xref: /freebsd/sys/netinet/raw_ip.c (revision a8445737e740901f5f2c8d24c12ef7fc8b00134e)
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  *	$Id: raw_ip.c,v 1.54 1998/05/15 20:11:34 wollman Exp $
35  */
36 
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/kernel.h>
40 #include <sys/malloc.h>
41 #include <sys/mbuf.h>
42 #include <sys/proc.h>
43 #include <sys/protosw.h>
44 #include <sys/socket.h>
45 #include <sys/socketvar.h>
46 #include <sys/sysctl.h>
47 
48 #include <vm/vm_zone.h>
49 
50 #include <net/if.h>
51 #include <net/route.h>
52 
53 #define _IP_VHL
54 #include <netinet/in.h>
55 #include <netinet/in_systm.h>
56 #include <netinet/ip.h>
57 #include <netinet/in_pcb.h>
58 #include <netinet/in_var.h>
59 #include <netinet/ip_var.h>
60 #include <netinet/ip_mroute.h>
61 
62 #include <netinet/ip_fw.h>
63 
64 #if !defined(COMPAT_IPFW) || COMPAT_IPFW == 1
65 #undef COMPAT_IPFW
66 #define COMPAT_IPFW 1
67 #else
68 #undef COMPAT_IPFW
69 #endif
70 
71 static struct inpcbhead ripcb;
72 static struct inpcbinfo ripcbinfo;
73 
74 /*
75  * Nominal space allocated to a raw ip socket.
76  */
77 #define	RIPSNDQ		8192
78 #define	RIPRCVQ		8192
79 
80 /*
81  * Raw interface to IP protocol.
82  */
83 
84 /*
85  * Initialize raw connection block q.
86  */
87 void
88 rip_init()
89 {
90 	LIST_INIT(&ripcb);
91 	ripcbinfo.listhead = &ripcb;
92 	/*
93 	 * XXX We don't use the hash list for raw IP, but it's easier
94 	 * to allocate a one entry hash list than it is to check all
95 	 * over the place for hashbase == NULL.
96 	 */
97 	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
98 	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
99 	ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb),
100 				   maxsockets, ZONE_INTERRUPT, 0);
101 }
102 
103 static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
104 /*
105  * Setup generic address and protocol structures
106  * for raw_input routine, then pass them along with
107  * mbuf chain.
108  */
109 void
110 rip_input(m, iphlen)
111 	struct mbuf *m;
112 	int iphlen;
113 {
114 	register struct ip *ip = mtod(m, struct ip *);
115 	register struct inpcb *inp;
116 	struct inpcb *last = 0;
117 	struct mbuf *opts = 0;
118 
119 	ripsrc.sin_addr = ip->ip_src;
120 	for (inp = ripcb.lh_first; inp != NULL; inp = inp->inp_list.le_next) {
121 		if (inp->inp_ip_p && inp->inp_ip_p != ip->ip_p)
122 			continue;
123 		if (inp->inp_laddr.s_addr &&
124                   inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
125 			continue;
126 		if (inp->inp_faddr.s_addr &&
127                   inp->inp_faddr.s_addr != ip->ip_src.s_addr)
128 			continue;
129 		if (last) {
130 			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
131 			if (n) {
132 				if (last->inp_flags & INP_CONTROLOPTS ||
133 				    last->inp_socket->so_options & SO_TIMESTAMP)
134 				    ip_savecontrol(last, &opts, ip, n);
135 				if (sbappendaddr(&last->inp_socket->so_rcv,
136 				    (struct sockaddr *)&ripsrc, n,
137 				    opts) == 0) {
138 					/* should notify about lost packet */
139 					m_freem(n);
140 					if (opts)
141 					    m_freem(opts);
142 				} else
143 					sorwakeup(last->inp_socket);
144 				opts = 0;
145 			}
146 		}
147 		last = inp;
148 	}
149 	if (last) {
150 		if (last->inp_flags & INP_CONTROLOPTS ||
151 		    last->inp_socket->so_options & SO_TIMESTAMP)
152 			ip_savecontrol(last, &opts, ip, m);
153 		if (sbappendaddr(&last->inp_socket->so_rcv,
154 		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
155 			m_freem(m);
156 			if (opts)
157 			    m_freem(opts);
158 		} else
159 			sorwakeup(last->inp_socket);
160 	} else {
161 		m_freem(m);
162               ipstat.ips_noproto++;
163               ipstat.ips_delivered--;
164       }
165 }
166 
167 /*
168  * Generate IP header and pass packet to ip_output.
