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 <machine/stdarg.h>
101 #include <net/if_types.h>
102 #include <net/if_var.h>
103 #include <net/if_dl.h>
104 #include <net/pfil.h>
105 #include <net/route.h>
106 #include <net/netisr.h>
108 #include <netinet/in.h>
109 #include <netinet/in_systm.h>
110 #include <netinet/in_var.h>
111 #include <netinet/ip.h>
112 #include <netinet/in_pcb.h>
113 #include <netinet/ip_var.h>
114 #include <netinet/ip_icmp.h>
115 #include <netinet/ip_divert.h>
116 #include <netinet/ip_flow.h>
118 #include <sys/thread2.h>
119 #include <sys/msgport2.h>
120 #include <net/netmsg2.h>
122 #include <sys/socketvar.h>
124 #include <net/ipfw/ip_fw.h>
125 #include <net/dummynet/ip_dummynet.h>
128 #include <netinet6/ipsec.h>
129 #include <netproto/key/key.h>
133 #include <netproto/ipsec/ipsec.h>
134 #include <netproto/ipsec/key.h>
138 static int ip_rsvp_on
;
139 struct socket
*ip_rsvpd
;
142 TUNABLE_INT("net.inet.ip.mpsafe", &ip_mpsafe
);
144 int ipforwarding
= 0;
145 SYSCTL_INT(_net_inet_ip
, IPCTL_FORWARDING
, forwarding
, CTLFLAG_RW
,
146 &ipforwarding
, 0, "Enable IP forwarding between interfaces");
148 static int ipsendredirects
= 1; /* XXX */
149 SYSCTL_INT(_net_inet_ip
, IPCTL_SENDREDIRECTS
, redirect
, CTLFLAG_RW
,
150 &ipsendredirects
, 0, "Enable sending IP redirects");
152 int ip_defttl
= IPDEFTTL
;
153 SYSCTL_INT(_net_inet_ip
, IPCTL_DEFTTL
, ttl
, CTLFLAG_RW
,
154 &ip_defttl
, 0, "Maximum TTL on IP packets");
156 static int ip_dosourceroute
= 0;
157 SYSCTL_INT(_net_inet_ip
, IPCTL_SOURCEROUTE
, sourceroute
, CTLFLAG_RW
,
158 &ip_dosourceroute
, 0, "Enable forwarding source routed IP packets");
160 static int ip_acceptsourceroute
= 0;
161 SYSCTL_INT(_net_inet_ip
, IPCTL_ACCEPTSOURCEROUTE
, accept_sourceroute
,
162 CTLFLAG_RW
, &ip_acceptsourceroute
, 0,
163 "Enable accepting source routed IP packets");
165 static int ip_keepfaith
= 0;
166 SYSCTL_INT(_net_inet_ip
, IPCTL_KEEPFAITH
, keepfaith
, CTLFLAG_RW
,
168 "Enable packet capture for FAITH IPv4->IPv6 translator daemon");
170 static int nipq
= 0; /* total # of reass queues */
172 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, maxfragpackets
, CTLFLAG_RW
,
174 "Maximum number of IPv4 fragment reassembly queue entries");
176 static int maxfragsperpacket
;
177 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, maxfragsperpacket
, CTLFLAG_RW
,
178 &maxfragsperpacket
, 0,
179 "Maximum number of IPv4 fragments allowed per packet");
181 static int ip_sendsourcequench
= 0;
182 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, sendsourcequench
, CTLFLAG_RW
,
183 &ip_sendsourcequench
, 0,
184 "Enable the transmission of source quench packets");
186 int ip_do_randomid
= 1;
187 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, random_id
, CTLFLAG_RW
,
189 "Assign random ip_id values");
191 * XXX - Setting ip_checkinterface mostly implements the receive side of
192 * the Strong ES model described in RFC 1122, but since the routing table
193 * and transmit implementation do not implement the Strong ES model,
194 * setting this to 1 results in an odd hybrid.
196 * XXX - ip_checkinterface currently must be disabled if you use ipnat
197 * to translate the destination address to another local interface.
199 * XXX - ip_checkinterface must be disabled if you add IP aliases
200 * to the loopback interface instead of the interface where the
201 * packets for those addresses are received.
203 static int ip_checkinterface
= 0;
204 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, check_interface
, CTLFLAG_RW
,
205 &ip_checkinterface
, 0, "Verify packet arrives on correct interface");
208 static int ipprintfs
= 0;
211 extern int udp_mpsafe_proto
;
212 extern int tcp_mpsafe_proto
;
214 extern struct domain inetdomain
;
215 extern struct protosw inetsw
[];
216 u_char ip_protox
[IPPROTO_MAX
];
217 struct in_ifaddrhead in_ifaddrheads
[MAXCPU
]; /* first inet address */
218 struct in_ifaddrhashhead
*in_ifaddrhashtbls
[MAXCPU
];
219 /* inet addr hash table */
220 u_long in_ifaddrhmask
; /* mask for hash table */
222 struct ip_stats ipstats_percpu
[MAXCPU
];
225 sysctl_ipstats(SYSCTL_HANDLER_ARGS
)
229 for (cpu
= 0; cpu
< ncpus
; ++cpu
) {
230 if ((error
= SYSCTL_OUT(req
, &ipstats_percpu
[cpu
],
231 sizeof(struct ip_stats
))))
233 if ((error
= SYSCTL_IN(req
, &ipstats_percpu
[cpu
],
234 sizeof(struct ip_stats
))))
240 SYSCTL_PROC(_net_inet_ip
, IPCTL_STATS
, stats
, (CTLTYPE_OPAQUE
| CTLFLAG_RW
),
241 0, 0, sysctl_ipstats
, "S,ip_stats", "IP statistics");
243 SYSCTL_STRUCT(_net_inet_ip
, IPCTL_STATS
, stats
, CTLFLAG_RW
,
244 &ipstat
, ip_stats
, "IP statistics");
247 /* Packet reassembly stuff */
248 #define IPREASS_NHASH_LOG2 6
249 #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
250 #define IPREASS_HMASK (IPREASS_NHASH - 1)
251 #define IPREASS_HASH(x,y) \
252 (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
254 static struct ipq ipq
[IPREASS_NHASH
];
257 SYSCTL_INT(_net_inet_ip
, IPCTL_DEFMTU
, mtu
, CTLFLAG_RW
,
258 &ip_mtu
, 0, "Default MTU");
262 static int ipstealth
= 0;
263 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, stealth
, CTLFLAG_RW
, &ipstealth
, 0, "");
265 static const int ipstealth
= 0;
268 struct mbuf
*(*ip_divert_p
)(struct mbuf
*, int, int);
270 struct pfil_head inet_pfil_hook
;
273 * struct ip_srcrt_opt is used to store packet state while it travels
276 * XXX Note that the code even makes assumptions on the size and
277 * alignment of fields inside struct ip_srcrt so e.g. adding some
278 * fields will break the code. This needs to be fixed.
280 * We need to save the IP options in case a protocol wants to respond
281 * to an incoming packet over the same route if the packet got here
282 * using IP source routing. This allows connection establishment and
283 * maintenance when the remote end is on a network that is not known
287 struct in_addr dst
; /* final destination */
288 char nop
; /* one NOP to align */
289 char srcopt
[IPOPT_OFFSET
+ 1]; /* OPTVAL, OLEN and OFFSET */
290 struct in_addr route
[MAX_IPOPTLEN
/sizeof(struct in_addr
)];
293 struct ip_srcrt_opt
{
295 struct ip_srcrt ip_srcrt
;
298 static MALLOC_DEFINE(M_IPQ
, "ipq", "IP Fragment Management");
299 static struct malloc_pipe ipq_mpipe
;
301 static void save_rte(struct mbuf
*, u_char
*, struct in_addr
);
302 static int ip_dooptions(struct mbuf
*m
, int, struct sockaddr_in
*);
303 static void ip_freef(struct ipq
*);
304 static void ip_input_handler(struct netmsg
*);
307 * IP initialization: fill in IP protocol switch table.
308 * All protocols not implemented in kernel go to raw IP protocol handler.
