xref: /freebsd-14-stable/sys/netinet6/raw_ip6.c (revision ce80cdeb1370705d72a7aa6c5784581592c89a61)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
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) 1982, 1986, 1988, 1993
34  *	The Regents of the University of California.
35  * All rights reserved.
36  *
37  * Redistribution and use in source and binary forms, with or without
38  * modification, are permitted provided that the following conditions
39  * are met:
40  * 1. Redistributions of source code must retain the above copyright
41  *    notice, this list of conditions and the following disclaimer.
42  * 2. Redistributions in binary form must reproduce the above copyright
43  *    notice, this list of conditions and the following disclaimer in the
44  *    documentation and/or other materials provided with the distribution.
45  * 3. Neither the name of the University nor the names of its contributors
46  *    may be used to endorse or promote products derived from this software
47  *    without specific prior written permission.
48  *
49  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
50  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
51  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
52  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
53  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
54  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
55  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
56  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
57  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
58  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
59  * SUCH DAMAGE.
60  *
61  *	@(#)raw_ip.c	8.2 (Berkeley) 1/4/94
62  */
63 
64 #include "opt_ipsec.h"
65 #include "opt_inet6.h"
66 #include "opt_route.h"
67 
68 #include <sys/param.h>
69 #include <sys/errno.h>
70 #include <sys/jail.h>
71 #include <sys/kernel.h>
72 #include <sys/lock.h>
73 #include <sys/malloc.h>
74 #include <sys/mbuf.h>
75 #include <sys/priv.h>
76 #include <sys/proc.h>
77 #include <sys/protosw.h>
78 #include <sys/signalvar.h>
79 #include <sys/socket.h>
80 #include <sys/socketvar.h>
81 #include <sys/sx.h>
82 #include <sys/syslog.h>
83 
84 #include <net/if.h>
85 #include <net/if_var.h>
86 #include <net/if_private.h>
87 #include <net/if_types.h>
88 #include <net/route.h>
89 #include <net/vnet.h>
90 
91 #include <netinet/in.h>
92 #include <netinet/in_var.h>
93 #include <netinet/in_systm.h>
94 #include <netinet/in_pcb.h>
95 
96 #include <netinet/icmp6.h>
97 #include <netinet/ip6.h>
98 #include <netinet/ip_var.h>
99 #include <netinet6/ip6_mroute.h>
100 #include <netinet6/in6_pcb.h>
101 #include <netinet6/ip6_var.h>
102 #include <netinet6/nd6.h>
103 #include <netinet6/raw_ip6.h>
104 #include <netinet6/in6_fib.h>
105 #include <netinet6/scope6_var.h>
106 #include <netinet6/send.h>
107 
108 #include <netipsec/ipsec_support.h>
109 
110 #include <machine/stdarg.h>
111 
112 #define	satosin6(sa)	((struct sockaddr_in6 *)(sa))
113 #define	ifatoia6(ifa)	((struct in6_ifaddr *)(ifa))
114 
115 /*
116  * Raw interface to IP6 protocol.
117  */
118 
119 VNET_DECLARE(struct inpcbinfo, ripcbinfo);
120 #define	V_ripcbinfo			VNET(ripcbinfo)
121 
122 VNET_DECLARE(int, rip_bind_all_fibs);
123 #define	V_rip_bind_all_fibs	VNET(rip_bind_all_fibs)
124 
125 extern u_long	rip_sendspace;
126 extern u_long	rip_recvspace;
127 
128 VNET_PCPUSTAT_DEFINE(struct rip6stat, rip6stat);
129 VNET_PCPUSTAT_SYSINIT(rip6stat);
130 
131 #ifdef VIMAGE
132 VNET_PCPUSTAT_SYSUNINIT(rip6stat);
133 #endif /* VIMAGE */
134 
135 /*
136  * Hooks for multicast routing. They all default to NULL, so leave them not
137  * initialized and rely on BSS being set to 0.
138  */
139 
140 /*
141  * The socket used to communicate with the multicast routing daemon.
142  */
143 VNET_DEFINE(struct socket *, ip6_mrouter);
144 
145 /*
146  * The various mrouter functions.
