/*
* Ethernet link layer output routine to send a raw frame to the device.
*
* This assumes that the 14 byte Ethernet header is present and contiguous
* in the first mbuf (if BRIDGE'ing).
*/
int
ether_output_frame(struct ifnet *ifp, struct mbuf *m)
{
int i;
if (PFIL_HOOKED(&V_link_pfil_hook)) {
i = pfil_run_hooks(&V_link_pfil_hook, &m, ifp, PFIL_OUT, NULL);
if (i != 0)
return (EACCES);
if (m == NULL)
return (0);
}
/*
* Queue message on interface, update output statistics if
* successful, and start output if interface not yet active.
*/
return ((ifp->if_transmit)(ifp, m));
}
/*
* 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;
struct m_tag *fwd_tag;
char ip6bufs[INET6_ADDRSTRLEN], ip6bufd[INET6_ADDRSTRLEN];
/*
* 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;
}
if (
#ifdef IPSTEALTH
V_ip6stealth == 0 &&
#endif
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;
}
/*
* 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_copym(m, 0, imin(m->m_pkthdr.len, ICMPV6_PLD_MAXLEN),
M_NOWAIT);
#ifdef IPSTEALTH
if (V_ip6stealth == 0)
#endif
ip6->ip6_hlim -= IPV6_HLIMDEC;
#if defined(IPSEC) || defined(IPSEC_SUPPORT)
if (IPSEC_ENABLED(ipv6)) {
if ((error = IPSEC_FORWARD(ipv6, m)) != 0) {
/* mbuf consumed by IPsec */
m_freem(mcopy);
if (error != EINPROGRESS)
IP6STAT_INC(ip6s_cantforward);
return;
}
/* No IPsec processing required */
}
#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));
rt = rin6.ro_rt;
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;
}
/*
* 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 (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;
else {
RTFREE(rt);
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);
RTFREE(rt);
goto again2;
}
pass:
/* See if the size was changed by the packet filter. */
if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
if (mcopy)
icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0,
IN6_LINKMTU(rt->rt_ifp));
goto bad;
}
error = nd6_output_ifp(rt->rt_ifp, origifp, m, dst, NULL);
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);
}
/*
* 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);
}
void
ip6_forward(struct mbuf *m, int srcrt)
{
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
struct sockaddr_in6 *dst;
struct rtentry *rt;
int error, type = 0, code = 0;
struct mbuf *mcopy = NULL;
struct ifnet *origifp; /* maybe unnecessary */
#ifdef IPSEC
struct secpolicy *sp = NULL;
#endif
#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)) {
#if !defined(FAST_IPSEC)
ipsec6stat.in_polvio++;
#endif
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 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.ip6s_cantforward++;
/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard) */
if (ip6_log_time + ip6_log_interval < time_second) {
ip6_log_time = time_second;
log(LOG_DEBUG,
"cannot forward "
"from %s to %s nxt %d received on %s\n",
ip6_sprintf(&ip6->ip6_src),
ip6_sprintf(&ip6->ip6_dst),
ip6->ip6_nxt,
if_name(m->m_pkthdr.rcvif));
}
m_freem(m);
return;
}
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;
/*
* 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 = ipsec6_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, IP_FORWARDING,
&error);
if (sp == NULL) {
ipsec6stat.out_inval++;
ip6stat.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.out_polvio++;
ip6stat.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 ? */
kprintf("ip6_forward: No IPsec request specified.\n");
ip6stat.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 ?? */
kprintf("ip6_forward: Invalid policy found. %d\n", sp->policy);
key_freesp(sp);
goto skip_ipsec;
}
{
struct ipsec_output_state state;
/*
* 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
*/
bzero(&state, sizeof(state));
state.m = m;
state.ro = NULL; /* update at ipsec6_output_tunnel() */
state.dst = NULL; /* update at ipsec6_output_tunnel() */
error = ipsec6_output_tunnel(&state, sp, 0);
m = state.m;
key_freesp(sp);
if (error) {
/* mbuf is already reclaimed in ipsec6_output_tunnel. */
switch (error) {
case EHOSTUNREACH:
case ENETUNREACH:
case EMSGSIZE:
case ENOBUFS:
case ENOMEM:
break;
default:
kprintf("ip6_output (ipsec): error code %d\n", error);
/* fall through */
case ENOENT:
/* don't show these error codes to the user */
break;
}
ip6stat.ip6s_cantforward++;
if (mcopy) {
#if 0
/* XXX: what icmp ? */
#else
m_freem(mcopy);
#endif
}
m_freem(m);
return;
}
}
skip_ipsec:
#endif /* IPSEC */
dst = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst;
if (!srcrt) {
/*
* ip6_forward_rt.ro_dst.sin6_addr is equal to ip6->ip6_dst
*/
if (ip6_forward_rt.ro_rt == NULL ||
!(ip6_forward_rt.ro_rt->rt_flags & RTF_UP)) {
if (ip6_forward_rt.ro_rt) {
RTFREE(ip6_forward_rt.ro_rt);
ip6_forward_rt.ro_rt = 0;
}
/* this probably fails but give it a try again */
rtalloc_ign((struct route *)&ip6_forward_rt,
RTF_PRCLONING);
}
if (ip6_forward_rt.ro_rt == NULL) {
ip6stat.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);
}
m_freem(m);
return;
}
} else if ((rt = ip6_forward_rt.ro_rt) == NULL ||
!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &dst->sin6_addr)) {
if (ip6_forward_rt.ro_rt) {
RTFREE(ip6_forward_rt.ro_rt);
ip6_forward_rt.ro_rt = 0;
}
bzero(dst, sizeof(*dst));
dst->sin6_len = sizeof(struct sockaddr_in6);
dst->sin6_family = AF_INET6;
dst->sin6_addr = ip6->ip6_dst;
rtalloc_ign((struct route *)&ip6_forward_rt, RTF_PRCLONING);
if (ip6_forward_rt.ro_rt == NULL) {
ip6stat.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);
}
m_freem(m);
return;
}
}
rt = ip6_forward_rt.ro_rt;
/*
* 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).
