raw_ip.c revision 76469 
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: head/sys/netinet/raw_ip.c 76469 2001-05-11 14:37:34Z ru $
35 */
36
37#include "opt_inet6.h"
38#include "opt_ipsec.h"
39
40#include <sys/param.h>
41#include <sys/systm.h>
42#include <sys/kernel.h>
43#include <sys/malloc.h>
44#include <sys/mbuf.h>
45#include <sys/protosw.h>
46#include <sys/socket.h>
47#include <sys/socketvar.h>
48#include <sys/sysctl.h>
49
50#include <vm/vm_zone.h>
51
52#include <net/if.h>
53#include <net/route.h>
54
55#define _IP_VHL
56#include <netinet/in.h>
57#include <netinet/in_systm.h>
58#include <netinet/ip.h>
59#include <netinet/in_pcb.h>
60#include <netinet/in_var.h>
61#include <netinet/ip_var.h>
62#include <netinet/ip_mroute.h>
63
64#include <netinet/ip_fw.h>
65
66#ifdef IPSEC
67#include <netinet6/ipsec.h>
68#endif /*IPSEC*/
69
70#include "opt_ipdn.h"
71#ifdef DUMMYNET
72#include <netinet/ip_dummynet.h>
73#endif
74
75struct	inpcbhead ripcb;
76struct	inpcbinfo ripcbinfo;
77
78/*
79 * Nominal space allocated to a raw ip socket.
80 */
81#define	RIPSNDQ		8192
82#define	RIPRCVQ		8192
83
84/*
85 * Raw interface to IP protocol.
86 */
87
88/*
89 * Initialize raw connection block q.
90 */
91void
92rip_init()
93{
94	LIST_INIT(&ripcb);
95	ripcbinfo.listhead = &ripcb;
96	/*
97	 * XXX We don't use the hash list for raw IP, but it's easier
98	 * to allocate a one entry hash list than it is to check all
99	 * over the place for hashbase == NULL.
100	 */
101	ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
102	ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
103	ripcbinfo.ipi_zone = zinit("ripcb", sizeof(struct inpcb),
104				   maxsockets, ZONE_INTERRUPT, 0);
105}
106
107static struct	sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
108/*
109 * Setup generic address and protocol structures
110 * for raw_input routine, then pass them along with
111 * mbuf chain.
112 */
113void
114rip_input(m, off, proto)
115	struct mbuf *m;
116	int off, proto;
117{
118	register struct ip *ip = mtod(m, struct ip *);
119	register struct inpcb *inp;
120	struct inpcb *last = 0;
121	struct mbuf *opts = 0;
122
123	ripsrc.sin_addr = ip->ip_src;
124	LIST_FOREACH(inp, &ripcb, inp_list) {
125#ifdef INET6
126		if ((inp->inp_vflag & INP_IPV4) == 0)
127			continue;
128#endif
129		if (inp->inp_ip_p && inp->inp_ip_p != proto)
130			continue;
131		if (inp->inp_laddr.s_addr &&
132                  inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
133			continue;
134		if (inp->inp_faddr.s_addr &&
135                  inp->inp_faddr.s_addr != ip->ip_src.s_addr)
136			continue;
137		if (last) {
138			struct mbuf *n = m_copy(m, 0, (int)M_COPYALL);
139			if (n) {
140				if (last->inp_flags & INP_CONTROLOPTS ||
141				    last->inp_socket->so_options & SO_TIMESTAMP)
142				    ip_savecontrol(last, &opts, ip, n);
143				if (sbappendaddr(&last->inp_socket->so_rcv,
144				    (struct sockaddr *)&ripsrc, n,
145				    opts) == 0) {
146					/* should notify about lost packet */
147					m_freem(n);
148					if (opts)
149					    m_freem(opts);
150				} else
151					sorwakeup(last->inp_socket);
152				opts = 0;
153			}
154		}
155		last = inp;
156	}
157	if (last) {
158		if (last->inp_flags & INP_CONTROLOPTS ||
159		    last->inp_socket->so_options & SO_TIMESTAMP)
160			ip_savecontrol(last, &opts, ip, m);
161		if (sbappendaddr(&last->inp_socket->so_rcv,
162		    (struct sockaddr *)&ripsrc, m, opts) == 0) {
163			m_freem(m);
164			if (opts)
165			    m_freem(opts);
166		} else
167			sorwakeup(last->inp_socket);
168	} else {
169		m_freem(m);
170              ipstat.ips_noproto++;
171              ipstat.ips_delivered--;
172      }
173}
174
175/*
176 * Generate IP header and pass packet to ip_output.
