/*
* Compute the MTU for a forwarded packet that gets IPSEC encapsulated.
* Called from ip_forward().
* Returns MTU suggestion for ICMP needfrag reply.
*/
int
ip_ipsec_mtu(struct mbuf *m, int mtu)
{
/*
* If the packet is routed over IPsec tunnel, tell the
* originator the tunnel MTU.
* tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
* XXX quickhack!!!
*/
return (mtu - ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, NULL));
}
/*
* Compute the MTU for a forwarded packet that gets IPSEC encapsulated.
* Called from ip_forward().
* Returns MTU suggestion for ICMP needfrag reply.
*/
int
ip6_ipsec_mtu(struct mbuf *m)
{
int mtu = 0;
/*
* If the packet is routed over IPsec tunnel, tell the
* originator the tunnel MTU.
* tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
* XXX quickhack!!!
*/
#ifdef IPSEC
struct secpolicy *sp = NULL;
int ipsecerror;
int ipsechdr;
struct route *ro;
sp = ipsec_getpolicybyaddr(m,
IPSEC_DIR_OUTBOUND,
IP_FORWARDING,
&ipsecerror);
if (sp != NULL) {
/* count IPsec header size */
ipsechdr = ipsec_hdrsiz(m, IPSEC_DIR_OUTBOUND, NULL);
/*
* find the correct route for outer IPv4
* header, compute tunnel MTU.
*/
if (sp->req != NULL &&
sp->req->sav != NULL &&
sp->req->sav->sah != NULL) {
ro = &sp->req->sav->sah->route_cache.sa_route;
if (ro->ro_rt && ro->ro_rt->rt_ifp) {
mtu =
ro->ro_rt->rt_rmx.rmx_mtu ?
ro->ro_rt->rt_rmx.rmx_mtu :
ro->ro_rt->rt_ifp->if_mtu;
mtu -= ipsechdr;
}
}
KEY_FREESP(&sp);
}
#endif /* IPSEC */
/* XXX else case missing. */
return mtu;
}
/*
* Forward a packet. If some error occurs return the sender
* an icmp packet. Note we can't always generate a meaningful
* icmp message because icmp doesn't have a large enough repertoire
* of codes and types.
*
* If not forwarding, just drop the packet. This could be confusing
* if ipforwarding was zero but some routing protocol was advancing
* us as a gateway to somewhere. However, we must let the routing
* protocol deal with that.
*
*/
void
ip6_forward(struct mbuf *m, int srcrt)
{
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
struct sockaddr_in6 *dst = NULL;
struct rtentry *rt = NULL;
struct route_in6 rin6;
int error, type = 0, code = 0;
struct mbuf *mcopy = NULL;
struct ifnet *origifp; /* maybe unnecessary */
u_int32_t inzone, outzone;
struct in6_addr src_in6, dst_in6, odst;
#ifdef IPSEC
struct secpolicy *sp = NULL;
#endif
#ifdef SCTP
int sw_csum;
#endif
struct m_tag *fwd_tag;
char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
#ifdef IPSEC
/*
* Check AH/ESP integrity.
*/
/*
* Don't increment ip6s_cantforward because this is the check
* before forwarding packet actually.
*/
if (ipsec6_in_reject(m, NULL)) {
IPSEC6STAT_INC(ips_in_polvio);
m_freem(m);
return;
}
#endif /* IPSEC */
/*
* Do not forward packets to multicast destination (should be handled
* by ip6_mforward().
* Do not forward packets with unspecified source. It was discussed
* in July 2000, on the ipngwg mailing list.
*/
if ((m->m_flags & (M_BCAST|M_MCAST)) != 0 ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
IP6STAT_INC(ip6s_cantforward);
/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
V_ip6_log_time = time_uptime;
log(LOG_DEBUG,
"cannot forward "
"from %s to %s nxt %d received on %s\n",
ip6_sprintf(ip6bufs, &ip6->ip6_src),
ip6_sprintf(ip6bufd, &ip6->ip6_dst),
ip6->ip6_nxt,
if_name(m->m_pkthdr.rcvif));
}
m_freem(m);
return;
}
#ifdef IPSTEALTH
if (!V_ip6stealth) {
#endif
if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
icmp6_error(m, ICMP6_TIME_EXCEEDED,
ICMP6_TIME_EXCEED_TRANSIT, 0);
return;
}
ip6->ip6_hlim -= IPV6_HLIMDEC;
#ifdef IPSTEALTH
}
#endif
/*
* Save at most ICMPV6_PLD_MAXLEN (= the min IPv6 MTU -
* size of IPv6 + ICMPv6 headers) bytes of the packet in case
* we need to generate an ICMP6 message to the src.
