xref: /openbsd/sys/netinet6/raw_ip6.c (revision ace0f189)
1 /*	$OpenBSD: raw_ip6.c,v 1.184 2024/04/17 20:48:51 bluhm Exp $	*/
2 /*	$KAME: raw_ip6.c,v 1.69 2001/03/04 15:55:44 itojun Exp $	*/
3 
4 /*
5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*
34  * Copyright (c) 1982, 1986, 1988, 1993
35  *	The Regents of the University of California.  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 "pf.h"
65 
66 #include <sys/param.h>
67 #include <sys/malloc.h>
68 #include <sys/mbuf.h>
69 #include <sys/socket.h>
70 #include <sys/protosw.h>
71 #include <sys/socketvar.h>
72 #include <sys/errno.h>
73 #include <sys/systm.h>
74 #include <sys/sysctl.h>
75 
76 #include <net/if.h>
77 #include <net/if_var.h>
78 #include <net/route.h>
79 
80 #include <netinet/in.h>
81 #include <netinet6/in6_var.h>
82 #include <netinet/ip6.h>
83 #include <netinet6/ip6_var.h>
84 #ifdef MROUTING
85 #include <netinet6/ip6_mroute.h>
86 #endif
87 #include <netinet/icmp6.h>
88 #include <netinet/ip.h>
89 #include <netinet/in_pcb.h>
90 #include <netinet6/nd6.h>
91 #include <netinet6/ip6protosw.h>
92 #include <netinet6/raw_ip6.h>
93 
94 #if NPF > 0
95 #include <net/pfvar.h>
96 #endif
97 
98 #include <sys/stdarg.h>
99 
100 /*
101  * Raw interface to IP6 protocol.
102  */
103 
104 struct	inpcbtable rawin6pcbtable;
105 
106 struct cpumem *rip6counters;
107 
108 const struct pr_usrreqs rip6_usrreqs = {
109 	.pru_attach	= rip6_attach,
110 	.pru_detach	= rip6_detach,
111 	.pru_lock	= rip6_lock,
112 	.pru_unlock	= rip6_unlock,
113 	.pru_locked	= rip6_locked,
114 	.pru_bind	= rip6_bind,
115 	.pru_connect	= rip6_connect,
116 	.pru_disconnect	= rip6_disconnect,
117 	.pru_shutdown	= rip6_shutdown,
118 	.pru_send	= rip6_send,
119 	.pru_control	= in6_control,
120 	.pru_sockaddr	= in6_sockaddr,
121 	.pru_peeraddr	= in6_peeraddr,
122 };
123 
124 /*
125  * Initialize raw connection block queue.
126  */
127 void
rip6_init(void)128 rip6_init(void)
129 {
130 	in_pcbinit(&rawin6pcbtable, 1);
131 	rip6counters = counters_alloc(rip6s_ncounters);
132 }
133 
134 int
rip6_input(struct mbuf ** mp,int * offp,int proto,int af)135 rip6_input(struct mbuf **mp, int *offp, int proto, int af)
136 {
137 	struct mbuf *m = *mp;
138 	struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
139 	struct inpcb *inp;
140 	SIMPLEQ_HEAD(, inpcb) inpcblist;
141 	struct in6_addr *key;
142 	struct sockaddr_in6 rip6src;
143 	uint8_t type;
144 
145 	KASSERT(af == AF_INET6);
146 
147 	if (proto == IPPROTO_ICMPV6) {
148 		struct icmp6_hdr *icmp6;
149 
150 		IP6_EXTHDR_GET(icmp6, struct icmp6_hdr *, m, *offp,
151 		    sizeof(*icmp6));
152 		if (icmp6 == NULL)
153 			return IPPROTO_DONE;
154 		type = icmp6->icmp6_type;
155 	} else
156 		rip6stat_inc(rip6s_ipackets);
157 
158 	memset(&rip6src, 0, sizeof(rip6src));
159 	rip6src.sin6_family = AF_INET6;
160 	rip6src.sin6_len = sizeof(rip6src);
161 	/* KAME hack: recover scopeid */
162 	in6_recoverscope(&rip6src, &ip6->ip6_src);
163 
164 	key = &ip6->ip6_dst;
165 #if NPF > 0
166 	if (m->m_pkthdr.pf.flags & PF_TAG_DIVERTED) {
167 		struct pf_divert *divert;
168 
169 		divert = pf_find_divert(m);
170 		KASSERT(divert != NULL);
171 		switch (divert->type) {
172 		case PF_DIVERT_TO:
173 			key = &divert->addr.v6;
174 			break;
175 		case PF_DIVERT_REPLY:
176 			break;
177 		default:
178 			panic("%s: unknown divert type %d, mbuf %p, divert %p",
179 			    __func__, divert->type, m, divert);
180 		}
181 	}
182 #endif
183 	SIMPLEQ_INIT(&inpcblist);
184 	rw_enter_write(&rawin6pcbtable.inpt_notify);
185 	mtx_enter(&rawin6pcbtable.inpt_mtx);
186 	TAILQ_FOREACH(inp, &rawin6pcbtable.inpt_queue, inp_queue) {
187 		KASSERT(ISSET(inp->inp_flags, INP_IPV6));
188 
189 		/*
190 		 * Packet must not be inserted after disconnected wakeup
191 		 * call.  To avoid race, check again when holding receive
192 		 * buffer mutex.