147  */
148 int (*ip6_mrouter_set)(struct socket *, struct sockopt *);
149 int (*ip6_mrouter_get)(struct socket *, struct sockopt *);
150 int (*ip6_mrouter_done)(void);
151 int (*ip6_mforward)(struct ip6_hdr *, struct ifnet *, struct mbuf *);
152 int (*mrt6_ioctl)(u_long, caddr_t);
153 
154 struct rip6_inp_match_ctx {
155 	struct ip6_hdr *ip6;
156 	int proto;
157 };
158 
159 static bool
rip6_inp_match(const struct inpcb * inp,void * v)160 rip6_inp_match(const struct inpcb *inp, void *v)
161 {
162 	struct rip6_inp_match_ctx *c = v;
163 	struct ip6_hdr *ip6 = c->ip6;
164 	int proto = c->proto;
165 
166 	/* XXX inp locking */
167 	if ((inp->inp_vflag & INP_IPV6) == 0)
168 		return (false);
169 	if (inp->inp_ip_p && inp->inp_ip_p != proto)
170 		return (false);
171 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
172 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr, &ip6->ip6_dst))
173 		return (false);
174 	if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr) &&
175 	    !IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr, &ip6->ip6_src))
176 		return (false);
177 
178 	return (true);
179 }
180 
181 /*
182  * Setup generic address and protocol structures for raw_input routine, then
183  * pass them along with mbuf chain.
184  */
185 int
rip6_input(struct mbuf ** mp,int * offp,int proto)186 rip6_input(struct mbuf **mp, int *offp, int proto)
187 {
188 	struct ifnet *ifp;
189 	struct mbuf *n, *m = *mp;
190 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
191 	struct inpcb *inp;
192 	struct mbuf *opts = NULL;
193 	struct sockaddr_in6 fromsa;
194 	struct rip6_inp_match_ctx ctx = { .ip6 = ip6, .proto = proto };
195 	struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
196 	    INPLOOKUP_RLOCKPCB, rip6_inp_match, &ctx);
197 	int delivered = 0, fib;
198 
199 	M_ASSERTPKTHDR(m);
200 	NET_EPOCH_ASSERT();
201 
202 	RIP6STAT_INC(rip6s_ipackets);
203 
204 	init_sin6(&fromsa, m, 0); /* general init */
205 
206 	fib = M_GETFIB(m);
207 	ifp = m->m_pkthdr.rcvif;
208 
209 	while ((inp = inp_next(&inpi)) != NULL) {
210 		INP_RLOCK_ASSERT(inp);
211 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
212 		/*
213 		 * Check AH/ESP integrity.
214 		 */
215 		if (IPSEC_ENABLED(ipv6) &&
216 		    IPSEC_CHECK_POLICY(ipv6, m, inp) != 0) {
217 			/* Do not inject data into pcb. */
218 			continue;
219 		}
220 #endif /* IPSEC */
221 		if (jailed_without_vnet(inp->inp_cred) &&
222 		    !IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) &&
223 		    prison_check_ip6(inp->inp_cred, &ip6->ip6_dst) != 0)
224 			/*
225 			 * Allow raw socket in jail to receive multicast;
226 			 * assume process had PRIV_NETINET_RAW at attach,
227 			 * and fall through into normal filter path if so.
228 			 */
229 			continue;
230 		if (V_rip_bind_all_fibs == 0 && fib != inp->inp_inc.inc_fibnum)
231 			/*
232 			 * Sockets bound to a specific FIB can only receive
233 			 * packets from that FIB.
234 			 */
235 			continue;
236 		if (inp->in6p_cksum != -1) {
237 			RIP6STAT_INC(rip6s_isum);
238 			if (m->m_pkthdr.len - (*offp + inp->in6p_cksum) < 2 ||
239 			    in6_cksum(m, proto, *offp,
240 			    m->m_pkthdr.len - *offp)) {
241 				RIP6STAT_INC(rip6s_badsum);
242 				/*
243 				 * Drop the received message, don't send an
244 				 * ICMP6 message. Set proto to IPPROTO_NONE
245 				 * to achieve that.
