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 1.71 2007/10/25 13:13:18 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>
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>
106 #include <net/intrq.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>
116 #include <sys/thread2.h>
117 #include <sys/msgport2.h>
118 #include <net/netmsg2.h>
120 #include <sys/socketvar.h>
122 #include <net/ipfw/ip_fw.h>
123 #include <net/dummynet/ip_dummynet.h>
126 #include <netinet6/ipsec.h>
127 #include <netproto/key/key.h>
131 #include <netproto/ipsec/ipsec.h>
132 #include <netproto/ipsec/key.h>
136 static int ip_rsvp_on
;
137 struct socket
*ip_rsvpd
;
139 int ipforwarding
= 0;
140 SYSCTL_INT(_net_inet_ip
, IPCTL_FORWARDING
, forwarding
, CTLFLAG_RW
,
141 &ipforwarding
, 0, "Enable IP forwarding between interfaces");
143 static int ipsendredirects
= 1; /* XXX */
144 SYSCTL_INT(_net_inet_ip
, IPCTL_SENDREDIRECTS
, redirect
, CTLFLAG_RW
,
145 &ipsendredirects
, 0, "Enable sending IP redirects");
147 int ip_defttl
= IPDEFTTL
;
148 SYSCTL_INT(_net_inet_ip
, IPCTL_DEFTTL
, ttl
, CTLFLAG_RW
,
149 &ip_defttl
, 0, "Maximum TTL on IP packets");
151 static int ip_dosourceroute
= 0;
152 SYSCTL_INT(_net_inet_ip
, IPCTL_SOURCEROUTE
, sourceroute
, CTLFLAG_RW
,
153 &ip_dosourceroute
, 0, "Enable forwarding source routed IP packets");
155 static int ip_acceptsourceroute
= 0;
156 SYSCTL_INT(_net_inet_ip
, IPCTL_ACCEPTSOURCEROUTE
, accept_sourceroute
,
157 CTLFLAG_RW
, &ip_acceptsourceroute
, 0,
158 "Enable accepting source routed IP packets");
160 static int ip_keepfaith
= 0;
161 SYSCTL_INT(_net_inet_ip
, IPCTL_KEEPFAITH
, keepfaith
, CTLFLAG_RW
,
163 "Enable packet capture for FAITH IPv4->IPv6 translater daemon");
165 static int nipq
= 0; /* total # of reass queues */
167 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, maxfragpackets
, CTLFLAG_RW
,
169 "Maximum number of IPv4 fragment reassembly queue entries");
171 static int maxfragsperpacket
;
172 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, maxfragsperpacket
, CTLFLAG_RW
,
173 &maxfragsperpacket
, 0,
174 "Maximum number of IPv4 fragments allowed per packet");
176 static int ip_sendsourcequench
= 0;
177 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, sendsourcequench
, CTLFLAG_RW
,
178 &ip_sendsourcequench
, 0,
179 "Enable the transmission of source quench packets");
181 int ip_do_randomid
= 0;
182 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, random_id
, CTLFLAG_RW
,
184 "Assign random ip_id values");
186 * XXX - Setting ip_checkinterface mostly implements the receive side of
187 * the Strong ES model described in RFC 1122, but since the routing table
188 * and transmit implementation do not implement the Strong ES model,
189 * setting this to 1 results in an odd hybrid.
191 * XXX - ip_checkinterface currently must be disabled if you use ipnat
192 * to translate the destination address to another local interface.
194 * XXX - ip_checkinterface must be disabled if you add IP aliases
195 * to the loopback interface instead of the interface where the
196 * packets for those addresses are received.