* [draft-ietf-ipngwg-icmp-v3-02.txt, Section 3.1]
*/
if (in6_addr2scopeid(m->m_pkthdr.rcvif, &ip6->ip6_src) !=
in6_addr2scopeid(rt->rt_ifp, &ip6->ip6_src)) {
ip6stat.ip6s_cantforward++;
ip6stat.ip6s_badscope++;
in6_ifstat_inc(rt->rt_ifp, ifs6_in_discard);
if (ip6_log_time + ip6_log_interval < time_second) {
ip6_log_time = time_second;
log(LOG_DEBUG,
"cannot forward "
"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
ip6_sprintf(&ip6->ip6_src),
ip6_sprintf(&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);
m_freem(m);
return;
}
if (m->m_pkthdr.len > rt->rt_ifp->if_mtu) {
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
mtu = rt->rt_ifp->if_mtu;
#ifdef IPSEC
/*
* When we do IPsec tunnel ingress, we need to play
* with if_mtu 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 = ipsec6_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND,
IP_FORWARDING, &ipsecerror);
if (sp) {
ipsechdrsiz = ipsec6_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
icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
}
m_freem(m);
return;
}
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 (rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt &&
(rt->rt_flags & (RTF_DYNAMIC | RTF_MODIFIED)) == 0) {
if (rt->rt_ifp->if_flags & IFF_POINTOPOINT) {
/*
* 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);
m_freem(m);
return;
}
type = ND_REDIRECT;
}
/*
* Check with the firewall...
*/
if (ip6_fw_enable && ip6_fw_chk_ptr) {
u_short port = 0;
/* If ipfw says divert, we have to just drop packet */
if ((*ip6_fw_chk_ptr)(&ip6, rt->rt_ifp, &port, &m)) {
m_freem(m);
goto freecopy;
}
if (!m)
goto freecopy;
}
/*
* 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) {
/*
* 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
{
kprintf("ip6_forward: outgoing interface is loopback. "
"src %s, dst %s, nxt %d, rcvif %s, outif %s\n",
ip6_sprintf(&ip6->ip6_src),
ip6_sprintf(&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);
/*
* Run through list of hooks for output packets.
*/
if (pfil_has_hooks(&inet6_pfil_hook)) {
error = pfil_run_hooks(&inet6_pfil_hook, &m, rt->rt_ifp,
PFIL_OUT);
if (error != 0)
goto senderr;
if (m == NULL)
goto freecopy;
ip6 = mtod(m, struct ip6_hdr *);
}
/*
* One helper hook function is used by any hook points.
* + from hhook_type we can determine the packet direction:
* HHOOK_TYPE_IPSEC_IN or HHOOK_TYPE_IPSEC_OUT;
* + from hhook_id we can determine address family: AF_INET or AF_INET6;
* + udata contains pointer to enc_softc;
* + ctx_data contains pointer to struct ipsec_ctx_data.
*/
static int
enc_hhook(int32_t hhook_type, int32_t hhook_id, void *udata, void *ctx_data,
void *hdata, struct osd *hosd)
{
struct enchdr hdr;
struct ipsec_ctx_data *ctx;
struct enc_softc *sc;
struct ifnet *ifp, *rcvif;
struct pfil_head *ph;
int pdir;
sc = (struct enc_softc *)udata;
ifp = sc->sc_ifp;
if ((ifp->if_flags & IFF_UP) == 0)
return (0);
ctx = (struct ipsec_ctx_data *)ctx_data;
/* XXX: wrong hook point was used by caller? */
if (ctx->af != hhook_id)
return (EPFNOSUPPORT);
if (((hhook_type == HHOOK_TYPE_IPSEC_IN &&
(ctx->enc & V_bpf_mask_in) != 0) ||
(hhook_type == HHOOK_TYPE_IPSEC_OUT &&
(ctx->enc & V_bpf_mask_out) != 0)) &&
bpf_peers_present(ifp->if_bpf) != 0) {
hdr.af = ctx->af;
hdr.spi = ctx->sav->spi;
hdr.flags = 0;
if (ctx->sav->alg_enc != SADB_EALG_NONE)
hdr.flags |= M_CONF;
if (ctx->sav->alg_auth != SADB_AALG_NONE)
hdr.flags |= M_AUTH;
