2 * Copyright (c) 2003, 2004 Jeffrey M. Hsu. All rights reserved.
3 * Copyright (c) 2003, 2004 The DragonFly Project. All rights reserved.
5 * This code is derived from software contributed to The DragonFly Project
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of The DragonFly Project nor the names of its
17 * contributors may be used to endorse or promote products derived
18 * from this software without specific, prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
24 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
25 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
27 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
30 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35 * Copyright (c) 1982, 1986, 1988, 1993
36 * The Regents of the University of California. All rights reserved.
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. All advertising materials mentioning features or use of this software
47 * must display the following acknowledgement:
48 * This product includes software developed by the University of
49 * California, Berkeley and its contributors.
50 * 4. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
66 * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
67 * $FreeBSD: src/sys/netinet/ip_input.c,v 1.130.2.52 2003/03/07 07:01:28 silby Exp $
68 * $DragonFly: src/sys/netinet/ip_input.c,v 1.115 2008/10/28 07:09:26 sephe Exp $
73 #include "opt_bootp.h"
76 #include "opt_ipdivert.h"
77 #include "opt_ipfilter.h"
78 #include "opt_ipstealth.h"
79 #include "opt_ipsec.h"
81 #include <sys/param.h>
82 #include <sys/systm.h>
84 #include <sys/malloc.h>
85 #include <sys/mpipe.h>
86 #include <sys/domain.h>
87 #include <sys/protosw.h>
88 #include <sys/socket.h>
90 #include <sys/globaldata.h>
91 #include <sys/thread.h>
92 #include <sys/kernel.h>
93 #include <sys/syslog.h>
94 #include <sys/sysctl.h>
95 #include <sys/in_cksum.h>
98 #include <sys/mplock2.h>
100 #include <machine/stdarg.h>
103 #include <net/if_types.h>
104 #include <net/if_var.h>
105 #include <net/if_dl.h>
106 #include <net/pfil.h>
107 #include <net/route.h>
108 #include <net/netisr.h>
110 #include <netinet/in.h>
111 #include <netinet/in_systm.h>
112 #include <netinet/in_var.h>
113 #include <netinet/ip.h>
114 #include <netinet/in_pcb.h>
115 #include <netinet/ip_var.h>
116 #include <netinet/ip_icmp.h>
117 #include <netinet/ip_divert.h>
118 #include <netinet/ip_flow.h>
120 #include <sys/thread2.h>
121 #include <sys/msgport2.h>
122 #include <net/netmsg2.h>
124 #include <sys/socketvar.h>
126 #include <net/ipfw/ip_fw.h>
127 #include <net/dummynet/ip_dummynet.h>
130 #include <netinet6/ipsec.h>
131 #include <netproto/key/key.h>
135 #include <netproto/ipsec/ipsec.h>
136 #include <netproto/ipsec/key.h>
140 static int ip_rsvp_on;
141 struct socket *ip_rsvpd;
144 TUNABLE_INT("net.inet.ip.mpsafe", &ip_mpsafe);
146 int ipforwarding = 0;
147 SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW,
148 &ipforwarding, 0, "Enable IP forwarding between interfaces");
150 static int ipsendredirects = 1; /* XXX */
151 SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW,
152 &ipsendredirects, 0, "Enable sending IP redirects");
154 int ip_defttl = IPDEFTTL;
155 SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW,
156 &ip_defttl, 0, "Maximum TTL on IP packets");
158 static int ip_dosourceroute = 0;
159 SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
160 &ip_dosourceroute, 0, "Enable forwarding source routed IP packets");
162 static int ip_acceptsourceroute = 0;
163 SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
164 CTLFLAG_RW, &ip_acceptsourceroute, 0,
165 "Enable accepting source routed IP packets");
167 static int ip_keepfaith = 0;
168 SYSCTL_INT(_net_inet_ip, IPCTL_KEEPFAITH, keepfaith, CTLFLAG_RW,
170 "Enable packet capture for FAITH IPv4->IPv6 translator daemon");
172 static int nipq = 0; /* total # of reass queues */
174 SYSCTL_INT(_net_inet_ip, OID_AUTO, maxfragpackets, CTLFLAG_RW,
176 "Maximum number of IPv4 fragment reassembly queue entries");
178 static int maxfragsperpacket;
179 SYSCTL_INT(_net_inet_ip, OID_AUTO, maxfragsperpacket, CTLFLAG_RW,
180 &maxfragsperpacket, 0,
181 "Maximum number of IPv4 fragments allowed per packet");
183 static int ip_sendsourcequench = 0;
184 SYSCTL_INT(_net_inet_ip, OID_AUTO, sendsourcequench, CTLFLAG_RW,
185 &ip_sendsourcequench, 0,
186 "Enable the transmission of source quench packets");
188 int ip_do_randomid = 1;
189 SYSCTL_INT(_net_inet_ip, OID_AUTO, random_id, CTLFLAG_RW,
191 "Assign random ip_id values");
193 * XXX - Setting ip_checkinterface mostly implements the receive side of
194 * the Strong ES model described in RFC 1122, but since the routing table
195 * and transmit implementation do not implement the Strong ES model,
196 * setting this to 1 results in an odd hybrid.
198 * XXX - ip_checkinterface currently must be disabled if you use ipnat
199 * to translate the destination address to another local interface.
201 * XXX - ip_checkinterface must be disabled if you add IP aliases
202 * to the loopback interface instead of the interface where the
203 * packets for those addresses are received.
205 static int ip_checkinterface = 0;
206 SYSCTL_INT(_net_inet_ip, OID_AUTO, check_interface, CTLFLAG_RW,
207 &ip_checkinterface, 0, "Verify packet arrives on correct interface");
209 static struct lwkt_token ipq_token = LWKT_TOKEN_MP_INITIALIZER(ipq_token);
212 static int ipprintfs = 0;
215 extern int udp_mpsafe_proto;
216 extern int tcp_mpsafe_proto;
218 extern struct domain inetdomain;
219 extern struct protosw inetsw[];
220 u_char ip_protox[IPPROTO_MAX];
221 struct in_ifaddrhead in_ifaddrheads[MAXCPU]; /* first inet address */
222 struct in_ifaddrhashhead *in_ifaddrhashtbls[MAXCPU];
223 /* inet addr hash table */
224 u_long in_ifaddrhmask; /* mask for hash table */
226 struct ip_stats ipstats_percpu[MAXCPU];
229 sysctl_ipstats(SYSCTL_HANDLER_ARGS)
233 for (cpu = 0; cpu < ncpus; ++cpu) {
234 if ((error = SYSCTL_OUT(req, &ipstats_percpu[cpu],
235 sizeof(struct ip_stats))))
237 if ((error = SYSCTL_IN(req, &ipstats_percpu[cpu],
238 sizeof(struct ip_stats))))
244 SYSCTL_PROC(_net_inet_ip, IPCTL_STATS, stats, (CTLTYPE_OPAQUE | CTLFLAG_RW),
245 0, 0, sysctl_ipstats, "S,ip_stats", "IP statistics");
247 SYSCTL_STRUCT(_net_inet_ip, IPCTL_STATS, stats, CTLFLAG_RW,
248 &ipstat, ip_stats, "IP statistics");
251 /* Packet reassembly stuff */
252 #define IPREASS_NHASH_LOG2 6
253 #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
254 #define IPREASS_HMASK (IPREASS_NHASH - 1)
255 #define IPREASS_HASH(x,y) \
256 (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
258 static struct ipq ipq[IPREASS_NHASH];
261 SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW,
262 &ip_mtu, 0, "Default MTU");
266 static int ipstealth = 0;
267 SYSCTL_INT(_net_inet_ip, OID_AUTO, stealth, CTLFLAG_RW, &ipstealth, 0, "");
269 static const int ipstealth = 0;
272 struct mbuf *(*ip_divert_p)(struct mbuf *, int, int);
274 struct pfil_head inet_pfil_hook;
277 * struct ip_srcrt_opt is used to store packet state while it travels
280 * XXX Note that the code even makes assumptions on the size and
281 * alignment of fields inside struct ip_srcrt so e.g. adding some
282 * fields will break the code. This needs to be fixed.
