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 $
72 #include "opt_bootp.h"
74 #include "opt_ipdivert.h"
75 #include "opt_ipstealth.h"
76 #include "opt_ipsec.h"
78 #include <sys/param.h>
79 #include <sys/systm.h>
81 #include <sys/malloc.h>
82 #include <sys/mpipe.h>
83 #include <sys/domain.h>
84 #include <sys/protosw.h>
85 #include <sys/socket.h>
87 #include <sys/globaldata.h>
88 #include <sys/thread.h>
89 #include <sys/kernel.h>
90 #include <sys/syslog.h>
91 #include <sys/sysctl.h>
92 #include <sys/in_cksum.h>
95 #include <sys/mplock2.h>
97 #include <machine/stdarg.h>
100 #include <net/if_types.h>
101 #include <net/if_var.h>
102 #include <net/if_dl.h>
103 #include <net/pfil.h>
104 #include <net/route.h>
105 #include <net/netisr.h>
107 #include <netinet/in.h>
108 #include <netinet/in_systm.h>
109 #include <netinet/in_var.h>
110 #include <netinet/ip.h>
111 #include <netinet/in_pcb.h>
112 #include <netinet/ip_var.h>
113 #include <netinet/ip_icmp.h>
114 #include <netinet/ip_divert.h>
115 #include <netinet/ip_flow.h>
117 #include <sys/thread2.h>
118 #include <sys/msgport2.h>
119 #include <net/netmsg2.h>
121 #include <sys/socketvar.h>
123 #include <net/ipfw/ip_fw.h>
124 #include <net/dummynet/ip_dummynet.h>
127 #include <netinet6/ipsec.h>
128 #include <netproto/key/key.h>
132 #include <netproto/ipsec/ipsec.h>
133 #include <netproto/ipsec/key.h>
137 static int ip_rsvp_on
;
138 struct socket
*ip_rsvpd
;
140 int ipforwarding
= 0;
141 SYSCTL_INT(_net_inet_ip
, IPCTL_FORWARDING
, forwarding
, CTLFLAG_RW
,
142 &ipforwarding
, 0, "Enable IP forwarding between interfaces");
144 static int ipsendredirects
= 1; /* XXX */
145 SYSCTL_INT(_net_inet_ip
, IPCTL_SENDREDIRECTS
, redirect
, CTLFLAG_RW
,
146 &ipsendredirects
, 0, "Enable sending IP redirects");
148 int ip_defttl
= IPDEFTTL
;
149 SYSCTL_INT(_net_inet_ip
, IPCTL_DEFTTL
, ttl
, CTLFLAG_RW
,
150 &ip_defttl
, 0, "Maximum TTL on IP packets");
152 static int ip_dosourceroute
= 0;
153 SYSCTL_INT(_net_inet_ip
, IPCTL_SOURCEROUTE
, sourceroute
, CTLFLAG_RW
,
154 &ip_dosourceroute
, 0, "Enable forwarding source routed IP packets");
156 static int ip_acceptsourceroute
= 0;
157 SYSCTL_INT(_net_inet_ip
, IPCTL_ACCEPTSOURCEROUTE
, accept_sourceroute
,
158 CTLFLAG_RW
, &ip_acceptsourceroute
, 0,
159 "Enable accepting source routed IP packets");
161 static int ip_keepfaith
= 0;
162 SYSCTL_INT(_net_inet_ip
, IPCTL_KEEPFAITH
, keepfaith
, CTLFLAG_RW
,
164 "Enable packet capture for FAITH IPv4->IPv6 translator daemon");
166 static int nipq
= 0; /* total # of reass queues */
168 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, maxfragpackets
, CTLFLAG_RW
,
170 "Maximum number of IPv4 fragment reassembly queue entries");
172 static int maxfragsperpacket
;
173 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, maxfragsperpacket
, CTLFLAG_RW
,
174 &maxfragsperpacket
, 0,
175 "Maximum number of IPv4 fragments allowed per packet");
177 static int ip_sendsourcequench
= 0;
178 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, sendsourcequench
, CTLFLAG_RW
,
179 &ip_sendsourcequench
, 0,
180 "Enable the transmission of source quench packets");
182 int ip_do_randomid
= 1;
183 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, random_id
, CTLFLAG_RW
,
185 "Assign random ip_id values");
187 * XXX - Setting ip_checkinterface mostly implements the receive side of
188 * the Strong ES model described in RFC 1122, but since the routing table
189 * and transmit implementation do not implement the Strong ES model,
190 * setting this to 1 results in an odd hybrid.
192 * XXX - ip_checkinterface currently must be disabled if you use ipnat
193 * to translate the destination address to another local interface.
195 * XXX - ip_checkinterface must be disabled if you add IP aliases
196 * to the loopback interface instead of the interface where the
197 * packets for those addresses are received.
199 static int ip_checkinterface
= 0;
200 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, check_interface
, CTLFLAG_RW
,
201 &ip_checkinterface
, 0, "Verify packet arrives on correct interface");
203 static int ip_dispatch_fast
= 0;
204 static int ip_dispatch_slow
= 0;
205 static int ip_dispatch_recheck
= 0;
206 static int ip_dispatch_software
= 0;
207 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, dispatch_fast_count
, CTLFLAG_RD
,
208 &ip_dispatch_fast
, 0,
209 "Number of IP dispatches handled on current CPU");
210 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, dispatch_slow_count
, CTLFLAG_RD
,
211 &ip_dispatch_slow
, 0,
212 "Number of IP dispatches messaged to another CPU");
213 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, dispatch_software_count
, CTLFLAG_RD
,
214 &ip_dispatch_software
, 0, "");
215 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, dispatch_recheck_count
, CTLFLAG_RD
,
216 &ip_dispatch_recheck
, 0, "");
218 static struct lwkt_token ipq_token
= LWKT_TOKEN_INITIALIZER(ipq_token
);
221 static int ipprintfs
= 0;
224 extern struct domain inetdomain
;
225 extern struct protosw inetsw
[];
226 u_char ip_protox
[IPPROTO_MAX
];
227 struct in_ifaddrhead in_ifaddrheads
[MAXCPU
]; /* first inet address */
228 struct in_ifaddrhashhead
*in_ifaddrhashtbls
[MAXCPU
];
229 /* inet addr hash table */
230 u_long in_ifaddrhmask
; /* mask for hash table */
232 struct ip_stats ipstats_percpu
[MAXCPU
];
235 sysctl_ipstats(SYSCTL_HANDLER_ARGS
)
239 for (cpu
= 0; cpu
< ncpus
; ++cpu
) {
240 if ((error
= SYSCTL_OUT(req
, &ipstats_percpu
[cpu
],
241 sizeof(struct ip_stats
))))
243 if ((error
= SYSCTL_IN(req
, &ipstats_percpu
[cpu
],
244 sizeof(struct ip_stats
))))
250 SYSCTL_PROC(_net_inet_ip
, IPCTL_STATS
, stats
, (CTLTYPE_OPAQUE
| CTLFLAG_RW
),
251 0, 0, sysctl_ipstats
, "S,ip_stats", "IP statistics");
253 SYSCTL_STRUCT(_net_inet_ip
, IPCTL_STATS
, stats
, CTLFLAG_RW
,
254 &ipstat
, ip_stats
, "IP statistics");
257 /* Packet reassembly stuff */
258 #define IPREASS_NHASH_LOG2 6
259 #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
260 #define IPREASS_HMASK (IPREASS_NHASH - 1)
261 #define IPREASS_HASH(x,y) \
262 (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
264 static TAILQ_HEAD(ipqhead
, ipq
) ipq
[IPREASS_NHASH
];
267 SYSCTL_INT(_net_inet_ip
, IPCTL_DEFMTU
, mtu
, CTLFLAG_RW
,
268 &ip_mtu
, 0, "Default MTU");
272 static int ipstealth
= 0;
273 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, stealth
, CTLFLAG_RW
, &ipstealth
, 0, "");
275 static const int ipstealth
= 0;
278 struct mbuf
*(*ip_divert_p
)(struct mbuf
*, int, int);
280 struct pfil_head inet_pfil_hook
;
283 * struct ip_srcrt_opt is used to store packet state while it travels
286 * XXX Note that the code even makes assumptions on the size and
287 * alignment of fields inside struct ip_srcrt so e.g. adding some
288 * fields will break the code. This needs to be fixed.