198 static int ip_checkinterface
= 0;
199 SYSCTL_INT(_net_inet_ip
, OID_AUTO
, check_interface
, CTLFLAG_RW
,
200 &ip_checkinterface
, 0, "Verify packet arrives on correct interface");
203 static int ipprintfs
= 0;
206 static struct ifqueue ipintrq
;
207 static int ipqmaxlen
= IFQ_MAXLEN
;
209 extern struct domain inetdomain
;
210 extern struct protosw inetsw
[];
211 u_char ip_protox
[IPPROTO_MAX
];
212 struct in_ifaddrhead in_ifaddrhead
; /* first inet address */
213 struct in_ifaddrhashhead
*in_ifaddrhashtbl
; /* inet addr hash table */
214 u_long in_ifaddrhmask
; /* mask for hash table */
216 SYSCTL_INT(_net_inet_ip
, IPCTL_INTRQMAXLEN
, intr_queue_maxlen
, CTLFLAG_RW
,
217 &ipintrq
.ifq_maxlen
, 0, "Maximum size of the IP input queue");
218 SYSCTL_INT(_net_inet_ip
, IPCTL_INTRQDROPS
, intr_queue_drops
, CTLFLAG_RD
,
219 &ipintrq
.ifq_drops
, 0, "Number of packets dropped from the IP input queue");
221 struct ip_stats ipstats_percpu
[MAXCPU
];
224 sysctl_ipstats(SYSCTL_HANDLER_ARGS
)
228 for (cpu
= 0; cpu
< ncpus
; ++cpu
) {
229 if ((error
= SYSCTL_OUT(req
, &ipstats_percpu
[cpu
],
230 sizeof(struct ip_stats
))))
232 if ((error
= SYSCTL_IN(req
, &ipstats_percpu
[cpu
],
233 sizeof(struct ip_stats
))))
239 SYSCTL_PROC(_net_inet_ip
, IPCTL_STATS
, stats
, (CTLTYPE_OPAQUE
| CTLFLAG_RW
),
240 0, 0, sysctl_ipstats
, "S,ip_stats", "IP statistics");
242 SYSCTL_STRUCT(_net_inet_ip
, IPCTL_STATS
, stats
, CTLFLAG_RW
,
243 &ipstat
, ip_stats
, "IP statistics");
246 /* Packet reassembly stuff */
247 #define IPREASS_NHASH_LOG2 6
248 #define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2)
249 #define IPREASS_HMASK (IPREASS_NHASH - 1)
250 #define IPREASS_HASH(x,y) \
251 (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK)
253 static struct ipq ipq
[IPREASS_NHASH
];
254 const int ipintrq_present
= 1;
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;
270 ip_fw_chk_t
*ip_fw_chk_ptr
;
275 ip_dn_io_t
*ip_dn_io_ptr
;
277 struct pfil_head inet_pfil_hook
;
280 * XXX this is ugly -- the following two global variables are
281 * used to store packet state while it travels through the stack.
282 * Note that the code even makes assumptions on the size and
283 * alignment of fields inside struct ip_srcrt so e.g. adding some
284 * fields will break the code. This needs to be fixed.
286 * We need to save the IP options in case a protocol wants to respond
287 * to an incoming packet over the same route if the packet got here
288 * using IP source routing. This allows connection establishment and
289 * maintenance when the remote end is on a network that is not known
292 static int ip_nhops
= 0;
294 static struct ip_srcrt
{
295 struct in_addr dst
; /* final destination */
296 char nop
; /* one NOP to align */
297 char srcopt
[IPOPT_OFFSET
+ 1]; /* OPTVAL, OLEN and OFFSET */
298 struct in_addr route
[MAX_IPOPTLEN
/sizeof(struct in_addr
)];
301 static MALLOC_DEFINE(M_IPQ
, "ipq", "IP Fragment Management");
302 static struct malloc_pipe ipq_mpipe
;
304 static void save_rte (u_char
*, struct in_addr
);
305 static int ip_dooptions (struct mbuf
*m
, int,
306 struct sockaddr_in
*next_hop
);
307 static void ip_forward (struct mbuf
*m
, boolean_t using_srcrt
,
308 struct sockaddr_in
*next_hop
);
309 static void ip_freef (struct ipq
*);
310 static void ip_input_handler (struct netmsg
*);
311 static struct mbuf
*ip_reass (struct mbuf
*, struct ipq
*,
312 struct ipq
*, u_int32_t
*);
315 * IP initialization: fill in IP protocol switch table.
316 * All protocols not implemented in kernel go to raw IP protocol handler.
328 * Make sure we can handle a reasonable number of fragments but
329 * cap it at 4000 (XXX).