284 * We need to save the IP options in case a protocol wants to respond
285 * to an incoming packet over the same route if the packet got here
286 * using IP source routing. This allows connection establishment and
287 * maintenance when the remote end is on a network that is not known
291 struct in_addr dst; /* final destination */
292 char nop; /* one NOP to align */
293 char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
294 struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
297 struct ip_srcrt_opt {
299 struct ip_srcrt ip_srcrt;
302 static MALLOC_DEFINE(M_IPQ, "ipq", "IP Fragment Management");
303 static struct malloc_pipe ipq_mpipe;
305 static void save_rte(struct mbuf *, u_char *, struct in_addr);
306 static int ip_dooptions(struct mbuf *m, int, struct sockaddr_in *);
307 static void ip_freef(struct ipq *);
308 static void ip_input_handler(struct netmsg *);
311 * IP initialization: fill in IP protocol switch table.
312 * All protocols not implemented in kernel go to raw IP protocol handler.
325 * Make sure we can handle a reasonable number of fragments but
326 * cap it at 4000 (XXX).
328 mpipe_init(&ipq_mpipe, M_IPQ, sizeof(struct ipq),
329 IFQ_MAXLEN, 4000, 0, NULL);
330 for (i = 0; i < ncpus; ++i) {
331 TAILQ_INIT(&in_ifaddrheads[i]);
332 in_ifaddrhashtbls[i] =
333 hashinit(INADDR_NHASH, M_IFADDR, &in_ifaddrhmask);
335 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
338 for (i = 0; i < IPPROTO_MAX; i++)
339 ip_protox[i] = pr - inetsw;
340 for (pr = inetdomain.dom_protosw;
341 pr < inetdomain.dom_protoswNPROTOSW; pr++) {
342 if (pr->pr_domain->dom_family == PF_INET && pr->pr_protocol) {
343 if (pr->pr_protocol != IPPROTO_RAW)
344 ip_protox[pr->pr_protocol] = pr - inetsw;
347 switch (pr->pr_protocol) {
349 if (tcp_mpsafe_proto)
350 pr->pr_flags |= PR_MPSAFE;
354 if (udp_mpsafe_proto)
355 pr->pr_flags |= PR_MPSAFE;
361 inet_pfil_hook.ph_type = PFIL_TYPE_AF;
362 inet_pfil_hook.ph_af = AF_INET;
363 if ((i = pfil_head_register(&inet_pfil_hook)) != 0) {
364 kprintf("%s: WARNING: unable to register pfil hook, "
365 "error %d\n", __func__, i);
368 for (i = 0; i < IPREASS_NHASH; i++)
369 ipq[i].next = ipq[i].prev = &ipq[i];
371 maxnipq = nmbclusters / 32;
372 maxfragsperpacket = 16;
374 ip_id = time_second & 0xffff;
377 * Initialize IP statistics counters for each CPU.
381 for (cpu = 0; cpu < ncpus; ++cpu) {
382 bzero(&ipstats_percpu[cpu], sizeof(struct ip_stats));
385 bzero(&ipstat, sizeof(struct ip_stats));
388 #if defined(IPSEC) || defined(FAST_IPSEC)
389 /* XXX IPSEC is not MPSAFE yet */
390 flags = NETISR_FLAG_NOTMPSAFE;
393 kprintf("ip: MPSAFE\n");
394 flags = NETISR_FLAG_MPSAFE;
396 flags = NETISR_FLAG_NOTMPSAFE;
399 netisr_register(NETISR_IP, ip_mport_in, ip_mport_pktinfo,
400 ip_input_handler, flags);
403 /* Do transport protocol processing. */
405 transport_processing_oncpu(struct mbuf *m, int hlen, struct ip *ip)
407 const struct protosw *pr = &inetsw[ip_protox[ip->ip_p]];
410 * Switch out to protocol's input routine.
413 pr->pr_input(m, hlen, ip->ip_p);
418 transport_processing_handler(netmsg_t netmsg)
420 struct netmsg_packet *pmsg = (struct netmsg_packet *)netmsg;
424 ip = mtod(pmsg->nm_packet, struct ip *);
425 hlen = pmsg->nm_netmsg.nm_lmsg.u.ms_result;
427 transport_processing_oncpu(pmsg->nm_packet, hlen, ip);
428 /* netmsg was embedded in the mbuf, do not reply! */
432 ip_input_handler(struct netmsg *msg0)
434 struct mbuf *m = ((struct netmsg_packet *)msg0)->nm_packet;
437 /* msg0 was embedded in the mbuf, do not reply! */
441 * IP input routine. Checksum and byte swap header. If fragmented
442 * try to reassemble. Process options. Pass to next level.
445 ip_input(struct mbuf *m)
448 struct in_ifaddr *ia = NULL;
449 struct in_ifaddr_container *iac;
452 struct in_addr pkt_dst;
453 boolean_t using_srcrt = FALSE; /* forward (by PFIL_HOOKS) */
454 boolean_t needredispatch = FALSE;
455 struct in_addr odst; /* original dst address(NAT) */
457 struct sockaddr_in *next_hop = NULL;
459 struct tdb_ident *tdbi;
460 struct secpolicy *sp;
466 if (m->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) {
468 mtag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
469 KKASSERT(mtag != NULL);
470 next_hop = m_tag_data(mtag);
473 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
474 /* dummynet already filtered us */
475 ip = mtod(m, struct ip *);
476 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
482 /* length checks already done in ip_mport() */
483 KASSERT(m->m_len >= sizeof(struct ip), ("IP header not in one mbuf"));
484 ip = mtod(m, struct ip *);
486 if (IP_VHL_V(ip->ip_vhl) != IPVERSION) {
487 ipstat.ips_badvers++;
491 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
492 /* length checks already done in ip_mport() */
493 KASSERT(hlen >= sizeof(struct ip), ("IP header len too small"));
494 KASSERT(m->m_len >= hlen, ("complete IP header not in one mbuf"));
496 /* 127/8 must not appear on wire - RFC1122 */
497 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
498 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
499 if (!(m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK)) {
500 ipstat.ips_badaddr++;
505 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
506 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
508 if (hlen == sizeof(struct ip))
509 sum = in_cksum_hdr(ip);
511 sum = in_cksum(m, hlen);
519 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
520 /* packet is dropped by traffic conditioner */
525 * Convert fields to host representation.