193 		 */
194 		if (ISSET(READ_ONCE(inp->inp_socket->so_rcv.sb_state),
195 		    SS_CANTRCVMORE))
196 			continue;
197 		if (rtable_l2(inp->inp_rtableid) !=
198 		    rtable_l2(m->m_pkthdr.ph_rtableid))
199 			continue;
200 
201 		if ((inp->inp_ipv6.ip6_nxt || proto == IPPROTO_ICMPV6) &&
202 		    inp->inp_ipv6.ip6_nxt != proto)
203 			continue;
204 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6) &&
205 		    !IN6_ARE_ADDR_EQUAL(&inp->inp_laddr6, key))
206 			continue;
207 		if (!IN6_IS_ADDR_UNSPECIFIED(&inp->inp_faddr6) &&
208 		    !IN6_ARE_ADDR_EQUAL(&inp->inp_faddr6, &ip6->ip6_src))
209 			continue;
210 		if (proto == IPPROTO_ICMPV6 && inp->inp_icmp6filt) {
211 			if (ICMP6_FILTER_WILLBLOCK(type, inp->inp_icmp6filt))
212 				continue;
213 		}
214 		if (proto != IPPROTO_ICMPV6 && inp->inp_cksum6 != -1) {
215 			rip6stat_inc(rip6s_isum);
216 			/*
217 			 * Although in6_cksum() does not need the position of
218 			 * the checksum field for verification, enforce that it
219 			 * is located within the packet.  Userland has given
220 			 * a checksum offset, a packet too short for that is
221 			 * invalid.  Avoid overflow with user supplied offset.
222 			 */
223 			if (m->m_pkthdr.len < *offp + 2 ||
224 			    m->m_pkthdr.len - *offp - 2 < inp->inp_cksum6 ||
225 			    in6_cksum(m, proto, *offp,
226 			    m->m_pkthdr.len - *offp)) {
227 				rip6stat_inc(rip6s_badsum);
228 				continue;
229 			}
230 		}
231 
232 		in_pcbref(inp);
233 		SIMPLEQ_INSERT_TAIL(&inpcblist, inp, inp_notify);
234 	}
235 	mtx_leave(&rawin6pcbtable.inpt_mtx);
236 
237 	if (SIMPLEQ_EMPTY(&inpcblist)) {
238 		struct counters_ref ref;
239 		uint64_t *counters;
240 
241 		rw_exit_write(&rawin6pcbtable.inpt_notify);
242 
243 		if (proto != IPPROTO_ICMPV6) {
244 			rip6stat_inc(rip6s_nosock);
245 			if (m->m_flags & M_MCAST)
246 				rip6stat_inc(rip6s_nosockmcast);
247 		}
248 		if (proto == IPPROTO_NONE || proto == IPPROTO_ICMPV6) {
249 			m_freem(m);
250 		} else {
251 			int prvnxt = ip6_get_prevhdr(m, *offp);
252 
253 			icmp6_error(m, ICMP6_PARAM_PROB,
254 			    ICMP6_PARAMPROB_NEXTHEADER, prvnxt);
255 		}
256 		counters = counters_enter(&ref, ip6counters);
257 		counters[ip6s_delivered]--;
258 		counters_leave(&ref, ip6counters);
259 
260 		return IPPROTO_DONE;
261 	}
262 
263 	while ((inp = SIMPLEQ_FIRST(&inpcblist)) != NULL) {
264 		struct mbuf *n, *opts = NULL;
265 
266 		SIMPLEQ_REMOVE_HEAD(&inpcblist, inp_notify);
267 		if (SIMPLEQ_EMPTY(&inpcblist))
268 			n = m;
269 		else
270 			n = m_copym(m, 0, M_COPYALL, M_NOWAIT);
271 		if (n != NULL) {
272 			struct socket *so = inp->inp_socket;
273 			int ret = 0;
274 
275 			if (inp->inp_flags & IN6P_CONTROLOPTS)