290 * We need to save the IP options in case a protocol wants to respond
291 * to an incoming packet over the same route if the packet got here
292 * using IP source routing. This allows connection establishment and
293 * maintenance when the remote end is on a network that is not known
297 struct in_addr dst
; /* final destination */
298 char nop
; /* one NOP to align */
299 char srcopt
[IPOPT_OFFSET
+ 1]; /* OPTVAL, OLEN and OFFSET */
300 struct in_addr route
[MAX_IPOPTLEN
/sizeof(struct in_addr
)];
303 struct ip_srcrt_opt
{
305 struct ip_srcrt ip_srcrt
;
308 static MALLOC_DEFINE(M_IPQ
, "ipq", "IP Fragment Management");
309 static struct malloc_pipe ipq_mpipe
;
311 static void save_rte(struct mbuf
*, u_char
*, struct in_addr
);
312 static int ip_dooptions(struct mbuf
*m
, int, struct sockaddr_in
*);
313 static void ip_freef(struct ipqhead
*, struct ipq
*);
314 static void ip_input_handler(netmsg_t
);
317 * IP initialization: fill in IP protocol switch table.
318 * All protocols not implemented in kernel go to raw IP protocol handler.
330 * Make sure we can handle a reasonable number of fragments but
331 * cap it at 4000 (XXX).
333 mpipe_init(&ipq_mpipe
, M_IPQ
, sizeof(struct ipq
),
334 IFQ_MAXLEN
, 4000, 0, NULL
, NULL
, NULL
);
335 for (i
= 0; i
< ncpus
; ++i
) {
336 TAILQ_INIT(&in_ifaddrheads
[i
]);
337 in_ifaddrhashtbls
[i
] =
338 hashinit(INADDR_NHASH
, M_IFADDR
, &in_ifaddrhmask
);
340 pr
= pffindproto(PF_INET
, IPPROTO_RAW
, SOCK_RAW
);
343 for (i
= 0; i
< IPPROTO_MAX
; i
++)
344 ip_protox
[i
] = pr
- inetsw
;
345 for (pr
= inetdomain
.dom_protosw
;
346 pr
< inetdomain
.dom_protoswNPROTOSW
; pr
++) {
347 if (pr
->pr_domain
->dom_family
== PF_INET
&& pr
->pr_protocol
) {
348 if (pr
->pr_protocol
!= IPPROTO_RAW
)
349 ip_protox
[pr
->pr_protocol
] = pr
- inetsw
;
353 inet_pfil_hook
.ph_type
= PFIL_TYPE_AF
;
354 inet_pfil_hook
.ph_af
= AF_INET
;
355 if ((i
= pfil_head_register(&inet_pfil_hook
)) != 0) {
356 kprintf("%s: WARNING: unable to register pfil hook, "
357 "error %d\n", __func__
, i
);
360 for (i
= 0; i
< IPREASS_NHASH
; i
++)
363 maxnipq
= nmbclusters
/ 32;
364 maxfragsperpacket
= 16;
366 ip_id
= time_second
& 0xffff;
369 * Initialize IP statistics counters for each CPU.
373 for (cpu
= 0; cpu
< ncpus
; ++cpu
) {
374 bzero(&ipstats_percpu
[cpu
], sizeof(struct ip_stats
));
377 bzero(&ipstat
, sizeof(struct ip_stats
));
380 netisr_register(NETISR_IP
, ip_input_handler
, ip_cpufn_in
);
381 netisr_register_hashcheck(NETISR_IP
, ip_hashcheck
);
384 /* Do transport protocol processing. */
386 transport_processing_oncpu(struct mbuf
*m
, int hlen
, struct ip
*ip
)
388 const struct protosw
*pr
= &inetsw
[ip_protox
[ip
->ip_p
]];
391 * Switch out to protocol's input routine.
394 pr
->pr_input(&m
, &hlen
, ip
->ip_p
);
399 transport_processing_handler(netmsg_t msg
)
401 struct netmsg_packet
*pmsg
= &msg
->packet
;
405 ip
= mtod(pmsg
->nm_packet
, struct ip
*);
406 hlen
= pmsg
->base
.lmsg
.u
.ms_result
;
408 transport_processing_oncpu(pmsg
->nm_packet
, hlen
, ip
);
409 /* msg was embedded in the mbuf, do not reply! */
413 ip_input_handler(netmsg_t msg
)
415 ip_input(msg
->packet
.nm_packet
);
416 /* msg was embedded in the mbuf, do not reply! */
420 * IP input routine. Checksum and byte swap header. If fragmented
421 * try to reassemble. Process options. Pass to next level.
424 ip_input(struct mbuf
*m
)
427 struct in_ifaddr
*ia
= NULL
;
428 struct in_ifaddr_container
*iac
;
431 struct in_addr pkt_dst
;
432 boolean_t using_srcrt
= FALSE
; /* forward (by PFIL_HOOKS) */
433 struct in_addr odst
; /* original dst address(NAT) */
435 struct sockaddr_in
*next_hop
= NULL
;
438 struct tdb_ident
*tdbi
;
439 struct secpolicy
*sp
;
446 * This routine is called from numerous places which may not have
447 * characterized the packet.
449 if ((m
->m_flags
& M_HASH
) == 0) {
450 ++ip_dispatch_software
;
451 ip_cpufn(&m
, 0, IP_MPORT_IN
);
454 KKASSERT(m
->m_flags
& M_HASH
);
456 ip
= mtod(m
, struct ip
*);
459 * Pull out certain tags
461 if (m
->m_pkthdr
.fw_flags
& IPFORWARD_MBUF_TAGGED
) {
463 mtag
= m_tag_find(m
, PACKET_TAG_IPFORWARD
, NULL
);
464 KKASSERT(mtag
!= NULL
);
465 next_hop
= m_tag_data(mtag
);
468 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
469 /* dummynet already filtered us */
470 ip
= mtod(m
, struct ip
*);
471 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
477 /* length checks already done in ip_cpufn() */
478 KASSERT(m
->m_len
>= sizeof(struct ip
), ("IP header not in one mbuf"));
480 if (IP_VHL_V(ip
->ip_vhl
) != IPVERSION
) {
481 ipstat
.ips_badvers
++;
485 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
486 /* length checks already done in ip_cpufn() */
487 KASSERT(hlen
>= sizeof(struct ip
), ("IP header len too small"));
488 KASSERT(m
->m_len
>= hlen
, ("complete IP header not in one mbuf"));
490 /* 127/8 must not appear on wire - RFC1122 */
491 if ((ntohl(ip
->ip_dst
.s_addr
) >> IN_CLASSA_NSHIFT
) == IN_LOOPBACKNET
||
492 (ntohl(ip
->ip_src
.s_addr
) >> IN_CLASSA_NSHIFT
) == IN_LOOPBACKNET
) {
493 if (!(m
->m_pkthdr
.rcvif
->if_flags
& IFF_LOOPBACK
)) {
494 ipstat
.ips_badaddr
++;
499 if (m
->m_pkthdr
.csum_flags
& CSUM_IP_CHECKED
) {
500 sum
= !(m
->m_pkthdr
.csum_flags
& CSUM_IP_VALID
);
502 if (hlen
== sizeof(struct ip
))
503 sum
= in_cksum_hdr(ip
);
505 sum
= in_cksum(m
, hlen
);
513 if (altq_input
!= NULL
&& (*altq_input
)(m
, AF_INET
) == 0) {
514 /* packet is dropped by traffic conditioner */
519 * Convert fields to host representation.
521 ip
->ip_len
= ntohs(ip
->ip_len
);
522 ip
->ip_off
= ntohs(ip
->ip_off
);
524 /* length checks already done in ip_cpufn() */
525 KASSERT(ip
->ip_len
>= hlen
, ("total length less then header length"));
526 KASSERT(m
->m_pkthdr
.len
>= ip
->ip_len
, ("mbuf too short"));
529 * Trim mbufs if longer than the IP header would have us expect.
531 if (m
->m_pkthdr
.len
> ip
->ip_len
) {
532 if (m
->m_len
== m
->m_pkthdr
.len
) {
533 m
->m_len
= ip
->ip_len
;
534 m
->m_pkthdr
.len
= ip
->ip_len
;
536 m_adj(m
, ip
->ip_len
- m
->m_pkthdr
.len
);
539 #if defined(IPSEC) && !defined(IPSEC_FILTERGIF)
541 * Bypass packet filtering for packets from a tunnel (gif).
