xref: /dragonfly/sys/netinet/raw_ip.c (revision 984263bc)
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. 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  *	@(#)raw_ip.c	8.7 (Berkeley) 5/15/95
34  * $FreeBSD: src/sys/netinet/raw_ip.c,v 1.64.2.15 2003/01/24 10:52:50 hsu Exp $
35  */
36 
37 #include "opt_inet6.h"
38 #include "opt_ipsec.h"
39 #include "opt_random_ip_id.h"
40 
41 #include <sys/param.h>
42 #include <sys/systm.h>
43 #include <sys/kernel.h>
44 #include <sys/malloc.h>
45 #include <sys/mbuf.h>
46 #include <sys/proc.h>
47 #include <sys/protosw.h>
48 #include <sys/socket.h>
49 #include <sys/socketvar.h>
50 #include <sys/sysctl.h>
51 
52 #include <vm/vm_zone.h>
53 
54 #include <net/if.h>
55 #include <net/route.h>
56 
57 #define _IP_VHL
58 #include <netinet/in.h>
59 #include <netinet/in_systm.h>
60 #include <netinet/ip.h>
61 #include <netinet/in_pcb.h>
62 #include <netinet/in_var.h>
63 #include <netinet/ip_var.h>
64 #include <netinet/ip_mroute.h>
65 
66 #include <netinet/ip_fw.h>
67 #include <netinet/ip_dummynet.h>
68 
69 #ifdef FAST_IPSEC
70 #include <netipsec/ipsec.h>
71 #endif /*FAST_IPSEC*/
72 
73 #ifdef IPSEC
74 #include <netinet6/ipsec.h>
75 #endif /*IPSEC*/
76 
77 struct	inpcbhead ripcb;
78 struct	inpcbinfo ripcbinfo;
79 
80 /* control hooks for ipfw and dummynet */
81 ip_fw_ctl_t *ip_fw_ctl_ptr;
82 ip_dn_ctl_t *ip_dn_ctl_ptr;
83 
84 /*
85  * hooks for multicast routing. They all default to NULL,
86  * so leave them not initialized and rely on BSS being set to 0.
87  */
88 
89 /* The socket used to communicate with the multicast routing daemon.  */
90 struct socket  *ip_mrouter;
91 
92 /* The various mrouter and rsvp functions */
93 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
94 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
95 int (*ip_mrouter_done)(void);
96 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
97 		struct ip_moptions *);
98 int (*mrt_ioctl)(int, caddr_t);
99 int (*legal_vif_num)(int);
100 u_long (*ip_mcast_src)(int);
101 
102 void (*rsvp_input_p)(struct mbuf *m, int off, int proto);
103 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
104 void (*ip_rsvp_force_done)(struct socket *);
105 
106 /*
107  * Nominal space allocated to a raw ip socket.
108  */
109 #define	RIPSNDQ		8192
110 #define	RIPRCVQ		8192
111 
112 /*
113  * Raw interface to IP protocol.
114  */
115 
116 /*
117  * Initialize raw connection block queue.
118  */
119 void
120 rip_init(void)
121 {
122 	LIST_INIT(&ripcb);
123 	ripcbinfo.listhead = &ripcb;
124 	/*
125 	 * XXX We don't use the hash list for raw IP, but it's easier
126 	 * to allocate a one entry hash list than it is to check all
127 	 * over the place for hashbase == NULL.
128 	 */
129 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
130 	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
131 	ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb),
132 				   maxsockets, ZONE_INTERRUPT, 0);
133 }
134 
135 /*
136  * XXX ripsrc is modified in rip_input, so we must be fix this
137  * when we want to make this code smp-friendly.
138  */
139 static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
140 
141 /*
142  * Setup generic address and protocol structures
143  * for raw_input routine, then pass them along with
144  * mbuf chain.