321 * Make sure we can handle a reasonable number of fragments but
322 * cap it at 4000 (XXX).
324 mpipe_init(&ipq_mpipe
, M_IPQ
, sizeof(struct ipq
),
325 IFQ_MAXLEN
, 4000, 0, NULL
);
326 for (i
= 0; i
< ncpus
; ++i
) {
327 TAILQ_INIT(&in_ifaddrheads
[i
]);
328 in_ifaddrhashtbls
[i
] =
329 hashinit(INADDR_NHASH
, M_IFADDR
, &in_ifaddrhmask
);
331 pr
= pffindproto(PF_INET
, IPPROTO_RAW
, SOCK_RAW
);
334 for (i
= 0; i
< IPPROTO_MAX
; i
++)
335 ip_protox
[i
] = pr
- inetsw
;
336 for (pr
= inetdomain
.dom_protosw
;
337 pr
< inetdomain
.dom_protoswNPROTOSW
; pr
++) {
338 if (pr
->pr_domain
->dom_family
== PF_INET
&& pr
->pr_protocol
) {
339 if (pr
->pr_protocol
!= IPPROTO_RAW
)
340 ip_protox
[pr
->pr_protocol
] = pr
- inetsw
;
343 switch (pr
->pr_protocol
) {
345 if (tcp_mpsafe_proto
)
346 pr
->pr_flags
|= PR_MPSAFE
;
350 if (udp_mpsafe_proto
)
351 pr
->pr_flags
|= PR_MPSAFE
;
357 inet_pfil_hook
.ph_type
= PFIL_TYPE_AF
;
358 inet_pfil_hook
.ph_af
= AF_INET
;
359 if ((i
= pfil_head_register(&inet_pfil_hook
)) != 0) {
360 kprintf("%s: WARNING: unable to register pfil hook, "
361 "error %d\n", __func__
, i
);
364 for (i
= 0; i
< IPREASS_NHASH
; i
++)
365 ipq
[i
].next
= ipq
[i
].prev
= &ipq
[i
];
367 maxnipq
= nmbclusters
/ 32;
368 maxfragsperpacket
= 16;
370 ip_id
= time_second
& 0xffff;
373 * Initialize IP statistics counters for each CPU.
377 for (cpu
= 0; cpu
< ncpus
; ++cpu
) {
378 bzero(&ipstats_percpu
[cpu
], sizeof(struct ip_stats
));
381 bzero(&ipstat
, sizeof(struct ip_stats
));
384 #if defined(IPSEC) || defined(FAST_IPSEC)
385 /* XXX IPSEC is not MPSAFE yet */
386 flags
= NETISR_FLAG_NOTMPSAFE
;
389 kprintf("ip: MPSAFE\n");
390 flags
= NETISR_FLAG_MPSAFE
;
392 flags
= NETISR_FLAG_NOTMPSAFE
;
395 netisr_register(NETISR_IP
, ip_mport_in
, ip_mport_pktinfo
,
396 ip_input_handler
, flags
);
399 /* Do transport protocol processing. */
401 transport_processing_oncpu(struct mbuf
*m
, int hlen
, struct ip
*ip
)
403 const struct protosw
*pr
= &inetsw
[ip_protox
[ip
->ip_p
]];
406 * Switch out to protocol's input routine.
409 pr
->pr_input(m
, hlen
, ip
->ip_p
);
414 transport_processing_handler(netmsg_t netmsg
)
416 struct netmsg_packet
*pmsg
= (struct netmsg_packet
*)netmsg
;
420 ip
= mtod(pmsg
->nm_packet
, struct ip
*);
421 hlen
= pmsg
->nm_netmsg
.nm_lmsg
.u
.ms_result
;
423 transport_processing_oncpu(pmsg
->nm_packet
, hlen
, ip
);
424 /* netmsg was embedded in the mbuf, do not reply! */
428 ip_input_handler(struct netmsg
*msg0
)
430 struct mbuf
*m
= ((struct netmsg_packet
*)msg0
)->nm_packet
;
433 /* msg0 was embedded in the mbuf, do not reply! */
437 * IP input routine. Checksum and byte swap header. If fragmented
438 * try to reassemble. Process options. Pass to next level.
441 ip_input(struct mbuf
*m
)
444 struct in_ifaddr
*ia
= NULL
;
445 struct in_ifaddr_container
*iac
;
448 struct in_addr pkt_dst
;
449 boolean_t using_srcrt
= FALSE
; /* forward (by PFIL_HOOKS) */
450 boolean_t needredispatch
= FALSE
;
451 struct in_addr odst
; /* original dst address(NAT) */
453 struct sockaddr_in
*next_hop
= NULL
;
455 struct tdb_ident
*tdbi
;
456 struct secpolicy
*sp
;
462 if (m
->m_pkthdr
.fw_flags
& IPFORWARD_MBUF_TAGGED
) {
464 mtag
= m_tag_find(m
, PACKET_TAG_IPFORWARD
, NULL
);
465 KKASSERT(mtag
!= NULL
);
466 next_hop
= m_tag_data(mtag
);
469 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
470 /* dummynet already filtered us */
471 ip
= mtod(m
, struct ip
*);
472 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
478 /* length checks already done in ip_mport() */
479 KASSERT(m
->m_len
>= sizeof(struct ip
), ("IP header not in one mbuf"));
480 ip
= mtod(m
, struct ip
*);
482 if (IP_VHL_V(ip
->ip_vhl
) != IPVERSION
) {
483 ipstat
.ips_badvers
++;
487 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
488 /* length checks already done in ip_mport() */
489 KASSERT(hlen
>= sizeof(struct ip
), ("IP header len too small"));
490 KASSERT(m
->m_len
>= hlen
, ("complete IP header not in one mbuf"));
492 /* 127/8 must not appear on wire - RFC1122 */
493 if ((ntohl(ip
->ip_dst
.s_addr
) >> IN_CLASSA_NSHIFT
) == IN_LOOPBACKNET
||
494 (ntohl(ip
->ip_src
.s_addr
) >> IN_CLASSA_NSHIFT
) == IN_LOOPBACKNET
) {
495 if (!(m
->m_pkthdr
.rcvif
->if_flags
& IFF_LOOPBACK
)) {
496 ipstat
.ips_badaddr
++;
501 if (m
->m_pkthdr
.csum_flags
& CSUM_IP_CHECKED
) {
502 sum
= !(m
->m_pkthdr
.csum_flags
& CSUM_IP_VALID
);
504 if (hlen
== sizeof(struct ip
))
505 sum
= in_cksum_hdr(ip
);
507 sum
= in_cksum(m
, hlen
);
515 if (altq_input
!= NULL
&& (*altq_input
)(m
, AF_INET
) == 0) {
516 /* packet is dropped by traffic conditioner */
521 * Convert fields to host representation.
523 ip
->ip_len
= ntohs(ip
->ip_len
);
524 ip
->ip_off
= ntohs(ip
->ip_off
);
526 /* length checks already done in ip_mport() */
527 KASSERT(ip
->ip_len
>= hlen
, ("total length less then header length"));
528 KASSERT(m
->m_pkthdr
.len
>= ip
->ip_len
, ("mbuf too short"));
531 * Trim mbufs if longer than the IP header would have us expect.
533 if (m
->m_pkthdr
.len
> ip
->ip_len
) {
534 if (m
->m_len
== m
->m_pkthdr
.len
) {
535 m
->m_len
= ip
->ip_len
;
536 m
->m_pkthdr
.len
= ip
->ip_len
;
538 m_adj(m
, ip
->ip_len
- m
->m_pkthdr
.len
);
541 #if defined(IPSEC) && !defined(IPSEC_FILTERGIF)
543 * Bypass packet filtering for packets from a tunnel (gif).
545 if (ipsec_gethist(m
, NULL
))
551 * Right now when no processing on packet has done
552 * and it is still fresh out of network we do our black
554 * - Firewall: deny/allow/divert
555 * - Xlate: translate packet's addr/port (NAT).
556 * - Pipe: pass pkt through dummynet.