246 				 */
247 				INP_RUNLOCK(inp);
248 				proto = IPPROTO_NONE;
249 				break;
250 			}
251 		}
252 		/*
253 		 * If this raw socket has multicast state, and we
254 		 * have received a multicast, check if this socket
255 		 * should receive it, as multicast filtering is now
256 		 * the responsibility of the transport layer.
257 		 */
258 		if (inp->in6p_moptions &&
259 		    IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
260 			/*
261 			 * If the incoming datagram is for MLD, allow it
262 			 * through unconditionally to the raw socket.
263 			 *
264 			 * Use the M_RTALERT_MLD flag to check for MLD
265 			 * traffic without having to inspect the mbuf chain
266 			 * more deeply, as all MLDv1/v2 host messages MUST
267 			 * contain the Router Alert option.
268 			 *
269 			 * In the case of MLDv1, we may not have explicitly
270 			 * joined the group, and may have set IFF_ALLMULTI
271 			 * on the interface. im6o_mc_filter() may discard
272 			 * control traffic we actually need to see.
273 			 *
274 			 * Userland multicast routing daemons should continue
275 			 * filter the control traffic appropriately.
276 			 */
277 			int blocked;
278 
279 			blocked = MCAST_PASS;
280 			if ((m->m_flags & M_RTALERT_MLD) == 0) {
281 				struct sockaddr_in6 mcaddr;
282 
283 				bzero(&mcaddr, sizeof(struct sockaddr_in6));
284 				mcaddr.sin6_len = sizeof(struct sockaddr_in6);
285 				mcaddr.sin6_family = AF_INET6;
286 				mcaddr.sin6_addr = ip6->ip6_dst;
287 
288 				blocked = im6o_mc_filter(inp->in6p_moptions,
289 				    ifp,
290 				    (struct sockaddr *)&mcaddr,
291 				    (struct sockaddr *)&fromsa);
292 			}
293 			if (blocked != MCAST_PASS) {
294 				IP6STAT_INC(ip6s_notmember);
295 				continue;
296 			}
297 		}
298 		if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
299 			continue;
300 		if (inp->inp_flags & INP_CONTROLOPTS ||
301 		    inp->inp_socket->so_options & SO_TIMESTAMP)
302 			ip6_savecontrol(inp, n, &opts);
303 		/* strip intermediate headers */
304 		m_adj(n, *offp);
305 		if (sbappendaddr(&inp->inp_socket->so_rcv,
306 		    (struct sockaddr *)&fromsa, n, opts) == 0) {
307 			soroverflow(inp->inp_socket);
308 			m_freem(n);
309 			if (opts)
310 				m_freem(opts);
311 			RIP6STAT_INC(rip6s_fullsock);
312 		} else {
313 			sorwakeup(inp->inp_socket);
314 			delivered++;
315 		}
316 		opts = NULL;
317 	}
318 	if (delivered == 0) {
319 		RIP6STAT_INC(rip6s_nosock);
320 		if (m->m_flags & M_MCAST)
321 			RIP6STAT_INC(rip6s_nosockmcast);
322 		if (proto == IPPROTO_NONE)
323 			m_freem(m);
324 		else
325 			icmp6_error(m, ICMP6_PARAM_PROB,
326 			    ICMP6_PARAMPROB_NEXTHEADER,
327 			    ip6_get_prevhdr(m, *offp));
328 		IP6STAT_DEC(ip6s_delivered);
329 	} else
330 		m_freem(m);
331 	return (IPPROTO_DONE);
332 }
333 
334 void
rip6_ctlinput(struct ip6ctlparam * ip6cp)335 rip6_ctlinput(struct ip6ctlparam *ip6cp)
336 {
337 	int errno;
338 
339 	if ((errno = icmp6_errmap(ip6cp->ip6c_icmp6)) != 0)
340 		in6_pcbnotify(&V_ripcbinfo, ip6cp->ip6c_finaldst, 0,
341 		    ip6cp->ip6c_src, 0, errno, ip6cp->ip6c_cmdarg,
342 		    in6_rtchange);
343 }
344 
345 /*
346  * Generate IPv6 header and pass packet to ip6_output.  Tack on options user
347  * may have setup with control call.