276 				ip6_savecontrol(inp, n, &opts);
277 			/* strip intermediate headers */
278 			m_adj(n, *offp);
279 
280 			mtx_enter(&so->so_rcv.sb_mtx);
281 			if (!ISSET(inp->inp_socket->so_rcv.sb_state,
282 			    SS_CANTRCVMORE)) {
283 				ret = sbappendaddr(so, &so->so_rcv,
284 				    sin6tosa(&rip6src), n, opts);
285 			}
286 			mtx_leave(&so->so_rcv.sb_mtx);
287 
288 			if (ret == 0) {
289 				m_freem(n);
290 				m_freem(opts);
291 				rip6stat_inc(rip6s_fullsock);
292 			} else
293 				sorwakeup(so);
294 		}
295 		in_pcbunref(inp);
296 	}
297 	rw_exit_write(&rawin6pcbtable.inpt_notify);
298 
299 	return IPPROTO_DONE;
300 }
301 
302 void
rip6_ctlinput(int cmd,struct sockaddr * sa,u_int rdomain,void * d)303 rip6_ctlinput(int cmd, struct sockaddr *sa, u_int rdomain, void *d)
304 {
305 	struct ip6_hdr *ip6;
306 	struct ip6ctlparam *ip6cp = NULL;
307 	struct sockaddr_in6 *sa6 = satosin6(sa);
308 	const struct sockaddr_in6 *sa6_src = NULL;
309 	void *cmdarg;
310 	void (*notify)(struct inpcb *, int) = in_rtchange;
311 	int nxt;
312 
313 	if (sa->sa_family != AF_INET6 ||
314 	    sa->sa_len != sizeof(struct sockaddr_in6))
315 		return;
316 
317 	if ((unsigned)cmd >= PRC_NCMDS)
318 		return;
319 	if (PRC_IS_REDIRECT(cmd))
320 		notify = in_rtchange, d = NULL;
321 	else if (cmd == PRC_HOSTDEAD)
322 		d = NULL;
323 	else if (cmd == PRC_MSGSIZE)
324 		; /* special code is present, see below */
325 	else if (inet6ctlerrmap[cmd] == 0)
326 		return;
327 
328 	/* if the parameter is from icmp6, decode it. */
329 	if (d != NULL) {
330 		ip6cp = (struct ip6ctlparam *)d;
331 		ip6 = ip6cp->ip6c_ip6;
332 		cmdarg = ip6cp->ip6c_cmdarg;
333 		sa6_src = ip6cp->ip6c_src;
334 		nxt = ip6cp->ip6c_nxt;
335 	} else {
336 		ip6 = NULL;
337 		cmdarg = NULL;
338 		sa6_src = &sa6_any;
339 		nxt = -1;
340 	}
341 
342 	if (ip6 && cmd == PRC_MSGSIZE) {
343 		int valid = 0;
344 		struct inpcb *inp;
345 
346 		/*
347 		 * Check to see if we have a valid raw IPv6 socket
348 		 * corresponding to the address in the ICMPv6 message
349 		 * payload, and the protocol (ip6_nxt) meets the socket.
350 		 * XXX chase extension headers, or pass final nxt value
351 		 * from icmp6_notify_error()
352 		 */
353 		inp = in6_pcblookup(&rawin6pcbtable, &sa6->sin6_addr, 0,
354 		    &sa6_src->sin6_addr, 0, rdomain);
355 
356 		if (inp && inp->inp_ipv6.ip6_nxt &&
357 		    inp->inp_ipv6.ip6_nxt == nxt)
358 			valid = 1;
359 
360 		/*
361 		 * Depending on the value of "valid" and routing table
362 		 * size (mtudisc_{hi,lo}wat), we will:
363 		 * - recalculate the new MTU and create the
364 		 *   corresponding routing entry, or
365 		 * - ignore the MTU change notification.