* Thanks to M_EXT, in most cases copy will not occur.
*
* It is important to save it before IPsec processing as IPsec
* processing may modify the mbuf.
*/
mcopy = m_copy(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN));
#ifdef IPSEC
/* get a security policy for this packet */
sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND,
IP_FORWARDING, &error);
if (sp == NULL) {
IPSEC6STAT_INC(ips_out_inval);
IP6STAT_INC(ip6s_cantforward);
if (mcopy) {
#if 0
/* XXX: what icmp ? */
#else
m_freem(mcopy);
#endif
}
m_freem(m);
return;
}
error = 0;
/* check policy */
switch (sp->policy) {
case IPSEC_POLICY_DISCARD:
/*
* This packet is just discarded.
*/
IPSEC6STAT_INC(ips_out_polvio);
IP6STAT_INC(ip6s_cantforward);
KEY_FREESP(&sp);
if (mcopy) {
#if 0
/* XXX: what icmp ? */
#else
m_freem(mcopy);
#endif
}
m_freem(m);
return;
case IPSEC_POLICY_BYPASS:
case IPSEC_POLICY_NONE:
/* no need to do IPsec. */
KEY_FREESP(&sp);
goto skip_ipsec;
case IPSEC_POLICY_IPSEC:
if (sp->req == NULL) {
/* XXX should be panic ? */
printf("ip6_forward: No IPsec request specified.\n");
IP6STAT_INC(ip6s_cantforward);
KEY_FREESP(&sp);
if (mcopy) {
#if 0
/* XXX: what icmp ? */
#else
m_freem(mcopy);
#endif
}
m_freem(m);
return;
}
/* do IPsec */
break;
case IPSEC_POLICY_ENTRUST:
default:
/* should be panic ?? */
printf("ip6_forward: Invalid policy found. %d\n", sp->policy);
KEY_FREESP(&sp);
goto skip_ipsec;
}
{
struct ipsecrequest *isr = NULL;
/*
* when the kernel forwards a packet, it is not proper to apply
* IPsec transport mode to the packet is not proper. this check
* avoid from this.
* at present, if there is even a transport mode SA request in the
* security policy, the kernel does not apply IPsec to the packet.
* this check is not enough because the following case is valid.
* ipsec esp/tunnel/xxx-xxx/require esp/transport//require;
*/
for (isr = sp->req; isr; isr = isr->next) {
if (isr->saidx.mode == IPSEC_MODE_ANY)
goto doipsectunnel;
if (isr->saidx.mode == IPSEC_MODE_TUNNEL)
goto doipsectunnel;
}
/*
* if there's no need for tunnel mode IPsec, skip.
*/
if (!isr)
goto skip_ipsec;
doipsectunnel:
/*
* All the extension headers will become inaccessible
* (since they can be encrypted).
* Don't panic, we need no more updates to extension headers
* on inner IPv6 packet (since they are now encapsulated).
*
* IPv6 [ESP|AH] IPv6 [extension headers] payload
*/
/*
* If we need to encapsulate the packet, do it here
* ipsec6_proces_packet will send the packet using ip6_output
*/
error = ipsec6_process_packet(m, sp->req);
KEY_FREESP(&sp);
if (error == EJUSTRETURN) {
/*
* We had a SP with a level of 'use' and no SA. We
* will just continue to process the packet without
* IPsec processing.
*/
error = 0;
goto skip_ipsec;
}
if (error) {
/* mbuf is already reclaimed in ipsec6_process_packet. */
switch (error) {
case EHOSTUNREACH:
case ENETUNREACH:
case EMSGSIZE:
case ENOBUFS:
case ENOMEM:
break;
default:
printf("ip6_output (ipsec): error code %d\n", error);
/* FALLTHROUGH */
case ENOENT:
/* don't show these error codes to the user */
break;
}
IP6STAT_INC(ip6s_cantforward);
if (mcopy) {
#if 0
/* XXX: what icmp ? */
#else
m_freem(mcopy);
#endif
}
return;
} else {
/*
* In the FAST IPSec case we have already
* re-injected the packet and it has been freed
* by the ipsec_done() function. So, just clean
* up after ourselves.