348  */
349 static int
rip6_send(struct socket * so,int flags,struct mbuf * m,struct sockaddr * nam,struct mbuf * control,struct thread * td)350 rip6_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
351     struct mbuf *control, struct thread *td)
352 {
353 	struct epoch_tracker et;
354 	struct inpcb *inp;
355 	struct sockaddr_in6 tmp, *dstsock;
356 	struct m_tag *mtag;
357 	struct ip6_hdr *ip6;
358 	u_int	plen = m->m_pkthdr.len;
359 	struct ip6_pktopts opt, *optp;
360 	struct ifnet *oifp = NULL;
361 	int error;
362 	int type = 0, code = 0;		/* for ICMPv6 output statistics only */
363 	int scope_ambiguous = 0;
364 	int use_defzone = 0;
365 	int hlim = 0;
366 	struct in6_addr in6a;
367 
368 	inp = sotoinpcb(so);
369 	KASSERT(inp != NULL, ("rip6_send: inp == NULL"));
370 
371 	/* Always copy sockaddr to avoid overwrites. */
372 	/* Unlocked read. */
373 	if (so->so_state & SS_ISCONNECTED) {
374 		if (nam) {
375 			error = EISCONN;
376 			goto release;
377 		}
378 		tmp = (struct sockaddr_in6 ){
379 			.sin6_family = AF_INET6,
380 			.sin6_len = sizeof(struct sockaddr_in6),
381 		};
382 		INP_RLOCK(inp);
383 		bcopy(&inp->in6p_faddr, &tmp.sin6_addr,
384 		    sizeof(struct in6_addr));
385 		INP_RUNLOCK(inp);
386 		dstsock = &tmp;
387 	} else {
388 		if (nam == NULL)
389 			error = ENOTCONN;
390 		else if (nam->sa_family != AF_INET6)
391 			error = EAFNOSUPPORT;
392 		else if (nam->sa_len != sizeof(struct sockaddr_in6))
393 			error = EINVAL;
394 		else
395 			error = 0;
396 		if (error != 0)
397 			goto release;
398 		dstsock = (struct sockaddr_in6 *)nam;
399 		if (dstsock->sin6_family != AF_INET6) {
400 			error = EAFNOSUPPORT;
401 			goto release;
402 		}
403 	}
404 
405 	INP_WLOCK(inp);
406 
407 	if (control != NULL) {
408 		NET_EPOCH_ENTER(et);
409 		error = ip6_setpktopts(control, &opt, inp->in6p_outputopts,
410 		    so->so_cred, inp->inp_ip_p);
411 		NET_EPOCH_EXIT(et);
412 
413 		if (error != 0) {
414 			goto bad;
415 		}
416 		optp = &opt;
417 	} else
418 		optp = inp->in6p_outputopts;
419 
420 	/*
421 	 * Check and convert scope zone ID into internal form.
422 	 *
423 	 * XXX: we may still need to determine the zone later.
424 	 */
425 	if (!(so->so_state & SS_ISCONNECTED)) {
426 		if (!optp || !optp->ip6po_pktinfo ||
427 		    !optp->ip6po_pktinfo->ipi6_ifindex)
428 			use_defzone = V_ip6_use_defzone;
429 		if (dstsock->sin6_scope_id == 0 && !use_defzone)
430 			scope_ambiguous = 1;
431 		if ((error = sa6_embedscope(dstsock, use_defzone)) != 0)
432 			goto bad;
433 	}
434 
435 	/*
436 	 * For an ICMPv6 packet, we should know its type and code to update
437 	 * statistics.