177 * Tack on options user may have setup with control call.
178 */
179int
180rip_output(m, so, dst)
181	struct mbuf *m;
182	struct socket *so;
183	u_long dst;
184{
185	register struct ip *ip;
186	register struct inpcb *inp = sotoinpcb(so);
187	int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
188
189	/*
190	 * If the user handed us a complete IP packet, use it.
191	 * Otherwise, allocate an mbuf for a header and fill it in.
192	 */
193	if ((inp->inp_flags & INP_HDRINCL) == 0) {
194		if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
195			m_freem(m);
196			return(EMSGSIZE);
197		}
198		M_PREPEND(m, sizeof(struct ip), M_TRYWAIT);
199		ip = mtod(m, struct ip *);
200		ip->ip_tos = inp->inp_ip_tos;
201		ip->ip_off = 0;
202		ip->ip_p = inp->inp_ip_p;
203		ip->ip_len = m->m_pkthdr.len;
204		ip->ip_src = inp->inp_laddr;
205		ip->ip_dst.s_addr = dst;
206		ip->ip_ttl = inp->inp_ip_ttl;
207	} else {
208		if (m->m_pkthdr.len > IP_MAXPACKET) {
209			m_freem(m);
210			return(EMSGSIZE);
211		}
212		ip = mtod(m, struct ip *);
213		/* don't allow both user specified and setsockopt options,
214		   and don't allow packet length sizes that will crash */
215		if (((IP_VHL_HL(ip->ip_vhl) != (sizeof (*ip) >> 2))
216		     && inp->inp_options)
217		    || (ip->ip_len > m->m_pkthdr.len)
218		    || (ip->ip_len < (IP_VHL_HL(ip->ip_vhl) << 2))) {
219			m_freem(m);
220			return EINVAL;
221		}
222		if (ip->ip_id == 0)
223			ip->ip_id = htons(ip_id++);
224		/* XXX prevent ip_output from overwriting header fields */
225		flags |= IP_RAWOUTPUT;
226		ipstat.ips_rawout++;
227	}
228
229#ifdef IPSEC
230	ipsec_setsocket(m, so);
231#endif /*IPSEC*/
232
233	return (ip_output(m, inp->inp_options, &inp->inp_route, flags,
234			  inp->inp_moptions));
235}
236
237/*
238 * Raw IP socket option processing.