bpf_mtap2(ifp->if_bpf, &hdr, sizeof(hdr), *ctx->mp);
}
switch (hhook_type) {
case HHOOK_TYPE_IPSEC_IN:
if (ctx->enc == IPSEC_ENC_BEFORE) {
/* Do accounting only once */
if_inc_counter(ifp, IFCOUNTER_IPACKETS, 1);
if_inc_counter(ifp, IFCOUNTER_IBYTES,
(*ctx->mp)->m_pkthdr.len);
}
if ((ctx->enc & V_filter_mask_in) == 0)
return (0); /* skip pfil processing */
pdir = PFIL_IN;
break;
case HHOOK_TYPE_IPSEC_OUT:
if (ctx->enc == IPSEC_ENC_BEFORE) {
/* Do accounting only once */
if_inc_counter(ifp, IFCOUNTER_OPACKETS, 1);
if_inc_counter(ifp, IFCOUNTER_OBYTES,
(*ctx->mp)->m_pkthdr.len);
}
if ((ctx->enc & V_filter_mask_out) == 0)
return (0); /* skip pfil processing */
pdir = PFIL_OUT;
break;
default:
return (EINVAL);
}
switch (hhook_id) {
#ifdef INET
case AF_INET:
ph = &V_inet_pfil_hook;
break;
#endif
#ifdef INET6
case AF_INET6:
ph = &V_inet6_pfil_hook;
break;
#endif
default:
ph = NULL;
}
if (ph == NULL || !PFIL_HOOKED(ph))
return (0);
/* Make a packet looks like it was received on enc(4) */
rcvif = (*ctx->mp)->m_pkthdr.rcvif;
(*ctx->mp)->m_pkthdr.rcvif = ifp;
if (pfil_run_hooks(ph, ctx->mp, ifp, pdir, NULL) != 0 ||
*ctx->mp == NULL) {
*ctx->mp = NULL; /* consumed by filter */
return (EACCES);
}
(*ctx->mp)->m_pkthdr.rcvif = rcvif;
return (0);
}
static
void
ip6_input(netmsg_t msg)
{
struct mbuf *m = msg->packet.nm_packet;
struct ip6_hdr *ip6;
int off = sizeof(struct ip6_hdr), nest;
u_int32_t plen;
u_int32_t rtalert = ~0;
int nxt, ours = 0, rh_present = 0;
struct ifnet *deliverifp = NULL;
struct in6_addr odst;
int srcrt = 0;
#ifdef IPSEC
/*
* should the inner packet be considered authentic?
* see comment in ah4_input().
*/
if (m) {
m->m_flags &= ~M_AUTHIPHDR;
m->m_flags &= ~M_AUTHIPDGM;
}
#endif
/*
* make sure we don't have onion peering information into m_aux.
*/
ip6_delaux(m);
/*
* mbuf statistics
*/
if (m->m_flags & M_EXT) {
if (m->m_next)
ip6stat.ip6s_mext2m++;
else
ip6stat.ip6s_mext1++;
} else {
#define M2MMAX NELEM(ip6stat.ip6s_m2m)
if (m->m_next) {
if (m->m_flags & M_LOOP) {
ip6stat.ip6s_m2m[loif[0].if_index]++; /* XXX */
} else if (m->m_pkthdr.rcvif->if_index < M2MMAX)
ip6stat.ip6s_m2m[m->m_pkthdr.rcvif->if_index]++;
else
ip6stat.ip6s_m2m[0]++;
} else
ip6stat.ip6s_m1++;
#undef M2MMAX
}
in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_receive);
ip6stat.ip6s_total++;
#ifndef PULLDOWN_TEST
/*
* L2 bridge code and some other code can return mbuf chain
* that does not conform to KAME requirement. too bad.
* XXX: fails to join if interface MTU > MCLBYTES. jumbogram?
*/
if (m && m->m_next != NULL && m->m_pkthdr.len < MCLBYTES) {
struct mbuf *n;
n = m_getb(m->m_pkthdr.len, MB_DONTWAIT, MT_HEADER, M_PKTHDR);
if (n == NULL)
goto bad;
M_MOVE_PKTHDR(n, m);
m_copydata(m, 0, n->m_pkthdr.len, mtod(n, caddr_t));
n->m_len = n->m_pkthdr.len;
m_freem(m);
m = n;
}
IP6_EXTHDR_CHECK_VOIDRET(m, 0, sizeof(struct ip6_hdr));
#endif
if (m->m_len < sizeof(struct ip6_hdr)) {
struct ifnet *inifp;
inifp = m->m_pkthdr.rcvif;
if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
ip6stat.ip6s_toosmall++;
in6_ifstat_inc(inifp, ifs6_in_hdrerr);
goto bad2;
}
}
ip6 = mtod(m, struct ip6_hdr *);
if ((ip6->ip6_vfc & IPV6_VERSION_MASK) != IPV6_VERSION) {
ip6stat.ip6s_badvers++;
in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr);
goto bad;
}
/*
* Run through list of hooks for input packets.
*
* NB: Beware of the destination address changing
* (e.g. by NAT rewriting). When this happens,
* tell ip6_forward to do the right thing.
