp->inp_moptions, mp);
614			break;
615
616		default:
617			error = ENOPROTOOPT;
618			break;
619		}
620		break;
621	}
622	return (error);
623}
624
625/*
626 * Set up IP options in pcb for insertion in output packets.
627 * Store in mbuf with pointer in pcbopt, adding pseudo-option
628 * with destination address if source routed.
629 */
630int
631#ifdef notyet
632ip_pcbopts(optname, pcbopt, m)
633	int optname;
634#else
635ip_pcbopts(pcbopt, m)
636#endif
637	struct mbuf **pcbopt;
638	register struct mbuf *m;
639{
640	register cnt, optlen;
641	register u_char *cp;
642	u_char opt;
643
644	/* turn off any old options */
645	if (*pcbopt)
646		(void)m_free(*pcbopt);
647	*pcbopt = 0;
648	if (m == (struct mbuf *)0 || m->m_len == 0) {
649		/*
650		 * Only turning off any previous options.
651		 */
652		if (m)
653			(void)m_free(m);
654		return (0);
655	}
656
657#ifndef	vax
658	if (m->m_len % sizeof(long))
659		goto bad;
660#endif
661	/*
662	 * IP first-hop destination address will be stored before
663	 * actual options; move other options back
664	 * and clear it when none present.
665	 */
666	if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
667		goto bad;
668	cnt = m->m_len;
669	m->m_len += sizeof(struct in_addr);
670	cp = mtod(m, u_char *) + sizeof(struct in_addr);
671	ovbcopy(mtod(m, caddr_t), (caddr_t)cp, (unsigned)cnt);
672	bzero(mtod(m, caddr_t), sizeof(struct in_addr));
673
674	for (; cnt > 0; cnt -= optlen, cp += optlen) {
675		opt = cp[IPOPT_OPTVAL];
676		if (opt == IPOPT_EOL)
677			break;
678		if (opt == IPOPT_NOP)
679			optlen = 1;
680		else {
681			optlen = cp[IPOPT_OLEN];
682			if (optlen <= IPOPT_OLEN || optlen > cnt)
683				goto bad;
684		}
685		switch (opt) {
686
687		default:
688			break;
689
690		case IPOPT_LSRR:
691		case IPOPT_SSRR:
692			/*
693			 * user process specifies route as:
694			 *	->A->B->C->D
695			 * D must be our final destination (but we can't
696			 * check that since we may not have connected yet).
697			 * A is first hop destination, which doesn't appear in
698			 * actual IP option, but is stored before the options.
699			 */
700			if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
701				goto bad;
702			m->m_len -= sizeof(struct in_addr);
703			cnt -= sizeof(struct in_addr);
704			optlen -= sizeof(struct in_addr);
705			cp[IPOPT_OLEN] = optlen;
706			/*
707			 * Move first hop before start of options.
708			 */
709			bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
710			    sizeof(struct in_addr));
711			/*
712			 * Then copy rest of options back
713			 * to close up the deleted entry.
714			 */
715			ovbcopy((caddr_t)(&cp[IPOPT_OFFSET+1] +
716			    sizeof(struct in_addr)),
717			    (caddr_t)&cp[IPOPT_OFFSET+1],
718			    (unsigned)cnt + sizeof(struct in_addr));
719			break;
720		}
721	}
722	if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
723		goto bad;
724	*pcbopt = m;
725	return (0);
726
727bad:
728	(void)m_free(m);
729	return (EINVAL);
730}
731
732/*
733 * Set the IP multicast options in response to user setsockopt().
734 */
735int
736ip_setmoptions(optname, imop, m)
737	int optname;
738	struct ip_moptions **imop;
739	struct mbuf *m;
740{
741	register int error = 0;
742	u_char loop;
743	register int i;
744	struct in_addr addr;
745	register struct ip_mreq *mreq;
746	register struct ifnet *ifp;
747	register struct ip_moptions *imo = *imop;
748	struct route ro;
749	register struct sockaddr_in *dst;
750
751	if (imo == NULL) {
752		/*
753		 * No multicast option buffer attached to the pcb;
754		 * allocate one and initialize to default values.
755		 */
756		imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS,
757		    M_WAITOK);
758
759		if (imo == NULL)
760			return (ENOBUFS);
761		*imop = imo;
762		imo->imo_multicast_ifp = NULL;
763		imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL;
764		imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP;
765		imo->imo_num_memberships = 0;
766	}
767
768	switch (optname) {
769
770	case IP_MULTICAST_IF:
771		/*
772		 * Select the interface for outgoing multicast packets.
773		 */
774		if (m == NULL || m->m_len != sizeof(struct in_addr)) {
775			error = EINVAL;
776			break;
777		}
778		addr = *(mtod(m, struct in_addr *));
779		/*
780		 * INADDR_ANY is used to remove a previous selection.
