xref: /dragonfly/sys/netinet/raw_ip.c (revision 896f2e3a)
1 /*
2  * Copyright (c) 1982, 1986, 1988, 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. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
30  * $FreeBSD: src/sys/netinet/raw_ip.c,v 1.64.2.16 2003/08/24 08:24:38 hsu Exp $
31  */
32 
33 #include "opt_inet6.h"
34 #include "opt_ipsec.h"
35 #include "opt_carp.h"
36 
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/kernel.h>
40 #include <sys/jail.h>
41 #include <sys/malloc.h>
42 #include <sys/mbuf.h>
43 #include <sys/proc.h>
44 #include <sys/priv.h>
45 #include <sys/protosw.h>
46 #include <sys/socket.h>
47 #include <sys/socketvar.h>
48 #include <sys/sysctl.h>
49 
50 #include <sys/thread2.h>
51 #include <sys/socketvar2.h>
52 #include <sys/msgport2.h>
53 
54 #include <machine/stdarg.h>
55 
56 #include <net/if.h>
57 #ifdef CARP
58 #include <net/if_types.h>
59 #endif
60 #include <net/route.h>
61 
62 #define _IP_VHL
63 #include <netinet/in.h>
64 #include <netinet/in_systm.h>
65 #include <netinet/ip.h>
66 #include <netinet/in_pcb.h>
67 #include <netinet/in_var.h>
68 #include <netinet/ip_var.h>
69 
70 #include <net/ip_mroute/ip_mroute.h>
71 #include <net/ipfw/ip_fw.h>
72 #include <net/ipfw3/ip_fw.h>
73 #include <net/dummynet/ip_dummynet.h>
74 #include <net/dummynet3/ip_dummynet3.h>
75 
76 #ifdef FAST_IPSEC
77 #include <netproto/ipsec/ipsec.h>
78 #endif /*FAST_IPSEC*/
79 
80 #ifdef IPSEC
81 #include <netinet6/ipsec.h>
82 #endif /*IPSEC*/
83 
84 struct	inpcbinfo ripcbinfo;
85 struct	inpcbportinfo ripcbportinfo;
86 
87 /* control hooks for ipfw and dummynet */
88 ip_fw_ctl_t *ip_fw_ctl_ptr;
89 ip_fw_ctl_t *ip_fw_ctl_x_ptr;
90 ip_dn_ctl_t *ip_dn_ctl_ptr;
91 
92 
93 /*
94  * hooks for multicast routing. They all default to NULL,
95  * so leave them not initialized and rely on BSS being set to 0.
96  */
97 
98 /* The socket used to communicate with the multicast routing daemon.  */
99 struct socket  *ip_mrouter;
100 
101 /* The various mrouter and rsvp functions */
102 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
103 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
104 int (*ip_mrouter_done)(void);
105 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
106 		struct ip_moptions *);
107 int (*mrt_ioctl)(int, caddr_t);
108 int (*legal_vif_num)(int);
109 u_long (*ip_mcast_src)(int);
110 
111 int (*rsvp_input_p)(struct mbuf **, int *, int);
112 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
113 void (*ip_rsvp_force_done)(struct socket *);
114 
115 /*
116  * Nominal space allocated to a raw ip socket.
117  */
118 #define	RIPSNDQ		8192
119 #define	RIPRCVQ		8192
120 
121 /*
122  * Raw interface to IP protocol.
123  */
124 
125 /*
126  * Initialize raw connection block queue.
127  */
128 void
129 rip_init(void)
130 {
131 	in_pcbinfo_init(&ripcbinfo, 0, FALSE);
132 	in_pcbportinfo_init(&ripcbportinfo, 1, FALSE, 0);
133 	/*
134 	 * XXX We don't use the hash list for raw IP, but it's easier
135 	 * to allocate a one entry hash list than it is to check all
136 	 * over the place for hashbase == NULL.
137 	 */
138 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
139 	ripcbinfo.portinfo = &ripcbportinfo;
140 	ripcbinfo.wildcardhashbase = hashinit(1, M_PCB,
141 					      &ripcbinfo.wildcardhashmask);
142 	ripcbinfo.ipi_size = sizeof(struct inpcb);
143 }
144 
145 /*
146  * Setup generic address and protocol structures
147  * for raw_input routine, then pass them along with
148  * mbuf chain.