*/
if (pfil_has_hooks(&inet6_pfil_hook)) {
odst = ip6->ip6_dst;
if (pfil_run_hooks(&inet6_pfil_hook, &m,
m->m_pkthdr.rcvif, PFIL_IN)) {
goto bad2;
}
if (m == NULL) /* consumed by filter */
goto bad2;
ip6 = mtod(m, struct ip6_hdr *);
srcrt = !IN6_ARE_ADDR_EQUAL(&odst, &ip6->ip6_dst);
}
static inline int
ip_output_pfil(struct mbuf **mp, struct ifnet *ifp, struct inpcb *inp,
struct sockaddr_in *dst, int *fibnum, int *error)
{
struct m_tag *fwd_tag = NULL;
struct mbuf *m;
struct in_addr odst;
struct ip *ip;
m = *mp;
ip = mtod(m, struct ip *);
/* Run through list of hooks for output packets. */
odst.s_addr = ip->ip_dst.s_addr;
*error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp, PFIL_OUT, inp);
m = *mp;
if ((*error) != 0 || m == NULL)
return 1; /* Finished */
ip = mtod(m, struct ip *);
/* See if destination IP address was changed by packet filter. */
if (odst.s_addr != ip->ip_dst.s_addr) {
m->m_flags |= M_SKIP_FIREWALL;
/* If destination is now ourself drop to ip_input(). */
if (in_localip(ip->ip_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) {
m->m_pkthdr.csum_flags |=
CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
m->m_pkthdr.csum_data = 0xffff;
}
m->m_pkthdr.csum_flags |=
CSUM_IP_CHECKED | CSUM_IP_VALID;
#ifdef SCTP
if (m->m_pkthdr.csum_flags & CSUM_SCTP)
m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
#endif
*error = netisr_queue(NETISR_IP, m);
return 1; /* Finished */
}
bzero(dst, sizeof(*dst));
dst->sin_family = AF_INET;
dst->sin_len = sizeof(*dst);
dst->sin_addr = ip->ip_dst;
return -1; /* Reloop */
}
/* See if fib was changed by packet filter. */
if ((*fibnum) != M_GETFIB(m)) {
m->m_flags |= M_SKIP_FIREWALL;
*fibnum = M_GETFIB(m);
return -1; /* Reloop for FIB change */
}
/* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */
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) {
m->m_pkthdr.csum_flags |=
CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
m->m_pkthdr.csum_data = 0xffff;
}
#ifdef SCTP
if (m->m_pkthdr.csum_flags & CSUM_SCTP)
m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
#endif
m->m_pkthdr.csum_flags |=
CSUM_IP_CHECKED | CSUM_IP_VALID;
*error = netisr_queue(NETISR_IP, m);
return 1; /* Finished */
}
/* Or forward to some other address? */
if ((m->m_flags & M_IP_NEXTHOP) &&
((fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL)) {
bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in));
m->m_flags |= M_SKIP_FIREWALL;
m->m_flags &= ~M_IP_NEXTHOP;
m_tag_delete(m, fwd_tag);
return -1; /* Reloop for CHANGE of dst */
}
return 0;
}
int
ipsec_filter(struct mbuf **mp, int dir, int flags)
{
int error, i;
struct ip *ip;
KASSERT(encif != NULL, ("%s: encif is null", __func__));
KASSERT(flags & (ENC_IN|ENC_OUT),
("%s: invalid flags: %04x", __func__, flags));
if ((encif->if_drv_flags & IFF_DRV_RUNNING) == 0)
return (0);
if (flags & ENC_IN) {
if ((flags & ipsec_filter_mask_in) == 0)
return (0);
} else {
if ((flags & ipsec_filter_mask_out) == 0)
return (0);
}
/* Skip pfil(9) if no filters are loaded */
if (1
#ifdef INET
&& !PFIL_HOOKED(&V_inet_pfil_hook)
#endif
#ifdef INET6
&& !PFIL_HOOKED(&V_inet6_pfil_hook)
#endif
) {
return (0);
}
i = min((*mp)->m_pkthdr.len, max_protohdr);
if ((*mp)->m_len < i) {
*mp = m_pullup(*mp, i);
if (*mp == NULL) {
printf("%s: m_pullup failed\n", __func__);
return (-1);
}
}
error = 0;
ip = mtod(*mp, struct ip *);
switch (ip->ip_v) {
#ifdef INET
case 4:
/*
* before calling the firewall, swap fields the same as
* IP does. here we assume the header is contiguous
*/
ip->ip_len = ntohs(ip->ip_len);
ip->ip_off = ntohs(ip->ip_off);
error = pfil_run_hooks(&V_inet_pfil_hook, mp,
encif, dir, NULL);
if (*mp == NULL || error != 0)
break;
/* restore byte ordering */
ip = mtod(*mp, struct ip *);
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
break;
#endif
#ifdef INET6
case 6:
error = pfil_run_hooks(&V_inet6_pfil_hook, mp,
encif, dir, NULL);
break;
#endif
default:
printf("%s: unknown IP version\n", __func__);
}
/*
* If the mbuf was consumed by the filter for requeueing (dummynet, etc)
* then error will be zero but we still want to return an error to our
* caller so the null mbuf isn't forwarded further.
*/
if (*mp == NULL && error == 0)
return (-1); /* Consumed by the filter */
if (*mp == NULL)
return (error);
if (error != 0)
goto bad;
return (error);
bad:
m_freem(*mp);
*mp = NULL;
return (error);
}
void
ip6_forward(struct mbuf *m, int srcrt)
{
struct ip6_hdr *ip6 = mtod(m, struct ip6_hdr *);
const struct sockaddr_in6 *dst;
struct rtentry *rt;
int error = 0, type = 0, code = 0;
struct mbuf *mcopy = NULL;
struct ifnet *origifp; /* maybe unnecessary */
uint32_t inzone, outzone;
struct in6_addr src_in6, dst_in6;
#ifdef IPSEC
int needipsec = 0;
struct secpolicy *sp = NULL;
#endif
/*
* Clear any in-bound checksum flags for this packet.
