xref: /src/sys/netinet/ip_encap.c

/*	$NetBSD: ip_encap.c,v 1.3 2000/07/05 21:32:51 thorpej Exp $	*/
/*	$KAME: ip_encap.c,v 1.30 2000/04/19 04:29:37 itojun Exp $	*/

/*
 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the project nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */
/*
 * My grandfather said that there's a devil inside tunnelling technology...
 *
 * We have surprisingly many protocols that want packets with IP protocol
 * #4 or #41.  Here's a list of protocols that want protocol #41:
 *	RFC1933 configured tunnel
 *	RFC1933 automatic tunnel
 *	RFC2401 IPsec tunnel
 *	RFC2473 IPv6 generic packet tunnelling
 *	RFC2529 6over4 tunnel
 *	mobile-ip6 (uses RFC2473)
 *	6to4 tunnel
 * Here's a list of protocol that want protocol #4:
 *	RFC1853 IPv4-in-IPv4 tunnelling
 *	RFC2003 IPv4 encapsulation within IPv4
 *	RFC2344 reverse tunnelling for mobile-ip4
 *	RFC2401 IPsec tunnel
 * Well, what can I say.  They impose different en/decapsulation mechanism
 * from each other, so they need separate protocol handler.  The only one
 * we can easily determine by protocol # is IPsec, which always has
 * AH/ESP/IPComp header right after outer IP header.
 *
 * So, clearly good old protosw does not work for protocol #4 and #41.
 * The code will let you match protocol via src/dst address pair.
 */
/* XXX is M_NETADDR correct? */

#ifdef __FreeBSD__
# include "opt_mrouting.h"
# if __FreeBSD__ == 3
#  include "opt_inet.h"
# endif
# if __FreeBSD__ >= 4
#  include "opt_inet.h"
#  include "opt_inet6.h"
# endif
#else
# ifdef __NetBSD__
#  include "opt_mrouting.h"
#  include "opt_inet.h"
# endif
#endif

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/mbuf.h>
#include <sys/errno.h>
#include <sys/protosw.h>

#include <net/if.h>
#include <net/route.h>

#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet/ip_encap.h>
#ifdef MROUTING
#include <netinet/ip_mroute.h>
#endif /* MROUTING */
#ifdef __OpenBSD__
#include <netinet/ip_ipsp.h>
#endif

#ifdef INET6
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/ip6protosw.h>
#endif

#include <machine/stdarg.h>

#ifdef __NetBSD__
# include "ipip.h"
# if NIPIP > 0
#  include <netinet/ip_ipip.h>
# else
#  ifdef MROUTING
#   include <netinet/ip_mroute.h>
#  endif
# endif
#endif

#include <net/net_osdep.h>

#if defined(__FreeBSD__) && __FreeBSD__ >= 3
#include <sys/kernel.h>
#include <sys/malloc.h>
MALLOC_DEFINE(M_NETADDR, "Export Host", "Export host address structure");
#endif

static void encap_add __P((struct encaptab *));
static int mask_match __P((const struct encaptab *, const struct sockaddr *,
		const struct sockaddr *));
static void encap_fillarg __P((struct mbuf *, const struct encaptab *));

/* rely upon BSS initialization */
LIST_HEAD(, encaptab) encaptab = LIST_HEAD_INITIALIZER(&encaptab);

void
encap_init()
{
#if 0
	/*
	 * we cannot use LIST_INIT() here, since drivers may want to call
	 * encap_attach(), on driver attach.  encap_init() wlil be called
	 * on AF_INET{,6} initialization, which happens after driver
	 * initialization - using LIST_INIT() here can nuke encap_attach()
	 * from drivers.
	 */
	LIST_INIT(&encaptab);
#endif
}