527 ip->ip_len = ntohs(ip->ip_len);
528 ip->ip_off = ntohs(ip->ip_off);
530 /* length checks already done in ip_mport() */
531 KASSERT(ip->ip_len >= hlen, ("total length less then header length"));
532 KASSERT(m->m_pkthdr.len >= ip->ip_len, ("mbuf too short"));
535 * Trim mbufs if longer than the IP header would have us expect.
537 if (m->m_pkthdr.len > ip->ip_len) {
538 if (m->m_len == m->m_pkthdr.len) {
539 m->m_len = ip->ip_len;
540 m->m_pkthdr.len = ip->ip_len;
542 m_adj(m, ip->ip_len - m->m_pkthdr.len);
545 #if defined(IPSEC) && !defined(IPSEC_FILTERGIF)
547 * Bypass packet filtering for packets from a tunnel (gif).
549 if (ipsec_gethist(m, NULL))
555 * Right now when no processing on packet has done
556 * and it is still fresh out of network we do our black
558 * - Firewall: deny/allow/divert
559 * - Xlate: translate packet's addr/port (NAT).
560 * - Pipe: pass pkt through dummynet.
561 * - Wrap: fake packet's addr/port <unimpl.>
562 * - Encapsulate: put it in another IP and send out. <unimp.>
567 * If we've been forwarded from the output side, then
568 * skip the firewall a second time
570 if (next_hop != NULL)
574 if (!pfil_has_hooks(&inet_pfil_hook)) {
575 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
577 * Strip dummynet tags from stranded packets
579 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
580 KKASSERT(mtag != NULL);
581 m_tag_delete(m, mtag);
582 m->m_pkthdr.fw_flags &= ~DUMMYNET_MBUF_TAGGED;
588 * Run through list of hooks for input packets.
590 * NB: Beware of the destination address changing (e.g.
591 * by NAT rewriting). When this happens, tell
592 * ip_forward to do the right thing.
595 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN))
597 if (m == NULL) /* consumed by filter */
599 ip = mtod(m, struct ip *);
600 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
601 using_srcrt = (odst.s_addr != ip->ip_dst.s_addr);
603 if (m->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) {
604 mtag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
605 KKASSERT(mtag != NULL);
606 next_hop = m_tag_data(mtag);
608 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
612 if (m->m_pkthdr.fw_flags & FW_MBUF_REDISPATCH) {
613 needredispatch = TRUE;
614 m->m_pkthdr.fw_flags &= ~FW_MBUF_REDISPATCH;
618 * Process options and, if not destined for us,
619 * ship it on. ip_dooptions returns 1 when an
620 * error was detected (causing an icmp message
621 * to be sent and the original packet to be freed).
623 if (hlen > sizeof(struct ip) && ip_dooptions(m, 0, next_hop))
626 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no
627 * matter if it is destined to another node, or whether it is
628 * a multicast one, RSVP wants it! and prevents it from being forwarded
629 * anywhere else. Also checks if the rsvp daemon is running before
630 * grabbing the packet.
632 if (rsvp_on && ip->ip_p == IPPROTO_RSVP)
636 * Check our list of addresses, to see if the packet is for us.
637 * If we don't have any addresses, assume any unicast packet
638 * we receive might be for us (and let the upper layers deal
641 if (TAILQ_EMPTY(&in_ifaddrheads[mycpuid]) &&
642 !(m->m_flags & (M_MCAST | M_BCAST)))
646 * Cache the destination address of the packet; this may be
647 * changed by use of 'ipfw fwd'.
649 pkt_dst = next_hop ? next_hop->sin_addr : ip->ip_dst;
652 * Enable a consistency check between the destination address
653 * and the arrival interface for a unicast packet (the RFC 1122
654 * strong ES model) if IP forwarding is disabled and the packet
655 * is not locally generated and the packet is not subject to
658 * XXX - Checking also should be disabled if the destination
659 * address is ipnat'ed to a different interface.
661 * XXX - Checking is incompatible with IP aliases added
662 * to the loopback interface instead of the interface where
663 * the packets are received.
665 checkif = ip_checkinterface &&
667 m->m_pkthdr.rcvif != NULL &&
668 !(m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) &&
672 * Check for exact addresses in the hash bucket.
674 LIST_FOREACH(iac, INADDR_HASH(pkt_dst.s_addr), ia_hash) {
678 * If the address matches, verify that the packet
679 * arrived via the correct interface if checking is
682 if (IA_SIN(ia)->sin_addr.s_addr == pkt_dst.s_addr &&
683 (!checkif || ia->ia_ifp == m->m_pkthdr.rcvif))
689 * Check for broadcast addresses.
691 * Only accept broadcast packets that arrive via the matching
692 * interface. Reception of forwarded directed broadcasts would
693 * be handled via ip_forward() and ether_output() with the loopback
694 * into the stack for SIMPLEX interfaces handled by ether_output().
696 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
697 struct ifaddr_container *ifac;
699 TAILQ_FOREACH(ifac, &m->m_pkthdr.rcvif->if_addrheads[mycpuid],
701 struct ifaddr *ifa = ifac->ifa;
703 if (ifa->ifa_addr == NULL) /* shutdown/startup race */
705 if (ifa->ifa_addr->sa_family != AF_INET)
708 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
711 if (ia->ia_netbroadcast.s_addr == pkt_dst.s_addr)
714 if (IA_SIN(ia)->sin_addr.s_addr == INADDR_ANY)
719 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
720 struct in_multi *inm;
722 /* XXX Multicast is not MPSAFE yet */
725 if (ip_mrouter != NULL) {
727 * If we are acting as a multicast router, all
728 * incoming multicast packets are passed to the
729 * kernel-level multicast forwarding function.
730 * The packet is returned (relatively) intact; if
731 * ip_mforward() returns a non-zero value, the packet
732 * must be discarded, else it may be accepted below.
734 if (ip_mforward != NULL &&
735 ip_mforward(ip, m->m_pkthdr.rcvif, m, NULL) != 0) {
737 ipstat.ips_cantforward++;
743 * The process-level routing daemon needs to receive
744 * all multicast IGMP packets, whether or not this
745 * host belongs to their destination groups.
747 if (ip->ip_p == IPPROTO_IGMP) {
751 ipstat.ips_forward++;
754 * See if we belong to the destination multicast group on the
757 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
760 ipstat.ips_notmember++;
768 if (ip->ip_dst.s_addr == INADDR_BROADCAST)
770 if (ip->ip_dst.s_addr == INADDR_ANY)
774 * FAITH(Firewall Aided Internet Translator)
776 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
778 if (ip->ip_p == IPPROTO_TCP || ip->ip_p == IPPROTO_ICMP)
786 * Not for us; forward if possible and desirable.
