xref: /freebsd/sys/netinet/raw_ip.c (revision 11afcc8f9f96d657b8e6f7547c02c1957331fc96)
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 1.53 1998/03/28 10:18:24 bde 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(op, so, level, optname, m, p)
229 	int op;
230 	struct socket *so;
231 	int level, optname;
232 	struct mbuf **m;
233 	struct proc *p;
234 {
235 	register struct inpcb *inp = sotoinpcb(so);
236 	register int error;
237 
238 	if (level != IPPROTO_IP) {
239 		if (op == PRCO_SETOPT && *m)
240 			(void)m_free(*m);
241 		return (EINVAL);
242 	}
243 
244 	switch (optname) {
245 
246 	case IP_HDRINCL:
247 		error = 0;
248 		if (op == PRCO_SETOPT) {
249 			if (m == 0 || *m == 0 || (*m)->m_len < sizeof (int))
250 				error = EINVAL;
251 			else if (*mtod(*m, int *))
252 				inp->inp_flags |= INP_HDRINCL;
253 			else
254 				inp->inp_flags &= ~INP_HDRINCL;
255 			if (*m)
256 				(void)m_free(*m);
257 		} else {
258 			*m = m_get(M_WAIT, MT_SOOPTS);
259 			(*m)->m_len = sizeof (int);
260 			*mtod(*m, int *) = inp->inp_flags & INP_HDRINCL;
261 		}
262 		return (error);
263 
264 #ifdef COMPAT_IPFW
265 	case IP_FW_GET:
266 		if (ip_fw_ctl_ptr == NULL || op == PRCO_SETOPT) {
267 			if (*m) (void)m_free(*m);
268 			return(EINVAL);
269 		}
270 		return (*ip_fw_ctl_ptr)(optname, m);
271 
272 	case IP_FW_ADD:
273 	case IP_FW_DEL:
274 	case IP_FW_FLUSH:
275 	case IP_FW_ZERO:
276 		if (ip_fw_ctl_ptr == NULL || op != PRCO_SETOPT) {
277 			if (*m) (void)m_free(*m);
278 			return(EINVAL);
279 		}
280 		return (*ip_fw_ctl_ptr)(optname, m);
281 
282 	case IP_NAT:
283 		if (ip_nat_ctl_ptr == NULL) {
284 			if (*m) (void)m_free(*m);
285 			return(EINVAL);
286 		}
287 		return (*ip_nat_ctl_ptr)(op, m);
288 
289 #endif
290 	case IP_RSVP_ON:
291 		return ip_rsvp_init(so);
292 		break;
293 
294 	case IP_RSVP_OFF:
295 		return ip_rsvp_done();
296 		break;
297 
298 	case IP_RSVP_VIF_ON:
299 		return ip_rsvp_vif_init(so, *m);
300 
301 	case IP_RSVP_VIF_OFF:
302 		return ip_rsvp_vif_done(so, *m);
303 
304 	case MRT_INIT:
305 	case MRT_DONE:
306 	case MRT_ADD_VIF:
307 	case MRT_DEL_VIF:
308 	case MRT_ADD_MFC:
309 	case MRT_DEL_MFC:
310 	case MRT_VERSION:
311 	case MRT_ASSERT:
312 		if (op == PRCO_SETOPT) {
313 			error = ip_mrouter_set(optname, so, *m);
314 			if (*m)
315 				(void)m_free(*m);
316 		} else if (op == PRCO_GETOPT) {
317 			error = ip_mrouter_get(optname, so, m);
318 		} else
319 			error = EINVAL;
320 		return (error);
321 	}
322 	return (ip_ctloutput(op, so, level, optname, m, p));
323 }
324 
325 /*
326  * This function exists solely to receive the PRC_IFDOWN messages which
327  * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
328  * and calls in_ifadown() to remove all routes corresponding to that address.
329  * It also receives the PRC_IFUP messages from if_up() and reinstalls the
330  * interface routes.