366 		 */
367 		icmp6_mtudisc_update((struct ip6ctlparam *)d, valid);
368 		in_pcbunref(inp);
369 
370 		/*
371 		 * regardless of if we called icmp6_mtudisc_update(),
372 		 * we need to call in6_pcbnotify(), to notify path
373 		 * MTU change to the userland (2292bis-02), because
374 		 * some unconnected sockets may share the same
375 		 * destination and want to know the path MTU.
376 		 */
377 	}
378 
379 	in6_pcbnotify(&rawin6pcbtable, sa6, 0,
380 	    sa6_src, 0, rdomain, cmd, cmdarg, notify);
381 }
382 
383 /*
384  * Generate IPv6 header and pass packet to ip6_output.
385  * Tack on options user may have setup with control call.
386  */
387 int
rip6_output(struct mbuf * m,struct socket * so,struct sockaddr * dstaddr,struct mbuf * control)388 rip6_output(struct mbuf *m, struct socket *so, struct sockaddr *dstaddr,
389     struct mbuf *control)
390 {
391 	struct in6_addr *dst;
392 	struct ip6_hdr *ip6;
393 	struct inpcb *inp;
394 	u_int	plen = m->m_pkthdr.len;
395 	int error = 0;
396 	struct ip6_pktopts opt, *optp = NULL;
397 	int type;		/* for ICMPv6 output statistics only */
398 	int priv = 0;
399 	int flags;
400 
401 	inp = sotoinpcb(so);
402 
403 	priv = 0;
404 	if ((so->so_state & SS_PRIV) != 0)
405 		priv = 1;
406 	if (control) {
407 		if ((error = ip6_setpktopts(control, &opt,
408 		    inp->inp_outputopts6,
409 		    priv, so->so_proto->pr_protocol)) != 0)
410 			goto bad;
411 		optp = &opt;
412 	} else
413 		optp = inp->inp_outputopts6;
414 
415 	if (dstaddr->sa_family != AF_INET6) {
416 		error = EAFNOSUPPORT;
417 		goto bad;
418 	}
419 	dst = &satosin6(dstaddr)->sin6_addr;
420 	if (IN6_IS_ADDR_V4MAPPED(dst)) {
421 		error = EADDRNOTAVAIL;
422 		goto bad;
423 	}
424 
425 	/*
426 	 * For an ICMPv6 packet, we should know its type and code
427 	 * to update statistics.
428 	 */
429 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
430 		struct icmp6_hdr *icmp6;
431 		if (m->m_len < sizeof(struct icmp6_hdr) &&
432 		    (m = m_pullup(m, sizeof(struct icmp6_hdr))) == NULL) {
433 			error = ENOBUFS;
434 			goto bad;
435 		}
436 		icmp6 = mtod(m, struct icmp6_hdr *);
437 		type = icmp6->icmp6_type;
438 	}
439 
440 	M_PREPEND(m, sizeof(*ip6), M_DONTWAIT);
441 	if (!m) {
442 		error = ENOBUFS;
443 		goto bad;
444 	}
445 	ip6 = mtod(m, struct ip6_hdr *);
446 
447 	/*
448 	 * Next header might not be ICMP6 but use its pseudo header anyway.
449 	 */
450 	ip6->ip6_dst = *dst;
451 
452 	/* KAME hack: embed scopeid */
453 	if (in6_embedscope(&ip6->ip6_dst, satosin6(dstaddr),
454 	    optp, inp->inp_moptions6) != 0) {
455 		error = EINVAL;
456 		goto bad;
457 	}
458 
459 	/*
460 	 * Source address selection.