789 ipstat.ips_cantforward++;
794 * Enforce inbound IPsec SPD.
796 if (ipsec4_in_reject(m, NULL)) {
797 ipsecstat.in_polvio++;
802 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
805 tdbi = (struct tdb_ident *)m_tag_data(mtag);
806 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
808 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
809 IP_FORWARDING, &error);
811 if (sp == NULL) { /* NB: can happen if error */
813 /*XXX error stat???*/
814 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
819 * Check security policy against packet attributes.
821 error = ipsec_in_reject(sp, m);
825 ipstat.ips_cantforward++;
829 ip_forward(m, using_srcrt, next_hop);
836 * IPSTEALTH: Process non-routing options only
837 * if the packet is destined for us.
840 hlen > sizeof(struct ip) &&
841 ip_dooptions(m, 1, next_hop))
844 /* Count the packet in the ip address stats */
846 ia->ia_ifa.if_ipackets++;
847 ia->ia_ifa.if_ibytes += m->m_pkthdr.len;
851 * If offset or IP_MF are set, must reassemble.
852 * Otherwise, nothing need be done.
853 * (We could look in the reassembly queue to see
854 * if the packet was previously fragmented,
855 * but it's not worth the time; just let them time out.)
857 if (ip->ip_off & (IP_MF | IP_OFFMASK)) {
859 * Attempt reassembly; if it succeeds, proceed.
860 * ip_reass() will return a different mbuf.
865 ip = mtod(m, struct ip *);
867 /* Get the header length of the reassembled packet */
868 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
870 needredispatch = TRUE;
877 * enforce IPsec policy checking if we are seeing last header.
878 * note that we do not visit this with protocols with pcb layer
879 * code - like udp/tcp/raw ip.
881 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) &&
882 ipsec4_in_reject(m, NULL)) {
883 ipsecstat.in_polvio++;
889 * enforce IPsec policy checking if we are seeing last header.
890 * note that we do not visit this with protocols with pcb layer
891 * code - like udp/tcp/raw ip.
893 if (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) {
895 * Check if the packet has already had IPsec processing
896 * done. If so, then just pass it along. This tag gets
897 * set during AH, ESP, etc. input handling, before the
898 * packet is returned to the ip input queue for delivery.
900 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
903 tdbi = (struct tdb_ident *)m_tag_data(mtag);
904 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
906 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
907 IP_FORWARDING, &error);
911 * Check security policy against packet attributes.
913 error = ipsec_in_reject(sp, m);
916 /* XXX error stat??? */
918 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
925 #endif /* FAST_IPSEC */
928 * NOTE: ip_len in host form and adjusted down by hlen for
929 * protocol processing.
931 ipstat.ips_delivered++;
932 if (needredispatch) {
933 struct netmsg_packet *pmsg;
936 ip->ip_off = htons(ip->ip_off);
937 ip->ip_len = htons(ip->ip_len + hlen);
938 port = ip_mport_in(&m);
942 pmsg = &m->m_hdr.mh_netmsg;
943 netmsg_init(&pmsg->nm_netmsg, NULL, &netisr_apanic_rport,
944 MSGF_MPSAFE, transport_processing_handler);
946 pmsg->nm_netmsg.nm_lmsg.u.ms_result = hlen;
948 ip = mtod(m, struct ip *);
949 ip->ip_len = ntohs(ip->ip_len) - hlen;
950 ip->ip_off = ntohs(ip->ip_off);
951 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg);
953 transport_processing_oncpu(m, hlen, ip);
962 * Take incoming datagram fragment and try to reassemble it into
963 * whole datagram. If a chain for reassembly of this datagram already
964 * exists, then it is given as fp; otherwise have to make a chain.
967 ip_reass(struct mbuf *m)
969 struct ip *ip = mtod(m, struct ip *);
970 struct mbuf *p = NULL, *q, *nq;
972 struct ipq *fp = NULL;
973 int hlen = IP_VHL_HL(ip->ip_vhl) << 2;
977 /* If maxnipq is 0, never accept fragments. */
979 ipstat.ips_fragments++;
980 ipstat.ips_fragdropped++;
985 sum = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
987 * Look for queue of fragments of this datagram.
989 lwkt_gettoken(&ipq_token);
990 for (fp = ipq[sum].next; fp != &ipq[sum]; fp = fp->next) {
991 if (ip->ip_id == fp->ipq_id &&
992 ip->ip_src.s_addr == fp->ipq_src.s_addr &&
993 ip->ip_dst.s_addr == fp->ipq_dst.s_addr &&
994 ip->ip_p == fp->ipq_p)
1001 * Enforce upper bound on number of fragmented packets
1002 * for which we attempt reassembly;
1003 * If maxnipq is -1, accept all fragments without limitation.
1005 if (nipq > maxnipq && maxnipq > 0) {
1007 * drop something from the tail of the current queue
1008 * before proceeding further
1010 if (ipq[sum].prev == &ipq[sum]) { /* gak */
1011 for (i = 0; i < IPREASS_NHASH; i++) {
1012 if (ipq[i].prev != &ipq[i]) {
1013 ipstat.ips_fragtimeout +=
1014 ipq[i].prev->ipq_nfrags;
1015 ip_freef(ipq[i].prev);
1020 ipstat.ips_fragtimeout +=
1021 ipq[sum].prev->ipq_nfrags;
1022 ip_freef(ipq[sum].prev);
1027 * Adjust ip_len to not reflect header,
1028 * convert offset of this to bytes.
1031 if (ip->ip_off & IP_MF) {
1033 * Make sure that fragments have a data length
1034 * that's a non-zero multiple of 8 bytes.
1036 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1037 ipstat.ips_toosmall++; /* XXX */
1041 m->m_flags |= M_FRAG;
1043 m->m_flags &= ~M_FRAG;
1047 ipstat.ips_fragments++;
1048 m->m_pkthdr.header = ip;
1051 * If the hardware has not done csum over this fragment
1052 * then csum_data is not valid at all.
1054 if ((m->m_pkthdr.csum_flags & (CSUM_FRAG_NOT_CHECKED | CSUM_DATA_VALID))
1055 == (CSUM_FRAG_NOT_CHECKED | CSUM_DATA_VALID)) {
1056 m->m_pkthdr.csum_data = 0;
1057 m->m_pkthdr.csum_flags &= ~(CSUM_DATA_VALID | CSUM_PSEUDO_HDR);
1061 * Presence of header sizes in mbufs
1062 * would confuse code below.
1068 * If first fragment to arrive, create a reassembly queue.
1071 if ((fp = mpipe_alloc_nowait(&ipq_mpipe)) == NULL)
1073 insque(fp, &ipq[sum]);
1076 fp->ipq_ttl = IPFRAGTTL;
1077 fp->ipq_p = ip->ip_p;
1078 fp->ipq_id = ip->ip_id;
1079 fp->ipq_src = ip->ip_src;
1080 fp->ipq_dst = ip->ip_dst;
1082 m->m_nextpkt = NULL;
1088 #define GETIP(m) ((struct ip*)((m)->m_pkthdr.header))
1091 * Find a segment which begins after this one does.