557 * - Wrap: fake packet's addr/port <unimpl.>
558 * - Encapsulate: put it in another IP and send out. <unimp.>
563 * If we've been forwarded from the output side, then
564 * skip the firewall a second time
566 if (next_hop
!= NULL
)
570 if (!pfil_has_hooks(&inet_pfil_hook
)) {
571 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
573 * Strip dummynet tags from stranded packets
575 mtag
= m_tag_find(m
, PACKET_TAG_DUMMYNET
, NULL
);
576 KKASSERT(mtag
!= NULL
);
577 m_tag_delete(m
, mtag
);
578 m
->m_pkthdr
.fw_flags
&= ~DUMMYNET_MBUF_TAGGED
;
584 * Run through list of hooks for input packets.
586 * NB: Beware of the destination address changing (e.g.
587 * by NAT rewriting). When this happens, tell
588 * ip_forward to do the right thing.
591 if (pfil_run_hooks(&inet_pfil_hook
, &m
, m
->m_pkthdr
.rcvif
, PFIL_IN
))
593 if (m
== NULL
) /* consumed by filter */
595 ip
= mtod(m
, struct ip
*);
596 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
597 using_srcrt
= (odst
.s_addr
!= ip
->ip_dst
.s_addr
);
599 if (m
->m_pkthdr
.fw_flags
& IPFORWARD_MBUF_TAGGED
) {
600 mtag
= m_tag_find(m
, PACKET_TAG_IPFORWARD
, NULL
);
601 KKASSERT(mtag
!= NULL
);
602 next_hop
= m_tag_data(mtag
);
604 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
608 if (m
->m_pkthdr
.fw_flags
& FW_MBUF_REDISPATCH
) {
609 needredispatch
= TRUE
;
610 m
->m_pkthdr
.fw_flags
&= ~FW_MBUF_REDISPATCH
;
614 * Process options and, if not destined for us,
615 * ship it on. ip_dooptions returns 1 when an
616 * error was detected (causing an icmp message
617 * to be sent and the original packet to be freed).
619 if (hlen
> sizeof(struct ip
) && ip_dooptions(m
, 0, next_hop
))
622 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no
623 * matter if it is destined to another node, or whether it is
624 * a multicast one, RSVP wants it! and prevents it from being forwarded
625 * anywhere else. Also checks if the rsvp daemon is running before
626 * grabbing the packet.
628 if (rsvp_on
&& ip
->ip_p
== IPPROTO_RSVP
)
632 * Check our list of addresses, to see if the packet is for us.
633 * If we don't have any addresses, assume any unicast packet
634 * we receive might be for us (and let the upper layers deal
637 if (TAILQ_EMPTY(&in_ifaddrheads
[mycpuid
]) &&
638 !(m
->m_flags
& (M_MCAST
| M_BCAST
)))
642 * Cache the destination address of the packet; this may be
643 * changed by use of 'ipfw fwd'.
645 pkt_dst
= next_hop
? next_hop
->sin_addr
: ip
->ip_dst
;
648 * Enable a consistency check between the destination address
649 * and the arrival interface for a unicast packet (the RFC 1122
650 * strong ES model) if IP forwarding is disabled and the packet
651 * is not locally generated and the packet is not subject to
654 * XXX - Checking also should be disabled if the destination
655 * address is ipnat'ed to a different interface.
657 * XXX - Checking is incompatible with IP aliases added
658 * to the loopback interface instead of the interface where
659 * the packets are received.
661 checkif
= ip_checkinterface
&&
663 m
->m_pkthdr
.rcvif
!= NULL
&&
664 !(m
->m_pkthdr
.rcvif
->if_flags
& IFF_LOOPBACK
) &&
668 * Check for exact addresses in the hash bucket.
670 LIST_FOREACH(iac
, INADDR_HASH(pkt_dst
.s_addr
), ia_hash
) {
674 * If the address matches, verify that the packet
675 * arrived via the correct interface if checking is
678 if (IA_SIN(ia
)->sin_addr
.s_addr
== pkt_dst
.s_addr
&&
679 (!checkif
|| ia
->ia_ifp
== m
->m_pkthdr
.rcvif
))
685 * Check for broadcast addresses.
687 * Only accept broadcast packets that arrive via the matching
688 * interface. Reception of forwarded directed broadcasts would
689 * be handled via ip_forward() and ether_output() with the loopback
690 * into the stack for SIMPLEX interfaces handled by ether_output().
692 if (m
->m_pkthdr
.rcvif
->if_flags
& IFF_BROADCAST
) {
693 struct ifaddr_container
*ifac
;
695 TAILQ_FOREACH(ifac
, &m
->m_pkthdr
.rcvif
->if_addrheads
[mycpuid
],
697 struct ifaddr
*ifa
= ifac
->ifa
;
699 if (ifa
->ifa_addr
== NULL
) /* shutdown/startup race */
701 if (ifa
->ifa_addr
->sa_family
!= AF_INET
)
704 if (satosin(&ia
->ia_broadaddr
)->sin_addr
.s_addr
==
707 if (ia
->ia_netbroadcast
.s_addr
== pkt_dst
.s_addr
)
710 if (IA_SIN(ia
)->sin_addr
.s_addr
== INADDR_ANY
)
715 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
))) {
716 struct in_multi
*inm
;
718 /* XXX Multicast is not MPSAFE yet */
721 if (ip_mrouter
!= NULL
) {
723 * If we are acting as a multicast router, all
724 * incoming multicast packets are passed to the
725 * kernel-level multicast forwarding function.
726 * The packet is returned (relatively) intact; if
727 * ip_mforward() returns a non-zero value, the packet
728 * must be discarded, else it may be accepted below.
730 if (ip_mforward
!= NULL
&&
731 ip_mforward(ip
, m
->m_pkthdr
.rcvif
, m
, NULL
) != 0) {
733 ipstat
.ips_cantforward
++;
739 * The process-level routing daemon needs to receive
740 * all multicast IGMP packets, whether or not this
741 * host belongs to their destination groups.
743 if (ip
->ip_p
== IPPROTO_IGMP
) {
747 ipstat
.ips_forward
++;
750 * See if we belong to the destination multicast group on the
753 IN_LOOKUP_MULTI(ip
->ip_dst
, m
->m_pkthdr
.rcvif
, inm
);
756 ipstat
.ips_notmember
++;
764 if (ip
->ip_dst
.s_addr
== INADDR_BROADCAST
)
766 if (ip
->ip_dst
.s_addr
== INADDR_ANY
)
770 * FAITH(Firewall Aided Internet Translator)
772 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
== IFT_FAITH
) {
774 if (ip
->ip_p
== IPPROTO_TCP
|| ip
->ip_p
== IPPROTO_ICMP
)
782 * Not for us; forward if possible and desirable.
785 ipstat
.ips_cantforward
++;
790 * Enforce inbound IPsec SPD.
792 if (ipsec4_in_reject(m
, NULL
)) {
793 ipsecstat
.in_polvio
++;
798 mtag
= m_tag_find(m
, PACKET_TAG_IPSEC_IN_DONE
, NULL
);
801 tdbi
= (struct tdb_ident
*)m_tag_data(mtag
);
802 sp
= ipsec_getpolicy(tdbi
, IPSEC_DIR_INBOUND
);
804 sp
= ipsec_getpolicybyaddr(m
, IPSEC_DIR_INBOUND
,
805 IP_FORWARDING
, &error
);
807 if (sp
== NULL
) { /* NB: can happen if error */
809 /*XXX error stat???*/
810 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
815 * Check security policy against packet attributes.
817 error
= ipsec_in_reject(sp
, m
);
821 ipstat
.ips_cantforward
++;
825 ip_forward(m
, using_srcrt
, next_hop
);
832 * IPSTEALTH: Process non-routing options only
833 * if the packet is destined for us.
836 hlen
> sizeof(struct ip
) &&
837 ip_dooptions(m
, 1, next_hop
))
840 /* Count the packet in the ip address stats */
842 ia
->ia_ifa
.if_ipackets
++;
843 ia
->ia_ifa
.if_ibytes
+= m
->m_pkthdr
.len
;
847 * If offset or IP_MF are set, must reassemble.