*/
m = NULL;
goto freecopy;
}
}
skip_ipsec:
#endif
again:
bzero(&rin6, sizeof(struct route_in6));
dst = (struct sockaddr_in6 *)&rin6.ro_dst;
dst->sin6_len = sizeof(struct sockaddr_in6);
dst->sin6_family = AF_INET6;
dst->sin6_addr = ip6->ip6_dst;
again2:
rin6.ro_rt = in6_rtalloc1((struct sockaddr *)dst, 0, 0, M_GETFIB(m));
if (rin6.ro_rt != NULL)
RT_UNLOCK(rin6.ro_rt);
else {
IP6STAT_INC(ip6s_noroute);
in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_noroute);
if (mcopy) {
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_NOROUTE, 0);
}
goto bad;
}
rt = rin6.ro_rt;
/*
* Source scope check: if a packet can't be delivered to its
* destination for the reason that the destination is beyond the scope
* of the source address, discard the packet and return an icmp6
* destination unreachable error with Code 2 (beyond scope of source
* address). We use a local copy of ip6_src, since in6_setscope()
* will possibly modify its first argument.
* [draft-ietf-ipngwg-icmp-v3-04.txt, Section 3.1]
*/
src_in6 = ip6->ip6_src;
if (in6_setscope(&src_in6, rt->rt_ifp, &outzone)) {
/* XXX: this should not happen */
IP6STAT_INC(ip6s_cantforward);
IP6STAT_INC(ip6s_badscope);
goto bad;
}
if (in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone)) {
IP6STAT_INC(ip6s_cantforward);
IP6STAT_INC(ip6s_badscope);
goto bad;
}
if (inzone != outzone) {
IP6STAT_INC(ip6s_cantforward);
IP6STAT_INC(ip6s_badscope);
in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
if (V_ip6_log_time + V_ip6_log_interval < time_uptime) {
V_ip6_log_time = time_uptime;
log(LOG_DEBUG,
"cannot forward "
"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
ip6_sprintf(ip6bufs, &ip6->ip6_src),
ip6_sprintf(ip6bufd, &ip6->ip6_dst),
ip6->ip6_nxt,
if_name(m->m_pkthdr.rcvif), if_name(rt->rt_ifp));
}
if (mcopy)
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
goto bad;
}
/*
* Destination scope check: if a packet is going to break the scope
* zone of packet's destination address, discard it. This case should
* usually be prevented by appropriately-configured routing table, but
* we need an explicit check because we may mistakenly forward the
* packet to a different zone by (e.g.) a default route.
*/
dst_in6 = ip6->ip6_dst;
if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
inzone != outzone) {
IP6STAT_INC(ip6s_cantforward);
IP6STAT_INC(ip6s_badscope);
goto bad;
}
if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
if (mcopy) {
u_long mtu;
#ifdef IPSEC
struct secpolicy *sp;
int ipsecerror;
size_t ipsechdrsiz;
#endif /* IPSEC */
mtu = IN6_LINKMTU(rt->rt_ifp);
#ifdef IPSEC
/*
* When we do IPsec tunnel ingress, we need to play
* with the link value (decrement IPsec header size
* from mtu value). The code is much simpler than v4
* case, as we have the outgoing interface for
* encapsulated packet as "rt->rt_ifp".
*/
sp = ipsec_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
IP_FORWARDING, &ipsecerror);
if (sp) {
ipsechdrsiz = ipsec_hdrsiz(mcopy,
IPSEC_DIR_OUTBOUND, NULL);
if (ipsechdrsiz < mtu)
mtu -= ipsechdrsiz;
}
/*
* if mtu becomes less than minimum MTU,
* tell minimum MTU (and I'll need to fragment it).
*/
if (mtu < IPV6_MMTU)
mtu = IPV6_MMTU;
#endif /* IPSEC */
icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
}
goto bad;
}
if (rt->rt_flags & RTF_GATEWAY)
dst = (struct sockaddr_in6 *)rt->rt_gateway;
/*
* If we are to forward the packet using the same interface
* as one we got the packet from, perhaps we should send a redirect
* to sender to shortcut a hop.
* Only send redirect if source is sending directly to us,
* and if packet was not source routed (or has any options).
* Also, don't send redirect if forwarding using a route
* modified by a redirect.
*/
if (V_ip6_sendredirects && rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
(rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) != 0) {
/*
* If the incoming interface is equal to the outgoing
* one, and the link attached to the interface is
* point-to-point, then it will be highly probable
* that a routing loop occurs. Thus, we immediately
* drop the packet and send an ICMPv6 error message.
*
* type/code is based on suggestion by Rich Draves.
* not sure if it is the best pick.
*/
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_ADDR, 0);
goto bad;
}
type = ND_REDIRECT;
}
/*
* Fake scoped addresses. Note that even link-local source or
* destinaion can appear, if the originating node just sends the
* packet to us (without address resolution for the destination).
* Since both icmp6_error and icmp6_redirect_output fill the embedded
* link identifiers, we can do this stuff after making a copy for
* returning an error.