543 if (ipsec_gethist(m
, NULL
))
549 * Right now when no processing on packet has done
550 * and it is still fresh out of network we do our black
552 * - Firewall: deny/allow/divert
553 * - Xlate: translate packet's addr/port (NAT).
554 * - Pipe: pass pkt through dummynet.
555 * - Wrap: fake packet's addr/port <unimpl.>
556 * - Encapsulate: put it in another IP and send out. <unimp.>
561 * If we've been forwarded from the output side, then
562 * skip the firewall a second time
564 if (next_hop
!= NULL
)
568 if (!pfil_has_hooks(&inet_pfil_hook
)) {
569 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
571 * Strip dummynet tags from stranded packets
573 mtag
= m_tag_find(m
, PACKET_TAG_DUMMYNET
, NULL
);
574 KKASSERT(mtag
!= NULL
);
575 m_tag_delete(m
, mtag
);
576 m
->m_pkthdr
.fw_flags
&= ~DUMMYNET_MBUF_TAGGED
;
582 * Run through list of hooks for input packets.
584 * NOTE! If the packet is rewritten pf/ipfw/whoever must
588 if (pfil_run_hooks(&inet_pfil_hook
, &m
, m
->m_pkthdr
.rcvif
, PFIL_IN
))
590 if (m
== NULL
) /* consumed by filter */
592 ip
= mtod(m
, struct ip
*);
593 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
594 using_srcrt
= (odst
.s_addr
!= ip
->ip_dst
.s_addr
);
596 if (m
->m_pkthdr
.fw_flags
& IPFORWARD_MBUF_TAGGED
) {
597 mtag
= m_tag_find(m
, PACKET_TAG_IPFORWARD
, NULL
);
598 KKASSERT(mtag
!= NULL
);
599 next_hop
= m_tag_data(mtag
);
601 if (m
->m_pkthdr
.fw_flags
& DUMMYNET_MBUF_TAGGED
) {
605 if (m
->m_pkthdr
.fw_flags
& FW_MBUF_REDISPATCH
) {
606 m
->m_pkthdr
.fw_flags
&= ~FW_MBUF_REDISPATCH
;
610 * Process options and, if not destined for us,
611 * ship it on. ip_dooptions returns 1 when an
612 * error was detected (causing an icmp message
613 * to be sent and the original packet to be freed).
615 if (hlen
> sizeof(struct ip
) && ip_dooptions(m
, 0, next_hop
))
618 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no
619 * matter if it is destined to another node, or whether it is
620 * a multicast one, RSVP wants it! and prevents it from being forwarded
621 * anywhere else. Also checks if the rsvp daemon is running before
622 * grabbing the packet.
624 if (rsvp_on
&& ip
->ip_p
== IPPROTO_RSVP
)
628 * Check our list of addresses, to see if the packet is for us.
629 * If we don't have any addresses, assume any unicast packet
630 * we receive might be for us (and let the upper layers deal
633 if (TAILQ_EMPTY(&in_ifaddrheads
[mycpuid
]) &&
634 !(m
->m_flags
& (M_MCAST
| M_BCAST
)))
638 * Cache the destination address of the packet; this may be
639 * changed by use of 'ipfw fwd'.
641 pkt_dst
= next_hop
? next_hop
->sin_addr
: ip
->ip_dst
;
644 * Enable a consistency check between the destination address
645 * and the arrival interface for a unicast packet (the RFC 1122
646 * strong ES model) if IP forwarding is disabled and the packet
647 * is not locally generated and the packet is not subject to
650 * XXX - Checking also should be disabled if the destination
651 * address is ipnat'ed to a different interface.
653 * XXX - Checking is incompatible with IP aliases added
654 * to the loopback interface instead of the interface where
655 * the packets are received.
657 checkif
= ip_checkinterface
&&
659 m
->m_pkthdr
.rcvif
!= NULL
&&
660 !(m
->m_pkthdr
.rcvif
->if_flags
& IFF_LOOPBACK
) &&
664 * Check for exact addresses in the hash bucket.
666 LIST_FOREACH(iac
, INADDR_HASH(pkt_dst
.s_addr
), ia_hash
) {
670 * If the address matches, verify that the packet
671 * arrived via the correct interface if checking is
674 if (IA_SIN(ia
)->sin_addr
.s_addr
== pkt_dst
.s_addr
&&
675 (!checkif
|| ia
->ia_ifp
== m
->m_pkthdr
.rcvif
))
681 * Check for broadcast addresses.
683 * Only accept broadcast packets that arrive via the matching
684 * interface. Reception of forwarded directed broadcasts would
685 * be handled via ip_forward() and ether_output() with the loopback
686 * into the stack for SIMPLEX interfaces handled by ether_output().
688 if (m
->m_pkthdr
.rcvif
->if_flags
& IFF_BROADCAST
) {
689 struct ifaddr_container
*ifac
;
691 TAILQ_FOREACH(ifac
, &m
->m_pkthdr
.rcvif
->if_addrheads
[mycpuid
],
693 struct ifaddr
*ifa
= ifac
->ifa
;
695 if (ifa
->ifa_addr
== NULL
) /* shutdown/startup race */
697 if (ifa
->ifa_addr
->sa_family
!= AF_INET
)
700 if (satosin(&ia
->ia_broadaddr
)->sin_addr
.s_addr
==
703 if (ia
->ia_netbroadcast
.s_addr
== pkt_dst
.s_addr
)
706 if (IA_SIN(ia
)->sin_addr
.s_addr
== INADDR_ANY
)
711 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
))) {
712 struct in_multi
*inm
;
714 /* XXX Multicast is not MPSAFE yet */
717 if (ip_mrouter
!= NULL
) {
719 * If we are acting as a multicast router, all
720 * incoming multicast packets are passed to the
721 * kernel-level multicast forwarding function.
722 * The packet is returned (relatively) intact; if
723 * ip_mforward() returns a non-zero value, the packet
724 * must be discarded, else it may be accepted below.
726 if (ip_mforward
!= NULL
&&
727 ip_mforward(ip
, m
->m_pkthdr
.rcvif
, m
, NULL
) != 0) {
729 ipstat
.ips_cantforward
++;
735 * The process-level routing daemon needs to receive
736 * all multicast IGMP packets, whether or not this
737 * host belongs to their destination groups.
739 if (ip
->ip_p
== IPPROTO_IGMP
) {
743 ipstat
.ips_forward
++;
746 * See if we belong to the destination multicast group on the
749 IN_LOOKUP_MULTI(ip
->ip_dst
, m
->m_pkthdr
.rcvif
, inm
);
752 ipstat
.ips_notmember
++;
760 if (ip
->ip_dst
.s_addr
== INADDR_BROADCAST
)
762 if (ip
->ip_dst
.s_addr
== INADDR_ANY
)
766 * FAITH(Firewall Aided Internet Translator)
768 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
== IFT_FAITH
) {
770 if (ip
->ip_p
== IPPROTO_TCP
|| ip
->ip_p
== IPPROTO_ICMP
)
778 * Not for us; forward if possible and desirable.
781 ipstat
.ips_cantforward
++;
786 * Enforce inbound IPsec SPD.
788 if (ipsec4_in_reject(m
, NULL
)) {
789 ipsecstat
.in_polvio
++;
794 mtag
= m_tag_find(m
, PACKET_TAG_IPSEC_IN_DONE
, NULL
);
797 tdbi
= (struct tdb_ident
*)m_tag_data(mtag
);
798 sp
= ipsec_getpolicy(tdbi
, IPSEC_DIR_INBOUND
);
800 sp
= ipsec_getpolicybyaddr(m
, IPSEC_DIR_INBOUND
,
801 IP_FORWARDING
, &error
);
803 if (sp
== NULL
) { /* NB: can happen if error */
805 /*XXX error stat???*/
806 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
811 * Check security policy against packet attributes.
813 error
= ipsec_in_reject(sp
, m
);
817 ipstat
.ips_cantforward
++;
821 ip_forward(m
, using_srcrt
, next_hop
);
828 * IPSTEALTH: Process non-routing options only
829 * if the packet is destined for us.
832 hlen
> sizeof(struct ip
) &&
833 ip_dooptions(m
, 1, next_hop
))
836 /* Count the packet in the ip address stats */
838 ia
->ia_ifa
.if_ipackets
++;
839 ia
->ia_ifa
.if_ibytes
+= m
->m_pkthdr
.len
;
843 * If offset or IP_MF are set, must reassemble.