239 */
240int
241rip_ctloutput(so, sopt)
242	struct socket *so;
243	struct sockopt *sopt;
244{
245	struct	inpcb *inp = sotoinpcb(so);
246	int	error, optval;
247
248	if (sopt->sopt_level != IPPROTO_IP)
249		return (EINVAL);
250
251	error = 0;
252
253	switch (sopt->sopt_dir) {
254	case SOPT_GET:
255		switch (sopt->sopt_name) {
256		case IP_HDRINCL:
257			optval = inp->inp_flags & INP_HDRINCL;
258			error = sooptcopyout(sopt, &optval, sizeof optval);
259			break;
260
261		case IP_FW_ADD:
262		case IP_FW_GET:
263			if (ip_fw_ctl_ptr == 0)
264				error = ENOPROTOOPT;
265			else
266				error = ip_fw_ctl_ptr(sopt);
267			break;
268
269#ifdef DUMMYNET
270		case IP_DUMMYNET_GET:
271			if (ip_dn_ctl_ptr == NULL)
272				error = ENOPROTOOPT ;
273			else
274				error = ip_dn_ctl_ptr(sopt);
275			break ;
276#endif /* DUMMYNET */
277
278		case MRT_INIT:
279		case MRT_DONE:
280		case MRT_ADD_VIF:
281		case MRT_DEL_VIF:
282		case MRT_ADD_MFC:
283		case MRT_DEL_MFC:
284		case MRT_VERSION:
285		case MRT_ASSERT:
286			error = ip_mrouter_get(so, sopt);
287			break;
288
289		default:
290			error = ip_ctloutput(so, sopt);
291			break;
292		}
293		break;
294
295	case SOPT_SET:
296		switch (sopt->sopt_name) {
297		case IP_HDRINCL:
298			error = sooptcopyin(sopt, &optval, sizeof optval,
299					    sizeof optval);
300			if (error)
301				break;
302			if (optval)
303				inp->inp_flags |= INP_HDRINCL;
304			else
305				inp->inp_flags &= ~INP_HDRINCL;
306			break;
307
308		case IP_FW_ADD:
309		case IP_FW_DEL:
310		case IP_FW_FLUSH:
311		case IP_FW_ZERO:
312		case IP_FW_RESETLOG:
313			if (ip_fw_ctl_ptr == 0)
314				error = ENOPROTOOPT;
315			else
316				error = ip_fw_ctl_ptr(sopt);
317			break;
318
319#ifdef DUMMYNET
320		case IP_DUMMYNET_CONFIGURE:
321		case IP_DUMMYNET_DEL:
322		case IP_DUMMYNET_FLUSH:
323			if (ip_dn_ctl_ptr == NULL)
324				error = ENOPROTOOPT ;
325			else
326				error = ip_dn_ctl_ptr(sopt);
327			break ;
328#endif
329
330		case IP_RSVP_ON:
331			error = ip_rsvp_init(so);
332			break;
333
334		case IP_RSVP_OFF:
335			error = ip_rsvp_done();
336			break;
337
338			/* XXX - should be combined */
339		case IP_RSVP_VIF_ON:
340			error = ip_rsvp_vif_init(so, sopt);
341			break;
342
343		case IP_RSVP_VIF_OFF:
344			error = ip_rsvp_vif_done(so, sopt);
345			break;
346
347		case MRT_INIT:
348		case MRT_DONE:
349		case MRT_ADD_VIF:
350		case MRT_DEL_VIF:
351		case MRT_ADD_MFC:
352		case MRT_DEL_MFC:
353		case MRT_VERSION:
354		case MRT_ASSERT:
355			error = ip_mrouter_set(so, sopt);
356			break;
357
358		default:
359			error = ip_ctloutput(so, sopt);
360			break;
361		}
362		break;
363	}
364
365	return (error);
366}
367
368/*
369 * This function exists solely to receive the PRC_IFDOWN messages which
370 * are sent by if_down().  It looks for an ifaddr whose ifa_addr is sa,
371 * and calls in_ifadown() to remove all routes corresponding to that address.
372 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
373 * interface routes.
374 */
375void
376rip_ctlinput(cmd, sa, vip)
377	int cmd;
378	struct sockaddr *sa;
379	void *vip;
380{
381	struct in_ifaddr *ia;
382	struct ifnet *ifp;
383	int err;
384	int flags;
385
386	switch (cmd) {
387	case PRC_IFDOWN:
388		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
389			if (ia->ia_ifa.ifa_addr == sa
390			    && (ia->ia_flags & IFA_ROUTE)) {
391				/*
392				 * in_ifscrub kills the interface route.