#if !(defined(__FreeBSD__) && __FreeBSD__ >= 4)
void
#if __STDC__
encap4_input(struct mbuf *m, ...)
#else
encap4_input(m, va_alist)
	struct mbuf *m;
	va_dcl
#endif
{
	int off, proto;
	struct ip *ip;
	struct sockaddr_in s, d;
	struct encaptab *ep, *match;
	va_list ap;
	int prio, matchprio;

	va_start(ap, m);
	off = va_arg(ap, int);
#ifndef __OpenBSD__
	proto = va_arg(ap, int);
#endif
	va_end(ap);

	ip = mtod(m, struct ip *);
#ifdef __OpenBSD__
	proto = ip->ip_p;
#endif

	bzero(&s, sizeof(s));
	s.sin_family = AF_INET;
	s.sin_len = sizeof(struct sockaddr_in);
	s.sin_addr = ip->ip_src;
	bzero(&d, sizeof(d));
	d.sin_family = AF_INET;
	d.sin_len = sizeof(struct sockaddr_in);
	d.sin_addr = ip->ip_dst;

	match = NULL;
	matchprio = 0;
	for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
		if (ep->af != AF_INET)
			continue;
		if (ep->proto >= 0 && ep->proto != proto)
			continue;
		if (ep->func)
			prio = (*ep->func)(m, off, proto, ep->arg);
		else {
			/*
			 * it's inbound traffic, we need to match in reverse
			 * order
			 */
			prio = mask_match(ep, (struct sockaddr *)&d,
			    (struct sockaddr *)&s);
		}

		/*
		 * We prioritize the matches by using bit length of the
		 * matches.  mask_match() and user-supplied matching function
		 * should return the bit length of the matches (for example,
		 * if both src/dst are matched for IPv4, 64 should be returned).
		 * 0 or negative return value means "it did not match".
		 *
		 * The question is, since we have two "mask" portion, we
		 * cannot really define total order between entries.
		 * For example, which of these should be preferred?
		 * mask_match() returns 48 (32 + 16) for both of them.
		 *	src=3ffe::/16, dst=3ffe:501::/32
		 *	src=3ffe:501::/32, dst=3ffe::/16
		 *
		 * We need to loop through all the possible candidates
		 * to get the best match - the search takes O(n) for
		 * n attachments (i.e. interfaces).
		 */
		if (prio <= 0)
			continue;
		if (prio > matchprio) {
			matchprio = prio;
			match = ep;
		}
	}

	if (match) {
		/* found a match, "match" has the best one */
		if (match->psw && match->psw->pr_input) {
			encap_fillarg(m, match);
			(*match->psw->pr_input)(m, off, proto);
		} else
			m_freem(m);
		return;
	}

	/* backward compatibility clauses */
#ifdef MROUTING
	if (proto == IPPROTO_IPV4 && mrt_ipip_input(m, off)) {
		/*
		 * Multicast routing code claimed this one.  No
		 * more processing at this level.
		 */
		return;
	}
#endif

	/* last resort: inject to raw socket */
	rip_input(m, off, proto);
}
#endif

#ifdef INET6
int
encap6_input(mp, offp, proto)
	struct mbuf **mp;
	int *offp;
	int proto;
{
	struct mbuf *m = *mp;
	struct ip6_hdr *ip6;
	struct sockaddr_in6 s, d;
	struct ip6protosw *psw;
	struct encaptab *ep, *match;
	int prio, matchprio;

	ip6 = mtod(m, struct ip6_hdr *);

	bzero(&s, sizeof(s));
	s.sin6_family = AF_INET6;
	s.sin6_len = sizeof(struct sockaddr_in6);
	s.sin6_addr = ip6->ip6_src;
	bzero(&d, sizeof(d));
	d.sin6_family = AF_INET6;
	d.sin6_len = sizeof(struct sockaddr_in6);
	d.sin6_addr = ip6->ip6_dst;

	match = NULL;
	matchprio = 0;
	for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
		if (ep->af != AF_INET6)
			continue;
		if (ep->proto >= 0 && ep->proto != proto)
			continue;
		if (ep->func)
			prio = (*ep->func)(m, *offp, proto, ep->arg);
		else {
			/*
			 * it's inbound traffic, we need to match in reverse
			 * order
			 */
			prio = mask_match(ep, (struct sockaddr *)&d,
			    (struct sockaddr *)&s);
		}