844 * Otherwise, nothing need be done.
845 * (We could look in the reassembly queue to see
846 * if the packet was previously fragmented,
847 * but it's not worth the time; just let them time out.)
849 if (ip
->ip_off
& (IP_MF
| IP_OFFMASK
)) {
851 * Attempt reassembly; if it succeeds, proceed. ip_reass()
852 * will return a different mbuf.
854 * NOTE: ip_reass() returns m with M_HASH cleared to force
855 * us to recharacterize the packet.
860 ip
= mtod(m
, struct ip
*);
862 /* Get the header length of the reassembled packet */
863 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
870 * enforce IPsec policy checking if we are seeing last header.
871 * note that we do not visit this with protocols with pcb layer
872 * code - like udp/tcp/raw ip.
874 if ((inetsw
[ip_protox
[ip
->ip_p
]].pr_flags
& PR_LASTHDR
) &&
875 ipsec4_in_reject(m
, NULL
)) {
876 ipsecstat
.in_polvio
++;
882 * enforce IPsec policy checking if we are seeing last header.
883 * note that we do not visit this with protocols with pcb layer
884 * code - like udp/tcp/raw ip.
886 if (inetsw
[ip_protox
[ip
->ip_p
]].pr_flags
& PR_LASTHDR
) {
888 * Check if the packet has already had IPsec processing
889 * done. If so, then just pass it along. This tag gets
890 * set during AH, ESP, etc. input handling, before the
891 * packet is returned to the ip input queue for delivery.
893 mtag
= m_tag_find(m
, PACKET_TAG_IPSEC_IN_DONE
, NULL
);
896 tdbi
= (struct tdb_ident
*)m_tag_data(mtag
);
897 sp
= ipsec_getpolicy(tdbi
, IPSEC_DIR_INBOUND
);
899 sp
= ipsec_getpolicybyaddr(m
, IPSEC_DIR_INBOUND
,
900 IP_FORWARDING
, &error
);
904 * Check security policy against packet attributes.
906 error
= ipsec_in_reject(sp
, m
);
909 /* XXX error stat??? */
911 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
918 #endif /* FAST_IPSEC */
921 * We must forward the packet to the correct protocol thread if
922 * we are not already in it.
924 * NOTE: ip_len is now in host form. ip_len is not adjusted
925 * further for protocol processing, instead we pass hlen
926 * to the protosw and let it deal with it.
928 ipstat
.ips_delivered
++;
930 if ((m
->m_flags
& M_HASH
) == 0) {
931 ++ip_dispatch_recheck
;
932 ip
->ip_len
= htons(ip
->ip_len
+ hlen
);
933 ip
->ip_off
= htons(ip
->ip_off
);
935 ip_cpufn(&m
, 0, IP_MPORT_IN
);
939 ip
= mtod(m
, struct ip
*);
940 ip
->ip_len
= ntohs(ip
->ip_len
) - hlen
;
941 ip
->ip_off
= ntohs(ip
->ip_off
);
942 KKASSERT(m
->m_flags
& M_HASH
);
944 port
= cpu_portfn(m
->m_pkthdr
.hash
);
946 if (port
!= &curthread
->td_msgport
) {
947 struct netmsg_packet
*pmsg
;
951 pmsg
= &m
->m_hdr
.mh_netmsg
;
952 netmsg_init(&pmsg
->base
, NULL
, &netisr_apanic_rport
,
953 0, transport_processing_handler
);
955 pmsg
->base
.lmsg
.u
.ms_result
= hlen
;
956 lwkt_sendmsg(port
, &pmsg
->base
.lmsg
);
959 transport_processing_oncpu(m
, hlen
, ip
);
968 * Take incoming datagram fragment and try to reassemble it into
969 * whole datagram. If a chain for reassembly of this datagram already
970 * exists, then it is given as fp; otherwise have to make a chain.
973 ip_reass(struct mbuf
*m
)
975 struct ip
*ip
= mtod(m
, struct ip
*);
976 struct mbuf
*p
= NULL
, *q
, *nq
;
978 struct ipq
*fp
= NULL
;
979 struct ipqhead
*head
;
980 int hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
984 /* If maxnipq is 0, never accept fragments. */
986 ipstat
.ips_fragments
++;
987 ipstat
.ips_fragdropped
++;
992 sum
= IPREASS_HASH(ip
->ip_src
.s_addr
, ip
->ip_id
);
994 * Look for queue of fragments of this datagram.
996 lwkt_gettoken(&ipq_token
);
998 TAILQ_FOREACH(fp
, head
, ipq_list
) {
999 if (ip
->ip_id
== fp
->ipq_id
&&
1000 ip
->ip_src
.s_addr
== fp
->ipq_src
.s_addr
&&
1001 ip
->ip_dst
.s_addr
== fp
->ipq_dst
.s_addr
&&
1002 ip
->ip_p
== fp
->ipq_p
)
1009 * Enforce upper bound on number of fragmented packets
1010 * for which we attempt reassembly;
1011 * If maxnipq is -1, accept all fragments without limitation.
1013 if (nipq
> maxnipq
&& maxnipq
> 0) {
1015 * drop something from the tail of the current queue
1016 * before proceeding further
1018 struct ipq
*q
= TAILQ_LAST(head
, ipqhead
);
1021 * The current queue is empty,
1022 * so drop from one of the others.
1024 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1025 struct ipq
*r
= TAILQ_LAST(&ipq
[i
], ipqhead
);
1027 ipstat
.ips_fragtimeout
+= r
->ipq_nfrags
;
1028 ip_freef(&ipq
[i
], r
);
1033 ipstat
.ips_fragtimeout
+= q
->ipq_nfrags
;
1039 * Adjust ip_len to not reflect header,
1040 * convert offset of this to bytes.
1043 if (ip
->ip_off
& IP_MF
) {
1045 * Make sure that fragments have a data length
1046 * that's a non-zero multiple of 8 bytes.
1048 if (ip
->ip_len
== 0 || (ip
->ip_len
& 0x7) != 0) {
1049 ipstat
.ips_toosmall
++; /* XXX */
1053 m
->m_flags
|= M_FRAG
;
1055 m
->m_flags
&= ~M_FRAG
;
1059 ipstat
.ips_fragments
++;
1060 m
->m_pkthdr
.header
= ip
;
1063 * If the hardware has not done csum over this fragment
1064 * then csum_data is not valid at all.
1066 if ((m
->m_pkthdr
.csum_flags
& (CSUM_FRAG_NOT_CHECKED
| CSUM_DATA_VALID
))
1067 == (CSUM_FRAG_NOT_CHECKED
| CSUM_DATA_VALID
)) {
1068 m
->m_pkthdr
.csum_data
= 0;
1069 m
->m_pkthdr
.csum_flags
&= ~(CSUM_DATA_VALID
| CSUM_PSEUDO_HDR
);
1073 * Presence of header sizes in mbufs
1074 * would confuse code below.
1080 * If first fragment to arrive, create a reassembly queue.
1083 if ((fp
= mpipe_alloc_nowait(&ipq_mpipe
)) == NULL
)
1085 TAILQ_INSERT_HEAD(head
, fp
, ipq_list
);
1088 fp
->ipq_ttl
= IPFRAGTTL
;
1089 fp
->ipq_p
= ip
->ip_p
;
1090 fp
->ipq_id
= ip
->ip_id
;
1091 fp
->ipq_src
= ip
->ip_src
;
1092 fp
->ipq_dst
= ip
->ip_dst
;
1094 m
->m_nextpkt
= NULL
;
1100 #define GETIP(m) ((struct ip*)((m)->m_pkthdr.header))
1103 * Find a segment which begins after this one does.
1105 for (p
= NULL
, q
= fp
->ipq_frags
; q
; p
= q
, q
= q
->m_nextpkt
) {
1106 if (GETIP(q
)->ip_off
> ip
->ip_off
)
1111 * If there is a preceding segment, it may provide some of
1112 * our data already. If so, drop the data from the incoming
1113 * segment. If it provides all of our data, drop us, otherwise
1114 * stick new segment in the proper place.
1116 * If some of the data is dropped from the the preceding
1117 * segment, then it's checksum is invalidated.