149  */
150 int
151 rip_input(struct mbuf **mp, int *offp, int proto)
152 {
153 	struct sockaddr_in ripsrc = { sizeof ripsrc, AF_INET };
154 	struct mbuf *m = *mp;
155 	struct ip *ip = mtod(m, struct ip *);
156 	struct inpcb *inp;
157 	struct inpcb *last = NULL;
158 	struct mbuf *opts = NULL;
159 
160 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
161 	    ("not in netisr0"));
162 
163 	*mp = NULL;
164 
165 	ripsrc.sin_addr = ip->ip_src;
166 	LIST_FOREACH(inp, &ripcbinfo.pcblisthead, inp_list) {
167 		if (inp->inp_flags & INP_PLACEMARKER)
168 			continue;
169 #ifdef INET6
170 		if (!INP_ISIPV4(inp))
171 			continue;
172 #endif
173 		if (inp->inp_ip_p && inp->inp_ip_p != proto)
174 			continue;
175 		if (inp->inp_laddr.s_addr != INADDR_ANY &&
176 		    inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
177 			continue;
178 		if (inp->inp_faddr.s_addr != INADDR_ANY &&
179 		    inp->inp_faddr.s_addr != ip->ip_src.s_addr)
180 			continue;
181 		if (last) {
182 			struct mbuf *n = m_copypacket(m, M_NOWAIT);
183 
184 #ifdef IPSEC
185 			/* check AH/ESP integrity. */
186 			if (n && ipsec4_in_reject_so(n, last->inp_socket)) {
187 				m_freem(n);
188 				ipsecstat.in_polvio++;
189 				/* do not inject data to pcb */
190 			} else
191 #endif /*IPSEC*/
192 #ifdef FAST_IPSEC
193 			/* check AH/ESP integrity. */
194 			if (ipsec4_in_reject(n, last)) {
195 				m_freem(n);
196 				/* do not inject data to pcb */
197 			} else
198 #endif /*FAST_IPSEC*/
199 			if (n) {
200 				lwkt_gettoken(&last->inp_socket->so_rcv.ssb_token);
201 				if (last->inp_flags & INP_CONTROLOPTS ||
202 				    last->inp_socket->so_options & SO_TIMESTAMP)
203 				    ip_savecontrol(last, &opts, ip, n);
204 				if (ssb_appendaddr(&last->inp_socket->so_rcv,
205 					    (struct sockaddr *)&ripsrc, n,
206 					    opts) == 0) {
207 					/* should notify about lost packet */
208 					m_freem(n);
209 					if (opts)
210 					    m_freem(opts);
211 				} else {
212 					sorwakeup(last->inp_socket);
213 				}
214 				lwkt_reltoken(&last->inp_socket->so_rcv.ssb_token);
215 				opts = NULL;
216 			}
217 		}
218 		last = inp;
219 	}
220 #ifdef IPSEC
221 	/* check AH/ESP integrity. */
222 	if (last && ipsec4_in_reject_so(m, last->inp_socket)) {
223 		m_freem(m);
224 		ipsecstat.in_polvio++;
225 		ipstat.ips_delivered--;
226 		/* do not inject data to pcb */
227 	} else
228 #endif /*IPSEC*/
229 #ifdef FAST_IPSEC
230 	/* check AH/ESP integrity. */
231 	if (last && ipsec4_in_reject(m, last)) {
232 		m_freem(m);
233 		ipstat.ips_delivered--;
234 		/* do not inject data to pcb */
235 	} else
236 #endif /*FAST_IPSEC*/
237 	/* Check the minimum TTL for socket. */
238 	if (last && ip->ip_ttl < last->inp_ip_minttl) {
239 		m_freem(opts);
240 		ipstat.ips_delivered--;
241 	} else if (last) {
242 		if (last->inp_flags & INP_CONTROLOPTS ||
243 		    last->inp_socket->so_options & SO_TIMESTAMP)
244 			ip_savecontrol(last, &opts, ip, m);
245 		lwkt_gettoken(&last->inp_socket->so_rcv.ssb_token);
246 		if (ssb_appendaddr(&last->inp_socket->so_rcv,
247 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
248 			m_freem(m);
249 			if (opts)
250 			    m_freem(opts);
251 		} else {
252 			sorwakeup(last->inp_socket);
253 		}
254 		lwkt_reltoken(&last->inp_socket->so_rcv.ssb_token);
255 	} else {
256 		m_freem(m);
257 		ipstat.ips_noproto++;
258 		ipstat.ips_delivered--;
259 	}
260 	return(IPPROTO_DONE);
261 }
262 
263 /*
264  * Generate IP header and pass packet to ip_output.