169  * Tack on options user may have setup with control call.
170  */
171 int
172 rip_output(m, so, dst)
173 	register struct mbuf *m;
174 	struct socket *so;
175 	u_long dst;
176 {
177 	register struct ip *ip;
178 	register struct inpcb *inp = sotoinpcb(so);
179 	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
180 
181 	/*
182 	 * If the user handed us a complete IP packet, use it.
183 	 * Otherwise, allocate an mbuf for a header and fill it in.
184 	 */
185 	if ((inp->inp_flags & INP_HDRINCL) == 0) {
186 		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
187 			m_freem(m);
188 			return(EMSGSIZE);
189 		}
190 		M_PREPEND(m, sizeof(struct ip), M_WAIT);
191 		ip = mtod(m, struct ip *);
192 		ip->ip_tos = 0;
193 		ip->ip_off = 0;
194 		ip->ip_p = inp->inp_ip_p;
195 		ip->ip_len = m->m_pkthdr.len;
196 		ip->ip_src = inp->inp_laddr;
197 		ip->ip_dst.s_addr = dst;
198 		ip->ip_ttl = MAXTTL;
199 	} else {
200 		if (m->m_pkthdr.len > IP_MAXPACKET) {
201 			m_freem(m);
202 			return(EMSGSIZE);
203 		}
204 		ip = mtod(m, struct ip *);
205 		/* don't allow both user specified and setsockopt options,
206 		   and don't allow packet length sizes that will crash */
207 		if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
208 		     && inp->inp_options)
209 		    || (ip->ip_len > m->m_pkthdr.len)
210 		    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
211 			m_freem(m);
212 			return EINVAL;
213 		}
214 		if (ip->ip_id == 0)
215 			ip->ip_id = htons(ip_id++);
216 		/* XXX prevent ip_output from overwriting header fields */
217 		flags |= IP_RAWOUTPUT;
218 		ipstat.ips_rawout++;
219 	}
220 	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
221 			  inp->inp_moptions));
222 }
223 
224 /*
225  * Raw IP socket option processing.
226  */
227 int
228 rip_ctloutput(so, sopt)
229 	struct socket *so;
230 	struct sockopt *sopt;
231 {
232 	struct	inpcb *inp = sotoinpcb(so);
233 	int	error, optval;
234 
235 	if (sopt->sopt_level != IPPROTO_IP)
236 		return (EINVAL);
237 
238 	error = 0;
239 
240 	switch (sopt->sopt_dir) {
241 	case SOPT_GET:
242 		switch (sopt->sopt_name) {
243 		case IP_HDRINCL:
244 			optval = inp->inp_flags & INP_HDRINCL;
245 			error = sooptcopyout(sopt, &optval, sizeof optval);
246 			break;
247 
248 #ifdef COMPAT_IPFW
249 		case IP_FW_GET:
250 			if (ip_fw_ctl_ptr == 0)
251 				error = ENOPROTOOPT;
252 			else
253 				error = ip_fw_ctl_ptr(sopt);
254 			break;
255 
256 		case IP_NAT:
257 			if (ip_nat_ctl_ptr == 0)
258 				error = ENOPROTOOPT;
259 			else
260 				error = ip_nat_ctl_ptr(sopt);
261 			break;
262 #endif /* COMPAT_IPFW */
263 
264 		case MRT_INIT:
265 		case MRT_DONE:
266 		case MRT_ADD_VIF:
267 		case MRT_DEL_VIF:
268 		case MRT_ADD_MFC:
269 		case MRT_DEL_MFC:
270 		case MRT_VERSION:
271 		case MRT_ASSERT:
272 			error = ip_mrouter_get(so, sopt);
273 			break;
274 
275 		default:
276 			error = ip_ctloutput(so, sopt);
277 			break;
278 		}
279 		break;
280 
281 	case SOPT_SET:
282 		switch (sopt->sopt_name) {
283 		case IP_HDRINCL:
284 			error = sooptcopyin(sopt, &optval, sizeof optval,
285 					    sizeof optval);
286 			if (error)
287 				break;
288 			if (optval)
289 				inp->inp_flags |= INP_HDRINCL;
290 			else
291 				inp->inp_flags &= ~INP_HDRINCL;
292 			break;
293 
294 #ifdef COMPAT_IPFW
295 		case IP_FW_ADD:
296 		case IP_FW_DEL:
297 		case IP_FW_FLUSH:
298 		case IP_FW_ZERO:
299 			if (ip_fw_ctl_ptr == 0)
300 				error = ENOPROTOOPT;
301 			else
302 				error = ip_fw_ctl_ptr(sopt);
303 			break;
304 
305 		case IP_NAT:
306 			if (ip_nat_ctl_ptr == 0)
307 				error = ENOPROTOOPT;
308 			else
309 				error = ip_nat_ctl_ptr(sopt);
310 			break;
311 #endif /* COMPAT_IPFW */
312 
313 		case IP_RSVP_ON:
314 			error = ip_rsvp_init(so);
315 			break;
316 
317 		case IP_RSVP_OFF:
318 			error = ip_rsvp_done();
319 			break;
320 
321 			/* XXX - should be combined */
322 		case IP_RSVP_VIF_ON:
323 			error = ip_rsvp_vif_init(so, sopt);
324 			break;
325 
326 		case IP_RSVP_VIF_OFF:
327 			error = ip_rsvp_vif_done(so, sopt);
328 			break;
329 
330 		case MRT_INIT:
331 		case MRT_DONE:
332 		case MRT_ADD_VIF:
333 		case MRT_DEL_VIF:
334 		case MRT_ADD_MFC:
335 		case MRT_DEL_MFC:
336 		case MRT_VERSION:
337 		case MRT_ASSERT:
338 			error = ip_mrouter_set(so, sopt);
339 			break;
340 
341 		default:
342 			error = ip_ctloutput(so, sopt);
343 			break;
344 		}
345 		break;
346 	}
347 
348 	return (error);
349 }
350 
351 /*
352  * This function exists solely to receive the PRC_IFDOWN messages which
353  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
354  * and calls in_ifadown() to remove all routes corresponding to that address.
355  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
356  * interface routes.
357  */
358 void
359 rip_ctlinput(cmd, sa, vip)
360 	int cmd;
361 	struct sockaddr *sa;
362 	void *vip;
363 {
364 	struct in_ifaddr *ia;
365 	struct ifnet *ifp;
366 	int err;
367 	int flags;
368 
369 	switch (cmd) {
370 	case PRC_IFDOWN:
371 		for (ia = in_ifaddrhead.tqh_first; ia;
372 		     ia = ia->ia_link.tqe_next) {
373 			if (ia->ia_ifa.ifa_addr == sa
374 			    && (ia->ia_flags & IFA_ROUTE)) {
375 				/*
376 				 * in_ifscrub kills the interface route.
377 				 */
378 				in_ifscrub(ia->ia_ifp, ia);
379 				/*
380 				 * in_ifadown gets rid of all the rest of
381 				 * the routes.  This is not quite the right
382 				 * thing to do, but at least if we are running
383 				 * a routing process they will come back.