393				 */
394				in_ifscrub(ia->ia_ifp, ia);
395				/*
396				 * in_ifadown gets rid of all the rest of
397				 * the routes.  This is not quite the right
398				 * thing to do, but at least if we are running
399				 * a routing process they will come back.
400				 */
401				in_ifadown(&ia->ia_ifa, 0);
402				break;
403			}
404		}
405		break;
406
407	case PRC_IFUP:
408		TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
409			if (ia->ia_ifa.ifa_addr == sa)
410				break;
411		}
412		if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
413			return;
414		flags = RTF_UP;
415		ifp = ia->ia_ifa.ifa_ifp;
416
417		if ((ifp->if_flags & IFF_LOOPBACK)
418		    || (ifp->if_flags & IFF_POINTOPOINT))
419			flags |= RTF_HOST;
420
421		err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
422		if (err == 0)
423			ia->ia_flags |= IFA_ROUTE;
424		break;
425	}
426}
427
428u_long	rip_sendspace = RIPSNDQ;
429u_long	rip_recvspace = RIPRCVQ;
430
431SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
432    &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
433SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
434    &rip_recvspace, 0, "Maximum incoming raw IP datagram size");
435
436static int
437rip_attach(struct socket *so, int proto, struct proc *p)
438{
439	struct inpcb *inp;
440	int error, s;
441
442	inp = sotoinpcb(so);
443	if (inp)
444		panic("rip_attach");
445	if (p && (error = suser(p)) != 0)
446		return error;
447
448	error = soreserve(so, rip_sendspace, rip_recvspace);
449	if (error)
450		return error;
451	s = splnet();
452	error = in_pcballoc(so, &ripcbinfo, p);
453	splx(s);
454	if (error)
455		return error;
456	inp = (struct inpcb *)so->so_pcb;
457	inp->inp_vflag |= INP_IPV4;
458	inp->inp_ip_p = proto;
459	inp->inp_ip_ttl = ip_defttl;
460#ifdef IPSEC
461	error = ipsec_init_policy(so, &inp->inp_sp);
462	if (error != 0) {
463		in_pcbdetach(inp);
464		return error;
465	}
466#endif /*IPSEC*/
467	return 0;
468}
469
470static int
471rip_detach(struct socket *so)
472{
473	struct inpcb *inp;
474
475	inp = sotoinpcb(so);
476	if (inp == 0)
477		panic("rip_detach");
478	if (so == ip_mrouter)
479		ip_mrouter_done();
480	ip_rsvp_force_done(so);
481	if (so == ip_rsvpd)
482		ip_rsvp_done();
483	in_pcbdetach(inp);
484	return 0;
485}
486
487static int
488rip_abort(struct socket *so)
489{
490	soisdisconnected(so);
491	return rip_detach(so);
492}
493
494static int
495rip_disconnect(struct socket *so)
496{
497	if ((so->so_state & SS_ISCONNECTED) == 0)
498		return ENOTCONN;
499	return rip_abort(so);
500}
501
502static int
503rip_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
504{
505	struct inpcb *inp = sotoinpcb(so);
506	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
507
508	if (nam->sa_len != sizeof(*addr))
509		return EINVAL;
510
511	if (TAILQ_EMPTY(&ifnet) || ((addr->sin_family != AF_INET) &&
512				    (addr->sin_family != AF_IMPLINK)) ||
513	    (addr->sin_addr.s_addr &&
514	     ifa_ifwithaddr((struct sockaddr *)addr) == 0))
515		return EADDRNOTAVAIL;
516	inp->inp_laddr = addr->sin_addr;
517	return 0;
518}
519
520static int
521rip_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
522{
523	struct inpcb *inp = sotoinpcb(so);
524	struct sockaddr_in *addr = (struct sockaddr_in *)nam;
525
526	if (nam->sa_len != sizeof(*addr))
527		return EINVAL;
528	if (TAILQ_EMPTY(&ifnet))
529		return EADDRNOTAVAIL;