265  * Tack on options user may have setup with control call.
266  */
267 int
268 rip_output(struct mbuf *m, struct socket *so, ...)
269 {
270 	struct ip *ip;
271 	struct inpcb *inp = so->so_pcb;
272 	__va_list ap;
273 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
274 	u_long dst;
275 
276 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
277 	    ("not in netisr0"));
278 
279 	__va_start(ap, so);
280 	dst = __va_arg(ap, u_long);
281 	__va_end(ap);
282 
283 	/*
284 	 * If the user handed us a complete IP packet, use it.
285 	 * Otherwise, allocate an mbuf for a header and fill it in.
286 	 */
287 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
288 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
289 			m_freem(m);
290 			return(EMSGSIZE);
291 		}
292 		M_PREPEND(m, sizeof(struct ip), M_WAITOK);
293 		if (m == NULL)
294 			return(ENOBUFS);
295 		ip = mtod(m, struct ip *);
296 		ip->ip_tos = inp->inp_ip_tos;
297 		ip->ip_off = 0;
298 		ip->ip_p = inp->inp_ip_p;
299 		ip->ip_len = m->m_pkthdr.len;
300 		ip->ip_src = inp->inp_laddr;
301 		ip->ip_dst.s_addr = dst;
302 		ip->ip_ttl = inp->inp_ip_ttl;
303 	} else {
304 		int hlen;
305 
306 		if (m->m_pkthdr.len > IP_MAXPACKET) {
307 			m_freem(m);
308 			return(EMSGSIZE);
309 		}
310 		if (m->m_len < sizeof(struct ip)) {
311 			m = m_pullup(m, sizeof(struct ip));
312 			if (m == NULL)
313 				return ENOBUFS;
314 		}
315 		ip = mtod(m, struct ip *);
316 		hlen = IP_VHL_HL(ip->ip_vhl) << 2;
317 
318 		/* Don't allow header length less than the minimum. */
319 		if (hlen < sizeof(struct ip)) {
320 			m_freem(m);
321 			return EINVAL;
322 		}
323 
324 		/*
325 		 * Don't allow both user specified and setsockopt options.
326 		 * Don't allow packet length sizes that will crash.
327 		 */
328 		if ((hlen != sizeof(struct ip) && inp->inp_options) ||
329 		    ip->ip_len > m->m_pkthdr.len || ip->ip_len < hlen) {
330 			m_freem(m);
331 			return EINVAL;
332 		}
333 		if (ip->ip_id == 0)
334 			ip->ip_id = ip_newid();
335 
336 		/* Prevent ip_output from overwriting header fields */
337 		flags |= IP_RAWOUTPUT;
338 		ipstat.ips_rawout++;
339 	}
340 
341 	return ip_output(m, inp->inp_options, &inp->inp_route, flags,
342 			 inp->inp_moptions, inp);
343 }
344 
345 /*
346  * Raw IP socket option processing.
347  */
348 void
349 rip_ctloutput(netmsg_t msg)
350 {
351 	struct socket *so = msg->base.nm_so;
352 	struct sockopt *sopt = msg->ctloutput.nm_sopt;
353 	struct	inpcb *inp = so->so_pcb;
354 	int	error, optval;
355 
356 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
357 	    ("not in netisr0"));
358 
359 	error = 0;
360 
361 	/* Get socket's owner cpuid hint */
362 	if (sopt->sopt_level == SOL_SOCKET &&
363 	    sopt->sopt_dir == SOPT_GET &&
364 	    sopt->sopt_name == SO_CPUHINT) {
365 		optval = mycpuid;
366 		soopt_from_kbuf(sopt, &optval, sizeof(optval));
367 		goto done;
368 	}
369 
370 	if (sopt->sopt_level != IPPROTO_IP) {
371 		error = EINVAL;
372 		goto done;
373 	}
374 
375 	switch (sopt->sopt_dir) {
376 	case SOPT_GET:
377 		switch (sopt->sopt_name) {
378 		case IP_HDRINCL:
379 			optval = inp->inp_flags & INP_HDRINCL;
380 			soopt_from_kbuf(sopt, &optval, sizeof optval);
381 			break;
382 		case IP_FW_X:
383 			if (IPFW3_LOADED)
384 				error = ip_fw3_sockopt(sopt);
385 			else
386 				error = ENOPROTOOPT;
387 			break;
388 		case IP_FW_ADD: /* ADD actually returns the body... */