331  */
332 void
333 rip_ctlinput(cmd, sa, vip)
334 	int cmd;
335 	struct sockaddr *sa;
336 	void *vip;
337 {
338 	struct in_ifaddr *ia;
339 	struct ifnet *ifp;
340 	int err;
341 	int flags;
342 
343 	switch(cmd) {
344 	case PRC_IFDOWN:
345 		for (ia = in_ifaddrhead.tqh_first; ia;
346 		     ia = ia->ia_link.tqe_next) {
347 			if (ia->ia_ifa.ifa_addr == sa
348 			    && (ia->ia_flags & IFA_ROUTE)) {
349 				/*
350 				 * in_ifscrub kills the interface route.
351 				 */
352 				in_ifscrub(ia->ia_ifp, ia);
353 				/*
354 				 * in_ifadown gets rid of all the rest of
355 				 * the routes.  This is not quite the right
356 				 * thing to do, but at least if we are running
357 				 * a routing process they will come back.
358 				 */
359 				in_ifadown(&ia->ia_ifa);
360 				break;
361 			}
362 		}
363 		break;
364 
365 	case PRC_IFUP:
366 		for (ia = in_ifaddrhead.tqh_first; ia;
367 		     ia = ia->ia_link.tqe_next) {
368 			if (ia->ia_ifa.ifa_addr == sa)
369 				break;
370 		}
371 		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
372 			return;
373 		flags = RTF_UP;
374 		ifp = ia->ia_ifa.ifa_ifp;
375 
376 		if ((ifp->if_flags & IFF_LOOPBACK)
377 		    || (ifp->if_flags & IFF_POINTOPOINT))
378 			flags |= RTF_HOST;
379 
380 		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
381 		if (err == 0)
382 			ia->ia_flags |= IFA_ROUTE;
383 		break;
384 	}
385 }
386 
387 static u_long	rip_sendspace = RIPSNDQ;
388 static u_long	rip_recvspace = RIPRCVQ;
389 
390 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW, &rip_sendspace,
391 	   0, "");
392 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW, &rip_recvspace,
393 	   0, "");
394 
395 static int
396 rip_attach(struct socket *so, int proto, struct proc *p)
397 {
398 	struct inpcb *inp;
399 	int error, s;
400 
401 	inp = sotoinpcb(so);
402 	if (inp)
403 		panic("rip_attach");
404 	if (p && (error = suser(p->p_ucred, &p->p_acflag)) != 0)
405 		return error;
406 
407 	s = splnet();
408 	error = in_pcballoc(so, &ripcbinfo, p);
409 	splx(s);
410 	if (error)
411 		return error;
412 	error = soreserve(so, rip_sendspace, rip_recvspace);
413 	if (error)
414 		return error;
415 	inp = (struct inpcb *)so->so_pcb;
416 	inp->inp_ip_p = proto;
417 	return 0;
418 }
419 
420 static int
421 rip_detach(struct socket *so)
422 {
423 	struct inpcb *inp;
424 
425 	inp = sotoinpcb(so);
426 	if (inp == 0)
427 		panic("rip_detach");
428 	if (so == ip_mrouter)
429 		ip_mrouter_done();
430 	ip_rsvp_force_done(so);
431 	if (so == ip_rsvpd)
432 		ip_rsvp_done();
433 	in_pcbdetach(inp);
434 	return 0;
435 }
436 
437 static int
438 rip_abort(struct socket *so)
439 {
440 	soisdisconnected(so);
441 	return rip_detach(so);
442 }
443 
444 static int
445 rip_disconnect(struct socket *so)
446 {
447 	if ((so->so_state & SS_ISCONNECTED) == 0)
448 		return ENOTCONN;
449 	return rip_abort(so);
450 }
451 
452 static int
453 rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
454 {
455 	struct inpcb *inp = sotoinpcb(so);
456 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
457 
458 	if (nam->sa_len != sizeof(*addr))
459 		return EINVAL;
460 
461 	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
462 				    (addr->sin_family != AF_IMPLINK)) ||
463 	    (addr->sin_addr.s_addr &&
464 	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
465 		return EADDRNOTAVAIL;
466 	inp->inp_laddr = addr->sin_addr;
467 	return 0;
468 }
469 
470 static int
471 rip_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
472 {
473 	struct inpcb *inp = sotoinpcb(so);
474 	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
475 