		/* see encap4_input() for issues here */
		if (prio <= 0)
			continue;
		if (prio > matchprio) {
			matchprio = prio;
			match = ep;
		}
	}

	if (match) {
		/* found a match */
		psw = (struct ip6protosw *)match->psw;
		if (psw && psw->pr_input) {
			encap_fillarg(m, match);
			return (*psw->pr_input)(mp, offp, proto);
		} else {
			m_freem(m);
			return IPPROTO_DONE;
		}
	}

	/* last resort: inject to raw socket */
	return rip6_input(mp, offp, proto);
}
#endif

static void
encap_add(ep)
	struct encaptab *ep;
{

	LIST_INSERT_HEAD(&encaptab, ep, chain);
}

/*
 * sp (src ptr) is always my side, and dp (dst ptr) is always remote side.
 * length of mask (sm and dm) is assumed to be same as sp/dp.
 * Return value will be necessary as input (cookie) for encap_detach().
 */
const struct encaptab *
encap_attach(af, proto, sp, sm, dp, dm, psw, arg)
	int af;
	int proto;
	const struct sockaddr *sp, *sm;
	const struct sockaddr *dp, *dm;
	const struct protosw *psw;
	void *arg;
{
	struct encaptab *ep;
	int error;
	int s;

#if defined(__NetBSD__) || defined(__OpenBSD__)
	s = splsoftnet();
#else
	s = splnet();
#endif
	/* sanity check on args */
	if (sp->sa_len > sizeof(ep->src) || dp->sa_len > sizeof(ep->dst)) {
		error = EINVAL;
		goto fail;
	}
	if (sp->sa_len != dp->sa_len) {
		error = EINVAL;
		goto fail;
	}
	if (af != sp->sa_family || af != dp->sa_family) {
		error = EINVAL;
		goto fail;
	}

	/* check if anyone have already attached with exactly same config */
	for (ep = LIST_FIRST(&encaptab); ep; ep = LIST_NEXT(ep, chain)) {
		if (ep->af != af)
			continue;
		if (ep->proto != proto)
			continue;
		if (ep->src.ss_len != sp->sa_len ||
		    bcmp(&ep->src, sp, sp->sa_len) != 0 ||
		    bcmp(&ep->srcmask, sm, sp->sa_len) != 0)
			continue;
		if (ep->dst.ss_len != dp->sa_len ||
		    bcmp(&ep->dst, dp, dp->sa_len) != 0 ||
		    bcmp(&ep->dstmask, dm, dp->sa_len) != 0)
			continue;

		error = EEXIST;
		goto fail;
	}

	/*
	 * XXX NEED TO CHECK viftable IN THE ip_mroute CODE!!!
	 * XXX Actually, that code needs to be replaced with
	 * XXX new code that uses `gif' tunnels.
	 */

	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
	if (ep == NULL) {
		error = ENOBUFS;
		goto fail;
	}
	bzero(ep, sizeof(*ep));

	ep->af = af;
	ep->proto = proto;
	bcopy(sp, &ep->src, sp->sa_len);
	bcopy(sm, &ep->srcmask, sp->sa_len);
	bcopy(dp, &ep->dst, dp->sa_len);
	bcopy(dm, &ep->dstmask, dp->sa_len);
	ep->psw = psw;
	ep->arg = arg;

	encap_add(ep);

	error = 0;
	splx(s);
	return ep;

fail:
	splx(s);
	return NULL;
}

const struct encaptab *
encap_attach_func(af, proto, func, psw, arg)
	int af;
	int proto;
	int (*func) __P((const struct mbuf *, int, int, void *));
	const struct protosw *psw;
	void *arg;
{
	struct encaptab *ep;
	int error;
	int s;