331 mpipe_init(&ipq_mpipe
, M_IPQ
, sizeof(struct ipq
),
332 IFQ_MAXLEN
, 4000, 0, NULL
);
333 TAILQ_INIT(&in_ifaddrhead
);
334 in_ifaddrhashtbl
= hashinit(INADDR_NHASH
, M_IFADDR
, &in_ifaddrhmask
);
335 pr
= pffindproto(PF_INET
, IPPROTO_RAW
, SOCK_RAW
);
338 for (i
= 0; i
< IPPROTO_MAX
; i
++)
339 ip_protox
[i
] = pr
- inetsw
;
340 for (pr
= inetdomain
.dom_protosw
;
341 pr
< inetdomain
.dom_protoswNPROTOSW
; pr
++)
342 if (pr
->pr_domain
->dom_family
== PF_INET
&&
343 pr
->pr_protocol
&& pr
->pr_protocol
!= IPPROTO_RAW
)
344 ip_protox
[pr
->pr_protocol
] = pr
- inetsw
;
346 inet_pfil_hook
.ph_type
= PFIL_TYPE_AF
;
347 inet_pfil_hook
.ph_af
= AF_INET
;
348 if ((i
= pfil_head_register(&inet_pfil_hook
)) != 0) {
349 kprintf("%s: WARNING: unable to register pfil hook, "
350 "error %d\n", __func__
, i
);
353 for (i
= 0; i
< IPREASS_NHASH
; i
++)
354 ipq
[i
].next
= ipq
[i
].prev
= &ipq
[i
];
356 maxnipq
= nmbclusters
/ 32;
357 maxfragsperpacket
= 16;
359 ip_id
= time_second
& 0xffff;
360 ipintrq
.ifq_maxlen
= ipqmaxlen
;
363 * Initialize IP statistics counters for each CPU.
367 for (cpu
= 0; cpu
< ncpus
; ++cpu
) {
368 bzero(&ipstats_percpu
[cpu
], sizeof(struct ip_stats
));
371 bzero(&ipstat
, sizeof(struct ip_stats
));
374 netisr_register(NETISR_IP
, ip_mport
, ip_input_handler
);
378 * XXX watch out this one. It is perhaps used as a cache for
379 * the most recently used route ? it is cleared in in_addroute()
380 * when a new route is successfully created.
382 struct route ipforward_rt
[MAXCPU
];
384 /* Do transport protocol processing. */
386 transport_processing_oncpu(struct mbuf
*m
, int hlen
, struct ip
*ip
,
387 struct sockaddr_in
*nexthop
)
390 * Switch out to protocol's input routine.
392 if (nexthop
&& ip
->ip_p
== IPPROTO_TCP
) {
393 /* TCP needs IPFORWARD info if available */
396 tag
.mh_type
= MT_TAG
;
397 tag
.mh_flags
= PACKET_TAG_IPFORWARD
;
398 tag
.mh_data
= (caddr_t
)nexthop
;
401 (*inetsw
[ip_protox
[ip
->ip_p
]].pr_input
)
402 ((struct mbuf
*)&tag
, hlen
, ip
->ip_p
);
404 (*inetsw
[ip_protox
[ip
->ip_p
]].pr_input
)(m
, hlen
, ip
->ip_p
);
408 struct netmsg_transport_packet
{
409 struct netmsg nm_netmsg
;
410 struct mbuf
*nm_mbuf
;
412 boolean_t nm_hasnexthop
;
413 struct sockaddr_in nm_nexthop
;
417 transport_processing_handler(netmsg_t netmsg
)
419 struct netmsg_transport_packet
*msg
= (void *)netmsg
;
420 struct sockaddr_in
*nexthop
;
423 ip
= mtod(msg
->nm_mbuf
, struct ip
*);
424 nexthop
= msg
->nm_hasnexthop
? &msg
->nm_nexthop
: NULL
;
425 transport_processing_oncpu(msg
->nm_mbuf
, msg
->nm_hlen
, ip
, nexthop
);
426 lwkt_replymsg(&msg
->nm_netmsg
.nm_lmsg
, 0);
430 ip_input_handler(struct netmsg
*msg0
)
432 struct mbuf
*m
= ((struct netmsg_packet
*)msg0
)->nm_packet
;
435 /* msg0 was embedded in the mbuf, do not reply! */
439 * IP input routine. Checksum and byte swap header. If fragmented
440 * try to reassemble. Process options. Pass to next level.