461 	 */
462 	{
463 		const struct in6_addr *in6a;
464 
465 		error = in6_pcbselsrc(&in6a, satosin6(dstaddr), inp, optp);
466 		if (error)
467 			goto bad;
468 
469 		ip6->ip6_src = *in6a;
470 	}
471 
472 	ip6->ip6_flow = inp->inp_flowinfo & IPV6_FLOWINFO_MASK;
473 	ip6->ip6_vfc  &= ~IPV6_VERSION_MASK;
474 	ip6->ip6_vfc  |= IPV6_VERSION;
475 #if 0				/* ip6_plen will be filled in ip6_output. */
476 	ip6->ip6_plen  = htons((u_short)plen);
477 #endif
478 	ip6->ip6_nxt   = inp->inp_ipv6.ip6_nxt;
479 	ip6->ip6_hlim = in6_selecthlim(inp);
480 
481 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6 ||
482 	    inp->inp_cksum6 != -1) {
483 		struct mbuf *n;
484 		int off;
485 		u_int16_t *sump;
486 		int sumoff;
487 
488 		/* compute checksum */
489 		if (so->so_proto->pr_protocol == IPPROTO_ICMPV6)
490 			off = offsetof(struct icmp6_hdr, icmp6_cksum);
491 		else
492 			off = inp->inp_cksum6;
493 		if (plen < 2 || plen - 2 < off) {
494 			error = EINVAL;
495 			goto bad;
496 		}
497 		off += sizeof(struct ip6_hdr);
498 
499 		n = m_pulldown(m, off, sizeof(*sump), &sumoff);
500 		if (n == NULL) {
501 			m = NULL;
502 			error = ENOBUFS;
503 			goto bad;
504 		}
505 		sump = (u_int16_t *)(mtod(n, caddr_t) + sumoff);
506 		*sump = 0;
507 		*sump = in6_cksum(m, ip6->ip6_nxt, sizeof(*ip6), plen);
508 	}
509 
510 	flags = 0;
511 	if (inp->inp_flags & IN6P_MINMTU)
512 		flags |= IPV6_MINMTU;
513 
514 	/* force routing table */
515 	m->m_pkthdr.ph_rtableid = inp->inp_rtableid;
516 
517 #if NPF > 0
518 	if (inp->inp_socket->so_state & SS_ISCONNECTED &&
519 	    so->so_proto->pr_protocol != IPPROTO_ICMPV6)
520 		pf_mbuf_link_inpcb(m, inp);
521 #endif
522 
523 	error = ip6_output(m, optp, &inp->inp_route, flags,
524 	    inp->inp_moptions6, &inp->inp_seclevel);
525 	if (so->so_proto->pr_protocol == IPPROTO_ICMPV6) {
526 		icmp6stat_inc(icp6s_outhist + type);
527 	} else
528 		rip6stat_inc(rip6s_opackets);
529 
530 	goto freectl;
531 
532  bad:
533 	m_freem(m);
534 
535  freectl:
536 	if (control) {
537 		ip6_clearpktopts(&opt, -1);
538 		m_freem(control);
539 	}
540 	return (error);
541 }
542 
543 /*
544  * Raw IPv6 socket option processing.
545  */
546 int
rip6_ctloutput(int op,struct socket * so,int level,int optname,struct mbuf * m)547 rip6_ctloutput(int op, struct socket *so, int level, int optname,
548     struct mbuf *m)
549 {
550 #ifdef MROUTING
551 	int error;
552 #endif
553 
554 	switch (level) {
555 	case IPPROTO_IPV6:
556 		switch (optname) {
557 #ifdef MROUTING
558 		case MRT6_INIT:
559 		case MRT6_DONE:
560 		case MRT6_ADD_MIF:
561 		case MRT6_DEL_MIF:
562 		case MRT6_ADD_MFC:
563 		case MRT6_DEL_MFC:
564 			if (op == PRCO_SETOPT) {
565 				error = ip6_mrouter_set(optname, so, m);
566 			} else if (op == PRCO_GETOPT)
567 				error = ip6_mrouter_get(optname, so, m);
568 			else
569 				error = EINVAL;
570 			return (error);
571 #endif
572 		case IPV6_CHECKSUM:
573 			return (ip6_raw_ctloutput(op, so, level, optname, m));
574 		default:
575 			return (ip6_ctloutput(op, so, level, optname, m));
576 		}
577 
578 	case IPPROTO_ICMPV6:
579 		/*
580 		 * XXX: is it better to call icmp6_ctloutput() directly
581 		 * from protosw?