1093 for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt) {
1094 if (GETIP(q)->ip_off > ip->ip_off)
1099 * If there is a preceding segment, it may provide some of
1100 * our data already. If so, drop the data from the incoming
1101 * segment. If it provides all of our data, drop us, otherwise
1102 * stick new segment in the proper place.
1104 * If some of the data is dropped from the the preceding
1105 * segment, then it's checksum is invalidated.
1108 i = GETIP(p)->ip_off + GETIP(p)->ip_len - ip->ip_off;
1110 if (i >= ip->ip_len)
1113 m->m_pkthdr.csum_flags = 0;
1117 m->m_nextpkt = p->m_nextpkt;
1120 m->m_nextpkt = fp->ipq_frags;
1125 * While we overlap succeeding segments trim them or,
1126 * if they are completely covered, dequeue them.
1128 for (; q != NULL && ip->ip_off + ip->ip_len > GETIP(q)->ip_off;
1130 i = (ip->ip_off + ip->ip_len) - GETIP(q)->ip_off;
1131 if (i < GETIP(q)->ip_len) {
1132 GETIP(q)->ip_len -= i;
1133 GETIP(q)->ip_off += i;
1135 q->m_pkthdr.csum_flags = 0;
1140 ipstat.ips_fragdropped++;
1142 q->m_nextpkt = NULL;
1148 * Check for complete reassembly and perform frag per packet
1151 * Frag limiting is performed here so that the nth frag has
1152 * a chance to complete the packet before we drop the packet.
1153 * As a result, n+1 frags are actually allowed per packet, but
1154 * only n will ever be stored. (n = maxfragsperpacket.)
1158 for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt) {
1159 if (GETIP(q)->ip_off != next) {
1160 if (fp->ipq_nfrags > maxfragsperpacket) {
1161 ipstat.ips_fragdropped += fp->ipq_nfrags;
1166 next += GETIP(q)->ip_len;
1168 /* Make sure the last packet didn't have the IP_MF flag */
1169 if (p->m_flags & M_FRAG) {
1170 if (fp->ipq_nfrags > maxfragsperpacket) {
1171 ipstat.ips_fragdropped += fp->ipq_nfrags;
1178 * Reassembly is complete. Make sure the packet is a sane size.
1182 if (next + (IP_VHL_HL(ip->ip_vhl) << 2) > IP_MAXPACKET) {
1183 ipstat.ips_toolong++;
1184 ipstat.ips_fragdropped += fp->ipq_nfrags;
1190 * Concatenate fragments.
1197 q->m_nextpkt = NULL;
1198 for (q = nq; q != NULL; q = nq) {
1200 q->m_nextpkt = NULL;
1201 m->m_pkthdr.csum_flags &= q->m_pkthdr.csum_flags;
1202 m->m_pkthdr.csum_data += q->m_pkthdr.csum_data;
1207 * Clean up the 1's complement checksum. Carry over 16 bits must
1208 * be added back. This assumes no more then 65535 packet fragments
1209 * were reassembled. A second carry can also occur (but not a third).
1211 m->m_pkthdr.csum_data = (m->m_pkthdr.csum_data & 0xffff) +
1212 (m->m_pkthdr.csum_data >> 16);
1213 if (m->m_pkthdr.csum_data > 0xFFFF)
1214 m->m_pkthdr.csum_data -= 0xFFFF;
1217 * Create header for new ip packet by
1218 * modifying header of first packet;
1219 * dequeue and discard fragment reassembly header.
1220 * Make header visible.
1223 ip->ip_src = fp->ipq_src;
1224 ip->ip_dst = fp->ipq_dst;
1227 mpipe_free(&ipq_mpipe, fp);
1228 m->m_len += (IP_VHL_HL(ip->ip_vhl) << 2);
1229 m->m_data -= (IP_VHL_HL(ip->ip_vhl) << 2);
1230 /* some debugging cruft by sklower, below, will go away soon */
1231 if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */
1234 for (n = m; n; n = n->m_next)
1236 m->m_pkthdr.len = plen;
1239 ipstat.ips_reassembled++;
1240 lwkt_reltoken(&ipq_token);
1244 ipstat.ips_fragdropped++;
1249 lwkt_reltoken(&ipq_token);
1256 * Free a fragment reassembly header and all
1257 * associated datagrams.
1259 * Called with ipq_token held.
1262 ip_freef(struct ipq *fp)
1267 * Remove first to protect against blocking
1272 * Clean out at our leisure
1274 while (fp->ipq_frags) {
1276 fp->ipq_frags = q->m_nextpkt;
1277 q->m_nextpkt = NULL;
1280 mpipe_free(&ipq_mpipe, fp);
1285 * IP timer processing;
1286 * if a timer expires on a reassembly
1287 * queue, discard it.
1295 lwkt_gettoken(&ipq_token);
1296 for (i = 0; i < IPREASS_NHASH; i++) {
1300 while (fp != &ipq[i]) {
1303 if (fp->prev->ipq_ttl == 0) {
1304 ipstat.ips_fragtimeout += fp->prev->ipq_nfrags;
1310 * If we are over the maximum number of fragments
1311 * (due to the limit being lowered), drain off
1312 * enough to get down to the new limit.
1314 if (maxnipq >= 0 && nipq > maxnipq) {
1315 for (i = 0; i < IPREASS_NHASH; i++) {
1316 while (nipq > maxnipq &&
1317 (ipq[i].next != &ipq[i])) {
1318 ipstat.ips_fragdropped +=
1319 ipq[i].next->ipq_nfrags;
1320 ip_freef(ipq[i].next);
1324 lwkt_reltoken(&ipq_token);
1329 * Drain off all datagram fragments.
1336 lwkt_gettoken(&ipq_token);
1337 for (i = 0; i < IPREASS_NHASH; i++) {
1338 while (ipq[i].next != &ipq[i]) {
1339 ipstat.ips_fragdropped += ipq[i].next->ipq_nfrags;
1340 ip_freef(ipq[i].next);
1343 lwkt_reltoken(&ipq_token);
1348 * Do option processing on a datagram,
1349 * possibly discarding it if bad options are encountered,
1350 * or forwarding it if source-routed.
1351 * The pass argument is used when operating in the IPSTEALTH
1352 * mode to tell what options to process:
1353 * [LS]SRR (pass 0) or the others (pass 1).
1354 * The reason for as many as two passes is that when doing IPSTEALTH,
1355 * non-routing options should be processed only if the packet is for us.
1356 * Returns 1 if packet has been forwarded/freed,
1357 * 0 if the packet should be processed further.
1360 ip_dooptions(struct mbuf *m, int pass, struct sockaddr_in *next_hop)
1362 struct sockaddr_in ipaddr = { sizeof ipaddr, AF_INET };
1363 struct ip *ip = mtod(m, struct ip *);
1365 struct in_ifaddr *ia;
1366 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB;
1367 boolean_t forward = FALSE;
1368 struct in_addr *sin, dst;
1372 cp = (u_char *)(ip + 1);
1373 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
1374 for (; cnt > 0; cnt -= optlen, cp += optlen) {
1375 opt = cp[IPOPT_OPTVAL];
1376 if (opt == IPOPT_EOL)
1378 if (opt == IPOPT_NOP)
1381 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1382 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1385 optlen = cp[IPOPT_OLEN];
1386 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1387 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1397 * Source routing with record.