848 * Otherwise, nothing need be done.
849 * (We could look in the reassembly queue to see
850 * if the packet was previously fragmented,
851 * but it's not worth the time; just let them time out.)
853 if (ip
->ip_off
& (IP_MF
| IP_OFFMASK
)) {
855 * Attempt reassembly; if it succeeds, proceed.
856 * ip_reass() will return a different mbuf.
861 ip
= mtod(m
, struct ip
*);
863 /* Get the header length of the reassembled packet */
864 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
866 needredispatch
= TRUE
;
873 * enforce IPsec policy checking if we are seeing last header.
874 * note that we do not visit this with protocols with pcb layer
875 * code - like udp/tcp/raw ip.
877 if ((inetsw
[ip_protox
[ip
->ip_p
]].pr_flags
& PR_LASTHDR
) &&
878 ipsec4_in_reject(m
, NULL
)) {
879 ipsecstat
.in_polvio
++;
885 * enforce IPsec policy checking if we are seeing last header.
886 * note that we do not visit this with protocols with pcb layer
887 * code - like udp/tcp/raw ip.
889 if (inetsw
[ip_protox
[ip
->ip_p
]].pr_flags
& PR_LASTHDR
) {
891 * Check if the packet has already had IPsec processing
892 * done. If so, then just pass it along. This tag gets
893 * set during AH, ESP, etc. input handling, before the
894 * packet is returned to the ip input queue for delivery.
896 mtag
= m_tag_find(m
, PACKET_TAG_IPSEC_IN_DONE
, NULL
);
899 tdbi
= (struct tdb_ident
*)m_tag_data(mtag
);
900 sp
= ipsec_getpolicy(tdbi
, IPSEC_DIR_INBOUND
);
902 sp
= ipsec_getpolicybyaddr(m
, IPSEC_DIR_INBOUND
,
903 IP_FORWARDING
, &error
);
907 * Check security policy against packet attributes.
909 error
= ipsec_in_reject(sp
, m
);
912 /* XXX error stat??? */
914 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
921 #endif /* FAST_IPSEC */
924 * NOTE: ip_len in host form and adjusted down by hlen for
925 * protocol processing.
927 ipstat
.ips_delivered
++;
928 if (needredispatch
) {
929 struct netmsg_packet
*pmsg
;
932 ip
->ip_off
= htons(ip
->ip_off
);
933 ip
->ip_len
= htons(ip
->ip_len
+ hlen
);
934 port
= ip_mport_in(&m
);
938 pmsg
= &m
->m_hdr
.mh_netmsg
;
939 netmsg_init(&pmsg
->nm_netmsg
, &netisr_apanic_rport
, MSGF_MPSAFE
,
940 transport_processing_handler
);
942 pmsg
->nm_netmsg
.nm_lmsg
.u
.ms_result
= hlen
;
944 ip
= mtod(m
, struct ip
*);
945 ip
->ip_len
= ntohs(ip
->ip_len
) - hlen
;
946 ip
->ip_off
= ntohs(ip
->ip_off
);
947 lwkt_sendmsg(port
, &pmsg
->nm_netmsg
.nm_lmsg
);
949 transport_processing_oncpu(m
, hlen
, ip
);
958 * Take incoming datagram fragment and try to reassemble it into
959 * whole datagram. If a chain for reassembly of this datagram already
960 * exists, then it is given as fp; otherwise have to make a chain.
963 ip_reass(struct mbuf
*m
)
965 struct ip
*ip
= mtod(m
, struct ip
*);
966 struct mbuf
*p
= NULL
, *q
, *nq
;
968 struct ipq
*fp
= NULL
;
969 int hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
973 /* If maxnipq is 0, never accept fragments. */
975 ipstat
.ips_fragments
++;
976 ipstat
.ips_fragdropped
++;
981 sum
= IPREASS_HASH(ip
->ip_src
.s_addr
, ip
->ip_id
);
983 * Look for queue of fragments of this datagram.
985 for (fp
= ipq
[sum
].next
; fp
!= &ipq
[sum
]; fp
= fp
->next
)
986 if (ip
->ip_id
== fp
->ipq_id
&&
987 ip
->ip_src
.s_addr
== fp
->ipq_src
.s_addr
&&
988 ip
->ip_dst
.s_addr
== fp
->ipq_dst
.s_addr
&&
989 ip
->ip_p
== fp
->ipq_p
)
995 * Enforce upper bound on number of fragmented packets
996 * for which we attempt reassembly;
997 * If maxnipq is -1, accept all fragments without limitation.
999 if (nipq
> maxnipq
&& maxnipq
> 0) {
1001 * drop something from the tail of the current queue
1002 * before proceeding further
1004 if (ipq
[sum
].prev
== &ipq
[sum
]) { /* gak */
1005 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1006 if (ipq
[i
].prev
!= &ipq
[i
]) {
1007 ipstat
.ips_fragtimeout
+=
1008 ipq
[i
].prev
->ipq_nfrags
;
1009 ip_freef(ipq
[i
].prev
);
1014 ipstat
.ips_fragtimeout
+=
1015 ipq
[sum
].prev
->ipq_nfrags
;
1016 ip_freef(ipq
[sum
].prev
);
1021 * Adjust ip_len to not reflect header,
1022 * convert offset of this to bytes.
1025 if (ip
->ip_off
& IP_MF
) {
1027 * Make sure that fragments have a data length
1028 * that's a non-zero multiple of 8 bytes.
1030 if (ip
->ip_len
== 0 || (ip
->ip_len
& 0x7) != 0) {
1031 ipstat
.ips_toosmall
++; /* XXX */
1035 m
->m_flags
|= M_FRAG
;
1037 m
->m_flags
&= ~M_FRAG
;
1040 ipstat
.ips_fragments
++;
1041 m
->m_pkthdr
.header
= ip
;
1044 * If the hardware has not done csum over this fragment
1045 * then csum_data is not valid at all.
1047 if ((m
->m_pkthdr
.csum_flags
& (CSUM_FRAG_NOT_CHECKED
| CSUM_DATA_VALID
))
1048 == (CSUM_FRAG_NOT_CHECKED
| CSUM_DATA_VALID
)) {
1049 m
->m_pkthdr
.csum_data
= 0;
1050 m
->m_pkthdr
.csum_flags
&= ~(CSUM_DATA_VALID
| CSUM_PSEUDO_HDR
);
1054 * Presence of header sizes in mbufs
1055 * would confuse code below.
1061 * If first fragment to arrive, create a reassembly queue.
1064 if ((fp
= mpipe_alloc_nowait(&ipq_mpipe
)) == NULL
)
1066 insque(fp
, &ipq
[sum
]);
1069 fp
->ipq_ttl
= IPFRAGTTL
;
1070 fp
->ipq_p
= ip
->ip_p
;
1071 fp
->ipq_id
= ip
->ip_id
;
1072 fp
->ipq_src
= ip
->ip_src
;
1073 fp
->ipq_dst
= ip
->ip_dst
;
1075 m
->m_nextpkt
= NULL
;
1081 #define GETIP(m) ((struct ip*)((m)->m_pkthdr.header))
1084 * Find a segment which begins after this one does.
1086 for (p
= NULL
, q
= fp
->ipq_frags
; q
; p
= q
, q
= q
->m_nextpkt
)
1087 if (GETIP(q
)->ip_off
> ip
->ip_off
)
1091 * If there is a preceding segment, it may provide some of
1092 * our data already. If so, drop the data from the incoming
1093 * segment. If it provides all of our data, drop us, otherwise
1094 * stick new segment in the proper place.
1096 * If some of the data is dropped from the the preceding
1097 * segment, then it's checksum is invalidated.
1100 i
= GETIP(p
)->ip_off
+ GETIP(p
)->ip_len
- ip
->ip_off
;
1102 if (i
>= ip
->ip_len
)
1105 m
->m_pkthdr
.csum_flags
= 0;
1109 m
->m_nextpkt
= p
->m_nextpkt
;
1112 m
->m_nextpkt
= fp
->ipq_frags
;
1117 * While we overlap succeeding segments trim them or,
1118 * if they are completely covered, dequeue them.