476 	if (nam->sa_len != sizeof(*addr))
477 		return EINVAL;
478 	if (TAILQ_EMPTY(&ifnet))
479 		return EADDRNOTAVAIL;
480 	if ((addr->sin_family != AF_INET) &&
481 	    (addr->sin_family != AF_IMPLINK))
482 		return EAFNOSUPPORT;
483 	inp->inp_faddr = addr->sin_addr;
484 	soisconnected(so);
485 	return 0;
486 }
487 
488 static int
489 rip_shutdown(struct socket *so)
490 {
491 	socantsendmore(so);
492 	return 0;
493 }
494 
495 static int
496 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
497 	 struct mbuf *control, struct proc *p)
498 {
499 	struct inpcb *inp = sotoinpcb(so);
500 	register u_long dst;
501 
502 	if (so->so_state & SS_ISCONNECTED) {
503 		if (nam) {
504 			m_freem(m);
505 			return EISCONN;
506 		}
507 		dst = inp->inp_faddr.s_addr;
508 	} else {
509 		if (nam == NULL) {
510 			m_freem(m);
511 			return ENOTCONN;
512 		}
513 		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
514 	}
515 	return rip_output(m, so, dst);
516 }
517 
518 static int
519 rip_pcblist SYSCTL_HANDLER_ARGS
520 {
521 	int error, i, n, s;
522 	struct inpcb *inp, **inp_list;
523 	inp_gen_t gencnt;
524 	struct xinpgen xig;
525 
526 	/*
527 	 * The process of preparing the TCB list is too time-consuming and
528 	 * resource-intensive to repeat twice on every request.
529 	 */
530 	if (req->oldptr == 0) {
531 		n = ripcbinfo.ipi_count;
532 		req->oldidx = 2 * (sizeof xig)
533 			+ (n + n/8) * sizeof(struct xinpcb);
534 		return 0;
535 	}
536 
537 	if (req->newptr != 0)
538 		return EPERM;
539 
540 	/*
541 	 * OK, now we're committed to doing something.
542 	 */
543 	s = splnet();
544 	gencnt = ripcbinfo.ipi_gencnt;
545 	n = ripcbinfo.ipi_count;
546 	splx(s);
547 
548 	xig.xig_len = sizeof xig;
549 	xig.xig_count = n;
550 	xig.xig_gen = gencnt;
551 	xig.xig_sogen = so_gencnt;
552 	error = SYSCTL_OUT(req, &xig, sizeof xig);
553 	if (error)
554 		return error;
555 
556 	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
557 	if (inp_list == 0)
558 		return ENOMEM;
559 
560 	s = splnet();
561 	for (inp = ripcbinfo.listhead->lh_first, i = 0; inp && i < n;
562 	     inp = inp->inp_list.le_next) {
563 		if (inp->inp_gencnt <= gencnt)
564 			inp_list[i++] = inp;
565 	}
566 	splx(s);
567 	n = i;
568 
569 	error = 0;
570 	for (i = 0; i < n; i++) {
571 		inp = inp_list[i];
572 		if (inp->inp_gencnt <= gencnt) {
573 			struct xinpcb xi;
574 			xi.xi_len = sizeof xi;
575 			/* XXX should avoid extra copy */
576 			bcopy(inp, &xi.xi_inp, sizeof *inp);
577 			if (inp->inp_socket)
578 				sotoxsocket(inp->inp_socket, &xi.xi_socket);
579 			error = SYSCTL_OUT(req, &xi, sizeof xi);
580 		}
581 	}
582 	if (!error) {
583 		/*
584 		 * Give the user an updated idea of our state.
585 		 * If the generation differs from what we told
586 		 * her before, she knows that something happened
587 		 * while we were processing this request, and it
588 		 * might be necessary to retry.
589 		 */
590 		s = splnet();
591 		xig.xig_gen = ripcbinfo.ipi_gencnt;
592 		xig.xig_sogen = so_gencnt;
593 		xig.xig_count = ripcbinfo.ipi_count;
594 		splx(s);
595 		error = SYSCTL_OUT(req, &xig, sizeof xig);
596 	}
597 	free(inp_list, M_TEMP);
598 	return error;
599 }
600 
601 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
602 	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
603 
604 struct pr_usrreqs rip_usrreqs = {
605 	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
606 	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
607 	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
608 	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
609 	in_setsockaddr, sosend, soreceive, sopoll
610 };
611