1120 i
= GETIP(p
)->ip_off
+ GETIP(p
)->ip_len
- ip
->ip_off
;
1122 if (i
>= ip
->ip_len
)
1125 m
->m_pkthdr
.csum_flags
= 0;
1129 m
->m_nextpkt
= p
->m_nextpkt
;
1132 m
->m_nextpkt
= fp
->ipq_frags
;
1137 * While we overlap succeeding segments trim them or,
1138 * if they are completely covered, dequeue them.
1140 for (; q
!= NULL
&& ip
->ip_off
+ ip
->ip_len
> GETIP(q
)->ip_off
;
1142 i
= (ip
->ip_off
+ ip
->ip_len
) - GETIP(q
)->ip_off
;
1143 if (i
< GETIP(q
)->ip_len
) {
1144 GETIP(q
)->ip_len
-= i
;
1145 GETIP(q
)->ip_off
+= i
;
1147 q
->m_pkthdr
.csum_flags
= 0;
1152 ipstat
.ips_fragdropped
++;
1154 q
->m_nextpkt
= NULL
;
1160 * Check for complete reassembly and perform frag per packet
1163 * Frag limiting is performed here so that the nth frag has
1164 * a chance to complete the packet before we drop the packet.
1165 * As a result, n+1 frags are actually allowed per packet, but
1166 * only n will ever be stored. (n = maxfragsperpacket.)
1170 for (p
= NULL
, q
= fp
->ipq_frags
; q
; p
= q
, q
= q
->m_nextpkt
) {
1171 if (GETIP(q
)->ip_off
!= next
) {
1172 if (fp
->ipq_nfrags
> maxfragsperpacket
) {
1173 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1178 next
+= GETIP(q
)->ip_len
;
1180 /* Make sure the last packet didn't have the IP_MF flag */
1181 if (p
->m_flags
& M_FRAG
) {
1182 if (fp
->ipq_nfrags
> maxfragsperpacket
) {
1183 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1190 * Reassembly is complete. Make sure the packet is a sane size.
1194 if (next
+ (IP_VHL_HL(ip
->ip_vhl
) << 2) > IP_MAXPACKET
) {
1195 ipstat
.ips_toolong
++;
1196 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1202 * Concatenate fragments.
1209 q
->m_nextpkt
= NULL
;
1210 for (q
= nq
; q
!= NULL
; q
= nq
) {
1212 q
->m_nextpkt
= NULL
;
1213 m
->m_pkthdr
.csum_flags
&= q
->m_pkthdr
.csum_flags
;
1214 m
->m_pkthdr
.csum_data
+= q
->m_pkthdr
.csum_data
;
1219 * Clean up the 1's complement checksum. Carry over 16 bits must
1220 * be added back. This assumes no more then 65535 packet fragments
1221 * were reassembled. A second carry can also occur (but not a third).
1223 m
->m_pkthdr
.csum_data
= (m
->m_pkthdr
.csum_data
& 0xffff) +
1224 (m
->m_pkthdr
.csum_data
>> 16);
1225 if (m
->m_pkthdr
.csum_data
> 0xFFFF)
1226 m
->m_pkthdr
.csum_data
-= 0xFFFF;
1229 * Create header for new ip packet by
1230 * modifying header of first packet;
1231 * dequeue and discard fragment reassembly header.
1232 * Make header visible.
1235 ip
->ip_src
= fp
->ipq_src
;
1236 ip
->ip_dst
= fp
->ipq_dst
;
1237 TAILQ_REMOVE(head
, fp
, ipq_list
);
1239 mpipe_free(&ipq_mpipe
, fp
);
1240 m
->m_len
+= (IP_VHL_HL(ip
->ip_vhl
) << 2);
1241 m
->m_data
-= (IP_VHL_HL(ip
->ip_vhl
) << 2);
1242 /* some debugging cruft by sklower, below, will go away soon */
1243 if (m
->m_flags
& M_PKTHDR
) { /* XXX this should be done elsewhere */
1246 for (n
= m
; n
; n
= n
->m_next
)
1248 m
->m_pkthdr
.len
= plen
;
1252 * Reassembly complete, return the next protocol.
1254 * Be sure to clear M_HASH to force the packet
1255 * to be re-characterized.
1257 * Clear M_FRAG, we are no longer a fragment.
1259 m
->m_flags
&= ~(M_HASH
| M_FRAG
);
1261 ipstat
.ips_reassembled
++;
1262 lwkt_reltoken(&ipq_token
);
1266 ipstat
.ips_fragdropped
++;
1271 lwkt_reltoken(&ipq_token
);
1278 * Free a fragment reassembly header and all
1279 * associated datagrams.
1281 * Called with ipq_token held.
1284 ip_freef(struct ipqhead
*fhp
, struct ipq
*fp
)
1289 * Remove first to protect against blocking
1291 TAILQ_REMOVE(fhp
, fp
, ipq_list
);
1294 * Clean out at our leisure
1296 while (fp
->ipq_frags
) {
1298 fp
->ipq_frags
= q
->m_nextpkt
;
1299 q
->m_nextpkt
= NULL
;
1302 mpipe_free(&ipq_mpipe
, fp
);
1307 * IP timer processing;
1308 * if a timer expires on a reassembly
1309 * queue, discard it.
1314 struct ipq
*fp
, *fp_temp
;
1315 struct ipqhead
*head
;
1318 lwkt_gettoken(&ipq_token
);
1319 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1321 TAILQ_FOREACH_MUTABLE(fp
, head
, ipq_list
, fp_temp
) {
1322 if (--fp
->ipq_ttl
== 0) {
1323 ipstat
.ips_fragtimeout
+= fp
->ipq_nfrags
;
1329 * If we are over the maximum number of fragments
1330 * (due to the limit being lowered), drain off
1331 * enough to get down to the new limit.
1333 if (maxnipq
>= 0 && nipq
> maxnipq
) {
1334 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1336 while (nipq
> maxnipq
&& !TAILQ_EMPTY(head
)) {
1337 ipstat
.ips_fragdropped
+=
1338 TAILQ_FIRST(head
)->ipq_nfrags
;
1339 ip_freef(head
, TAILQ_FIRST(head
));
1343 lwkt_reltoken(&ipq_token
);
1348 * Drain off all datagram fragments.
1353 struct ipqhead
*head
;
1356 lwkt_gettoken(&ipq_token
);
1357 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1359 while (!TAILQ_EMPTY(head
)) {
1360 ipstat
.ips_fragdropped
+= TAILQ_FIRST(head
)->ipq_nfrags
;
1361 ip_freef(head
, TAILQ_FIRST(head
));
1364 lwkt_reltoken(&ipq_token
);
1369 * Do option processing on a datagram,
1370 * possibly discarding it if bad options are encountered,
1371 * or forwarding it if source-routed.
1372 * The pass argument is used when operating in the IPSTEALTH
1373 * mode to tell what options to process:
1374 * [LS]SRR (pass 0) or the others (pass 1).
1375 * The reason for as many as two passes is that when doing IPSTEALTH,
1376 * non-routing options should be processed only if the packet is for us.
1377 * Returns 1 if packet has been forwarded/freed,
1378 * 0 if the packet should be processed further.
1381 ip_dooptions(struct mbuf
*m
, int pass
, struct sockaddr_in
*next_hop
)
1383 struct sockaddr_in ipaddr
= { sizeof ipaddr
, AF_INET
};
1384 struct ip
*ip
= mtod(m
, struct ip
*);
1386 struct in_ifaddr
*ia
;
1387 int opt
, optlen
, cnt
, off
, code
, type
= ICMP_PARAMPROB
;
1388 boolean_t forward
= FALSE
;
1389 struct in_addr
*sin
, dst
;
1393 cp
= (u_char
*)(ip
+ 1);
1394 cnt
= (IP_VHL_HL(ip
->ip_vhl
) << 2) - sizeof(struct ip
);
1395 for (; cnt
> 0; cnt
-= optlen
, cp
+= optlen
) {
1396 opt
= cp
[IPOPT_OPTVAL
];
1397 if (opt
== IPOPT_EOL
)
1399 if (opt
== IPOPT_NOP
)
1402 if (cnt
< IPOPT_OLEN
+ sizeof(*cp
)) {
1403 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1406 optlen
= cp
[IPOPT_OLEN
];
1407 if (optlen
< IPOPT_OLEN
+ sizeof(*cp
) || optlen
> cnt
) {
1408 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1418 * Source routing with record.
1419 * Find interface with current destination address.
1420 * If none on this machine then drop if strictly routed,
1421 * or do nothing if loosely routed.