#if defined(__NetBSD__) || defined(__OpenBSD__)
	s = splsoftnet();
#else
	s = splnet();
#endif
	/* sanity check on args */
	if (!func) {
		error = EINVAL;
		goto fail;
	}

	ep = malloc(sizeof(*ep), M_NETADDR, M_NOWAIT);	/*XXX*/
	if (ep == NULL) {
		error = ENOBUFS;
		goto fail;
	}
	bzero(ep, sizeof(*ep));

	ep->af = af;
	ep->proto = proto;
	ep->func = func;
	ep->psw = psw;
	ep->arg = arg;

	encap_add(ep);

	error = 0;
	splx(s);
	return ep;

fail:
	splx(s);
	return NULL;
}

int
encap_detach(cookie)
	const struct encaptab *cookie;
{
	const struct encaptab *ep = cookie;
	struct encaptab *p;

	for (p = LIST_FIRST(&encaptab); p; p = LIST_NEXT(p, chain)) {
		if (p == ep) {
			LIST_REMOVE(p, chain);
			free(p, M_NETADDR);	/*XXX*/
			return 0;
		}
	}

	return EINVAL;
}

static int
mask_match(ep, sp, dp)
	const struct encaptab *ep;
	const struct sockaddr *sp;
	const struct sockaddr *dp;
{
	struct sockaddr_storage s;
	struct sockaddr_storage d;
	int i;
	u_int8_t *p, *q, *r;
	int matchlen;

	if (sp->sa_len > sizeof(s) || dp->sa_len > sizeof(d))
		return 0;
	if (sp->sa_family != ep->af || dp->sa_family != ep->af)
		return 0;
	if (sp->sa_len != ep->src.ss_len || dp->sa_len != ep->dst.ss_len)
		return 0;

	matchlen = 0;

	p = (u_int8_t *)sp;
	q = (u_int8_t *)&ep->srcmask;
	r = (u_int8_t *)&s;
	for (i = 0 ; i < sp->sa_len; i++) {
		r[i] = p[i] & q[i];
		/* XXX estimate */
		matchlen += (q[i] ? 8 : 0);
	}

	p = (u_int8_t *)dp;
	q = (u_int8_t *)&ep->dstmask;
	r = (u_int8_t *)&d;
	for (i = 0 ; i < dp->sa_len; i++) {
		r[i] = p[i] & q[i];
		/* XXX rough estimate */
		matchlen += (q[i] ? 8 : 0);
	}

	/* need to overwrite len/family portion as we don't compare them */
	s.ss_len = sp->sa_len;
	s.ss_family = sp->sa_family;
	d.ss_len = dp->sa_len;
	d.ss_family = dp->sa_family;

	if (bcmp(&s, &ep->src, ep->src.ss_len) == 0 &&
	    bcmp(&d, &ep->dst, ep->dst.ss_len) == 0) {
		return matchlen;
	} else
		return 0;
}

static void
encap_fillarg(m, ep)
	struct mbuf *m;
	const struct encaptab *ep;
{
#if 0
	m->m_pkthdr.aux = ep->arg;
#else
	struct mbuf *n;

	n = m_aux_add(m, AF_INET, IPPROTO_IPV4);
	if (n) {
		*mtod(n, void **) = ep->arg;
		n->m_len = sizeof(void *);
	}
#endif
}

void *
encap_getarg(m)
	struct mbuf *m;
{
	void *p;
#if 0
	p = m->m_pkthdr.aux;
	m->m_pkthdr.aux = NULL;
	return p;
#else
	struct mbuf *n;

	p = NULL;
	n = m_aux_find(m, AF_INET, IPPROTO_IPV4);
	if (n) {
		if (n->m_len == sizeof(void *))
			p = *mtod(n, void **);
		m_aux_delete(m, n);
	}
	return p;
#endif
}