145  */
146 void
147 rip_input(struct mbuf *m, int off, int proto)
148 {
149 	struct ip *ip = mtod(m, struct ip *);
150 	struct inpcb *inp;
151 	struct inpcb *last = NULL;
152 	struct mbuf *opts = NULL;
153 
154 	ripsrc.sin_addr = ip->ip_src;
155 	LIST_FOREACH(inp, &ripcb, inp_list) {
156 #ifdef INET6
157 		if ((inp->inp_vflag & INP_IPV4) == 0)
158 			continue;
159 #endif
160 		if (inp->inp_ip_p && inp->inp_ip_p != proto)
161 			continue;
162 		if (inp->inp_laddr.s_addr != INADDR_ANY &&
163 		    inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
164 			continue;
165 		if (inp->inp_faddr.s_addr != INADDR_ANY &&
166 		    inp->inp_faddr.s_addr != ip->ip_src.s_addr)
167 			continue;
168 		if (last) {
169 			struct mbuf *n = m_copypacket(m, M_DONTWAIT);
170 
171 #ifdef IPSEC
172 			/* check AH/ESP integrity. */
173 			if (n && ipsec4_in_reject_so(n, last->inp_socket)) {
174 				m_freem(n);
175 				ipsecstat.in_polvio++;
176 				/* do not inject data to pcb */
177 			} else
178 #endif /*IPSEC*/
179 #ifdef FAST_IPSEC
180 			/* check AH/ESP integrity. */
181 			if (ipsec4_in_reject(n, last)) {
182 				m_freem(n);
183 				/* do not inject data to pcb */
184 			} else
185 #endif /*FAST_IPSEC*/
186 			if (n) {
187 				if (last->inp_flags & INP_CONTROLOPTS ||
188 				    last->inp_socket->so_options & SO_TIMESTAMP)
189 				    ip_savecontrol(last, &opts, ip, n);
190 				if (sbappendaddr(&last->inp_socket->so_rcv,
191 				    (struct sockaddr *)&ripsrc, n,
192 				    opts) == 0) {
193 					/* should notify about lost packet */
194 					m_freem(n);
195 					if (opts)
196 					    m_freem(opts);
197 				} else
198 					sorwakeup(last->inp_socket);
199 				opts = 0;
200 			}
201 		}
202 		last = inp;
203 	}
204 #ifdef IPSEC
205 	/* check AH/ESP integrity. */
206 	if (last && ipsec4_in_reject_so(m, last->inp_socket)) {
207 		m_freem(m);
208 		ipsecstat.in_polvio++;
209 		ipstat.ips_delivered--;
210 		/* do not inject data to pcb */
211 	} else
212 #endif /*IPSEC*/
213 #ifdef FAST_IPSEC
214 	/* check AH/ESP integrity. */
215 	if (last && ipsec4_in_reject(m, last)) {
216 		m_freem(m);
217 		ipstat.ips_delivered--;
218 		/* do not inject data to pcb */
219 	} else
220 #endif /*FAST_IPSEC*/
221 	if (last) {
222 		if (last->inp_flags & INP_CONTROLOPTS ||
223 		    last->inp_socket->so_options & SO_TIMESTAMP)
224 			ip_savecontrol(last, &opts, ip, m);
225 		if (sbappendaddr(&last->inp_socket->so_rcv,
226 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
227 			m_freem(m);
228 			if (opts)
229 			    m_freem(opts);
230 		} else
231 			sorwakeup(last->inp_socket);
232 	} else {
233 		m_freem(m);
234 		ipstat.ips_noproto++;
235 		ipstat.ips_delivered--;
236 	}
237 }
238 
239 /*
240  * Generate IP header and pass packet to ip_output.
241  * Tack on options user may have setup with control call.
242  */
243 int
244 rip_output(struct mbuf *m, struct socket *so, u_long dst)
245 {
246 	struct ip *ip;
247 	struct inpcb *inp = sotoinpcb(so);
248 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
249 
250 	/*
251 	 * If the user handed us a complete IP packet, use it.
252 	 * Otherwise, allocate an mbuf for a header and fill it in.