389 		case IP_FW_GET:
390 			if (IPFW_LOADED)
391 				error = ip_fw_sockopt(sopt);
392 			else
393 				error = ENOPROTOOPT;
394 			break;
395 
396 		case IP_DUMMYNET_GET:
397 			error = ip_dn_sockopt(sopt);
398 			break ;
399 
400 		case MRT_INIT:
401 		case MRT_DONE:
402 		case MRT_ADD_VIF:
403 		case MRT_DEL_VIF:
404 		case MRT_ADD_MFC:
405 		case MRT_DEL_MFC:
406 		case MRT_VERSION:
407 		case MRT_ASSERT:
408 		case MRT_API_SUPPORT:
409 		case MRT_API_CONFIG:
410 		case MRT_ADD_BW_UPCALL:
411 		case MRT_DEL_BW_UPCALL:
412 			error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
413 				EOPNOTSUPP;
414 			break;
415 
416 		default:
417 			ip_ctloutput(msg);
418 			/* msg invalid now */
419 			return;
420 		}
421 		break;
422 
423 	case SOPT_SET:
424 		switch (sopt->sopt_name) {
425 		case IP_HDRINCL:
426 			error = soopt_to_kbuf(sopt, &optval, sizeof optval,
427 					      sizeof optval);
428 			if (error)
429 				break;
430 			if (optval)
431 				inp->inp_flags |= INP_HDRINCL;
432 			else
433 				inp->inp_flags &= ~INP_HDRINCL;
434 			break;
435 		case IP_FW_X:
436 			if (IPFW3_LOADED)
437 				error = ip_fw_ctl_x_ptr(sopt);
438 			else
439 				 error = ENOPROTOOPT;
440 			break;
441 		case IP_FW_ADD:
442 		case IP_FW_DEL:
443 		case IP_FW_FLUSH:
444 		case IP_FW_ZERO:
445 		case IP_FW_RESETLOG:
446 			if (IPFW_LOADED)
447 				error = ip_fw_ctl_ptr(sopt);
448 			else
449 				error = ENOPROTOOPT;
450 			break;
451 
452 		case IP_DUMMYNET_CONFIGURE:
453 		case IP_DUMMYNET_DEL:
454 		case IP_DUMMYNET_FLUSH:
455 			error = ip_dn_sockopt(sopt);
456 			break ;
457 
458 		case IP_RSVP_ON:
459 			error = ip_rsvp_init(so);
460 			break;
461 
462 		case IP_RSVP_OFF:
463 			error = ip_rsvp_done();
464 			break;
465 
466 		case IP_RSVP_VIF_ON:
467 		case IP_RSVP_VIF_OFF:
468 			error = ip_rsvp_vif ?
469 				ip_rsvp_vif(so, sopt) : EINVAL;
470 			break;
471 
472 		case MRT_INIT:
473 		case MRT_DONE:
474 		case MRT_ADD_VIF:
475 		case MRT_DEL_VIF:
476 		case MRT_ADD_MFC:
477 		case MRT_DEL_MFC:
478 		case MRT_VERSION:
479 		case MRT_ASSERT:
480 		case MRT_API_SUPPORT:
481 		case MRT_API_CONFIG:
482 		case MRT_ADD_BW_UPCALL:
483 		case MRT_DEL_BW_UPCALL:
484 			error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
485 					EOPNOTSUPP;
486 			break;
487 
488 		default:
489 			ip_ctloutput(msg);
490 			/* msg invalid now */
491 			return;
492 		}
493 		break;
494 	}
495 done:
496 	lwkt_replymsg(&msg->lmsg, error);
497 }
498 
499 /*
500  * This function exists solely to receive the PRC_IFDOWN messages which
501  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
502  * and calls in_ifadown() to remove all routes corresponding to that address.
503  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
504  * interface routes.
505  */
506 void
507 rip_ctlinput(netmsg_t msg)
508 {
509 	int cmd = msg->ctlinput.nm_cmd;
510 	struct sockaddr *sa = msg->ctlinput.nm_arg;
511 	struct in_ifaddr *ia;
512 	struct in_ifaddr_container *iac;
513 	struct ifnet *ifp;
514 	int err;
515 	int flags;
516 
517 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
518 	    ("not in netisr0"));
519 
520 	switch (cmd) {
521 	case PRC_IFDOWN:
522 		TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) {
523 			ia = iac->ia;
524 
525 			if (ia->ia_ifa.ifa_addr == sa &&
526 			    (ia->ia_flags & IFA_ROUTE)) {
527 				/*
528 				 * in_ifscrub kills the interface route.