*/
m->m_pkthdr.csum_flags = 0;
/*
* 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 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)) {
ip6_cantforward(ip6, m->m_pkthdr.rcvif, NULL,
((m->m_flags & (M_BCAST|M_MCAST)) != 0) ? "bcast/mcast" :
IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ? "mcast/dst" :
"unspec/src");
m_freem(m);
return;
}
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;
/*
* 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
if (ipsec_used) {
/* Check the security policy (SP) for the packet */
sp = ipsec6_check_policy(m, NULL, 0, &needipsec, &error);
if (error != 0) {
/*
* Hack: -EINVAL is used to signal that a packet
* should be silently discarded. This is typically
* because we asked key management for an SA and
* it was delayed (e.g. kicked up to IKE).
*/
if (error == -EINVAL)
error = 0;
goto freecopy;
}
}
#endif /* IPSEC */
if (srcrt) {
union {
struct sockaddr dst;
struct sockaddr_in6 dst6;
} u;
sockaddr_in6_init(&u.dst6, &ip6->ip6_dst, 0, 0, 0);
if ((rt = rtcache_lookup(&ip6_forward_rt, &u.dst)) == NULL) {
IP6_STATINC(IP6_STAT_NOROUTE);
/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_noroute) */
if (mcopy) {
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_NOROUTE, 0);
}
m_freem(m);
return;
}
} else if ((rt = rtcache_validate(&ip6_forward_rt)) == NULL &&
(rt = rtcache_update(&ip6_forward_rt, 1)) == NULL) {
/*
* rtcache_getdst(ip6_forward_rt)->sin6_addr was equal to
* ip6->ip6_dst
*/
IP6_STATINC(IP6_STAT_NOROUTE);
/* XXX in6_ifstat_inc(rt->rt_ifp, ifs6_in_noroute) */
if (mcopy) {
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_NOROUTE, 0);
}
m_freem(m);
return;
}
dst = satocsin6(rtcache_getdst(&ip6_forward_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-v3-07, Section 3.1]
*/
src_in6 = ip6->ip6_src;
inzone = outzone = ~0;
if (in6_setscope(&src_in6, rt->rt_ifp, &outzone) != 0 ||
in6_setscope(&src_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
inzone != outzone) {
ip6_cantforward(ip6, m->m_pkthdr.rcvif, rt->rt_ifp,
"src[%s] inzone %d outzone %d",
in6_getscopename(&ip6->ip6_src), inzone, outzone);
if (mcopy)
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
m_freem(m);
return;
}
#ifdef IPSEC
/*
* If we need to encapsulate the packet, do it here
* ipsec6_proces_packet will send the packet using ip6_output
*/
if (needipsec) {
int s = splsoftnet();
error = ipsec6_process_packet(m, sp->req);
splx(s);
if (mcopy)
goto freecopy;
}
#endif
/*
* 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;
inzone = outzone = ~0;
if (in6_setscope(&dst_in6, m->m_pkthdr.rcvif, &inzone) != 0 ||
in6_setscope(&dst_in6, rt->rt_ifp, &outzone) != 0 ||
inzone != outzone) {
ip6_cantforward(ip6, m->m_pkthdr.rcvif, rt->rt_ifp,
"dst[%s] inzone %d outzone %d",
in6_getscopename(&ip6->ip6_dst), inzone, outzone);
if (mcopy)
icmp6_error(mcopy, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_BEYONDSCOPE, 0);
m_freem(m);
return;
}
if (m->m_pkthdr.len > IN6_LINKMTU(rt->rt_ifp)) {
in6_ifstat_inc(rt->rt_ifp, ifs6_in_toobig);
if (mcopy) {
u_long mtu;
mtu = IN6_LINKMTU(rt->rt_ifp);
icmp6_error(mcopy, ICMP6_PACKET_TOO_BIG, 0, mtu);
}
m_freem(m);
return;
}
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 (rt->rt_ifp == m->m_pkthdr.rcvif && !srcrt && ip6_sendredirects &&
(rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0) {
if ((rt->rt_ifp->if_flags & IFF_POINTOPOINT) &&
nd6_is_addr_neighbor(
satocsin6(rtcache_getdst(&ip6_forward_rt)),
rt->rt_ifp)) {
/*
* If the incoming interface is equal to the outgoing
* one, the link attached to the interface is
* point-to-point, and the IPv6 destination is
* regarded as on-link on the link, then it will be
* highly probable that the destination address does
* not exist on the link and that the packet is going
* to loop. Thus, we immediately drop the packet and
* send an ICMPv6 error message.
* For other routing loops, we dare to let the packet
* go to the loop, so that a remote diagnosing host
* can detect the loop by traceroute.
* 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);
m_freem(m);
return;
}
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(&ip6->ip6_src),
ip6_sprintf(&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);
/*
* Run through list of hooks for output packets.