1120 for (; q
!= NULL
&& ip
->ip_off
+ ip
->ip_len
> GETIP(q
)->ip_off
;
1122 i
= (ip
->ip_off
+ ip
->ip_len
) - GETIP(q
)->ip_off
;
1123 if (i
< GETIP(q
)->ip_len
) {
1124 GETIP(q
)->ip_len
-= i
;
1125 GETIP(q
)->ip_off
+= i
;
1127 q
->m_pkthdr
.csum_flags
= 0;
1132 ipstat
.ips_fragdropped
++;
1134 q
->m_nextpkt
= NULL
;
1140 * Check for complete reassembly and perform frag per packet
1143 * Frag limiting is performed here so that the nth frag has
1144 * a chance to complete the packet before we drop the packet.
1145 * As a result, n+1 frags are actually allowed per packet, but
1146 * only n will ever be stored. (n = maxfragsperpacket.)
1150 for (p
= NULL
, q
= fp
->ipq_frags
; q
; p
= q
, q
= q
->m_nextpkt
) {
1151 if (GETIP(q
)->ip_off
!= next
) {
1152 if (fp
->ipq_nfrags
> maxfragsperpacket
) {
1153 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1158 next
+= GETIP(q
)->ip_len
;
1160 /* Make sure the last packet didn't have the IP_MF flag */
1161 if (p
->m_flags
& M_FRAG
) {
1162 if (fp
->ipq_nfrags
> maxfragsperpacket
) {
1163 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1170 * Reassembly is complete. Make sure the packet is a sane size.
1174 if (next
+ (IP_VHL_HL(ip
->ip_vhl
) << 2) > IP_MAXPACKET
) {
1175 ipstat
.ips_toolong
++;
1176 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1182 * Concatenate fragments.
1189 q
->m_nextpkt
= NULL
;
1190 for (q
= nq
; q
!= NULL
; q
= nq
) {
1192 q
->m_nextpkt
= NULL
;
1193 m
->m_pkthdr
.csum_flags
&= q
->m_pkthdr
.csum_flags
;
1194 m
->m_pkthdr
.csum_data
+= q
->m_pkthdr
.csum_data
;
1199 * Clean up the 1's complement checksum. Carry over 16 bits must
1200 * be added back. This assumes no more then 65535 packet fragments
1201 * were reassembled. A second carry can also occur (but not a third).
1203 m
->m_pkthdr
.csum_data
= (m
->m_pkthdr
.csum_data
& 0xffff) +
1204 (m
->m_pkthdr
.csum_data
>> 16);
1205 if (m
->m_pkthdr
.csum_data
> 0xFFFF)
1206 m
->m_pkthdr
.csum_data
-= 0xFFFF;
1209 * Create header for new ip packet by
1210 * modifying header of first packet;
1211 * dequeue and discard fragment reassembly header.
1212 * Make header visible.
1215 ip
->ip_src
= fp
->ipq_src
;
1216 ip
->ip_dst
= fp
->ipq_dst
;
1219 mpipe_free(&ipq_mpipe
, fp
);
1220 m
->m_len
+= (IP_VHL_HL(ip
->ip_vhl
) << 2);
1221 m
->m_data
-= (IP_VHL_HL(ip
->ip_vhl
) << 2);
1222 /* some debugging cruft by sklower, below, will go away soon */
1223 if (m
->m_flags
& M_PKTHDR
) { /* XXX this should be done elsewhere */
1226 for (n
= m
; n
; n
= n
->m_next
)
1228 m
->m_pkthdr
.len
= plen
;
1231 ipstat
.ips_reassembled
++;
1235 ipstat
.ips_fragdropped
++;
1245 * Free a fragment reassembly header and all
1246 * associated datagrams.
1249 ip_freef(struct ipq
*fp
)
1253 while (fp
->ipq_frags
) {
1255 fp
->ipq_frags
= q
->m_nextpkt
;
1256 q
->m_nextpkt
= NULL
;
1260 mpipe_free(&ipq_mpipe
, fp
);
1265 * IP timer processing;
1266 * if a timer expires on a reassembly
1267 * queue, discard it.
1276 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1280 while (fp
!= &ipq
[i
]) {
1283 if (fp
->prev
->ipq_ttl
== 0) {
1284 ipstat
.ips_fragtimeout
+= fp
->prev
->ipq_nfrags
;
1290 * If we are over the maximum number of fragments
1291 * (due to the limit being lowered), drain off
1292 * enough to get down to the new limit.
1294 if (maxnipq
>= 0 && nipq
> maxnipq
) {
1295 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1296 while (nipq
> maxnipq
&&
1297 (ipq
[i
].next
!= &ipq
[i
])) {
1298 ipstat
.ips_fragdropped
+=
1299 ipq
[i
].next
->ipq_nfrags
;
1300 ip_freef(ipq
[i
].next
);
1309 * Drain off all datagram fragments.
1316 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1317 while (ipq
[i
].next
!= &ipq
[i
]) {
1318 ipstat
.ips_fragdropped
+= ipq
[i
].next
->ipq_nfrags
;
1319 ip_freef(ipq
[i
].next
);
1326 * Do option processing on a datagram,
1327 * possibly discarding it if bad options are encountered,
1328 * or forwarding it if source-routed.
1329 * The pass argument is used when operating in the IPSTEALTH
1330 * mode to tell what options to process:
1331 * [LS]SRR (pass 0) or the others (pass 1).
1332 * The reason for as many as two passes is that when doing IPSTEALTH,
1333 * non-routing options should be processed only if the packet is for us.
1334 * Returns 1 if packet has been forwarded/freed,
1335 * 0 if the packet should be processed further.
1338 ip_dooptions(struct mbuf
*m
, int pass
, struct sockaddr_in
*next_hop
)
1340 struct sockaddr_in ipaddr
= { sizeof ipaddr
, AF_INET
};
1341 struct ip
*ip
= mtod(m
, struct ip
*);
1343 struct in_ifaddr
*ia
;
1344 int opt
, optlen
, cnt
, off
, code
, type
= ICMP_PARAMPROB
;
1345 boolean_t forward
= FALSE
;
1346 struct in_addr
*sin
, dst
;
1350 cp
= (u_char
*)(ip
+ 1);
1351 cnt
= (IP_VHL_HL(ip
->ip_vhl
) << 2) - sizeof(struct ip
);
1352 for (; cnt
> 0; cnt
-= optlen
, cp
+= optlen
) {
1353 opt
= cp
[IPOPT_OPTVAL
];
1354 if (opt
== IPOPT_EOL
)
1356 if (opt
== IPOPT_NOP
)
1359 if (cnt
< IPOPT_OLEN
+ sizeof(*cp
)) {
1360 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1363 optlen
= cp
[IPOPT_OLEN
];
1364 if (optlen
< IPOPT_OLEN
+ sizeof(*cp
) || optlen
> cnt
) {
1365 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1375 * Source routing with record.
1376 * Find interface with current destination address.
1377 * If none on this machine then drop if strictly routed,
1378 * or do nothing if loosely routed.
1379 * Record interface address and bring up next address
1380 * component. If strictly routed make sure next
1381 * address is on directly accessible net.
1385 if (ipstealth
&& pass
> 0)
1387 if (optlen
< IPOPT_OFFSET
+ sizeof(*cp
)) {
1388 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1391 if ((off
= cp
[IPOPT_OFFSET
]) < IPOPT_MINOFF
) {
1392 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1395 ipaddr
.sin_addr
= ip
->ip_dst
;
1396 ia
= (struct in_ifaddr
*)
1397 ifa_ifwithaddr((struct sockaddr
*)&ipaddr
);
1399 if (opt
== IPOPT_SSRR
) {
1400 type
= ICMP_UNREACH
;
1401 code
= ICMP_UNREACH_SRCFAIL
;
1404 if (!ip_dosourceroute
)
1405 goto nosourcerouting
;
1407 * Loose routing, and not at next destination
1408 * yet; nothing to do except forward.