582 		 */
583 		return (icmp6_ctloutput(op, so, level, optname, m));
584 
585 	default:
586 		return EINVAL;
587 	}
588 }
589 
590 extern	u_long rip6_sendspace;
591 extern	u_long rip6_recvspace;
592 
593 int
rip6_attach(struct socket * so,int proto,int wait)594 rip6_attach(struct socket *so, int proto, int wait)
595 {
596 	struct inpcb *inp;
597 	int error;
598 
599 	if (so->so_pcb)
600 		panic("%s", __func__);
601 	if ((so->so_state & SS_PRIV) == 0)
602 		return (EACCES);
603 	if (proto < 0 || proto >= IPPROTO_MAX)
604 		return EPROTONOSUPPORT;
605 
606 	if ((error = soreserve(so, rip6_sendspace, rip6_recvspace)))
607 		return error;
608 	NET_ASSERT_LOCKED();
609 	if ((error = in_pcballoc(so, &rawin6pcbtable, wait)))
610 		return error;
611 
612 	inp = sotoinpcb(so);
613 	inp->inp_ipv6.ip6_nxt = proto;
614 	inp->inp_cksum6 = -1;
615 
616 	inp->inp_icmp6filt = malloc(sizeof(struct icmp6_filter), M_PCB,
617 	    wait == M_WAIT ? M_WAITOK : M_NOWAIT);
618 	if (inp->inp_icmp6filt == NULL) {
619 		in_pcbdetach(inp);
620 		return ENOMEM;
621 	}
622 	ICMP6_FILTER_SETPASSALL(inp->inp_icmp6filt);
623 	return 0;
624 }
625 
626 int
rip6_detach(struct socket * so)627 rip6_detach(struct socket *so)
628 {
629 	struct inpcb *inp = sotoinpcb(so);
630 
631 	soassertlocked(so);
632 
633 	if (inp == NULL)
634 		panic("%s", __func__);
635 #ifdef MROUTING
636 	if (so == ip6_mrouter[inp->inp_rtableid])
637 		ip6_mrouter_done(so);
638 #endif
639 	free(inp->inp_icmp6filt, M_PCB, sizeof(struct icmp6_filter));
640 	inp->inp_icmp6filt = NULL;
641 
642 	in_pcbdetach(inp);
643 
644 	return (0);
645 }
646 
647 void
rip6_lock(struct socket * so)648 rip6_lock(struct socket *so)
649 {
650 	struct inpcb *inp = sotoinpcb(so);
651 
652 	NET_ASSERT_LOCKED();
653 	mtx_enter(&inp->inp_mtx);
654 }
655 
656 void
rip6_unlock(struct socket * so)657 rip6_unlock(struct socket *so)
658 {
659 	struct inpcb *inp = sotoinpcb(so);
660 
661 	NET_ASSERT_LOCKED();
662 	mtx_leave(&inp->inp_mtx);
663 }
664 
665 int
rip6_locked(struct socket * so)666 rip6_locked(struct socket *so)
667 {
668 	struct inpcb *inp = sotoinpcb(so);
669 
670 	return mtx_owned(&inp->inp_mtx);
671 }
672 
673 int
rip6_bind(struct socket * so,struct mbuf * nam,struct proc * p)674 rip6_bind(struct socket *so, struct mbuf *nam, struct proc *p)
675 {
676 	struct inpcb *inp = sotoinpcb(so);
677 	struct sockaddr_in6 *addr;
678 	int error;
679 
680 	soassertlocked(so);
681 
682 	if ((error = in6_nam2sin6(nam, &addr)))
683 		return (error);
684 
685 	/*
686 	 * Make sure to not enter in_pcblookup_local(), local ports
687 	 * are non-sensical for raw sockets.