384 				 */
385 				in_ifadown(&ia->ia_ifa);
386 				break;
387 			}
388 		}
389 		break;
390 
391 	case PRC_IFUP:
392 		for (ia = in_ifaddrhead.tqh_first; ia;
393 		     ia = ia->ia_link.tqe_next) {
394 			if (ia->ia_ifa.ifa_addr == sa)
395 				break;
396 		}
397 		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
398 			return;
399 		flags = RTF_UP;
400 		ifp = ia->ia_ifa.ifa_ifp;
401 
402 		if ((ifp->if_flags & IFF_LOOPBACK)
403 		    || (ifp->if_flags & IFF_POINTOPOINT))
404 			flags |= RTF_HOST;
405 
406 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
407 		if (err == 0)
408 			ia->ia_flags |= IFA_ROUTE;
409 		break;
410 	}
411 }
412 
413 static u_long	rip_sendspace = RIPSNDQ;
414 static u_long	rip_recvspace = RIPRCVQ;
415 
416 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, &rip_sendspace,
417 	   0, "");
418 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, &rip_recvspace,
419 	   0, "");
420 
421 static int
422 rip_attach(struct socket *so, int proto, struct proc *p)
423 {
424 	struct inpcb *inp;
425 	int error, s;
426 
427 	inp = sotoinpcb(so);
428 	if (inp)
429 		panic("rip_attach");
430 	if (p && (error = suser(p->p_ucred, &p->p_acflag)) != 0)
431 		return error;
432 
433 	s = splnet();
434 	error = in_pcballoc(so, &ripcbinfo, p);
435 	splx(s);
436 	if (error)
437 		return error;
438 	error = soreserve(so, rip_sendspace, rip_recvspace);
439 	if (error)
440 		return error;
441 	inp = (struct inpcb *)so->so_pcb;
442 	inp->inp_ip_p = proto;
443 	return 0;
444 }
445 
446 static int
447 rip_detach(struct socket *so)
448 {
449 	struct inpcb *inp;
450 
451 	inp = sotoinpcb(so);
452 	if (inp == 0)
453 		panic("rip_detach");
454 	if (so == ip_mrouter)
455 		ip_mrouter_done();
456 	ip_rsvp_force_done(so);
457 	if (so == ip_rsvpd)
458 		ip_rsvp_done();
459 	in_pcbdetach(inp);
460 	return 0;
461 }
462 
463 static int
464 rip_abort(struct socket *so)
465 {
466 	soisdisconnected(so);
467 	return rip_detach(so);
468 }
469 
470 static int
471 rip_disconnect(struct socket *so)
472 {
473 	if ((so->so_state & SS_ISCONNECTED) == 0)
474 		return ENOTCONN;
475 	return rip_abort(so);
476 }
477 
478 static int
479 rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
480 {
481 	struct inpcb *inp = sotoinpcb(so);
482 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
483 
484 	if (nam->sa_len != sizeof(*addr))
485 		return EINVAL;
486 
487 	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
488 				    (addr->sin_family != AF_IMPLINK)) ||
489 	    (addr->sin_addr.s_addr &&
490 	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
491 		return EADDRNOTAVAIL;
492 	inp->inp_laddr = addr->sin_addr;
493 	return 0;
494 }
495 
496 static int
497 rip_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
498 {
499 	struct inpcb *inp = sotoinpcb(so);
500 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
501 
502 	if (nam->sa_len != sizeof(*addr))
503 		return EINVAL;
504 	if (TAILQ_EMPTY(&ifnet))
505 		return EADDRNOTAVAIL;
506 	if ((addr->sin_family != AF_INET) &&
507 	    (addr->sin_family != AF_IMPLINK))
508 		return EAFNOSUPPORT;
509 	inp->inp_faddr = addr->sin_addr;
510 	soisconnected(so);
511 	return 0;
512 }
513 
514 static int
515 rip_shutdown(struct socket *so)
516 {
517 	socantsendmore(so);
518 	return 0;
519 }
520 
521 static int
522 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
523 	 struct mbuf *control, struct proc *p)
524 {
525 	struct inpcb *inp = sotoinpcb(so);
526 	register u_long dst;
527 
528 	if (so->so_state & SS_ISCONNECTED) {
529 		if (nam) {
530 			m_freem(m);
531 			return EISCONN;
532 		}
533 		dst = inp->inp_faddr.s_addr;
534 	} else {
535 		if (nam == NULL) {
536 			m_freem(m);
537 			return ENOTCONN;
538 		}
539 		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
540 	}
541 	return rip_output(m, so, dst);
542 }
543 
544 static int
545 rip_pcblist SYSCTL_HANDLER_ARGS
546 {
547 	int error, i, n, s;
548 	struct inpcb *inp, **inp_list;
549 	inp_gen_t gencnt;
550 	struct xinpgen xig;
551 
552 	/*
553 	 * The process of preparing the TCB list is too time-consuming and
554 	 * resource-intensive to repeat twice on every request.