438 	 */
439 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
440 		struct icmp6_hdr *icmp6;
441 		if (m->m_len < sizeof(struct icmp6_hdr) &&
442 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
443 			error = ENOBUFS;
444 			goto bad;
445 		}
446 		icmp6 = mtod(m, struct icmp6_hdr *);
447 		type = icmp6->icmp6_type;
448 		code = icmp6->icmp6_code;
449 	}
450 
451 	M_PREPEND(m, sizeof(*ip6), M_NOWAIT);
452 	if (m == NULL) {
453 		error = ENOBUFS;
454 		goto bad;
455 	}
456 	ip6 = mtod(m, struct ip6_hdr *);
457 
458 #ifdef ROUTE_MPATH
459 	if (CALC_FLOWID_OUTBOUND) {
460 		uint32_t hash_type, hash_val;
461 
462 		hash_val = fib6_calc_software_hash(&inp->in6p_laddr,
463 		    &dstsock->sin6_addr, 0, 0, inp->inp_ip_p, &hash_type);
464 		inp->inp_flowid = hash_val;
465 		inp->inp_flowtype = hash_type;
466 	}
467 #endif
468 	/*
469 	 * Source address selection.
470 	 */
471 	NET_EPOCH_ENTER(et);
472 	error = in6_selectsrc_socket(dstsock, optp, inp, so->so_cred,
473 	    scope_ambiguous, &in6a, &hlim);
474 	NET_EPOCH_EXIT(et);
475 
476 	if (error)
477 		goto bad;
478 	error = prison_check_ip6(inp->inp_cred, &in6a);
479 	if (error != 0)
480 		goto bad;
481 	ip6->ip6_src = in6a;
482 
483 	ip6->ip6_dst = dstsock->sin6_addr;
484 
485 	/*
486 	 * Fill in the rest of the IPv6 header fields.
487 	 */
488 	ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
489 	    (inp->inp_flow & IPV6_FLOWINFO_MASK);
490 	ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
491 	    (IPV6_VERSION & IPV6_VERSION_MASK);
492 
493 	/*
494 	 * ip6_plen will be filled in ip6_output, so not fill it here.
495 	 */
496 	ip6->ip6_nxt = inp->inp_ip_p;
497 	ip6->ip6_hlim = hlim;
498 
499 	if (inp->inp_ip_p == IPPROTO_ICMPV6 || inp->in6p_cksum != -1) {
500 		struct mbuf *n;
501 		int off;
502 		u_int16_t *p;
503 
504 		/* Compute checksum. */
505 		if (inp->inp_ip_p == IPPROTO_ICMPV6)
506 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
507 		else
508 			off = inp->in6p_cksum;
509 		if (plen < off + 2) {
510 			error = EINVAL;
511 			goto bad;
512 		}
513 		off += sizeof(struct ip6_hdr);
514 
515 		n = m;
516 		while (n && n->m_len <= off) {
517 			off -= n->m_len;
518 			n = n->m_next;
519 		}
520 		if (!n)
521 			goto bad;
522 		p = (u_int16_t *)(mtod(n, caddr_t) + off);
523 		*p = 0;
524 		*p = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
525 	}
526 
527 	/*
528 	 * Send RA/RS messages to user land for protection, before sending
529 	 * them to rtadvd/rtsol.
530 	 */
531 	if ((send_sendso_input_hook != NULL) &&
532 	    inp->inp_ip_p == IPPROTO_ICMPV6) {
533 		switch (type) {
534 		case ND_ROUTER_ADVERT:
535 		case ND_ROUTER_SOLICIT:
536 			mtag = m_tag_get(PACKET_TAG_ND_OUTGOING,
537 				sizeof(unsigned short), M_NOWAIT);
538 			if (mtag == NULL)
539 				goto bad;
540 			m_tag_prepend(m, mtag);
541 		}
542 	}
543 
544 	NET_EPOCH_ENTER(et);
545 	error = ip6_output(m, optp, NULL, 0, inp->in6p_moptions, &oifp, inp);
546 	NET_EPOCH_EXIT(et);
547 	if (inp->inp_ip_p == IPPROTO_ICMPV6) {
548 		if (oifp)
549 			icmp6_ifoutstat_inc(oifp, type, code);
550 		ICMP6STAT_INC(icp6s_outhist[type]);
551 	} else
552 		RIP6STAT_INC(rip6s_opackets);
553 
554 	goto freectl;
555 
556  bad:
557 	if (m)
558 		m_freem(m);
559 
560  freectl:
561 	if (control != NULL) {
562 		ip6_clearpktopts(&opt, -1);
563 		m_freem(control);
564 	}
565 	INP_WUNLOCK(inp);
566 	return (error);
567 
568 release:
569 	if (control != NULL)
570 		m_freem(control);
571 	m_freem(m);
572 	return (error);
573 }
574 
575 /*
576  * Raw IPv6 socket option processing.