529 				 */
530 				in_ifscrub(ia->ia_ifp, ia);
531 				/*
532 				 * in_ifadown gets rid of all the rest of
533 				 * the routes.  This is not quite the right
534 				 * thing to do, but at least if we are running
535 				 * a routing process they will come back.
536 				 */
537 				in_ifadown(&ia->ia_ifa, 0);
538 				break;
539 			}
540 		}
541 		break;
542 
543 	case PRC_IFUP:
544 		ia = NULL;
545 		TAILQ_FOREACH(iac, &in_ifaddrheads[mycpuid], ia_link) {
546 			if (iac->ia->ia_ifa.ifa_addr == sa) {
547 				ia = iac->ia;
548 				break;
549 			}
550 		}
551 		if (ia == NULL || (ia->ia_flags & IFA_ROUTE))
552 			goto done;
553 		flags = RTF_UP;
554 		ifp = ia->ia_ifa.ifa_ifp;
555 
556 #ifdef CARP
557 		/*
558 		 * Don't add prefix routes for CARP interfaces.
559 		 * Prefix routes creation is handled by CARP
560 		 * interfaces themselves.
561 		 */
562 		if (ifp->if_type == IFT_CARP)
563 			goto done;
564 #endif
565 
566 		if ((ifp->if_flags & IFF_LOOPBACK) ||
567 		    (ifp->if_flags & IFF_POINTOPOINT))
568 			flags |= RTF_HOST;
569 
570 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
571 		if (err == 0)
572 			ia->ia_flags |= IFA_ROUTE;
573 		break;
574 	}
575 done:
576 	lwkt_replymsg(&msg->lmsg, 0);
577 }
578 
579 u_long	rip_sendspace = RIPSNDQ;
580 u_long	rip_recvspace = RIPRCVQ;
581 
582 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
583     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
584 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
585     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
586 
587 static void
588 rip_attach(netmsg_t msg)
589 {
590 	struct socket *so = msg->base.nm_so;
591 	int proto = msg->attach.nm_proto;
592 	struct pru_attach_info *ai = msg->attach.nm_ai;
593 	struct inpcb *inp;
594 	int error;
595 
596 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
597 	    ("not in netisr0"));
598 
599 	inp = so->so_pcb;
600 	if (inp)
601 		panic("rip_attach");
602 	error = priv_check_cred(ai->p_ucred, PRIV_NETINET_RAW, NULL_CRED_OKAY);
603 	if (error)
604 		goto done;
605 
606 	error = soreserve(so, rip_sendspace, rip_recvspace, ai->sb_rlimit);
607 	if (error)
608 		goto done;
609 
610 	error = in_pcballoc(so, &ripcbinfo);
611 	if (error == 0) {
612 		inp = (struct inpcb *)so->so_pcb;
613 		inp->inp_ip_p = proto;
614 		inp->inp_ip_ttl = ip_defttl;
615 	}
616 done:
617 	lwkt_replymsg(&msg->lmsg, error);
618 }
619 
620 static void
621 rip_detach(netmsg_t msg)
622 {
623 	struct socket *so = msg->base.nm_so;
624 	struct inpcb *inp;
625 
626 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
627 	    ("not in netisr0"));
628 
629 	inp = so->so_pcb;
630 	if (inp == NULL)
631 		panic("rip_detach");
632 	if (so == ip_mrouter && ip_mrouter_done)
633 		ip_mrouter_done();
634 	if (ip_rsvp_force_done)
635 		ip_rsvp_force_done(so);
636 	if (so == ip_rsvpd)
637 		ip_rsvp_done();
638 	in_pcbdetach(inp);
639 	lwkt_replymsg(&msg->lmsg, 0);
640 }
641 
642 static void
643 rip_abort(netmsg_t msg)
644 {
645 	/*
646 	 * Raw socket does not support listen(2),
647 	 * so this should never be called.