555 	 */
556 	if (req->oldptr == 0) {
557 		n = ripcbinfo.ipi_count;
558 		req->oldidx = 2 * (sizeof xig)
559 			+ (n + n/8) * sizeof(struct xinpcb);
560 		return 0;
561 	}
562 
563 	if (req->newptr != 0)
564 		return EPERM;
565 
566 	/*
567 	 * OK, now we're committed to doing something.
568 	 */
569 	s = splnet();
570 	gencnt = ripcbinfo.ipi_gencnt;
571 	n = ripcbinfo.ipi_count;
572 	splx(s);
573 
574 	xig.xig_len = sizeof xig;
575 	xig.xig_count = n;
576 	xig.xig_gen = gencnt;
577 	xig.xig_sogen = so_gencnt;
578 	error = SYSCTL_OUT(req, &xig, sizeof xig);
579 	if (error)
580 		return error;
581 
582 	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
583 	if (inp_list == 0)
584 		return ENOMEM;
585 
586 	s = splnet();
587 	for (inp = ripcbinfo.listhead->lh_first, i = 0; inp && i < n;
588 	     inp = inp->inp_list.le_next) {
589 		if (inp->inp_gencnt <= gencnt)
590 			inp_list[i++] = inp;
591 	}
592 	splx(s);
593 	n = i;
594 
595 	error = 0;
596 	for (i = 0; i < n; i++) {
597 		inp = inp_list[i];
598 		if (inp->inp_gencnt <= gencnt) {
599 			struct xinpcb xi;
600 			xi.xi_len = sizeof xi;
601 			/* XXX should avoid extra copy */
602 			bcopy(inp, &xi.xi_inp, sizeof *inp);
603 			if (inp->inp_socket)
604 				sotoxsocket(inp->inp_socket, &xi.xi_socket);
605 			error = SYSCTL_OUT(req, &xi, sizeof xi);
606 		}
607 	}
608 	if (!error) {
609 		/*
610 		 * Give the user an updated idea of our state.
611 		 * If the generation differs from what we told
612 		 * her before, she knows that something happened
613 		 * while we were processing this request, and it
614 		 * might be necessary to retry.
615 		 */
616 		s = splnet();
617 		xig.xig_gen = ripcbinfo.ipi_gencnt;
618 		xig.xig_sogen = so_gencnt;
619 		xig.xig_count = ripcbinfo.ipi_count;
620 		splx(s);
621 		error = SYSCTL_OUT(req, &xig, sizeof xig);
622 	}
623 	free(inp_list, M_TEMP);
624 	return error;
625 }
626 
627 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
628 	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
629 
630 struct pr_usrreqs rip_usrreqs = {
631 	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
632 	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
633 	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
634 	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
635 	in_setsockaddr, sosend, soreceive, sopoll
636 };
637