577  */
578 int
rip6_ctloutput(struct socket * so,struct sockopt * sopt)579 rip6_ctloutput(struct socket *so, struct sockopt *sopt)
580 {
581 	struct inpcb *inp = sotoinpcb(so);
582 	int error;
583 
584 	if (sopt->sopt_level == IPPROTO_ICMPV6)
585 		/*
586 		 * XXX: is it better to call icmp6_ctloutput() directly
587 		 * from protosw?
588 		 */
589 		return (icmp6_ctloutput(so, sopt));
590 	else if (sopt->sopt_level != IPPROTO_IPV6) {
591 		if (sopt->sopt_dir == SOPT_SET &&
592 		    sopt->sopt_level == SOL_SOCKET &&
593 		    sopt->sopt_name == SO_SETFIB)
594 			return (ip6_ctloutput(so, sopt));
595 		return (EINVAL);
596 	}
597 
598 	error = 0;
599 
600 	switch (sopt->sopt_dir) {
601 	case SOPT_GET:
602 		switch (sopt->sopt_name) {
603 		case MRT6_INIT:
604 		case MRT6_DONE:
605 		case MRT6_ADD_MIF:
606 		case MRT6_DEL_MIF:
607 		case MRT6_ADD_MFC:
608 		case MRT6_DEL_MFC:
609 		case MRT6_PIM:
610 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
611 				return (EOPNOTSUPP);
612 			error = ip6_mrouter_get ?  ip6_mrouter_get(so, sopt) :
613 			    EOPNOTSUPP;
614 			break;
615 		case IPV6_CHECKSUM:
616 			error = ip6_raw_ctloutput(so, sopt);
617 			break;
618 		default:
619 			error = ip6_ctloutput(so, sopt);
620 			break;
621 		}
622 		break;
623 
624 	case SOPT_SET:
625 		switch (sopt->sopt_name) {
626 		case MRT6_INIT:
627 		case MRT6_DONE:
628 		case MRT6_ADD_MIF:
629 		case MRT6_DEL_MIF:
630 		case MRT6_ADD_MFC:
631 		case MRT6_DEL_MFC:
632 		case MRT6_PIM:
633 			if (inp->inp_ip_p != IPPROTO_ICMPV6)
634 				return (EOPNOTSUPP);
635 			error = ip6_mrouter_set ?  ip6_mrouter_set(so, sopt) :
636 			    EOPNOTSUPP;
637 			break;
638 		case IPV6_CHECKSUM:
639 			error = ip6_raw_ctloutput(so, sopt);
640 			break;
641 		default:
642 			error = ip6_ctloutput(so, sopt);
643 			break;
644 		}
645 		break;
646 	}
647 
648 	return (error);
649 }
650 
651 static int
rip6_attach(struct socket * so,int proto,struct thread * td)652 rip6_attach(struct socket *so, int proto, struct thread *td)
653 {
654 	struct inpcb *inp;
655 	struct icmp6_filter *filter;
656 	int error;
657 
658 	inp = sotoinpcb(so);
659 	KASSERT(inp == NULL, ("rip6_attach: inp != NULL"));
660 
661 	error = priv_check(td, PRIV_NETINET_RAW);
662 	if (error)
663 		return (error);
664 	if (proto >= IPPROTO_MAX || proto < 0)
665 		return (EPROTONOSUPPORT);
666 	error = soreserve(so, rip_sendspace, rip_recvspace);
667 	if (error)
668 		return (error);
669 	filter = malloc(sizeof(struct icmp6_filter), M_PCB, M_NOWAIT);
670 	if (filter == NULL)
671 		return (ENOMEM);
672 	error = in_pcballoc(so, &V_ripcbinfo);
673 	if (error) {
674 		free(filter, M_PCB);
675 		return (error);
676 	}
677 	inp = (struct inpcb *)so->so_pcb;
678 	inp->inp_ip_p = proto;
679 	inp->in6p_cksum = -1;