253 	 */
254 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
255 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
256 			m_freem(m);
257 			return(EMSGSIZE);
258 		}
259 		M_PREPEND(m, sizeof(struct ip), M_WAIT);
260 		ip = mtod(m, struct ip *);
261 		ip->ip_tos = inp->inp_ip_tos;
262 		ip->ip_off = 0;
263 		ip->ip_p = inp->inp_ip_p;
264 		ip->ip_len = m->m_pkthdr.len;
265 		ip->ip_src = inp->inp_laddr;
266 		ip->ip_dst.s_addr = dst;
267 		ip->ip_ttl = inp->inp_ip_ttl;
268 	} else {
269 		if (m->m_pkthdr.len > IP_MAXPACKET) {
270 			m_freem(m);
271 			return(EMSGSIZE);
272 		}
273 		ip = mtod(m, struct ip *);
274 		/* don't allow both user specified and setsockopt options,
275 		   and don't allow packet length sizes that will crash */
276 		if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
277 		     && inp->inp_options)
278 		    || (ip->ip_len > m->m_pkthdr.len)
279 		    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
280 			m_freem(m);
281 			return EINVAL;
282 		}
283 		if (ip->ip_id == 0)
284 #ifdef RANDOM_IP_ID
285 			ip->ip_id = ip_randomid();
286 #else
287 			ip->ip_id = htons(ip_id++);
288 #endif
289 		/* XXX prevent ip_output from overwriting header fields */
290 		flags |= IP_RAWOUTPUT;
291 		ipstat.ips_rawout++;
292 	}
293 
294 	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
295 			  inp->inp_moptions, inp));
296 }
297 
298 /*
299  * Raw IP socket option processing.
300  */
301 int
302 rip_ctloutput(struct socket *so, struct sockopt *sopt)
303 {
304 	struct	inpcb *inp = sotoinpcb(so);
305 	int	error, optval;
306 
307 	if (sopt->sopt_level != IPPROTO_IP)
308 		return (EINVAL);
309 
310 	error = 0;
311 
312 	switch (sopt->sopt_dir) {
313 	case SOPT_GET:
314 		switch (sopt->sopt_name) {
315 		case IP_HDRINCL:
316 			optval = inp->inp_flags & INP_HDRINCL;
317 			error = sooptcopyout(sopt, &optval, sizeof optval);
318 			break;
319 
320 		case IP_FW_ADD: /* ADD actually returns the body... */
321 		case IP_FW_GET:
322 			if (IPFW_LOADED)
323 				error = ip_fw_ctl_ptr(sopt);
324 			else
325 				error = ENOPROTOOPT;
326 			break;
327 
328 		case IP_DUMMYNET_GET:
329 			if (DUMMYNET_LOADED)
330 				error = ip_dn_ctl_ptr(sopt);
331 			else
332 				error = ENOPROTOOPT;
333 			break ;
334 
335 		case MRT_INIT:
336 		case MRT_DONE:
337 		case MRT_ADD_VIF:
338 		case MRT_DEL_VIF:
339 		case MRT_ADD_MFC:
340 		case MRT_DEL_MFC:
341 		case MRT_VERSION:
342 		case MRT_ASSERT:
343 			error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
344 				EOPNOTSUPP;
345 			break;
346 
347 		default:
348 			error = ip_ctloutput(so, sopt);
349 			break;
350 		}
351 		break;
352 
353 	case SOPT_SET:
354 		switch (sopt->sopt_name) {
355 		case IP_HDRINCL:
356 			error = sooptcopyin(sopt, &optval, sizeof optval,
357 					    sizeof optval);
358 			if (error)
359 				break;
360 			if (optval)
361 				inp->inp_flags |= INP_HDRINCL;
362 			else
363 				inp->inp_flags &= ~INP_HDRINCL;
364 			break;
365 
366 		case IP_FW_ADD:
367 		case IP_FW_DEL:
368 		case IP_FW_FLUSH:
369 		case IP_FW_ZERO:
370 		case IP_FW_RESETLOG:
371 			if (IPFW_LOADED)
372 				error = ip_fw_ctl_ptr(sopt);
373 			else
374 				error = ENOPROTOOPT;
375 			break;
376 
377 		case IP_DUMMYNET_CONFIGURE:
378 		case IP_DUMMYNET_DEL:
379 		case IP_DUMMYNET_FLUSH:
380 			if (DUMMYNET_LOADED)
381 				error = ip_dn_ctl_ptr(sopt);
382 			else
383 				error = ENOPROTOOPT ;
384 			break ;
385 
386 		case IP_RSVP_ON:
387 			error = ip_rsvp_init(so);
388 			break;
389 
390 		case IP_RSVP_OFF:
391 			error = ip_rsvp_done();
392 			break;
393 
394 		case IP_RSVP_VIF_ON:
395 		case IP_RSVP_VIF_OFF:
396 			error = ip_rsvp_vif ?
397 				ip_rsvp_vif(so, sopt) : EINVAL;
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 			error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
409 					EOPNOTSUPP;
410 			break;
411 
412 		default:
413 			error = ip_ctloutput(so, sopt);
414 			break;
415 		}
416 		break;
417 	}
418 
419 	return (error);
420 }
421 
422 /*
423  * This function exists solely to receive the PRC_IFDOWN messages which
424  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
425  * and calls in_ifadown() to remove all routes corresponding to that address.