1412 off
--; /* 0 origin */
1413 if (off
> optlen
- (int)sizeof(struct in_addr
)) {
1415 * End of source route. Should be for us.
1417 if (!ip_acceptsourceroute
)
1418 goto nosourcerouting
;
1419 save_rte(m
, cp
, ip
->ip_src
);
1424 if (!ip_dosourceroute
) {
1426 char buf
[sizeof "aaa.bbb.ccc.ddd"];
1429 * Acting as a router, so generate ICMP
1432 strcpy(buf
, inet_ntoa(ip
->ip_dst
));
1434 "attempted source route from %s to %s\n",
1435 inet_ntoa(ip
->ip_src
), buf
);
1436 type
= ICMP_UNREACH
;
1437 code
= ICMP_UNREACH_SRCFAIL
;
1441 * Not acting as a router,
1445 ipstat
.ips_cantforward
++;
1452 * locate outgoing interface
1454 memcpy(&ipaddr
.sin_addr
, cp
+ off
,
1455 sizeof ipaddr
.sin_addr
);
1457 if (opt
== IPOPT_SSRR
) {
1458 #define INA struct in_ifaddr *
1459 #define SA struct sockaddr *
1460 if ((ia
= (INA
)ifa_ifwithdstaddr((SA
)&ipaddr
))
1462 ia
= (INA
)ifa_ifwithnet((SA
)&ipaddr
);
1464 ia
= ip_rtaddr(ipaddr
.sin_addr
, NULL
);
1467 type
= ICMP_UNREACH
;
1468 code
= ICMP_UNREACH_SRCFAIL
;
1471 ip
->ip_dst
= ipaddr
.sin_addr
;
1472 memcpy(cp
+ off
, &IA_SIN(ia
)->sin_addr
,
1473 sizeof(struct in_addr
));
1474 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1476 * Let ip_intr's mcast routing check handle mcast pkts
1478 forward
= !IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
));
1482 if (ipstealth
&& pass
== 0)
1484 if (optlen
< IPOPT_OFFSET
+ sizeof(*cp
)) {
1485 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1488 if ((off
= cp
[IPOPT_OFFSET
]) < IPOPT_MINOFF
) {
1489 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1493 * If no space remains, ignore.
1495 off
--; /* 0 origin */
1496 if (off
> optlen
- (int)sizeof(struct in_addr
))
1498 memcpy(&ipaddr
.sin_addr
, &ip
->ip_dst
,
1499 sizeof ipaddr
.sin_addr
);
1501 * locate outgoing interface; if we're the destination,
1502 * use the incoming interface (should be same).
1504 if ((ia
= (INA
)ifa_ifwithaddr((SA
)&ipaddr
)) == NULL
&&
1505 (ia
= ip_rtaddr(ipaddr
.sin_addr
, NULL
)) == NULL
) {
1506 type
= ICMP_UNREACH
;
1507 code
= ICMP_UNREACH_HOST
;
1510 memcpy(cp
+ off
, &IA_SIN(ia
)->sin_addr
,
1511 sizeof(struct in_addr
));
1512 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1516 if (ipstealth
&& pass
== 0)
1518 code
= cp
- (u_char
*)ip
;
1519 if (optlen
< 4 || optlen
> 40) {
1520 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1523 if ((off
= cp
[IPOPT_OFFSET
]) < 5) {
1524 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1527 if (off
> optlen
- (int)sizeof(int32_t)) {
1528 cp
[IPOPT_OFFSET
+ 1] += (1 << 4);
1529 if ((cp
[IPOPT_OFFSET
+ 1] & 0xf0) == 0) {
1530 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1535 off
--; /* 0 origin */
1536 sin
= (struct in_addr
*)(cp
+ off
);
1537 switch (cp
[IPOPT_OFFSET
+ 1] & 0x0f) {
1539 case IPOPT_TS_TSONLY
:
1542 case IPOPT_TS_TSANDADDR
:
1543 if (off
+ sizeof(n_time
) +
1544 sizeof(struct in_addr
) > optlen
) {
1545 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1548 ipaddr
.sin_addr
= dst
;
1549 ia
= (INA
)ifaof_ifpforaddr((SA
)&ipaddr
,
1553 memcpy(sin
, &IA_SIN(ia
)->sin_addr
,
1554 sizeof(struct in_addr
));
1555 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1556 off
+= sizeof(struct in_addr
);
1559 case IPOPT_TS_PRESPEC
:
1560 if (off
+ sizeof(n_time
) +
1561 sizeof(struct in_addr
) > optlen
) {
1562 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1565 memcpy(&ipaddr
.sin_addr
, sin
,
1566 sizeof(struct in_addr
));
1567 if (ifa_ifwithaddr((SA
)&ipaddr
) == NULL
)
1569 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1570 off
+= sizeof(struct in_addr
);
1574 code
= &cp
[IPOPT_OFFSET
+ 1] - (u_char
*)ip
;
1578 memcpy(cp
+ off
, &ntime
, sizeof(n_time
));
1579 cp
[IPOPT_OFFSET
] += sizeof(n_time
);
1582 if (forward
&& ipforwarding
) {
1583 ip_forward(m
, TRUE
, next_hop
);
1588 icmp_error(m
, type
, code
, 0, 0);
1589 ipstat
.ips_badoptions
++;
1594 * Given address of next destination (final or next hop),
1595 * return internet address info of interface to be used to get there.
1598 ip_rtaddr(struct in_addr dst
, struct route
*ro0
)
1600 struct route sro
, *ro
;
1601 struct sockaddr_in
*sin
;
1602 struct in_ifaddr
*ia
;
1607 bzero(&sro
, sizeof(sro
));
1611 sin
= (struct sockaddr_in
*)&ro
->ro_dst
;
1613 if (ro
->ro_rt
== NULL
|| dst
.s_addr
!= sin
->sin_addr
.s_addr
) {
1614 if (ro
->ro_rt
!= NULL
) {
1618 sin
->sin_family
= AF_INET
;
1619 sin
->sin_len
= sizeof *sin
;
1620 sin
->sin_addr
= dst
;
1621 rtalloc_ign(ro
, RTF_PRCLONING
);
1624 if (ro
->ro_rt
== NULL
)
1627 ia
= ifatoia(ro
->ro_rt
->rt_ifa
);
1635 * Save incoming source route for use in replies,
1636 * to be picked up later by ip_srcroute if the receiver is interested.
1639 save_rte(struct mbuf
*m
, u_char
*option
, struct in_addr dst
)
1642 struct ip_srcrt_opt
*opt
;
1645 mtag
= m_tag_get(PACKET_TAG_IPSRCRT
, sizeof(*opt
), MB_DONTWAIT
);
1648 opt
= m_tag_data(mtag
);
1650 olen
= option
[IPOPT_OLEN
];
1653 kprintf("save_rte: olen %d\n", olen
);
1655 if (olen
> sizeof(opt
->ip_srcrt
) - (1 + sizeof(dst
))) {
1659 bcopy(option
, opt
->ip_srcrt
.srcopt
, olen
);
1660 opt
->ip_nhops
= (olen
- IPOPT_OFFSET
- 1) / sizeof(struct in_addr
);
1661 opt
->ip_srcrt
.dst
= dst
;
1662 m_tag_prepend(m
, mtag
);
1666 * Retrieve incoming source route for use in replies,
1667 * in the same form used by setsockopt.
1668 * The first hop is placed before the options, will be removed later.