680 	inp->in6p_icmp6filt = filter;
681 	ICMP6_FILTER_SETPASSALL(inp->in6p_icmp6filt);
682 	INP_WUNLOCK(inp);
683 	return (0);
684 }
685 
686 static void
rip6_detach(struct socket * so)687 rip6_detach(struct socket *so)
688 {
689 	struct inpcb *inp;
690 
691 	inp = sotoinpcb(so);
692 	KASSERT(inp != NULL, ("rip6_detach: inp == NULL"));
693 
694 	if (so == V_ip6_mrouter && ip6_mrouter_done)
695 		ip6_mrouter_done();
696 	/* xxx: RSVP */
697 	INP_WLOCK(inp);
698 	free(inp->in6p_icmp6filt, M_PCB);
699 	in_pcbfree(inp);
700 }
701 
702 /* XXXRW: This can't ever be called. */
703 static void
rip6_abort(struct socket * so)704 rip6_abort(struct socket *so)
705 {
706 	struct inpcb *inp __diagused;
707 
708 	inp = sotoinpcb(so);
709 	KASSERT(inp != NULL, ("rip6_abort: inp == NULL"));
710 
711 	soisdisconnected(so);
712 }
713 
714 static void
rip6_close(struct socket * so)715 rip6_close(struct socket *so)
716 {
717 	struct inpcb *inp __diagused;
718 
719 	inp = sotoinpcb(so);
720 	KASSERT(inp != NULL, ("rip6_close: inp == NULL"));
721 
722 	soisdisconnected(so);
723 }
724 
725 static int
rip6_disconnect(struct socket * so)726 rip6_disconnect(struct socket *so)
727 {
728 	struct inpcb *inp;
729 
730 	inp = sotoinpcb(so);
731 	KASSERT(inp != NULL, ("rip6_disconnect: inp == NULL"));
732 
733 	if ((so->so_state & SS_ISCONNECTED) == 0)
734 		return (ENOTCONN);
735 	inp->in6p_faddr = in6addr_any;
736 	rip6_abort(so);
737 	return (0);
738 }
739 
740 static int
rip6_bind(struct socket * so,struct sockaddr * nam,struct thread * td)741 rip6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
742 {
743 	struct epoch_tracker et;
744 	struct inpcb *inp;
745 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
746 	struct ifaddr *ifa = NULL;
747 	int error = 0;
748 
749 	inp = sotoinpcb(so);
750 	KASSERT(inp != NULL, ("rip6_bind: inp == NULL"));
751 
752 	if (nam->sa_family != AF_INET6)
753 		return (EAFNOSUPPORT);
754 	if (nam->sa_len != sizeof(*addr))
755 		return (EINVAL);
756 	if ((error = prison_check_ip6(td->td_ucred, &addr->sin6_addr)) != 0)
757 		return (error);
758 	if (CK_STAILQ_EMPTY(&V_ifnet) || addr->sin6_family != AF_INET6)
759 		return (EADDRNOTAVAIL);
760 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
761 		return (error);
762 
763 	NET_EPOCH_ENTER(et);
764 	if (!IN6_IS_ADDR_UNSPECIFIED(&addr->sin6_addr) &&
765 	    (ifa = ifa_ifwithaddr((struct sockaddr *)addr)) == NULL) {
766 		NET_EPOCH_EXIT(et);
767 		return (EADDRNOTAVAIL);
768 	}
769 	if (ifa != NULL &&
770 	    ((struct in6_ifaddr *)ifa)->ia6_flags &
771 	    (IN6_IFF_ANYCAST|IN6_IFF_NOTREADY|
772 	     IN6_IFF_DETACHED|IN6_IFF_DEPRECATED)) {
773 		NET_EPOCH_EXIT(et);
774 		return (EADDRNOTAVAIL);
775 	}
776 	NET_EPOCH_EXIT(et);
777 	INP_WLOCK(inp);
778 	INP_INFO_WLOCK(&V_ripcbinfo);