648 	 */
649 	panic("rip_abort is called");
650 }
651 
652 static void
653 rip_disconnect(netmsg_t msg)
654 {
655 	struct socket *so = msg->base.nm_so;
656 	int error;
657 
658 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
659 	    ("not in netisr0"));
660 
661 	if (so->so_state & SS_ISCONNECTED) {
662 		soisdisconnected(so);
663 		error = 0;
664 	} else {
665 		error = ENOTCONN;
666 	}
667 	lwkt_replymsg(&msg->lmsg, error);
668 }
669 
670 static void
671 rip_bind(netmsg_t msg)
672 {
673 	struct socket *so = msg->base.nm_so;
674 	struct sockaddr *nam = msg->bind.nm_nam;
675 	struct inpcb *inp = so->so_pcb;
676 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
677 	int error;
678 
679 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
680 	    ("not in netisr0"));
681 
682 	if (nam->sa_len == sizeof(*addr)) {
683 		if (ifnet_array_isempty() ||
684 		    ((addr->sin_family != AF_INET) &&
685 		     (addr->sin_family != AF_IMPLINK)) ||
686 		    (addr->sin_addr.s_addr != INADDR_ANY &&
687 		     ifa_ifwithaddr((struct sockaddr *)addr) == 0)) {
688 			error = EADDRNOTAVAIL;
689 		} else {
690 			inp->inp_laddr = addr->sin_addr;
691 			error = 0;
692 		}
693 	} else {
694 		error = EINVAL;
695 	}
696 	lwkt_replymsg(&msg->lmsg, error);
697 }
698 
699 static void
700 rip_connect(netmsg_t msg)
701 {
702 	struct socket *so = msg->base.nm_so;
703 	struct sockaddr *nam = msg->connect.nm_nam;
704 	struct inpcb *inp = so->so_pcb;
705 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
706 	int error;
707 
708 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
709 	    ("not in netisr0"));
710 
711 	if (nam->sa_len != sizeof(*addr)) {
712 		error = EINVAL;
713 	} else if (ifnet_array_isempty()) {
714 		error = EADDRNOTAVAIL;
715 	} else {
716 		if ((addr->sin_family != AF_INET) &&
717 		    (addr->sin_family != AF_IMPLINK)) {
718 			error = EAFNOSUPPORT;
719 		} else {
720 			inp->inp_faddr = addr->sin_addr;
721 			soisconnected(so);
722 			error = 0;
723 		}
724 	}
725 	lwkt_replymsg(&msg->lmsg, error);
726 }
727 
728 static void
729 rip_shutdown(netmsg_t msg)
730 {
731 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
732 	    ("not in netisr0"));
733 
734 	socantsendmore(msg->base.nm_so);
735 	lwkt_replymsg(&msg->lmsg, 0);
736 }
737 
738 static void
739 rip_send(netmsg_t msg)
740 {
741 	struct socket *so = msg->base.nm_so;
742 	struct mbuf *m = msg->send.nm_m;
743 	/*struct mbuf *control = msg->send.nm_control;*/
744 	struct sockaddr *nam = msg->send.nm_addr;
745 	/*int flags = msg->send.nm_flags;*/
746 	struct inpcb *inp = so->so_pcb;
747 	u_long dst;
748 	int error;
749 
750 	KASSERT(&curthread->td_msgport == netisr_cpuport(0),
751 	    ("not in netisr0"));
752 
753 	if (so->so_state & SS_ISCONNECTED) {
754 		if (nam) {
755 			m_freem(m);
756 			error = EISCONN;
757 		} else {
758 			dst = inp->inp_faddr.s_addr;
759 			error = rip_output(m, so, dst);
760 		}
761 	} else {
762 		if (nam == NULL) {
763 			m_freem(m);
764 			error = ENOTCONN;
765 		} else {
766 			dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
767 			error = rip_output(m, so, dst);
768 		}
769 	}
770 	lwkt_replymsg(&msg->lmsg, error);
771 }
772 
773 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, &ripcbinfo, 1,
774 	    in_pcblist_global, "S,xinpcb", "List of active raw IP sockets");
775 
776 struct pr_usrreqs rip_usrreqs = {
777 	.pru_abort = rip_abort,
778 	.pru_accept = pr_generic_notsupp,
779 	.pru_attach = rip_attach,
780 	.pru_bind = rip_bind,
781 	.pru_connect = rip_connect,
782 	.pru_connect2 = pr_generic_notsupp,
783 	.pru_control = in_control_dispatch,
784 	.pru_detach = rip_detach,
785 	.pru_disconnect = rip_disconnect,
786 	.pru_listen = pr_generic_notsupp,
787 	.pru_peeraddr = in_setpeeraddr_dispatch,
788 	.pru_rcvd = pr_generic_notsupp,
789 	.pru_rcvoob = pr_generic_notsupp,
790 	.pru_send = rip_send,
791 	.pru_sense = pru_sense_null,
792 	.pru_shutdown = rip_shutdown,
793 	.pru_sockaddr = in_setsockaddr_dispatch,
794 	.pru_sosend = sosend,
795 	.pru_soreceive = soreceive
796 };
797