1422 * Record interface address and bring up next address
1423 * component. If strictly routed make sure next
1424 * address is on directly accessible net.
1428 if (ipstealth
&& pass
> 0)
1430 if (optlen
< IPOPT_OFFSET
+ sizeof(*cp
)) {
1431 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1434 if ((off
= cp
[IPOPT_OFFSET
]) < IPOPT_MINOFF
) {
1435 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1438 ipaddr
.sin_addr
= ip
->ip_dst
;
1439 ia
= (struct in_ifaddr
*)
1440 ifa_ifwithaddr((struct sockaddr
*)&ipaddr
);
1442 if (opt
== IPOPT_SSRR
) {
1443 type
= ICMP_UNREACH
;
1444 code
= ICMP_UNREACH_SRCFAIL
;
1447 if (!ip_dosourceroute
)
1448 goto nosourcerouting
;
1450 * Loose routing, and not at next destination
1451 * yet; nothing to do except forward.
1455 off
--; /* 0 origin */
1456 if (off
> optlen
- (int)sizeof(struct in_addr
)) {
1458 * End of source route. Should be for us.
1460 if (!ip_acceptsourceroute
)
1461 goto nosourcerouting
;
1462 save_rte(m
, cp
, ip
->ip_src
);
1467 if (!ip_dosourceroute
) {
1469 char buf
[sizeof "aaa.bbb.ccc.ddd"];
1472 * Acting as a router, so generate ICMP
1475 strcpy(buf
, inet_ntoa(ip
->ip_dst
));
1477 "attempted source route from %s to %s\n",
1478 inet_ntoa(ip
->ip_src
), buf
);
1479 type
= ICMP_UNREACH
;
1480 code
= ICMP_UNREACH_SRCFAIL
;
1484 * Not acting as a router,
1488 ipstat
.ips_cantforward
++;
1495 * locate outgoing interface
1497 memcpy(&ipaddr
.sin_addr
, cp
+ off
,
1498 sizeof ipaddr
.sin_addr
);
1500 if (opt
== IPOPT_SSRR
) {
1501 #define INA struct in_ifaddr *
1502 #define SA struct sockaddr *
1503 if ((ia
= (INA
)ifa_ifwithdstaddr((SA
)&ipaddr
))
1505 ia
= (INA
)ifa_ifwithnet((SA
)&ipaddr
);
1507 ia
= ip_rtaddr(ipaddr
.sin_addr
, NULL
);
1510 type
= ICMP_UNREACH
;
1511 code
= ICMP_UNREACH_SRCFAIL
;
1514 ip
->ip_dst
= ipaddr
.sin_addr
;
1515 memcpy(cp
+ off
, &IA_SIN(ia
)->sin_addr
,
1516 sizeof(struct in_addr
));
1517 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1519 * Let ip_intr's mcast routing check handle mcast pkts
1521 forward
= !IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
));
1525 if (ipstealth
&& pass
== 0)
1527 if (optlen
< IPOPT_OFFSET
+ sizeof(*cp
)) {
1528 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1531 if ((off
= cp
[IPOPT_OFFSET
]) < IPOPT_MINOFF
) {
1532 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1536 * If no space remains, ignore.
1538 off
--; /* 0 origin */
1539 if (off
> optlen
- (int)sizeof(struct in_addr
))
1541 memcpy(&ipaddr
.sin_addr
, &ip
->ip_dst
,
1542 sizeof ipaddr
.sin_addr
);
1544 * locate outgoing interface; if we're the destination,
1545 * use the incoming interface (should be same).
1547 if ((ia
= (INA
)ifa_ifwithaddr((SA
)&ipaddr
)) == NULL
&&
1548 (ia
= ip_rtaddr(ipaddr
.sin_addr
, NULL
)) == NULL
) {
1549 type
= ICMP_UNREACH
;
1550 code
= ICMP_UNREACH_HOST
;
1553 memcpy(cp
+ off
, &IA_SIN(ia
)->sin_addr
,
1554 sizeof(struct in_addr
));
1555 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1559 if (ipstealth
&& pass
== 0)
1561 code
= cp
- (u_char
*)ip
;
1562 if (optlen
< 4 || optlen
> 40) {
1563 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1566 if ((off
= cp
[IPOPT_OFFSET
]) < 5) {
1567 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1570 if (off
> optlen
- (int)sizeof(int32_t)) {
1571 cp
[IPOPT_OFFSET
+ 1] += (1 << 4);
1572 if ((cp
[IPOPT_OFFSET
+ 1] & 0xf0) == 0) {
1573 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1578 off
--; /* 0 origin */
1579 sin
= (struct in_addr
*)(cp
+ off
);
1580 switch (cp
[IPOPT_OFFSET
+ 1] & 0x0f) {
1582 case IPOPT_TS_TSONLY
:
1585 case IPOPT_TS_TSANDADDR
:
1586 if (off
+ sizeof(n_time
) +
1587 sizeof(struct in_addr
) > optlen
) {
1588 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1591 ipaddr
.sin_addr
= dst
;
1592 ia
= (INA
)ifaof_ifpforaddr((SA
)&ipaddr
,
1596 memcpy(sin
, &IA_SIN(ia
)->sin_addr
,
1597 sizeof(struct in_addr
));
1598 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1599 off
+= sizeof(struct in_addr
);
1602 case IPOPT_TS_PRESPEC
:
1603 if (off
+ sizeof(n_time
) +
1604 sizeof(struct in_addr
) > optlen
) {
1605 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1608 memcpy(&ipaddr
.sin_addr
, sin
,
1609 sizeof(struct in_addr
));
1610 if (ifa_ifwithaddr((SA
)&ipaddr
) == NULL
)
1612 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1613 off
+= sizeof(struct in_addr
);
1617 code
= &cp
[IPOPT_OFFSET
+ 1] - (u_char
*)ip
;
1621 memcpy(cp
+ off
, &ntime
, sizeof(n_time
));
1622 cp
[IPOPT_OFFSET
] += sizeof(n_time
);
1625 if (forward
&& ipforwarding
) {
1626 ip_forward(m
, TRUE
, next_hop
);
1631 icmp_error(m
, type
, code
, 0, 0);
1632 ipstat
.ips_badoptions
++;
1637 * Given address of next destination (final or next hop),
1638 * return internet address info of interface to be used to get there.
1641 ip_rtaddr(struct in_addr dst
, struct route
*ro0
)
1643 struct route sro
, *ro
;
1644 struct sockaddr_in
*sin
;
1645 struct in_ifaddr
*ia
;
1650 bzero(&sro
, sizeof(sro
));
1654 sin
= (struct sockaddr_in
*)&ro
->ro_dst
;
1656 if (ro
->ro_rt
== NULL
|| dst
.s_addr
!= sin
->sin_addr
.s_addr
) {
1657 if (ro
->ro_rt
!= NULL
) {
1661 sin
->sin_family
= AF_INET
;
1662 sin
->sin_len
= sizeof *sin
;
1663 sin
->sin_addr
= dst
;
1664 rtalloc_ign(ro
, RTF_PRCLONING
);
1667 if (ro
->ro_rt
== NULL
)
1670 ia
= ifatoia(ro
->ro_rt
->rt_ifa
);
1678 * Save incoming source route for use in replies,
1679 * to be picked up later by ip_srcroute if the receiver is interested.
1682 save_rte(struct mbuf
*m
, u_char
*option
, struct in_addr dst
)
1685 struct ip_srcrt_opt
*opt
;
1688 mtag
= m_tag_get(PACKET_TAG_IPSRCRT
, sizeof(*opt
), MB_DONTWAIT
);
1691 opt
= m_tag_data(mtag
);
1693 olen
= option
[IPOPT_OLEN
];
1696 kprintf("save_rte: olen %d\n", olen
);
1698 if (olen
> sizeof(opt
->ip_srcrt
) - (1 + sizeof(dst
))) {
1702 bcopy(option
, opt
->ip_srcrt
.srcopt
, olen
);
1703 opt
->ip_nhops
= (olen
- IPOPT_OFFSET
- 1) / sizeof(struct in_addr
);
1704 opt
->ip_srcrt
.dst
= dst
;
1705 m_tag_prepend(m
, mtag
);
1709 * Retrieve incoming source route for use in replies,
1710 * in the same form used by setsockopt.
1711 * The first hop is placed before the options, will be removed later.