1398 * Find interface with current destination address.
1399 * If none on this machine then drop if strictly routed,
1400 * or do nothing if loosely routed.
1401 * Record interface address and bring up next address
1402 * component. If strictly routed make sure next
1403 * address is on directly accessible net.
1407 if (ipstealth && pass > 0)
1409 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1410 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1413 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1414 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1417 ipaddr.sin_addr = ip->ip_dst;
1418 ia = (struct in_ifaddr *)
1419 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
1421 if (opt == IPOPT_SSRR) {
1422 type = ICMP_UNREACH;
1423 code = ICMP_UNREACH_SRCFAIL;
1426 if (!ip_dosourceroute)
1427 goto nosourcerouting;
1429 * Loose routing, and not at next destination
1430 * yet; nothing to do except forward.
1434 off--; /* 0 origin */
1435 if (off > optlen - (int)sizeof(struct in_addr)) {
1437 * End of source route. Should be for us.
1439 if (!ip_acceptsourceroute)
1440 goto nosourcerouting;
1441 save_rte(m, cp, ip->ip_src);
1446 if (!ip_dosourceroute) {
1448 char buf[sizeof "aaa.bbb.ccc.ddd"];
1451 * Acting as a router, so generate ICMP
1454 strcpy(buf, inet_ntoa(ip->ip_dst));
1456 "attempted source route from %s to %s\n",
1457 inet_ntoa(ip->ip_src), buf);
1458 type = ICMP_UNREACH;
1459 code = ICMP_UNREACH_SRCFAIL;
1463 * Not acting as a router,
1467 ipstat.ips_cantforward++;
1474 * locate outgoing interface
1476 memcpy(&ipaddr.sin_addr, cp + off,
1477 sizeof ipaddr.sin_addr);
1479 if (opt == IPOPT_SSRR) {
1480 #define INA struct in_ifaddr *
1481 #define SA struct sockaddr *
1482 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr))
1484 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
1486 ia = ip_rtaddr(ipaddr.sin_addr, NULL);
1489 type = ICMP_UNREACH;
1490 code = ICMP_UNREACH_SRCFAIL;
1493 ip->ip_dst = ipaddr.sin_addr;
1494 memcpy(cp + off, &IA_SIN(ia)->sin_addr,
1495 sizeof(struct in_addr));
1496 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1498 * Let ip_intr's mcast routing check handle mcast pkts
1500 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
1504 if (ipstealth && pass == 0)
1506 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1507 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1510 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1511 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1515 * If no space remains, ignore.
1517 off--; /* 0 origin */
1518 if (off > optlen - (int)sizeof(struct in_addr))
1520 memcpy(&ipaddr.sin_addr, &ip->ip_dst,
1521 sizeof ipaddr.sin_addr);
1523 * locate outgoing interface; if we're the destination,
1524 * use the incoming interface (should be same).
1526 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == NULL &&
1527 (ia = ip_rtaddr(ipaddr.sin_addr, NULL)) == NULL) {
1528 type = ICMP_UNREACH;
1529 code = ICMP_UNREACH_HOST;
1532 memcpy(cp + off, &IA_SIN(ia)->sin_addr,
1533 sizeof(struct in_addr));
1534 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1538 if (ipstealth && pass == 0)
1540 code = cp - (u_char *)ip;
1541 if (optlen < 4 || optlen > 40) {
1542 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1545 if ((off = cp[IPOPT_OFFSET]) < 5) {
1546 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1549 if (off > optlen - (int)sizeof(int32_t)) {
1550 cp[IPOPT_OFFSET + 1] += (1 << 4);
1551 if ((cp[IPOPT_OFFSET + 1] & 0xf0) == 0) {
1552 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1557 off--; /* 0 origin */
1558 sin = (struct in_addr *)(cp + off);
1559 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
1561 case IPOPT_TS_TSONLY:
1564 case IPOPT_TS_TSANDADDR:
1565 if (off + sizeof(n_time) +
1566 sizeof(struct in_addr) > optlen) {
1567 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1570 ipaddr.sin_addr = dst;
1571 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
1575 memcpy(sin, &IA_SIN(ia)->sin_addr,
1576 sizeof(struct in_addr));
1577 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1578 off += sizeof(struct in_addr);
1581 case IPOPT_TS_PRESPEC:
1582 if (off + sizeof(n_time) +
1583 sizeof(struct in_addr) > optlen) {
1584 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1587 memcpy(&ipaddr.sin_addr, sin,
1588 sizeof(struct in_addr));
1589 if (ifa_ifwithaddr((SA)&ipaddr) == NULL)
1591 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1592 off += sizeof(struct in_addr);
1596 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
1600 memcpy(cp + off, &ntime, sizeof(n_time));
1601 cp[IPOPT_OFFSET] += sizeof(n_time);
1604 if (forward && ipforwarding) {
1605 ip_forward(m, TRUE, next_hop);
1610 icmp_error(m, type, code, 0, 0);
1611 ipstat.ips_badoptions++;
1616 * Given address of next destination (final or next hop),
1617 * return internet address info of interface to be used to get there.
1620 ip_rtaddr(struct in_addr dst, struct route *ro0)
1622 struct route sro, *ro;
1623 struct sockaddr_in *sin;
1624 struct in_ifaddr *ia;
1629 bzero(&sro, sizeof(sro));
1633 sin = (struct sockaddr_in *)&ro->ro_dst;
1635 if (ro->ro_rt == NULL || dst.s_addr != sin->sin_addr.s_addr) {
1636 if (ro->ro_rt != NULL) {
1640 sin->sin_family = AF_INET;
1641 sin->sin_len = sizeof *sin;
1642 sin->sin_addr = dst;
1643 rtalloc_ign(ro, RTF_PRCLONING);
1646 if (ro->ro_rt == NULL)
1649 ia = ifatoia(ro->ro_rt->rt_ifa);
1657 * Save incoming source route for use in replies,
1658 * to be picked up later by ip_srcroute if the receiver is interested.
1661 save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
1664 struct ip_srcrt_opt *opt;
1667 mtag = m_tag_get(PACKET_TAG_IPSRCRT, sizeof(*opt), MB_DONTWAIT);
1670 opt = m_tag_data(mtag);
1672 olen = option[IPOPT_OLEN];
1675 kprintf("save_rte: olen %d\n", olen);
1677 if (olen > sizeof(opt->ip_srcrt) - (1 + sizeof(dst))) {
1681 bcopy(option, opt->ip_srcrt.srcopt, olen);
1682 opt->ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1683 opt->ip_srcrt.dst = dst;
1684 m_tag_prepend(m, mtag);
1688 * Retrieve incoming source route for use in replies,
1689 * in the same form used by setsockopt.