688 	 */
689 	addr->sin6_port = 0;
690 
691 	if ((error = in6_pcbaddrisavail(inp, addr, 0, p)))
692 		return (error);
693 
694 	mtx_enter(&rawin6pcbtable.inpt_mtx);
695 	inp->inp_laddr6 = addr->sin6_addr;
696 	mtx_leave(&rawin6pcbtable.inpt_mtx);
697 
698 	return (0);
699 }
700 
701 int
rip6_connect(struct socket * so,struct mbuf * nam)702 rip6_connect(struct socket *so, struct mbuf *nam)
703 {
704 	struct inpcb *inp = sotoinpcb(so);
705 	struct sockaddr_in6 *addr;
706 	const struct in6_addr *in6a;
707 	int error;
708 
709 	soassertlocked(so);
710 
711 	if ((error = in6_nam2sin6(nam, &addr)))
712 		return (error);
713 
714 	/* Source address selection. XXX: need pcblookup? */
715 	error = in6_pcbselsrc(&in6a, addr, inp, inp->inp_outputopts6);
716 	if (error)
717 		return (error);
718 
719 	mtx_enter(&rawin6pcbtable.inpt_mtx);
720 	inp->inp_laddr6 = *in6a;
721 	inp->inp_faddr6 = addr->sin6_addr;
722 	mtx_leave(&rawin6pcbtable.inpt_mtx);
723 	soisconnected(so);
724 
725 	return (0);
726 }
727 
728 int
rip6_disconnect(struct socket * so)729 rip6_disconnect(struct socket *so)
730 {
731 	struct inpcb *inp = sotoinpcb(so);
732 
733 	soassertlocked(so);
734 
735 	if ((so->so_state & SS_ISCONNECTED) == 0)
736 		return (ENOTCONN);
737 
738 	soisdisconnected(so);
739 	mtx_enter(&rawin6pcbtable.inpt_mtx);
740 	inp->inp_faddr6 = in6addr_any;
741 	mtx_leave(&rawin6pcbtable.inpt_mtx);
742 
743 	return (0);
744 }
745 
746 int
rip6_shutdown(struct socket * so)747 rip6_shutdown(struct socket *so)
748 {
749 	/*
750 	 * Mark the connection as being incapable of further input.
751 	 */
752 	soassertlocked(so);
753 	socantsendmore(so);
754 	return (0);
755 }
756 
757 int
rip6_send(struct socket * so,struct mbuf * m,struct mbuf * nam,struct mbuf * control)758 rip6_send(struct socket *so, struct mbuf *m, struct mbuf *nam,
759 	struct mbuf *control)
760 {
761 	struct inpcb *inp = sotoinpcb(so);
762 	struct sockaddr_in6 dst;
763 	int error;
764 
765 	soassertlocked(so);
766 
767 	/*
768 	 * Ship a packet out. The appropriate raw output
769 	 * routine handles any messaging necessary.
770 	 */
771 
772 	/* always copy sockaddr to avoid overwrites */
773 	memset(&dst, 0, sizeof(dst));
774 	dst.sin6_family = AF_INET6;
775 	dst.sin6_len = sizeof(dst);
776 	if (so->so_state & SS_ISCONNECTED) {
777 		if (nam) {
778 			error = EISCONN;
779 			goto out;
780 		}
781 		dst.sin6_addr = inp->inp_faddr6;
782 	} else {
783 		struct sockaddr_in6 *addr6;
784 
785 		if (nam == NULL) {
786 			error = ENOTCONN;
787 			goto out;
788 		}
789 		if ((error = in6_nam2sin6(nam, &addr6)))
790 			goto out;
791 		dst.sin6_addr = addr6->sin6_addr;
792 		dst.sin6_scope_id = addr6->sin6_scope_id;
793 	}
794 	error = rip6_output(m, so, sin6tosa(&dst), control);
795 	control = NULL;
796 	m = NULL;
797 
798 out:
799 	m_freem(control);
800 	m_freem(m);
801 
802 	return (error);
803 }
804 
805 int
rip6_sysctl_rip6stat(void * oldp,size_t * oldplen,void * newp)806 rip6_sysctl_rip6stat(void *oldp, size_t *oldplen, void *newp)
807 {
808 	struct rip6stat rip6stat;
809 
810 	CTASSERT(sizeof(rip6stat) == rip6s_ncounters * sizeof(uint64_t));
811 	counters_read(rip6counters, (uint64_t *)&rip6stat, rip6s_ncounters,
812 	    NULL);
813 
814 	return (sysctl_rdstruct(oldp, oldplen, newp,
815 	    &rip6stat, sizeof(rip6stat)));
816 }
817 
818 int
rip6_sysctl(int * name,u_int namelen,void * oldp,size_t * oldlenp,void * newp,size_t newlen)819 rip6_sysctl(int *name, u_int namelen, void *oldp, size_t *oldlenp,
820     void *newp, size_t newlen)
821 {
822 	/* All sysctl names at this level are terminal. */
823 	if (namelen != 1)
824 		return ENOTDIR;
825 
826 	switch (name[0]) {
827 	case RIPV6CTL_STATS:
828 		return (rip6_sysctl_rip6stat(oldp, oldlenp, newp));
829 	default:
830 		return (EOPNOTSUPP);
831 	}
832 	/* NOTREACHED */
833 }
834