*/
if ((rt->rt_ifp->if_flags & IFF_LOOPBACK) != 0) {
/*
* See corresponding comments in ip6_output.
* XXX: but is it possible that ip6_forward() sends a packet
* to a loopback interface? I don't think so, and thus
* I bark here. ([email protected])
* XXX: it is common to route invalid packets to loopback.
* also, the codepath will be visited on use of ::1 in
* rthdr. (itojun)
*/
#if 1
if (0)
#else
if ((rt->rt_flags & (RTF_BLACKHOLE|RTF_REJECT)) == 0)
#endif
{
printf("ip6_forward: outgoing interface is loopback. "
"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
ip6_sprintf(ip6bufs, &ip6->ip6_src),
ip6_sprintf(ip6bufd, &ip6->ip6_dst),
ip6->ip6_nxt, if_name(m->m_pkthdr.rcvif),
if_name(rt->rt_ifp));
}
/* we can just use rcvif in forwarding. */
origifp = m->m_pkthdr.rcvif;
}
else
origifp = rt->rt_ifp;
/*
* clear embedded scope identifiers if necessary.
* in6_clearscope will touch the addresses only when necessary.
*/
in6_clearscope(&ip6->ip6_src);
in6_clearscope(&ip6->ip6_dst);
/* Jump over all PFIL processing if hooks are not active. */
if (!PFIL_HOOKED(&V_inet6_pfil_hook))
goto pass;
odst = ip6->ip6_dst;
/* Run through list of hooks for output packets. */
error = pfil_run_hooks(&V_inet6_pfil_hook, &m, rt->rt_ifp, PFIL_OUT, NULL);
if (error != 0 || m == NULL)
goto freecopy; /* consumed by filter */
ip6 = mtod(m, struct ip6_hdr *);
/* See if destination IP address was changed by packet filter. */
if (!IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst)) {
m->m_flags |= M_SKIP_FIREWALL;
/* If destination is now ourself drop to ip6_input(). */
if (in6_localip(&ip6->ip6_dst)) {
m->m_flags |= M_FASTFWD_OURS;
if (m->m_pkthdr.rcvif == NULL)
m->m_pkthdr.rcvif = V_loif;
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
m->m_pkthdr.csum_flags |=
CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
m->m_pkthdr.csum_data = 0xffff;
}
#ifdef SCTP
if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
#endif
error = netisr_queue(NETISR_IPV6, m);
goto out;
} else
goto again; /* Redo the routing table lookup. */
}
/* See if local, if yes, send it to netisr. */
if (m->m_flags & M_FASTFWD_OURS) {
if (m->m_pkthdr.rcvif == NULL)
m->m_pkthdr.rcvif = V_loif;
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA_IPV6) {
m->m_pkthdr.csum_flags |=
CSUM_DATA_VALID_IPV6 | CSUM_PSEUDO_HDR;
m->m_pkthdr.csum_data = 0xffff;
}
#ifdef SCTP
if (m->m_pkthdr.csum_flags & CSUM_SCTP_IPV6)
m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
#endif
error = netisr_queue(NETISR_IPV6, m);
goto out;
}
/* Or forward to some other address? */
if ((m->m_flags & M_IP6_NEXTHOP) &&
(fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
dst = (struct sockaddr_in6 *)&rin6.ro_dst;
bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in6));
m->m_flags |= M_SKIP_FIREWALL;
m->m_flags &= ~M_IP6_NEXTHOP;
m_tag_delete(m, fwd_tag);
goto again2;
}
pass:
error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
if (error) {
in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
IP6STAT_INC(ip6s_cantforward);
} else {
IP6STAT_INC(ip6s_forward);
in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
if (type)
IP6STAT_INC(ip6s_redirectsent);
else {
if (mcopy)
goto freecopy;
}
}
if (mcopy == NULL)
goto out;
switch (error) {
case 0:
if (type == ND_REDIRECT) {
icmp6_redirect_output(mcopy, rt);
goto out;
}
goto freecopy;
case EMSGSIZE:
/* xxx MTU is constant in PPP? */
goto freecopy;
case ENOBUFS:
/* Tell source to slow down like source quench in IP? */
goto freecopy;
case ENETUNREACH: /* shouldn't happen, checked above */
case EHOSTUNREACH:
case ENETDOWN:
case EHOSTDOWN:
default:
type = ICMP6_DST_UNREACH;
code = ICMP6_DST_UNREACH_ADDR;
break;
}
icmp6_error(mcopy, type, code, 0);
goto out;
freecopy:
m_freem(mcopy);
goto out;
bad:
m_freem(m);
out:
if (rt != NULL)
RTFREE(rt);
}