1690 * The first hop is placed before the options, will be removed later.
1693 ip_srcroute(struct mbuf *m0)
1695 struct in_addr *p, *q;
1698 struct ip_srcrt_opt *opt;
1703 mtag = m_tag_find(m0, PACKET_TAG_IPSRCRT, NULL);
1706 opt = m_tag_data(mtag);
1708 if (opt->ip_nhops == 0)
1710 m = m_get(MB_DONTWAIT, MT_HEADER);
1714 #define OPTSIZ (sizeof(opt->ip_srcrt.nop) + sizeof(opt->ip_srcrt.srcopt))
1716 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1717 m->m_len = opt->ip_nhops * sizeof(struct in_addr) +
1718 sizeof(struct in_addr) + OPTSIZ;
1721 kprintf("ip_srcroute: nhops %d mlen %d",
1722 opt->ip_nhops, m->m_len);
1727 * First save first hop for return route
1729 p = &opt->ip_srcrt.route[opt->ip_nhops - 1];
1730 *(mtod(m, struct in_addr *)) = *p--;
1733 kprintf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
1737 * Copy option fields and padding (nop) to mbuf.
1739 opt->ip_srcrt.nop = IPOPT_NOP;
1740 opt->ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1741 memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), &opt->ip_srcrt.nop,
1743 q = (struct in_addr *)(mtod(m, caddr_t) +
1744 sizeof(struct in_addr) + OPTSIZ);
1747 * Record return path as an IP source route,
1748 * reversing the path (pointers are now aligned).
1750 while (p >= opt->ip_srcrt.route) {
1753 kprintf(" %x", ntohl(q->s_addr));
1758 * Last hop goes to final destination.
1760 *q = opt->ip_srcrt.dst;
1761 m_tag_delete(m0, mtag);
1764 kprintf(" %x\n", ntohl(q->s_addr));
1770 * Strip out IP options.
1773 ip_stripoptions(struct mbuf *m)
1776 struct ip *ip = mtod(m, struct ip *);
1780 optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
1781 opts = (caddr_t)(ip + 1);
1782 datalen = m->m_len - (sizeof(struct ip) + optlen);
1783 bcopy(opts + optlen, opts, datalen);
1785 if (m->m_flags & M_PKTHDR)
1786 m->m_pkthdr.len -= optlen;
1787 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2);
1790 u_char inetctlerrmap[PRC_NCMDS] = {
1792 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1793 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1794 EMSGSIZE, EHOSTUNREACH, 0, 0,
1796 ENOPROTOOPT, ECONNREFUSED
1800 * Forward a packet. If some error occurs return the sender
1801 * an icmp packet. Note we can't always generate a meaningful
1802 * icmp message because icmp doesn't have a large enough repertoire
1803 * of codes and types.
1805 * If not forwarding, just drop the packet. This could be confusing
1806 * if ipforwarding was zero but some routing protocol was advancing
1807 * us as a gateway to somewhere. However, we must let the routing
1808 * protocol deal with that.
1810 * The using_srcrt parameter indicates whether the packet is being forwarded
1811 * via a source route.
1814 ip_forward(struct mbuf *m, boolean_t using_srcrt, struct sockaddr_in *next_hop)
1816 struct ip *ip = mtod(m, struct ip *);
1818 struct route fwd_ro;
1819 int error, type = 0, code = 0, destmtu = 0;
1822 struct in_addr pkt_dst;
1826 * Cache the destination address of the packet; this may be
1827 * changed by use of 'ipfw fwd'.
1829 pkt_dst = (next_hop != NULL) ? next_hop->sin_addr : ip->ip_dst;
1833 kprintf("forward: src %x dst %x ttl %x\n",
1834 ip->ip_src.s_addr, pkt_dst.s_addr, ip->ip_ttl);
1837 if (m->m_flags & (M_BCAST | M_MCAST) || !in_canforward(pkt_dst)) {
1838 ipstat.ips_cantforward++;
1842 if (!ipstealth && ip->ip_ttl <= IPTTLDEC) {
1843 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1847 bzero(&fwd_ro, sizeof(fwd_ro));
1848 ip_rtaddr(pkt_dst, &fwd_ro);
1849 if (fwd_ro.ro_rt == NULL) {
1850 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1856 * Save the IP header and at most 8 bytes of the payload,
1857 * in case we need to generate an ICMP message to the src.
1859 * XXX this can be optimized a lot by saving the data in a local
1860 * buffer on the stack (72 bytes at most), and only allocating the
1861 * mbuf if really necessary. The vast majority of the packets
1862 * are forwarded without having to send an ICMP back (either
1863 * because unnecessary, or because rate limited), so we are
1864 * really we are wasting a lot of work here.
1866 * We don't use m_copy() because it might return a reference
1867 * to a shared cluster. Both this function and ip_output()
1868 * assume exclusive access to the IP header in `m', so any
1869 * data in a cluster may change before we reach icmp_error().
1871 MGETHDR(mcopy, MB_DONTWAIT, m->m_type);
1872 if (mcopy != NULL && !m_dup_pkthdr(mcopy, m, MB_DONTWAIT)) {
1874 * It's probably ok if the pkthdr dup fails (because
1875 * the deep copy of the tag chain failed), but for now
1876 * be conservative and just discard the copy since
1877 * code below may some day want the tags.
1882 if (mcopy != NULL) {
1883 mcopy->m_len = imin((IP_VHL_HL(ip->ip_vhl) << 2) + 8,
1885 mcopy->m_pkthdr.len = mcopy->m_len;
1886 m_copydata(m, 0, mcopy->m_len, mtod(mcopy, caddr_t));
1890 ip->ip_ttl -= IPTTLDEC;
1893 * If forwarding packet using same interface that it came in on,
1894 * perhaps should send a redirect to sender to shortcut a hop.
1895 * Only send redirect if source is sending directly to us,
1896 * and if packet was not source routed (or has any options).
1897 * Also, don't send redirect if forwarding using a default route
1898 * or a route modified by a redirect.
1900 if (rt->rt_ifp == m->m_pkthdr.rcvif &&
1901 !(rt->rt_flags & (RTF_DYNAMIC | RTF_MODIFIED)) &&
1902 satosin(rt_key(rt))->sin_addr.s_addr != INADDR_ANY &&
1903 ipsendredirects && !using_srcrt && next_hop == NULL) {
1904 u_long src = ntohl(ip->ip_src.s_addr);
1905 struct in_ifaddr *rt_ifa = (struct in_ifaddr *)rt->rt_ifa;
1907 if (rt_ifa != NULL &&
1908 (src & rt_ifa->ia_subnetmask) == rt_ifa->ia_subnet) {
1909 if (rt->rt_flags & RTF_GATEWAY)
1910 dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1912 dest = pkt_dst.s_addr;
1914 * Router requirements says to only send
1917 type = ICMP_REDIRECT;
1918 code = ICMP_REDIRECT_HOST;
1921 kprintf("redirect (%d) to %x\n", code, dest);
1926 error = ip_output(m, NULL, &fwd_ro, IP_FORWARDING, NULL, NULL);
1928 ipstat.ips_forward++;
1931 ipflow_create(&fwd_ro, mcopy);
1936 ipstat.ips_redirectsent++;
1939 ipstat.ips_cantforward++;
1946 * Send ICMP message.