1714 ip_srcroute(struct mbuf
*m0
)
1716 struct in_addr
*p
, *q
;
1719 struct ip_srcrt_opt
*opt
;
1724 mtag
= m_tag_find(m0
, PACKET_TAG_IPSRCRT
, NULL
);
1727 opt
= m_tag_data(mtag
);
1729 if (opt
->ip_nhops
== 0)
1731 m
= m_get(MB_DONTWAIT
, MT_HEADER
);
1735 #define OPTSIZ (sizeof(opt->ip_srcrt.nop) + sizeof(opt->ip_srcrt.srcopt))
1737 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1738 m
->m_len
= opt
->ip_nhops
* sizeof(struct in_addr
) +
1739 sizeof(struct in_addr
) + OPTSIZ
;
1742 kprintf("ip_srcroute: nhops %d mlen %d",
1743 opt
->ip_nhops
, m
->m_len
);
1748 * First save first hop for return route
1750 p
= &opt
->ip_srcrt
.route
[opt
->ip_nhops
- 1];
1751 *(mtod(m
, struct in_addr
*)) = *p
--;
1754 kprintf(" hops %x", ntohl(mtod(m
, struct in_addr
*)->s_addr
));
1758 * Copy option fields and padding (nop) to mbuf.
1760 opt
->ip_srcrt
.nop
= IPOPT_NOP
;
1761 opt
->ip_srcrt
.srcopt
[IPOPT_OFFSET
] = IPOPT_MINOFF
;
1762 memcpy(mtod(m
, caddr_t
) + sizeof(struct in_addr
), &opt
->ip_srcrt
.nop
,
1764 q
= (struct in_addr
*)(mtod(m
, caddr_t
) +
1765 sizeof(struct in_addr
) + OPTSIZ
);
1768 * Record return path as an IP source route,
1769 * reversing the path (pointers are now aligned).
1771 while (p
>= opt
->ip_srcrt
.route
) {
1774 kprintf(" %x", ntohl(q
->s_addr
));
1779 * Last hop goes to final destination.
1781 *q
= opt
->ip_srcrt
.dst
;
1782 m_tag_delete(m0
, mtag
);
1785 kprintf(" %x\n", ntohl(q
->s_addr
));
1791 * Strip out IP options.
1794 ip_stripoptions(struct mbuf
*m
)
1797 struct ip
*ip
= mtod(m
, struct ip
*);
1801 optlen
= (IP_VHL_HL(ip
->ip_vhl
) << 2) - sizeof(struct ip
);
1802 opts
= (caddr_t
)(ip
+ 1);
1803 datalen
= m
->m_len
- (sizeof(struct ip
) + optlen
);
1804 bcopy(opts
+ optlen
, opts
, datalen
);
1806 if (m
->m_flags
& M_PKTHDR
)
1807 m
->m_pkthdr
.len
-= optlen
;
1808 ip
->ip_vhl
= IP_MAKE_VHL(IPVERSION
, sizeof(struct ip
) >> 2);
1811 u_char inetctlerrmap
[PRC_NCMDS
] = {
1813 0, EMSGSIZE
, EHOSTDOWN
, EHOSTUNREACH
,
1814 EHOSTUNREACH
, EHOSTUNREACH
, ECONNREFUSED
, ECONNREFUSED
,
1815 EMSGSIZE
, EHOSTUNREACH
, 0, 0,
1817 ENOPROTOOPT
, ECONNREFUSED
1821 * Forward a packet. If some error occurs return the sender
1822 * an icmp packet. Note we can't always generate a meaningful
1823 * icmp message because icmp doesn't have a large enough repertoire
1824 * of codes and types.
1826 * If not forwarding, just drop the packet. This could be confusing
1827 * if ipforwarding was zero but some routing protocol was advancing
1828 * us as a gateway to somewhere. However, we must let the routing
1829 * protocol deal with that.
1831 * The using_srcrt parameter indicates whether the packet is being forwarded
1832 * via a source route.
1835 ip_forward(struct mbuf
*m
, boolean_t using_srcrt
, struct sockaddr_in
*next_hop
)
1837 struct ip
*ip
= mtod(m
, struct ip
*);
1839 struct route fwd_ro
;
1840 int error
, type
= 0, code
= 0, destmtu
= 0;
1843 struct in_addr pkt_dst
;
1847 * Cache the destination address of the packet; this may be
1848 * changed by use of 'ipfw fwd'.
1850 pkt_dst
= (next_hop
!= NULL
) ? next_hop
->sin_addr
: ip
->ip_dst
;
1854 kprintf("forward: src %x dst %x ttl %x\n",
1855 ip
->ip_src
.s_addr
, pkt_dst
.s_addr
, ip
->ip_ttl
);
1858 if (m
->m_flags
& (M_BCAST
| M_MCAST
) || !in_canforward(pkt_dst
)) {
1859 ipstat
.ips_cantforward
++;
1863 if (!ipstealth
&& ip
->ip_ttl
<= IPTTLDEC
) {
1864 icmp_error(m
, ICMP_TIMXCEED
, ICMP_TIMXCEED_INTRANS
, dest
, 0);
1868 bzero(&fwd_ro
, sizeof(fwd_ro
));
1869 ip_rtaddr(pkt_dst
, &fwd_ro
);
1870 if (fwd_ro
.ro_rt
== NULL
) {
1871 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_HOST
, dest
, 0);
1877 * Save the IP header and at most 8 bytes of the payload,
1878 * in case we need to generate an ICMP message to the src.
1880 * XXX this can be optimized a lot by saving the data in a local
1881 * buffer on the stack (72 bytes at most), and only allocating the
1882 * mbuf if really necessary. The vast majority of the packets
1883 * are forwarded without having to send an ICMP back (either
1884 * because unnecessary, or because rate limited), so we are
1885 * really we are wasting a lot of work here.
1887 * We don't use m_copy() because it might return a reference
1888 * to a shared cluster. Both this function and ip_output()
1889 * assume exclusive access to the IP header in `m', so any
1890 * data in a cluster may change before we reach icmp_error().
1892 MGETHDR(mcopy
, MB_DONTWAIT
, m
->m_type
);
1893 if (mcopy
!= NULL
&& !m_dup_pkthdr(mcopy
, m
, MB_DONTWAIT
)) {
1895 * It's probably ok if the pkthdr dup fails (because
1896 * the deep copy of the tag chain failed), but for now
1897 * be conservative and just discard the copy since
1898 * code below may some day want the tags.
1903 if (mcopy
!= NULL
) {
1904 mcopy
->m_len
= imin((IP_VHL_HL(ip
->ip_vhl
) << 2) + 8,
1906 mcopy
->m_pkthdr
.len
= mcopy
->m_len
;
1907 m_copydata(m
, 0, mcopy
->m_len
, mtod(mcopy
, caddr_t
));
1911 ip
->ip_ttl
-= IPTTLDEC
;
1914 * If forwarding packet using same interface that it came in on,
1915 * perhaps should send a redirect to sender to shortcut a hop.
1916 * Only send redirect if source is sending directly to us,
1917 * and if packet was not source routed (or has any options).
1918 * Also, don't send redirect if forwarding using a default route
1919 * or a route modified by a redirect.
1921 if (rt
->rt_ifp
== m
->m_pkthdr
.rcvif
&&
1922 !(rt
->rt_flags
& (RTF_DYNAMIC
| RTF_MODIFIED
)) &&
1923 satosin(rt_key(rt
))->sin_addr
.s_addr
!= INADDR_ANY
&&
1924 ipsendredirects
&& !using_srcrt
&& next_hop
== NULL
) {
1925 u_long src
= ntohl(ip
->ip_src
.s_addr
);
1926 struct in_ifaddr
*rt_ifa
= (struct in_ifaddr
*)rt
->rt_ifa
;
1928 if (rt_ifa
!= NULL
&&
1929 (src
& rt_ifa
->ia_subnetmask
) == rt_ifa
->ia_subnet
) {
1930 if (rt
->rt_flags
& RTF_GATEWAY
)
1931 dest
= satosin(rt
->rt_gateway
)->sin_addr
.s_addr
;
1933 dest
= pkt_dst
.s_addr
;
1935 * Router requirements says to only send
1938 type
= ICMP_REDIRECT
;
1939 code
= ICMP_REDIRECT_HOST
;
1942 kprintf("redirect (%d) to %x\n", code
, dest
);
1947 error
= ip_output(m
, NULL
, &fwd_ro
, IP_FORWARDING
, NULL
, NULL
);
1949 ipstat
.ips_forward
++;
1952 ipflow_create(&fwd_ro
, mcopy
);
1957 ipstat
.ips_redirectsent
++;
1960 ipstat
.ips_cantforward
++;
1967 * Send ICMP message.