781		 * When no interface is selected, a default one is
782		 * chosen every time a multicast packet is sent.
783		 */
784		if (addr.s_addr == INADDR_ANY) {
785			imo->imo_multicast_ifp = NULL;
786			break;
787		}
788		/*
789		 * The selected interface is identified by its local
790		 * IP address.  Find the interface and confirm that
791		 * it supports multicasting.
792		 */
793		INADDR_TO_IFP(addr, ifp);
794		if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
795			error = EADDRNOTAVAIL;
796			break;
797		}
798		imo->imo_multicast_ifp = ifp;
799		break;
800
801	case IP_MULTICAST_TTL:
802		/*
803		 * Set the IP time-to-live for outgoing multicast packets.
804		 */
805		if (m == NULL || m->m_len != 1) {
806			error = EINVAL;
807			break;
808		}
809		imo->imo_multicast_ttl = *(mtod(m, u_char *));
810		break;
811
812	case IP_MULTICAST_LOOP:
813		/*
814		 * Set the loopback flag for outgoing multicast packets.
815		 * Must be zero or one.
816		 */
817		if (m == NULL || m->m_len != 1 ||
818		   (loop = *(mtod(m, u_char *))) > 1) {
819			error = EINVAL;
820			break;
821		}
822		imo->imo_multicast_loop = loop;
823		break;
824
825	case IP_ADD_MEMBERSHIP:
826		/*
827		 * Add a multicast group membership.
828		 * Group must be a valid IP multicast address.
829		 */
830		if (m == NULL || m->m_len != sizeof(struct ip_mreq)) {
831			error = EINVAL;
832			break;
833		}
834		mreq = mtod(m, struct ip_mreq *);
835		if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) {
836			error = EINVAL;
837			break;
838		}
839		/*
840		 * If no interface address was provided, use the interface of
841		 * the route to the given multicast address.
842		 */
843		if (mreq->imr_interface.s_addr == INADDR_ANY) {
844			ro.ro_rt = NULL;
845			dst = (struct sockaddr_in *)&ro.ro_dst;
846			dst->sin_len = sizeof(*dst);
847			dst->sin_family = AF_INET;
848			dst->sin_addr = mreq->imr_multiaddr;
849			rtalloc(&ro);
850			if (ro.ro_rt == NULL) {
851				error = EADDRNOTAVAIL;
852				break;
853			}
854			ifp = ro.ro_rt->rt_ifp;
855			rtfree(ro.ro_rt);
856		}
857		else {
858			INADDR_TO_IFP(mreq->imr_interface, ifp);
859		}
860		/*
861		 * See if we found an interface, and confirm that it
862		 * supports multicast.
863		 */
864		if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
865			error = EADDRNOTAVAIL;
866			break;
867		}
868		/*
869		 * See if the membership already exists or if all the
870		 * membership slots are full.
871		 */
872		for (i = 0; i < imo->imo_num_memberships; ++i) {
873			if (imo->imo_membership[i]->inm_ifp == ifp &&
874			    imo->imo_membership[i]->inm_addr.s_addr
875						== mreq->imr_multiaddr.s_addr)
876				break;
877		}
878		if (i < imo->imo_num_memberships) {
879			error = EADDRINUSE;
880			break;
881		}
882		if (i == IP_MAX_MEMBERSHIPS) {
883			error = ETOOMANYREFS;
884			break;
885		}
886		/*
887		 * Everything looks good; add a new record to the multicast
888		 * address list for the given interface.
889		 */
890		if ((imo->imo_membership[i] =
891		    in_addmulti(&mreq->imr_multiaddr, ifp)) == NULL) {
892			error = ENOBUFS;
893			break;
894		}
895		++imo->imo_num_memberships;
896		break;
897
898	case IP_DROP_MEMBERSHIP:
899		/*
900		 * Drop a multicast group membership.
901		 * Group must be a valid IP multicast address.
902		 */
903		if (m == NULL || m->m_len != sizeof(struct ip_mreq)) {
904			error = EINVAL;
905			break;
906		}
907		mreq = mtod(m, struct ip_mreq *);
908		if (!IN_MULTICAST(ntohl(mreq->imr_multiaddr.s_addr))) {
909			error = EINVAL;
910			break;
911		}
912		/*
913		 * If an interface address was specified, get a pointer
914		 * to its ifnet structure.
915		 */
916		if (mreq->imr_interface.s_addr == INADDR_ANY)
917			ifp = NULL;
918		else {
919			INADDR_TO_IFP(mreq->imr_interface, ifp);
920			if (ifp == NULL) {
921				error = EADDRNOTAVAIL;
922				break;
923			}
924		}
925		/*
926		 * Find the membership in the membership array.