426  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
427  * interface routes.
428  */
429 void
430 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
431 {
432 	struct in_ifaddr *ia;
433 	struct ifnet *ifp;
434 	int err;
435 	int flags;
436 
437 	switch (cmd) {
438 	case PRC_IFDOWN:
439 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
440 			if (ia->ia_ifa.ifa_addr == sa
441 			    && (ia->ia_flags & IFA_ROUTE)) {
442 				/*
443 				 * in_ifscrub kills the interface route.
444 				 */
445 				in_ifscrub(ia->ia_ifp, ia);
446 				/*
447 				 * in_ifadown gets rid of all the rest of
448 				 * the routes.  This is not quite the right
449 				 * thing to do, but at least if we are running
450 				 * a routing process they will come back.
451 				 */
452 				in_ifadown(&ia->ia_ifa, 0);
453 				break;
454 			}
455 		}
456 		break;
457 
458 	case PRC_IFUP:
459 		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
460 			if (ia->ia_ifa.ifa_addr == sa)
461 				break;
462 		}
463 		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
464 			return;
465 		flags = RTF_UP;
466 		ifp = ia->ia_ifa.ifa_ifp;
467 
468 		if ((ifp->if_flags & IFF_LOOPBACK)
469 		    || (ifp->if_flags & IFF_POINTOPOINT))
470 			flags |= RTF_HOST;
471 
472 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
473 		if (err == 0)
474 			ia->ia_flags |= IFA_ROUTE;
475 		break;
476 	}
477 }
478 
479 u_long	rip_sendspace = RIPSNDQ;
480 u_long	rip_recvspace = RIPRCVQ;
481 
482 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
483     &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
484 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
485     &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
486 
487 static int
488 rip_attach(struct socket *so, int proto, struct proc *p)
489 {
490 	struct inpcb *inp;
491 	int error, s;
492 
493 	inp = sotoinpcb(so);
494 	if (inp)
495 		panic("rip_attach");
496 	if (p && (error = suser(p)) != 0)
497 		return error;
498 
499 	error = soreserve(so, rip_sendspace, rip_recvspace);
500 	if (error)
501 		return error;
502 	s = splnet();
503 	error = in_pcballoc(so, &ripcbinfo, p);
504 	splx(s);
505 	if (error)
506 		return error;
507 	inp = (struct inpcb *)so->so_pcb;
508 	inp->inp_vflag |= INP_IPV4;
509 	inp->inp_ip_p = proto;
510 	inp->inp_ip_ttl = ip_defttl;
511 	return 0;
512 }
513 
514 static int
515 rip_detach(struct socket *so)
516 {
517 	struct inpcb *inp;
518 
519 	inp = sotoinpcb(so);
520 	if (inp == 0)
521 		panic("rip_detach");
522 	if (so == ip_mrouter && ip_mrouter_done)
523 		ip_mrouter_done();
524 	if (ip_rsvp_force_done)
525 		ip_rsvp_force_done(so);
526 	if (so == ip_rsvpd)
527 		ip_rsvp_done();
528 	in_pcbdetach(inp);
529 	return 0;
530 }
531 
532 static int
533 rip_abort(struct socket *so)
534 {
535 	soisdisconnected(so);
536 	if (so->so_state & SS_NOFDREF)
537 		return rip_detach(so);
538 	return 0;
539 }
540 
541 static int
542 rip_disconnect(struct socket *so)
543 {
544 	if ((so->so_state & SS_ISCONNECTED) == 0)
545 		return ENOTCONN;
546 	return rip_abort(so);
547 }
548 
549 static int
550 rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
551 {
552 	struct inpcb *inp = sotoinpcb(so);
553 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
554 
555 	if (nam->sa_len != sizeof(*addr))
556 		return EINVAL;
557 
558 	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
559 				    (addr->sin_family != AF_IMPLINK)) ||
560 	    (addr->sin_addr.s_addr != INADDR_ANY &&
561 	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
562 		return EADDRNOTAVAIL;
563 	inp->inp_laddr = addr->sin_addr;
564 	return 0;
565 }
566 
567 static int
568 rip_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
569 {