1671 ip_srcroute(struct mbuf
*m0
)
1673 struct in_addr
*p
, *q
;
1676 struct ip_srcrt_opt
*opt
;
1681 mtag
= m_tag_find(m0
, PACKET_TAG_IPSRCRT
, NULL
);
1684 opt
= m_tag_data(mtag
);
1686 if (opt
->ip_nhops
== 0)
1688 m
= m_get(MB_DONTWAIT
, MT_HEADER
);
1692 #define OPTSIZ (sizeof(opt->ip_srcrt.nop) + sizeof(opt->ip_srcrt.srcopt))
1694 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1695 m
->m_len
= opt
->ip_nhops
* sizeof(struct in_addr
) +
1696 sizeof(struct in_addr
) + OPTSIZ
;
1699 kprintf("ip_srcroute: nhops %d mlen %d",
1700 opt
->ip_nhops
, m
->m_len
);
1705 * First save first hop for return route
1707 p
= &opt
->ip_srcrt
.route
[opt
->ip_nhops
- 1];
1708 *(mtod(m
, struct in_addr
*)) = *p
--;
1711 kprintf(" hops %x", ntohl(mtod(m
, struct in_addr
*)->s_addr
));
1715 * Copy option fields and padding (nop) to mbuf.
1717 opt
->ip_srcrt
.nop
= IPOPT_NOP
;
1718 opt
->ip_srcrt
.srcopt
[IPOPT_OFFSET
] = IPOPT_MINOFF
;
1719 memcpy(mtod(m
, caddr_t
) + sizeof(struct in_addr
), &opt
->ip_srcrt
.nop
,
1721 q
= (struct in_addr
*)(mtod(m
, caddr_t
) +
1722 sizeof(struct in_addr
) + OPTSIZ
);
1725 * Record return path as an IP source route,
1726 * reversing the path (pointers are now aligned).
1728 while (p
>= opt
->ip_srcrt
.route
) {
1731 kprintf(" %x", ntohl(q
->s_addr
));
1736 * Last hop goes to final destination.
1738 *q
= opt
->ip_srcrt
.dst
;
1739 m_tag_delete(m0
, mtag
);
1742 kprintf(" %x\n", ntohl(q
->s_addr
));
1748 * Strip out IP options.
1751 ip_stripoptions(struct mbuf
*m
)
1754 struct ip
*ip
= mtod(m
, struct ip
*);
1758 optlen
= (IP_VHL_HL(ip
->ip_vhl
) << 2) - sizeof(struct ip
);
1759 opts
= (caddr_t
)(ip
+ 1);
1760 datalen
= m
->m_len
- (sizeof(struct ip
) + optlen
);
1761 bcopy(opts
+ optlen
, opts
, datalen
);
1763 if (m
->m_flags
& M_PKTHDR
)
1764 m
->m_pkthdr
.len
-= optlen
;
1765 ip
->ip_vhl
= IP_MAKE_VHL(IPVERSION
, sizeof(struct ip
) >> 2);
1768 u_char inetctlerrmap
[PRC_NCMDS
] = {
1770 0, EMSGSIZE
, EHOSTDOWN
, EHOSTUNREACH
,
1771 EHOSTUNREACH
, EHOSTUNREACH
, ECONNREFUSED
, ECONNREFUSED
,
1772 EMSGSIZE
, EHOSTUNREACH
, 0, 0,
1774 ENOPROTOOPT
, ECONNREFUSED
1778 * Forward a packet. If some error occurs return the sender
1779 * an icmp packet. Note we can't always generate a meaningful
1780 * icmp message because icmp doesn't have a large enough repertoire
1781 * of codes and types.
1783 * If not forwarding, just drop the packet. This could be confusing
1784 * if ipforwarding was zero but some routing protocol was advancing
1785 * us as a gateway to somewhere. However, we must let the routing
1786 * protocol deal with that.
1788 * The using_srcrt parameter indicates whether the packet is being forwarded
1789 * via a source route.
1792 ip_forward(struct mbuf
*m
, boolean_t using_srcrt
, struct sockaddr_in
*next_hop
)
1794 struct ip
*ip
= mtod(m
, struct ip
*);
1796 struct route fwd_ro
;
1797 int error
, type
= 0, code
= 0, destmtu
= 0;
1800 struct in_addr pkt_dst
;
1804 * Cache the destination address of the packet; this may be
1805 * changed by use of 'ipfw fwd'.
1807 pkt_dst
= (next_hop
!= NULL
) ? next_hop
->sin_addr
: ip
->ip_dst
;
1811 kprintf("forward: src %x dst %x ttl %x\n",
1812 ip
->ip_src
.s_addr
, pkt_dst
.s_addr
, ip
->ip_ttl
);
1815 if (m
->m_flags
& (M_BCAST
| M_MCAST
) || !in_canforward(pkt_dst
)) {
1816 ipstat
.ips_cantforward
++;
1820 if (!ipstealth
&& ip
->ip_ttl
<= IPTTLDEC
) {
1821 icmp_error(m
, ICMP_TIMXCEED
, ICMP_TIMXCEED_INTRANS
, dest
, 0);
1825 bzero(&fwd_ro
, sizeof(fwd_ro
));
1826 ip_rtaddr(pkt_dst
, &fwd_ro
);
1827 if (fwd_ro
.ro_rt
== NULL
) {
1828 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_HOST
, dest
, 0);
1834 * Save the IP header and at most 8 bytes of the payload,
1835 * in case we need to generate an ICMP message to the src.
1837 * XXX this can be optimized a lot by saving the data in a local
1838 * buffer on the stack (72 bytes at most), and only allocating the
1839 * mbuf if really necessary. The vast majority of the packets
1840 * are forwarded without having to send an ICMP back (either
1841 * because unnecessary, or because rate limited), so we are
1842 * really we are wasting a lot of work here.
1844 * We don't use m_copy() because it might return a reference
1845 * to a shared cluster. Both this function and ip_output()
1846 * assume exclusive access to the IP header in `m', so any
1847 * data in a cluster may change before we reach icmp_error().
1849 MGETHDR(mcopy
, MB_DONTWAIT
, m
->m_type
);
1850 if (mcopy
!= NULL
&& !m_dup_pkthdr(mcopy
, m
, MB_DONTWAIT
)) {
1852 * It's probably ok if the pkthdr dup fails (because
1853 * the deep copy of the tag chain failed), but for now
1854 * be conservative and just discard the copy since
1855 * code below may some day want the tags.
1860 if (mcopy
!= NULL
) {
1861 mcopy
->m_len
= imin((IP_VHL_HL(ip
->ip_vhl
) << 2) + 8,
1863 mcopy
->m_pkthdr
.len
= mcopy
->m_len
;
1864 m_copydata(m
, 0, mcopy
->m_len
, mtod(mcopy
, caddr_t
));
1868 ip
->ip_ttl
-= IPTTLDEC
;
1871 * If forwarding packet using same interface that it came in on,
1872 * perhaps should send a redirect to sender to shortcut a hop.
1873 * Only send redirect if source is sending directly to us,
1874 * and if packet was not source routed (or has any options).
1875 * Also, don't send redirect if forwarding using a default route
1876 * or a route modified by a redirect.
1878 if (rt
->rt_ifp
== m
->m_pkthdr
.rcvif
&&
1879 !(rt
->rt_flags
& (RTF_DYNAMIC
| RTF_MODIFIED
)) &&
1880 satosin(rt_key(rt
))->sin_addr
.s_addr
!= INADDR_ANY
&&
1881 ipsendredirects
&& !using_srcrt
&& next_hop
== NULL
) {
1882 u_long src
= ntohl(ip
->ip_src
.s_addr
);
1883 struct in_ifaddr
*rt_ifa
= (struct in_ifaddr
*)rt
->rt_ifa
;
1885 if (rt_ifa
!= NULL
&&
1886 (src
& rt_ifa
->ia_subnetmask
) == rt_ifa
->ia_subnet
) {
1887 if (rt
->rt_flags
& RTF_GATEWAY
)
1888 dest
= satosin(rt
->rt_gateway
)->sin_addr
.s_addr
;
1890 dest
= pkt_dst
.s_addr
;
1892 * Router requirements says to only send
1895 type
= ICMP_REDIRECT
;
1896 code
= ICMP_REDIRECT_HOST
;
1899 kprintf("redirect (%d) to %x\n", code
, dest
);
1904 error
= ip_output(m
, NULL
, &fwd_ro
, IP_FORWARDING
, NULL
, NULL
);
1906 ipstat
.ips_forward
++;
1909 ipflow_create(&fwd_ro
, mcopy
);
1914 ipstat
.ips_redirectsent
++;
1917 ipstat
.ips_cantforward
++;
1924 * Send ICMP message.