*/
if ((error = pfil_run_hooks(inet6_pfil_hook, &m, rt->rt_ifp,
PFIL_OUT)) != 0)
goto senderr;
if (m == NULL)
goto freecopy;
ip6 = mtod(m, struct ip6_hdr *);
error = nd6_output(rt->rt_ifp, origifp, m, dst, rt);
if (error) {
in6_ifstat_inc(rt->rt_ifp, ifs6_out_discard);
IP6_STATINC(IP6_STAT_CANTFORWARD);
} else {
IP6_STATINC(IP6_STAT_FORWARD);
in6_ifstat_inc(rt->rt_ifp, ifs6_out_forward);
if (type)
IP6_STATINC(IP6_STAT_REDIRECTSENT);
else {
#ifdef GATEWAY
if (m->m_flags & M_CANFASTFWD)
ip6flow_create(&ip6_forward_rt, m);
#endif
if (mcopy)
goto freecopy;
}
}
senderr:
if (mcopy == NULL)
return;
switch (error) {
case 0:
if (type == ND_REDIRECT) {
icmp6_redirect_output(mcopy, rt);
return;
}
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);
return;
freecopy:
m_freem(mcopy);
return;
}
struct mbuf*
ip6_tryforward(struct mbuf *m)
{
struct sockaddr_in6 dst;
struct nhop6_basic nh;
struct m_tag *fwd_tag;
struct ip6_hdr *ip6;
struct ifnet *rcvif;
uint32_t plen;
int error;
/*
* Fallback conditions to ip6_input for slow path processing.
*/
ip6 = mtod(m, struct ip6_hdr *);
if (ip6->ip6_nxt == IPPROTO_HOPOPTS ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_dst) ||
IN6_IS_ADDR_LINKLOCAL(&ip6->ip6_src) ||
IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src) ||
in6_localip(&ip6->ip6_dst))
return (m);
/*
* Check that the amount of data in the buffers
* is as at least much as the IPv6 header would have us expect.
* Trim mbufs if longer than we expect.
* Drop packet if shorter than we expect.
*/
rcvif = m->m_pkthdr.rcvif;
plen = ntohs(ip6->ip6_plen);
if (plen == 0) {
/*
* Jumbograms must have hop-by-hop header and go via
* slow path.
*/
IP6STAT_INC(ip6s_badoptions);
goto dropin;
}
if (m->m_pkthdr.len - sizeof(struct ip6_hdr) < plen) {
IP6STAT_INC(ip6s_tooshort);
in6_ifstat_inc(rcvif, ifs6_in_truncated);
goto dropin;
}
if (m->m_pkthdr.len > sizeof(struct ip6_hdr) + plen) {
if (m->m_len == m->m_pkthdr.len) {
m->m_len = sizeof(struct ip6_hdr) + plen;
m->m_pkthdr.len = sizeof(struct ip6_hdr) + plen;
} else
m_adj(m, sizeof(struct ip6_hdr) + plen -
m->m_pkthdr.len);
}
/*
* Hop limit.
*/
#ifdef IPSTEALTH
if (!V_ip6stealth)
#endif
if (ip6->ip6_hlim <= IPV6_HLIMDEC) {
icmp6_error(m, ICMP6_TIME_EXCEEDED,
ICMP6_TIME_EXCEED_TRANSIT, 0);
m = NULL;
goto dropin;
}
bzero(&dst, sizeof(dst));
dst.sin6_family = AF_INET6;
dst.sin6_len = sizeof(dst);
dst.sin6_addr = ip6->ip6_dst;
/*
* Incoming packet firewall processing.
*/
if (!PFIL_HOOKED(&V_inet6_pfil_hook))
goto passin;
if (pfil_run_hooks(&V_inet6_pfil_hook, &m, rcvif, PFIL_IN,
NULL) != 0 || m == NULL)
goto dropin;
/*
* If packet filter sets the M_FASTFWD_OURS flag, this means
* that new destination or next hop is our local address.
* So, we can just go back to ip6_input.
* XXX: should we decrement ip6_hlim in such case?
*
* Also it can forward packet to another destination, e.g.
* M_IP6_NEXTHOP flag is set and fwd_tag is attached to mbuf.
*/
if (m->m_flags & M_FASTFWD_OURS)
return (m);
ip6 = mtod(m, struct ip6_hdr *);
if ((m->m_flags & M_IP6_NEXTHOP) &&
(fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
/*
* Now we will find route to forwarded by pfil destination.
*/
bcopy((fwd_tag + 1), &dst, sizeof(dst));
m->m_flags &= ~M_IP6_NEXTHOP;
m_tag_delete(m, fwd_tag);
} else {
/* Update dst since pfil could change it */
dst.sin6_addr = ip6->ip6_dst;
}
passin:
/*
* Find route to destination.
*/
if (ip6_findroute(&nh, &dst, m) != 0) {
m = NULL;
in6_ifstat_inc(rcvif, ifs6_in_noroute);
goto dropin;
}
/*
* We used slow path processing for packets with scoped addresses.
* So, scope checks aren't needed here.
*/
if (m->m_pkthdr.len > nh.nh_mtu) {
in6_ifstat_inc(nh.nh_ifp, ifs6_in_toobig);
icmp6_error(m, ICMP6_PACKET_TOO_BIG, 0, nh.nh_mtu);
m = NULL;
goto dropout;
}
/*
* Outgoing packet firewall processing.
*/
if (!PFIL_HOOKED(&V_inet6_pfil_hook))
goto passout;
if (pfil_run_hooks(&V_inet6_pfil_hook, &m, nh.nh_ifp, PFIL_OUT,
NULL) != 0 || m == NULL)
goto dropout;
/*
* If packet filter sets the M_FASTFWD_OURS flag, this means
* that new destination or next hop is our local address.
* So, we can just go back to ip6_input.
*
* Also it can forward packet to another destination, e.g.
* M_IP6_NEXTHOP flag is set and fwd_tag is attached to mbuf.