1951 case 0: /* forwarded, but need redirect */
1952 /* type, code set above */
1955 case ENETUNREACH: /* shouldn't happen, checked above */
1960 type = ICMP_UNREACH;
1961 code = ICMP_UNREACH_HOST;
1965 type = ICMP_UNREACH;
1966 code = ICMP_UNREACH_NEEDFRAG;
1969 * If the packet is routed over IPsec tunnel, tell the
1970 * originator the tunnel MTU.
1971 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1974 if (fwd_ro.ro_rt != NULL) {
1975 struct secpolicy *sp = NULL;
1980 sp = ipsec4_getpolicybyaddr(mcopy,
1986 destmtu = fwd_ro.ro_rt->rt_ifp->if_mtu;
1988 /* count IPsec header size */
1989 ipsechdr = ipsec4_hdrsiz(mcopy,
1994 * find the correct route for outer IPv4
1995 * header, compute tunnel MTU.
1998 if (sp->req != NULL && sp->req->sav != NULL &&
1999 sp->req->sav->sah != NULL) {
2000 ro = &sp->req->sav->sah->sa_route;
2001 if (ro->ro_rt != NULL &&
2002 ro->ro_rt->rt_ifp != NULL) {
2004 ro->ro_rt->rt_ifp->if_mtu;
2005 destmtu -= ipsechdr;
2014 * If the packet is routed over IPsec tunnel, tell the
2015 * originator the tunnel MTU.
2016 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
2019 if (fwd_ro.ro_rt != NULL) {
2020 struct secpolicy *sp = NULL;
2025 sp = ipsec_getpolicybyaddr(mcopy,
2031 destmtu = fwd_ro.ro_rt->rt_ifp->if_mtu;
2033 /* count IPsec header size */
2034 ipsechdr = ipsec4_hdrsiz(mcopy,
2039 * find the correct route for outer IPv4
2040 * header, compute tunnel MTU.
2043 if (sp->req != NULL &&
2044 sp->req->sav != NULL &&
2045 sp->req->sav->sah != NULL) {
2046 ro = &sp->req->sav->sah->sa_route;
2047 if (ro->ro_rt != NULL &&
2048 ro->ro_rt->rt_ifp != NULL) {
2050 ro->ro_rt->rt_ifp->if_mtu;
2051 destmtu -= ipsechdr;
2058 #else /* !IPSEC && !FAST_IPSEC */
2059 if (fwd_ro.ro_rt != NULL)
2060 destmtu = fwd_ro.ro_rt->rt_ifp->if_mtu;
2062 ipstat.ips_cantfrag++;
2067 * A router should not generate ICMP_SOURCEQUENCH as
2068 * required in RFC1812 Requirements for IP Version 4 Routers.
2069 * Source quench could be a big problem under DoS attacks,
2070 * or if the underlying interface is rate-limited.
2071 * Those who need source quench packets may re-enable them
2072 * via the net.inet.ip.sendsourcequench sysctl.
2074 if (!ip_sendsourcequench) {
2078 type = ICMP_SOURCEQUENCH;
2083 case EACCES: /* ipfw denied packet */
2087 icmp_error(mcopy, type, code, dest, destmtu);
2089 if (fwd_ro.ro_rt != NULL)
2090 RTFREE(fwd_ro.ro_rt);
2094 ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2097 if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2101 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
2102 SCM_TIMESTAMP, SOL_SOCKET);
2104 mp = &(*mp)->m_next;
2106 if (inp->inp_flags & INP_RECVDSTADDR) {
2107 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
2108 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2110 mp = &(*mp)->m_next;
2112 if (inp->inp_flags & INP_RECVTTL) {
2113 *mp = sbcreatecontrol((caddr_t) &ip->ip_ttl,
2114 sizeof(u_char), IP_RECVTTL, IPPROTO_IP);
2116 mp = &(*mp)->m_next;
2120 * Moving these out of udp_input() made them even more broken
2121 * than they already were.
2123 /* options were tossed already */
2124 if (inp->inp_flags & INP_RECVOPTS) {
2125 *mp = sbcreatecontrol((caddr_t) opts_deleted_above,
2126 sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP);
2128 mp = &(*mp)->m_next;
2130 /* ip_srcroute doesn't do what we want here, need to fix */
2131 if (inp->inp_flags & INP_RECVRETOPTS) {
2132 *mp = sbcreatecontrol((caddr_t) ip_srcroute(m),
2133 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2135 mp = &(*mp)->m_next;
2138 if (inp->inp_flags & INP_RECVIF) {
2141 struct sockaddr_dl sdl;
2144 struct sockaddr_dl *sdp;
2145 struct sockaddr_dl *sdl2 = &sdlbuf.sdl;
2147 if (((ifp = m->m_pkthdr.rcvif)) &&
2148 ((ifp->if_index != 0) && (ifp->if_index <= if_index))) {
2149 sdp = IF_LLSOCKADDR(ifp);
2151 * Change our mind and don't try copy.
2153 if ((sdp->sdl_family != AF_LINK) ||
2154 (sdp->sdl_len > sizeof(sdlbuf))) {
2157 bcopy(sdp, sdl2, sdp->sdl_len);
2161 offsetof(struct sockaddr_dl, sdl_data[0]);
2162 sdl2->sdl_family = AF_LINK;
2163 sdl2->sdl_index = 0;
2164 sdl2->sdl_nlen = sdl2->sdl_alen = sdl2->sdl_slen = 0;
2166 *mp = sbcreatecontrol((caddr_t) sdl2, sdl2->sdl_len,
2167 IP_RECVIF, IPPROTO_IP);
2169 mp = &(*mp)->m_next;
2174 * XXX these routines are called from the upper part of the kernel.
2176 * They could also be moved to ip_mroute.c, since all the RSVP
2177 * handling is done there already.
2180 ip_rsvp_init(struct socket *so)
2182 if (so->so_type != SOCK_RAW ||
2183 so->so_proto->pr_protocol != IPPROTO_RSVP)
2186 if (ip_rsvpd != NULL)
2191 * This may seem silly, but we need to be sure we don't over-increment
2192 * the RSVP counter, in case something slips up.
2207 * This may seem silly, but we need to be sure we don't over-decrement
2208 * the RSVP counter, in case something slips up.
2218 rsvp_input(struct mbuf *m, ...) /* XXX must fixup manually */
2224 off = __va_arg(ap, int);
2225 proto = __va_arg(ap, int);
2228 if (rsvp_input_p) { /* call the real one if loaded */
2229 rsvp_input_p(m, off, proto);
2233 /* Can still get packets with rsvp_on = 0 if there is a local member
2234 * of the group to which the RSVP packet is addressed. But in this
2235 * case we want to throw the packet away.
2243 if (ip_rsvpd != NULL) {
2244 rip_input(m, off, proto);
2247 /* Drop the packet */