570 	struct inpcb *inp = sotoinpcb(so);
571 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
572 
573 	if (nam->sa_len != sizeof(*addr))
574 		return EINVAL;
575 	if (TAILQ_EMPTY(&ifnet))
576 		return EADDRNOTAVAIL;
577 	if ((addr->sin_family != AF_INET) &&
578 	    (addr->sin_family != AF_IMPLINK))
579 		return EAFNOSUPPORT;
580 	inp->inp_faddr = addr->sin_addr;
581 	soisconnected(so);
582 	return 0;
583 }
584 
585 static int
586 rip_shutdown(struct socket *so)
587 {
588 	socantsendmore(so);
589 	return 0;
590 }
591 
592 static int
593 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
594 	 struct mbuf *control, struct proc *p)
595 {
596 	struct inpcb *inp = sotoinpcb(so);
597 	u_long dst;
598 
599 	if (so->so_state & SS_ISCONNECTED) {
600 		if (nam) {
601 			m_freem(m);
602 			return EISCONN;
603 		}
604 		dst = inp->inp_faddr.s_addr;
605 	} else {
606 		if (nam == NULL) {
607 			m_freem(m);
608 			return ENOTCONN;
609 		}
610 		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
611 	}
612 	return rip_output(m, so, dst);
613 }
614 
615 static int
616 rip_pcblist(SYSCTL_HANDLER_ARGS)
617 {
618 	int error, i, n, s;
619 	struct inpcb *inp, **inp_list;
620 	inp_gen_t gencnt;
621 	struct xinpgen xig;
622 
623 	/*
624 	 * The process of preparing the TCB list is too time-consuming and
625 	 * resource-intensive to repeat twice on every request.
626 	 */
627 	if (req->oldptr == 0) {
628 		n = ripcbinfo.ipi_count;
629 		req->oldidx = 2 * (sizeof xig)
630 			+ (n + n/8) * sizeof(struct xinpcb);
631 		return 0;
632 	}
633 
634 	if (req->newptr != 0)
635 		return EPERM;
636 
637 	/*
638 	 * OK, now we're committed to doing something.
639 	 */
640 	s = splnet();
641 	gencnt = ripcbinfo.ipi_gencnt;
642 	n = ripcbinfo.ipi_count;
643 	splx(s);
644 
645 	xig.xig_len = sizeof xig;
646 	xig.xig_count = n;
647 	xig.xig_gen = gencnt;
648 	xig.xig_sogen = so_gencnt;
649 	error = SYSCTL_OUT(req, &xig, sizeof xig);
650 	if (error)
651 		return error;
652 
653 	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
654 	if (inp_list == 0)
655 		return ENOMEM;
656 
657 	s = splnet();
658 	for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
659 	     inp = LIST_NEXT(inp, inp_list)) {
660 		if (inp->inp_gencnt <= gencnt)
661 			inp_list[i++] = inp;
662 	}
663 	splx(s);
664 	n = i;
665 
666 	error = 0;
667 	for (i = 0; i < n; i++) {
668 		inp = inp_list[i];
669 		if (inp->inp_gencnt <= gencnt) {
670 			struct xinpcb xi;
671 			xi.xi_len = sizeof xi;
672 			/* XXX should avoid extra copy */
673 			bcopy(inp, &xi.xi_inp, sizeof *inp);
674 			if (inp->inp_socket)
675 				sotoxsocket(inp->inp_socket, &xi.xi_socket);
676 			error = SYSCTL_OUT(req, &xi, sizeof xi);
677 		}
678 	}
679 	if (!error) {
680 		/*
681 		 * Give the user an updated idea of our state.
682 		 * If the generation differs from what we told
683 		 * her before, she knows that something happened
684 		 * while we were processing this request, and it
685 		 * might be necessary to retry.
686 		 */
687 		s = splnet();
688 		xig.xig_gen = ripcbinfo.ipi_gencnt;
689 		xig.xig_sogen = so_gencnt;
690 		xig.xig_count = ripcbinfo.ipi_count;
691 		splx(s);
692 		error = SYSCTL_OUT(req, &xig, sizeof xig);
693 	}
694 	free(inp_list, M_TEMP);
695 	return error;
696 }
697 
698 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
699 	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
700 
701 struct pr_usrreqs rip_usrreqs = {
702 	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
703 	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
704 	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
705 	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
706 	in_setsockaddr, sosend, soreceive, sopoll
707 };
708