779 	inp->in6p_laddr = addr->sin6_addr;
780 	INP_INFO_WUNLOCK(&V_ripcbinfo);
781 	INP_WUNLOCK(inp);
782 	return (0);
783 }
784 
785 static int
rip6_connect(struct socket * so,struct sockaddr * nam,struct thread * td)786 rip6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
787 {
788 	struct inpcb *inp;
789 	struct sockaddr_in6 *addr = (struct sockaddr_in6 *)nam;
790 	struct in6_addr in6a;
791 	struct epoch_tracker et;
792 	int error = 0, scope_ambiguous = 0;
793 
794 	inp = sotoinpcb(so);
795 	KASSERT(inp != NULL, ("rip6_connect: inp == NULL"));
796 
797 	if (nam->sa_len != sizeof(*addr))
798 		return (EINVAL);
799 	if (CK_STAILQ_EMPTY(&V_ifnet))
800 		return (EADDRNOTAVAIL);
801 	if (addr->sin6_family != AF_INET6)
802 		return (EAFNOSUPPORT);
803 
804 	/*
805 	 * Application should provide a proper zone ID or the use of default
806 	 * zone IDs should be enabled.  Unfortunately, some applications do
807 	 * not behave as it should, so we need a workaround.  Even if an
808 	 * appropriate ID is not determined, we'll see if we can determine
809 	 * the outgoing interface.  If we can, determine the zone ID based on
810 	 * the interface below.
811 	 */
812 	if (addr->sin6_scope_id == 0 && !V_ip6_use_defzone)
813 		scope_ambiguous = 1;
814 	if ((error = sa6_embedscope(addr, V_ip6_use_defzone)) != 0)
815 		return (error);
816 
817 	INP_WLOCK(inp);
818 	INP_INFO_WLOCK(&V_ripcbinfo);
819 	/* Source address selection. XXX: need pcblookup? */
820 	NET_EPOCH_ENTER(et);
821 	error = in6_selectsrc_socket(addr, inp->in6p_outputopts,
822 	    inp, so->so_cred, scope_ambiguous, &in6a, NULL);
823 	NET_EPOCH_EXIT(et);
824 	if (error) {
825 		INP_INFO_WUNLOCK(&V_ripcbinfo);
826 		INP_WUNLOCK(inp);
827 		return (error);
828 	}
829 
830 	inp->in6p_faddr = addr->sin6_addr;
831 	inp->in6p_laddr = in6a;
832 	soisconnected(so);
833 	INP_INFO_WUNLOCK(&V_ripcbinfo);
834 	INP_WUNLOCK(inp);
835 	return (0);
836 }
837 
838 static int
rip6_shutdown(struct socket * so)839 rip6_shutdown(struct socket *so)
840 {
841 	struct inpcb *inp;
842 
843 	inp = sotoinpcb(so);
844 	KASSERT(inp != NULL, ("rip6_shutdown: inp == NULL"));
845 
846 	INP_WLOCK(inp);
847 	socantsendmore(so);
848 	INP_WUNLOCK(inp);
849 	return (0);
850 }
851 
852 struct protosw rip6_protosw = {
853 	.pr_type =		SOCK_RAW,
854 	.pr_flags =		PR_ATOMIC|PR_ADDR,
855 	.pr_ctloutput =		rip6_ctloutput,
856 	.pr_abort =		rip6_abort,
857 	.pr_attach =		rip6_attach,
858 	.pr_bind =		rip6_bind,
859 	.pr_connect =		rip6_connect,
860 	.pr_control =		in6_control,
861 	.pr_detach =		rip6_detach,
862 	.pr_disconnect =	rip6_disconnect,
863 	.pr_peeraddr =		in6_getpeeraddr,
864 	.pr_send =		rip6_send,
865 	.pr_shutdown =		rip6_shutdown,
866 	.pr_sockaddr =		in6_getsockaddr,
867 	.pr_close =		rip6_close
868 };
869