*/
if (m->m_flags & M_FASTFWD_OURS) {
/*
* XXX: we did one hop and should decrement hop limit. But
* now we are the destination and just don't pay attention.
*/
return (m);
}
/*
* Again. A packet filter could change the destination address.
*/
ip6 = mtod(m, struct ip6_hdr *);
if (m->m_flags & M_IP6_NEXTHOP)
fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
else
fwd_tag = NULL;
if (fwd_tag != NULL ||
!IN6_ARE_ADDR_EQUAL(&dst.sin6_addr, &ip6->ip6_dst)) {
if (fwd_tag != NULL) {
bcopy((fwd_tag + 1), &dst, sizeof(dst));
m->m_flags &= ~M_IP6_NEXTHOP;
m_tag_delete(m, fwd_tag);
} else
dst.sin6_addr = ip6->ip6_dst;
/*
* Redo route lookup with new destination address
*/
if (ip6_findroute(&nh, &dst, m) != 0) {
m = NULL;
goto dropout;
}
}
passout:
#ifdef IPSTEALTH
if (!V_ip6stealth)
#endif
{
ip6->ip6_hlim -= IPV6_HLIMDEC;
}
m_clrprotoflags(m); /* Avoid confusing lower layers. */
IP_PROBE(send, NULL, NULL, ip6, nh.nh_ifp, NULL, ip6);
/*
* XXX: we need to use destination address with embedded scope
* zone id, because LLTABLE uses such form of addresses for lookup.
*/
dst.sin6_addr = nh.nh_addr;
if (IN6_IS_SCOPE_LINKLOCAL(&dst.sin6_addr))
dst.sin6_addr.s6_addr16[1] = htons(nh.nh_ifp->if_index & 0xffff);
error = (*nh.nh_ifp->if_output)(nh.nh_ifp, m,
(struct sockaddr *)&dst, NULL);
if (error != 0) {
in6_ifstat_inc(nh.nh_ifp, ifs6_out_discard);
IP6STAT_INC(ip6s_cantforward);
} else {
in6_ifstat_inc(nh.nh_ifp, ifs6_out_forward);
IP6STAT_INC(ip6s_forward);
}
return (NULL);
dropin:
in6_ifstat_inc(rcvif, ifs6_in_discard);
goto drop;
dropout:
in6_ifstat_inc(nh.nh_ifp, ifs6_out_discard);
drop:
if (m != NULL)
m_freem(m);
return (NULL);
}
/*
* Ip input routine. Checksum and byte swap header. If fragmented
* try to reassemble. Process options. Pass to next level.
*/
void
ipintr()
{
register struct ip *ip;
register struct mbuf *m;
register struct ipq *fp;
register struct in_ifaddr *ia;
int hlen, s;
int srcrt = 0;
next:
/*
* Get next datagram off input queue and get IP header
* in first mbuf.
*/
s = splimp();
IF_DEQUEUE(&ipintrq, m);
splx(s);
if (m == 0)
return;
#ifdef DIAGNOSTIC
if ((m->m_flags & M_PKTHDR) == 0)
panic("ipintr no HDR");
#endif
/*
* If no IP addresses have been set yet but the interfaces
* are receiving, can't do anything with incoming packets yet.
*/
if (in_ifaddr == NULL)
goto bad;
ipstat.ips_total++;
if (m->m_len < sizeof (struct ip) &&
(m = m_pullup(m, sizeof (struct ip))) == 0) {
ipstat.ips_toosmall++;
goto next;
}
ip = mtod(m, struct ip *);
if (ip->ip_v != IPVERSION) {
ipstat.ips_badvers++;
goto bad;
}
hlen = ip->ip_hl << 2;
if (hlen < sizeof(struct ip)) { /* minimum header length */
ipstat.ips_badhlen++;
goto bad;
}
if (hlen > m->m_len) {
if ((m = m_pullup(m, hlen)) == 0) {
ipstat.ips_badhlen++;
goto next;
}
ip = mtod(m, struct ip *);
}
/*
if (ip->ip_sum = in_cksum(m, hlen)) {
ipstat.ips_badsum++;
goto bad;
}
*/
/*
* Convert fields to host representation.
*/
NTOHS(ip->ip_len);
if (ip->ip_len < hlen) {
ipstat.ips_badlen++;
goto bad;
}
NTOHS(ip->ip_id);
NTOHS(ip->ip_off);
/*
* Check that the amount of data in the buffers
* is as at least much as the IP header would have us expect.
* Trim mbufs if longer than we expect.
* Drop packet if shorter than we expect.
*/
if (m->m_pkthdr.len < ip->ip_len) {
ipstat.ips_tooshort++;
goto bad;
}
if (m->m_pkthdr.len > ip->ip_len) {
if (m->m_len == m->m_pkthdr.len) {
m->m_len = ip->ip_len;
m->m_pkthdr.len = ip->ip_len;
} else
m_adj(m, ip->ip_len - m->m_pkthdr.len);
}
#ifdef PFIL_HOOKS
/*
* Run through list of hooks for input packets. If there are any
* filters which require that additional packets in the flow are
* not fast-forwarded, they must clear the M_CANFASTFWD flag.
* Note that filters must _never_ set this flag, as another filter
* in the list may have previously cleared it.
*/
/*
* let ipfilter look at packet on the wire,
* not the decapsulated packet.