927		 */
928		for (i = 0; i < imo->imo_num_memberships; ++i) {
929			if ((ifp == NULL ||
930			     imo->imo_membership[i]->inm_ifp == ifp) &&
931			     imo->imo_membership[i]->inm_addr.s_addr ==
932			     mreq->imr_multiaddr.s_addr)
933				break;
934		}
935		if (i == imo->imo_num_memberships) {
936			error = EADDRNOTAVAIL;
937			break;
938		}
939		/*
940		 * Give up the multicast address record to which the
941		 * membership points.
942		 */
943		in_delmulti(imo->imo_membership[i]);
944		/*
945		 * Remove the gap in the membership array.
946		 */
947		for (++i; i < imo->imo_num_memberships; ++i)
948			imo->imo_membership[i-1] = imo->imo_membership[i];
949		--imo->imo_num_memberships;
950		break;
951
952	default:
953		error = EOPNOTSUPP;
954		break;
955	}
956
957	/*
958	 * If all options have default values, no need to keep the mbuf.
959	 */
960	if (imo->imo_multicast_ifp == NULL &&
961	    imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL &&
962	    imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP &&
963	    imo->imo_num_memberships == 0) {
964		free(*imop, M_IPMOPTS);
965		*imop = NULL;
966	}
967
968	return (error);
969}
970
971/*
972 * Return the IP multicast options in response to user getsockopt().
973 */
974int
975ip_getmoptions(optname, imo, mp)
976	int optname;
977	register struct ip_moptions *imo;
978	register struct mbuf **mp;
979{
980	u_char *ttl;
981	u_char *loop;
982	struct in_addr *addr;
983	struct in_ifaddr *ia;
984
985	*mp = m_get(M_WAIT, MT_SOOPTS);
986
987	switch (optname) {
988
989	case IP_MULTICAST_IF:
990		addr = mtod(*mp, struct in_addr *);
991		(*mp)->m_len = sizeof(struct in_addr);
992		if (imo == NULL || imo->imo_multicast_ifp == NULL)
993			addr->s_addr = INADDR_ANY;
994		else {
995			IFP_TO_IA(imo->imo_multicast_ifp, ia);
996			addr->s_addr = (ia == NULL) ? INADDR_ANY
997					: IA_SIN(ia)->sin_addr.s_addr;
998		}
999		return (0);
1000
1001	case IP_MULTICAST_TTL:
1002		ttl = mtod(*mp, u_char *);
1003		(*mp)->m_len = 1;
1004		*ttl = (imo == NULL) ? IP_DEFAULT_MULTICAST_TTL
1005				     : imo->imo_multicast_ttl;
1006		return (0);
1007
1008	case IP_MULTICAST_LOOP:
1009		loop = mtod(*mp, u_char *);
1010		(*mp)->m_len = 1;
1011		*loop = (imo == NULL) ? IP_DEFAULT_MULTICAST_LOOP
1012				      : imo->imo_multicast_loop;
1013		return (0);
1014
1015	default:
1016		return (EOPNOTSUPP);
1017	}
1018}
1019
1020/*
1021 * Discard the IP multicast options.
1022 */
1023void
1024ip_freemoptions(imo)
1025	register struct ip_moptions *imo;
1026{
1027	register int i;
1028
1029	if (imo != NULL) {
1030		for (i = 0; i < imo->imo_num_memberships; ++i)
1031			in_delmulti(imo->imo_membership[i]);
1032		free(imo, M_IPMOPTS);
1033	}
1034}
1035
1036/*
1037 * Routine called from ip_output() to loop back a copy of an IP multicast
1038 * packet to the input queue of a specified interface.  Note that this
1039 * calls the output routine of the loopback "driver", but with an interface
1040 * pointer that might NOT be &loif -- easier than replicating that code here.
1041 */
1042static void
1043ip_mloopback(ifp, m, dst)
1044	struct ifnet *ifp;
1045	register struct mbuf *m;
1046	register struct sockaddr_in *dst;
1047{
1048	register struct ip *ip;
1049	struct mbuf *copym;
1050
1051	copym = m_copy(m, 0, M_COPYALL);
1052	if (copym != NULL) {
1053		/*
1054		 * We don't bother to fragment if the IP length is greater
1055		 * than the interface's MTU.  Can this possibly matter?
1056		 */
1057		ip = mtod(copym, struct ip *);
1058		ip->ip_len = htons((u_short)ip->ip_len);
1059		ip->ip_off = htons((u_short)ip->ip_off);
1060		ip->ip_sum = 0;
1061		ip->ip_sum = in_cksum(copym, ip->ip_hl << 2);
1062		(void) looutput(ifp, copym, (struct sockaddr *)dst, NULL);
1063	}
1064}
1065