1929 case 0: /* forwarded, but need redirect */
1930 /* type, code set above */
1933 case ENETUNREACH
: /* shouldn't happen, checked above */
1938 type
= ICMP_UNREACH
;
1939 code
= ICMP_UNREACH_HOST
;
1943 type
= ICMP_UNREACH
;
1944 code
= ICMP_UNREACH_NEEDFRAG
;
1947 * If the packet is routed over IPsec tunnel, tell the
1948 * originator the tunnel MTU.
1949 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1952 if (fwd_ro
.ro_rt
!= NULL
) {
1953 struct secpolicy
*sp
= NULL
;
1958 sp
= ipsec4_getpolicybyaddr(mcopy
,
1964 destmtu
= fwd_ro
.ro_rt
->rt_ifp
->if_mtu
;
1966 /* count IPsec header size */
1967 ipsechdr
= ipsec4_hdrsiz(mcopy
,
1972 * find the correct route for outer IPv4
1973 * header, compute tunnel MTU.
1976 if (sp
->req
!= NULL
&& sp
->req
->sav
!= NULL
&&
1977 sp
->req
->sav
->sah
!= NULL
) {
1978 ro
= &sp
->req
->sav
->sah
->sa_route
;
1979 if (ro
->ro_rt
!= NULL
&&
1980 ro
->ro_rt
->rt_ifp
!= NULL
) {
1982 ro
->ro_rt
->rt_ifp
->if_mtu
;
1983 destmtu
-= ipsechdr
;
1992 * If the packet is routed over IPsec tunnel, tell the
1993 * originator the tunnel MTU.
1994 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1997 if (fwd_ro
.ro_rt
!= NULL
) {
1998 struct secpolicy
*sp
= NULL
;
2003 sp
= ipsec_getpolicybyaddr(mcopy
,
2009 destmtu
= fwd_ro
.ro_rt
->rt_ifp
->if_mtu
;
2011 /* count IPsec header size */
2012 ipsechdr
= ipsec4_hdrsiz(mcopy
,
2017 * find the correct route for outer IPv4
2018 * header, compute tunnel MTU.
2021 if (sp
->req
!= NULL
&&
2022 sp
->req
->sav
!= NULL
&&
2023 sp
->req
->sav
->sah
!= NULL
) {
2024 ro
= &sp
->req
->sav
->sah
->sa_route
;
2025 if (ro
->ro_rt
!= NULL
&&
2026 ro
->ro_rt
->rt_ifp
!= NULL
) {
2028 ro
->ro_rt
->rt_ifp
->if_mtu
;
2029 destmtu
-= ipsechdr
;
2036 #else /* !IPSEC && !FAST_IPSEC */
2037 if (fwd_ro
.ro_rt
!= NULL
)
2038 destmtu
= fwd_ro
.ro_rt
->rt_ifp
->if_mtu
;
2040 ipstat
.ips_cantfrag
++;
2045 * A router should not generate ICMP_SOURCEQUENCH as
2046 * required in RFC1812 Requirements for IP Version 4 Routers.
2047 * Source quench could be a big problem under DoS attacks,
2048 * or if the underlying interface is rate-limited.
2049 * Those who need source quench packets may re-enable them
2050 * via the net.inet.ip.sendsourcequench sysctl.
2052 if (!ip_sendsourcequench
) {
2056 type
= ICMP_SOURCEQUENCH
;
2061 case EACCES
: /* ipfw denied packet */
2065 icmp_error(mcopy
, type
, code
, dest
, destmtu
);
2067 if (fwd_ro
.ro_rt
!= NULL
)
2068 RTFREE(fwd_ro
.ro_rt
);
2072 ip_savecontrol(struct inpcb
*inp
, struct mbuf
**mp
, struct ip
*ip
,
2075 if (inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
2079 *mp
= sbcreatecontrol((caddr_t
) &tv
, sizeof(tv
),
2080 SCM_TIMESTAMP
, SOL_SOCKET
);
2082 mp
= &(*mp
)->m_next
;
2084 if (inp
->inp_flags
& INP_RECVDSTADDR
) {
2085 *mp
= sbcreatecontrol((caddr_t
) &ip
->ip_dst
,
2086 sizeof(struct in_addr
), IP_RECVDSTADDR
, IPPROTO_IP
);
2088 mp
= &(*mp
)->m_next
;
2090 if (inp
->inp_flags
& INP_RECVTTL
) {
2091 *mp
= sbcreatecontrol((caddr_t
) &ip
->ip_ttl
,
2092 sizeof(u_char
), IP_RECVTTL
, IPPROTO_IP
);
2094 mp
= &(*mp
)->m_next
;
2098 * Moving these out of udp_input() made them even more broken
2099 * than they already were.
2101 /* options were tossed already */
2102 if (inp
->inp_flags
& INP_RECVOPTS
) {
2103 *mp
= sbcreatecontrol((caddr_t
) opts_deleted_above
,
2104 sizeof(struct in_addr
), IP_RECVOPTS
, IPPROTO_IP
);
2106 mp
= &(*mp
)->m_next
;
2108 /* ip_srcroute doesn't do what we want here, need to fix */
2109 if (inp
->inp_flags
& INP_RECVRETOPTS
) {
2110 *mp
= sbcreatecontrol((caddr_t
) ip_srcroute(m
),
2111 sizeof(struct in_addr
), IP_RECVRETOPTS
, IPPROTO_IP
);
2113 mp
= &(*mp
)->m_next
;
2116 if (inp
->inp_flags
& INP_RECVIF
) {
2119 struct sockaddr_dl sdl
;
2122 struct sockaddr_dl
*sdp
;
2123 struct sockaddr_dl
*sdl2
= &sdlbuf
.sdl
;
2125 if (((ifp
= m
->m_pkthdr
.rcvif
)) &&
2126 ((ifp
->if_index
!= 0) && (ifp
->if_index
<= if_index
))) {
2127 sdp
= IF_LLSOCKADDR(ifp
);
2129 * Change our mind and don't try copy.
2131 if ((sdp
->sdl_family
!= AF_LINK
) ||
2132 (sdp
->sdl_len
> sizeof(sdlbuf
))) {
2135 bcopy(sdp
, sdl2
, sdp
->sdl_len
);
2139 offsetof(struct sockaddr_dl
, sdl_data
[0]);
2140 sdl2
->sdl_family
= AF_LINK
;
2141 sdl2
->sdl_index
= 0;
2142 sdl2
->sdl_nlen
= sdl2
->sdl_alen
= sdl2
->sdl_slen
= 0;
2144 *mp
= sbcreatecontrol((caddr_t
) sdl2
, sdl2
->sdl_len
,
2145 IP_RECVIF
, IPPROTO_IP
);
2147 mp
= &(*mp
)->m_next
;
2152 * XXX these routines are called from the upper part of the kernel.
2154 * They could also be moved to ip_mroute.c, since all the RSVP
2155 * handling is done there already.
2158 ip_rsvp_init(struct socket
*so
)
2160 if (so
->so_type
!= SOCK_RAW
||
2161 so
->so_proto
->pr_protocol
!= IPPROTO_RSVP
)
2164 if (ip_rsvpd
!= NULL
)
2169 * This may seem silly, but we need to be sure we don't over-increment
2170 * the RSVP counter, in case something slips up.
2185 * This may seem silly, but we need to be sure we don't over-decrement
2186 * the RSVP counter, in case something slips up.
2196 rsvp_input(struct mbuf
*m
, ...) /* XXX must fixup manually */
2202 off
= __va_arg(ap
, int);
2203 proto
= __va_arg(ap
, int);
2206 if (rsvp_input_p
) { /* call the real one if loaded */
2207 rsvp_input_p(m
, off
, proto
);
2211 /* Can still get packets with rsvp_on = 0 if there is a local member
2212 * of the group to which the RSVP packet is addressed. But in this
2213 * case we want to throw the packet away.
2221 if (ip_rsvpd
!= NULL
) {
2222 rip_input(m
, off
, proto
);
2225 /* Drop the packet */