*/
{
struct in_addr odst;
odst = ip->ip_dst;
if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif,
PFIL_IN) != 0)
return;
if (m == NULL)
return;
ip = mtod(m, struct ip *);
hlen = ip->ip_hl << 2;
/*
* XXX The setting of "srcrt" here is to prevent ip_forward()
* from generating ICMP redirects for packets that have
* been redirected by a hook back out on to the same LAN that
* they came from and is not an indication that the packet
* is being inffluenced by source routing options. This
* allows things like
* "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
* where tlp0 is both on the 1.1.1.0/24 network and is the
* default route for hosts on 1.1.1.0/24. Of course this
* also requires a "map tlp0 ..." to complete the story.
* One might argue whether or not this kind of network config.
* should be supported in this manner...
*/
srcrt = (odst.s_addr != ip->ip_dst.s_addr);
}
#endif /* PFIL_HOOKS */
/*
* Process options and, if not destined for us,
* ship it on. ip_dooptions returns 1 when an
* error was detected (causing an icmp message
* to be sent and the original packet to be freed).
*/
ip_nhops = 0; /* for source routed packets */
if (hlen > sizeof (struct ip) && ip_dooptions(m))
goto next;
/*
* Check our list of addresses, to see if the packet is for us.
*/
for (ia = in_ifaddr; ia; ia = ia->ia_next) {
#define satosin(sa) ((struct sockaddr_in *)(sa))
if (IA_SIN(ia)->sin_addr.s_addr == ip->ip_dst.s_addr)
goto ours;
if (
#ifdef DIRECTED_BROADCAST
ia->ia_ifp == m->m_pkthdr.rcvif &&
#endif
(ia->ia_ifp->if_flags & IFF_BROADCAST)) {
u_long t;
if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
ip->ip_dst.s_addr)
goto ours;
if (ip->ip_dst.s_addr == ia->ia_netbroadcast.s_addr)
goto ours;
/*
* Look for all-0's host part (old broadcast addr),
* either for subnet or net.
*/
t = ntohl(ip->ip_dst.s_addr);
if (t == ia->ia_subnet)
goto ours;
if (t == ia->ia_net)
goto ours;
}
}
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
struct in_multi *inm;
#ifdef MROUTING
extern struct socket *ip_mrouter;
if (ip_mrouter) {
/*
* If we are acting as a multicast router, all
* incoming multicast packets are passed to the
* kernel-level multicast forwarding function.
* The packet is returned (relatively) intact; if
* ip_mforward() returns a non-zero value, the packet
* must be discarded, else it may be accepted below.
*
* (The IP ident field is put in the same byte order
* as expected when ip_mforward() is called from
* ip_output().)
*/
ip->ip_id = htons(ip->ip_id);
if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) {
ipstat.ips_cantforward++;
m_freem(m);
goto next;
}
ip->ip_id = ntohs(ip->ip_id);
/*
* The process-level routing demon needs to receive
* all multicast IGMP packets, whether or not this
* host belongs to their destination groups.
*/
if (ip->ip_p == IPPROTO_IGMP)
goto ours;
ipstat.ips_forward++;
}
#endif
/*
* See if we belong to the destination multicast group on the
* arrival interface.
*/
IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
if (inm == NULL) {
ipstat.ips_cantforward++;
m_freem(m);
goto next;
}
goto ours;
}
if (ip->ip_dst.s_addr == (u_long)INADDR_BROADCAST)
goto ours;
if (ip->ip_dst.s_addr == INADDR_ANY)
goto ours;
/*
* Not for us; forward if possible and desirable.
*/
if (ipforwarding == 0) {
ipstat.ips_cantforward++;
m_freem(m);
} else
ip_forward(m, 0);
goto next;
ours:
/*
* If offset or IP_MF are set, must reassemble.
* Otherwise, nothing need be done.
* (We could look in the reassembly queue to see
* if the packet was previously fragmented,
* but it's not worth the time; just let them time out.)
*/
if (ip->ip_off &~ IP_DF) {
if (m->m_flags & M_EXT) { /* XXX */
if ((m = m_pullup(m, sizeof (struct ip))) == 0) {
ipstat.ips_toosmall++;
goto next;
}
ip = mtod(m, struct ip *);
}
/*
* Look for queue of fragments
* of this datagram.
*/
for (fp = ipq.next; fp != &ipq; fp = fp->next)
if (ip->ip_id == fp->ipq_id &&
ip->ip_src.s_addr == fp->ipq_src.s_addr &&
ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
ip->ip_p == fp->ipq_p)
goto found;
fp = 0;
found:
/*
* Adjust ip_len to not reflect header,
* set ip_mff if more fragments are expected,
* convert offset of this to bytes.
*/
ip->ip_len -= hlen;
((struct ipasfrag *)ip)->ipf_mff &= ~1;
if (ip->ip_off & IP_MF)
((struct ipasfrag *)ip)->ipf_mff |= 1;
ip->ip_off <<= 3;
/*
* If datagram marked as having more fragments
* or if this is not the first fragment,
* attempt reassembly; if it succeeds, proceed.
*/
if (((struct ipasfrag *)ip)->ipf_mff & 1 || ip->ip_off) {
ipstat.ips_fragments++;
ip = ip_reass((struct ipasfrag *)ip, fp);
if (ip == 0)
goto next;
ipstat.ips_reassembled++;
m = dtom(ip);
} else if (fp)
ip_freef(fp);
} else