530	if ((addr->sin_family != AF_INET) &&
531	    (addr->sin_family != AF_IMPLINK))
532		return EAFNOSUPPORT;
533	inp->inp_faddr = addr->sin_addr;
534	soisconnected(so);
535	return 0;
536}
537
538static int
539rip_shutdown(struct socket *so)
540{
541	socantsendmore(so);
542	return 0;
543}
544
545static int
546rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
547	 struct mbuf *control, struct proc *p)
548{
549	struct inpcb *inp = sotoinpcb(so);
550	register u_long dst;
551
552	if (so->so_state & SS_ISCONNECTED) {
553		if (nam) {
554			m_freem(m);
555			return EISCONN;
556		}
557		dst = inp->inp_faddr.s_addr;
558	} else {
559		if (nam == NULL) {
560			m_freem(m);
561			return ENOTCONN;
562		}
563		dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
564	}
565	return rip_output(m, so, dst);
566}
567
568static int
569rip_pcblist(SYSCTL_HANDLER_ARGS)
570{
571	int error, i, n, s;
572	struct inpcb *inp, **inp_list;
573	inp_gen_t gencnt;
574	struct xinpgen xig;
575
576	/*
577	 * The process of preparing the TCB list is too time-consuming and
578	 * resource-intensive to repeat twice on every request.
579	 */
580	if (req->oldptr == 0) {
581		n = ripcbinfo.ipi_count;
582		req->oldidx = 2 * (sizeof xig)
583			+ (n + n/8) * sizeof(struct xinpcb);
584		return 0;
585	}
586
587	if (req->newptr != 0)
588		return EPERM;
589
590	/*
591	 * OK, now we're committed to doing something.
592	 */
593	s = splnet();
594	gencnt = ripcbinfo.ipi_gencnt;
595	n = ripcbinfo.ipi_count;
596	splx(s);
597
598	xig.xig_len = sizeof xig;
599	xig.xig_count = n;
600	xig.xig_gen = gencnt;
601	xig.xig_sogen = so_gencnt;
602	error = SYSCTL_OUT(req, &xig, sizeof xig);
603	if (error)
604		return error;
605
606	inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
607	if (inp_list == 0)
608		return ENOMEM;
609
610	s = splnet();
611	for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
612	     inp = LIST_NEXT(inp, inp_list)) {
613		if (inp->inp_gencnt <= gencnt)
614			inp_list[i++] = inp;
615	}
616	splx(s);
617	n = i;
618
619	error = 0;
620	for (i = 0; i < n; i++) {
621		inp = inp_list[i];
622		if (inp->inp_gencnt <= gencnt) {
623			struct xinpcb xi;
624			xi.xi_len = sizeof xi;
625			/* XXX should avoid extra copy */
626			bcopy(inp, &xi.xi_inp, sizeof *inp);
627			if (inp->inp_socket)
628				sotoxsocket(inp->inp_socket, &xi.xi_socket);
629			error = SYSCTL_OUT(req, &xi, sizeof xi);
630		}
631	}
632	if (!error) {
633		/*
634		 * Give the user an updated idea of our state.
635		 * If the generation differs from what we told
636		 * her before, she knows that something happened
637		 * while we were processing this request, and it
638		 * might be necessary to retry.
639		 */
640		s = splnet();
641		xig.xig_gen = ripcbinfo.ipi_gencnt;
642		xig.xig_sogen = so_gencnt;
643		xig.xig_count = ripcbinfo.ipi_count;
644		splx(s);
645		error = SYSCTL_OUT(req, &xig, sizeof xig);
646	}
647	free(inp_list, M_TEMP);
648	return error;
649}
650
651SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
652	    rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
653
654struct pr_usrreqs rip_usrreqs = {
655	rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
656	pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
657	pru_listen_notsupp, in_setpeeraddr, pru_rcvd_notsupp,
658	pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
659	in_setsockaddr, sosend, soreceive, sopoll
660};
661