1972 case 0: /* forwarded, but need redirect */
1973 /* type, code set above */
1976 case ENETUNREACH
: /* shouldn't happen, checked above */
1981 type
= ICMP_UNREACH
;
1982 code
= ICMP_UNREACH_HOST
;
1986 type
= ICMP_UNREACH
;
1987 code
= ICMP_UNREACH_NEEDFRAG
;
1990 * If the packet is routed over IPsec tunnel, tell the
1991 * originator the tunnel MTU.
1992 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
1995 if (fwd_ro
.ro_rt
!= NULL
) {
1996 struct secpolicy
*sp
= NULL
;
2001 sp
= ipsec4_getpolicybyaddr(mcopy
,
2007 destmtu
= fwd_ro
.ro_rt
->rt_ifp
->if_mtu
;
2009 /* count IPsec header size */
2010 ipsechdr
= ipsec4_hdrsiz(mcopy
,
2015 * find the correct route for outer IPv4
2016 * header, compute tunnel MTU.
2019 if (sp
->req
!= NULL
&& sp
->req
->sav
!= NULL
&&
2020 sp
->req
->sav
->sah
!= NULL
) {
2021 ro
= &sp
->req
->sav
->sah
->sa_route
;
2022 if (ro
->ro_rt
!= NULL
&&
2023 ro
->ro_rt
->rt_ifp
!= NULL
) {
2025 ro
->ro_rt
->rt_ifp
->if_mtu
;
2026 destmtu
-= ipsechdr
;
2035 * If the packet is routed over IPsec tunnel, tell the
2036 * originator the tunnel MTU.
2037 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
2040 if (fwd_ro
.ro_rt
!= NULL
) {
2041 struct secpolicy
*sp
= NULL
;
2046 sp
= ipsec_getpolicybyaddr(mcopy
,
2052 destmtu
= fwd_ro
.ro_rt
->rt_ifp
->if_mtu
;
2054 /* count IPsec header size */
2055 ipsechdr
= ipsec4_hdrsiz(mcopy
,
2060 * find the correct route for outer IPv4
2061 * header, compute tunnel MTU.
2064 if (sp
->req
!= NULL
&&
2065 sp
->req
->sav
!= NULL
&&
2066 sp
->req
->sav
->sah
!= NULL
) {
2067 ro
= &sp
->req
->sav
->sah
->sa_route
;
2068 if (ro
->ro_rt
!= NULL
&&
2069 ro
->ro_rt
->rt_ifp
!= NULL
) {
2071 ro
->ro_rt
->rt_ifp
->if_mtu
;
2072 destmtu
-= ipsechdr
;
2079 #else /* !IPSEC && !FAST_IPSEC */
2080 if (fwd_ro
.ro_rt
!= NULL
)
2081 destmtu
= fwd_ro
.ro_rt
->rt_ifp
->if_mtu
;
2083 ipstat
.ips_cantfrag
++;
2088 * A router should not generate ICMP_SOURCEQUENCH as
2089 * required in RFC1812 Requirements for IP Version 4 Routers.
2090 * Source quench could be a big problem under DoS attacks,
2091 * or if the underlying interface is rate-limited.
2092 * Those who need source quench packets may re-enable them
2093 * via the net.inet.ip.sendsourcequench sysctl.
2095 if (!ip_sendsourcequench
) {
2099 type
= ICMP_SOURCEQUENCH
;
2104 case EACCES
: /* ipfw denied packet */
2108 icmp_error(mcopy
, type
, code
, dest
, destmtu
);
2110 if (fwd_ro
.ro_rt
!= NULL
)
2111 RTFREE(fwd_ro
.ro_rt
);
2115 ip_savecontrol(struct inpcb
*inp
, struct mbuf
**mp
, struct ip
*ip
,
2118 if (inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
2122 *mp
= sbcreatecontrol((caddr_t
) &tv
, sizeof(tv
),
2123 SCM_TIMESTAMP
, SOL_SOCKET
);
2125 mp
= &(*mp
)->m_next
;
2127 if (inp
->inp_flags
& INP_RECVDSTADDR
) {
2128 *mp
= sbcreatecontrol((caddr_t
) &ip
->ip_dst
,
2129 sizeof(struct in_addr
), IP_RECVDSTADDR
, IPPROTO_IP
);
2131 mp
= &(*mp
)->m_next
;
2133 if (inp
->inp_flags
& INP_RECVTTL
) {
2134 *mp
= sbcreatecontrol((caddr_t
) &ip
->ip_ttl
,
2135 sizeof(u_char
), IP_RECVTTL
, IPPROTO_IP
);
2137 mp
= &(*mp
)->m_next
;
2141 * Moving these out of udp_input() made them even more broken
2142 * than they already were.
2144 /* options were tossed already */
2145 if (inp
->inp_flags
& INP_RECVOPTS
) {
2146 *mp
= sbcreatecontrol((caddr_t
) opts_deleted_above
,
2147 sizeof(struct in_addr
), IP_RECVOPTS
, IPPROTO_IP
);
2149 mp
= &(*mp
)->m_next
;
2151 /* ip_srcroute doesn't do what we want here, need to fix */
2152 if (inp
->inp_flags
& INP_RECVRETOPTS
) {
2153 *mp
= sbcreatecontrol((caddr_t
) ip_srcroute(m
),
2154 sizeof(struct in_addr
), IP_RECVRETOPTS
, IPPROTO_IP
);
2156 mp
= &(*mp
)->m_next
;
2159 if (inp
->inp_flags
& INP_RECVIF
) {
2162 struct sockaddr_dl sdl
;
2165 struct sockaddr_dl
*sdp
;
2166 struct sockaddr_dl
*sdl2
= &sdlbuf
.sdl
;
2168 if (((ifp
= m
->m_pkthdr
.rcvif
)) &&
2169 ((ifp
->if_index
!= 0) && (ifp
->if_index
<= if_index
))) {
2170 sdp
= IF_LLSOCKADDR(ifp
);
2172 * Change our mind and don't try copy.
2174 if ((sdp
->sdl_family
!= AF_LINK
) ||
2175 (sdp
->sdl_len
> sizeof(sdlbuf
))) {
2178 bcopy(sdp
, sdl2
, sdp
->sdl_len
);
2182 offsetof(struct sockaddr_dl
, sdl_data
[0]);
2183 sdl2
->sdl_family
= AF_LINK
;
2184 sdl2
->sdl_index
= 0;
2185 sdl2
->sdl_nlen
= sdl2
->sdl_alen
= sdl2
->sdl_slen
= 0;
2187 *mp
= sbcreatecontrol((caddr_t
) sdl2
, sdl2
->sdl_len
,
2188 IP_RECVIF
, IPPROTO_IP
);
2190 mp
= &(*mp
)->m_next
;
2195 * XXX these routines are called from the upper part of the kernel.
2197 * They could also be moved to ip_mroute.c, since all the RSVP
2198 * handling is done there already.
2201 ip_rsvp_init(struct socket
*so
)
2203 if (so
->so_type
!= SOCK_RAW
||
2204 so
->so_proto
->pr_protocol
!= IPPROTO_RSVP
)
2207 if (ip_rsvpd
!= NULL
)
2212 * This may seem silly, but we need to be sure we don't over-increment
2213 * the RSVP counter, in case something slips up.
2228 * This may seem silly, but we need to be sure we don't over-decrement
2229 * the RSVP counter, in case something slips up.
2239 rsvp_input(struct mbuf
**mp
, int *offp
, int proto
)
2241 struct mbuf
*m
= *mp
;
2247 if (rsvp_input_p
) { /* call the real one if loaded */
2249 rsvp_input_p(mp
, offp
, proto
);
2250 return(IPPROTO_DONE
);
2253 /* Can still get packets with rsvp_on = 0 if there is a local member
2254 * of the group to which the RSVP packet is addressed. But in this
2255 * case we want to throw the packet away.
2260 return(IPPROTO_DONE
);
2263 if (ip_rsvpd
!= NULL
) {
2265 rip_input(mp
, offp
, proto
);
2266 return(IPPROTO_DONE
);
2268 /* Drop the packet */
2270 return(IPPROTO_DONE
);