443 ip_input(struct mbuf
*m
)
447 struct in_ifaddr
*ia
= NULL
;
449 int i
, hlen
, checkif
;
451 struct in_addr pkt_dst
;
452 u_int32_t divert_info
= 0; /* packet divert/tee info */
453 struct ip_fw_args args
;
454 boolean_t using_srcrt
= FALSE
; /* forward (by PFIL_HOOKS) */
455 boolean_t needredispatch
= FALSE
;
456 struct in_addr odst
; /* original dst address(NAT) */
459 struct tdb_ident
*tdbi
;
460 struct secpolicy
*sp
;
467 args
.next_hop
= NULL
;
469 /* Grab info from MT_TAG mbufs prepended to the chain. */
470 while (m
!= NULL
&& m
->m_type
== MT_TAG
) {
471 switch(m
->_m_tag_id
) {
472 case PACKET_TAG_IPFORWARD
:
473 args
.next_hop
= (struct sockaddr_in
*)m
->m_hdr
.mh_data
;
476 kprintf("ip_input: unrecognised MT_TAG tag %d\n",
482 KASSERT(m
!= NULL
&& (m
->m_flags
& M_PKTHDR
), ("ip_input: no HDR"));
484 /* Extract info from dummynet tag */
485 mtag
= m_tag_find(m
, PACKET_TAG_DUMMYNET
, NULL
);
487 args
.rule
= ((struct dn_pkt
*)m_tag_data(mtag
))->rule
;
489 m_tag_delete(m
, mtag
);
493 if (args
.rule
!= NULL
) { /* dummynet already filtered us */
494 ip
= mtod(m
, struct ip
*);
495 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
501 /* length checks already done in ip_demux() */
502 KASSERT(m
->m_len
>= sizeof(ip
), ("IP header not in one mbuf"));
504 ip
= mtod(m
, struct ip
*);
506 if (IP_VHL_V(ip
->ip_vhl
) != IPVERSION
) {
507 ipstat
.ips_badvers
++;
511 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
512 /* length checks already done in ip_demux() */
513 KASSERT(hlen
>= sizeof(struct ip
), ("IP header len too small"));
514 KASSERT(m
->m_len
>= hlen
, ("packet shorter than IP header length"));
516 /* 127/8 must not appear on wire - RFC1122 */
517 if ((ntohl(ip
->ip_dst
.s_addr
) >> IN_CLASSA_NSHIFT
) == IN_LOOPBACKNET
||
518 (ntohl(ip
->ip_src
.s_addr
) >> IN_CLASSA_NSHIFT
) == IN_LOOPBACKNET
) {
519 if (!(m
->m_pkthdr
.rcvif
->if_flags
& IFF_LOOPBACK
)) {
520 ipstat
.ips_badaddr
++;
525 if (m
->m_pkthdr
.csum_flags
& CSUM_IP_CHECKED
) {
526 sum
= !(m
->m_pkthdr
.csum_flags
& CSUM_IP_VALID
);
528 if (hlen
== sizeof(struct ip
)) {
529 sum
= in_cksum_hdr(ip
);
531 sum
= in_cksum(m
, hlen
);
540 if (altq_input
!= NULL
&& (*altq_input
)(m
, AF_INET
) == 0) {
541 /* packet is dropped by traffic conditioner */
546 * Convert fields to host representation.
548 ip
->ip_len
= ntohs(ip
->ip_len
);
549 if (ip
->ip_len
< hlen
) {
553 ip
->ip_off
= ntohs(ip
->ip_off
);
556 * Check that the amount of data in the buffers
557 * is as at least much as the IP header would have us expect.
558 * Trim mbufs if longer than we expect.
559 * Drop packet if shorter than we expect.
561 if (m
->m_pkthdr
.len
< ip
->ip_len
) {
562 ipstat
.ips_tooshort
++;
565 if (m
->m_pkthdr
.len
> ip
->ip_len
) {
566 if (m
->m_len
== m
->m_pkthdr
.len
) {
567 m
->m_len
= ip
->ip_len
;
568 m
->m_pkthdr
.len
= ip
->ip_len
;
570 m_adj(m
, ip
->ip_len
- m
->m_pkthdr
.len
);
572 #if defined(IPSEC) && !defined(IPSEC_FILTERGIF)
574 * Bypass packet filtering for packets from a tunnel (gif).
576 if (ipsec_gethist(m
, NULL
))
582 * Right now when no processing on packet has done
583 * and it is still fresh out of network we do our black
585 * - Firewall: deny/allow/divert
586 * - Xlate: translate packet's addr/port (NAT).
587 * - Pipe: pass pkt through dummynet.
588 * - Wrap: fake packet's addr/port <unimpl.>
589 * - Encapsulate: put it in another IP and send out. <unimp.>
595 * Run through list of hooks for input packets.
597 * NB: Beware of the destination address changing (e.g.
598 * by NAT rewriting). When this happens, tell
599 * ip_forward to do the right thing.
601 if (pfil_has_hooks(&inet_pfil_hook
)) {
603 if (pfil_run_hooks(&inet_pfil_hook
, &m
,
604 m
->m_pkthdr
.rcvif
, PFIL_IN
)) {
607 if (m
== NULL
) /* consumed by filter */
609 ip
= mtod(m
, struct ip
*);
610 using_srcrt
= (odst
.s_addr
!= ip
->ip_dst
.s_addr
);
613 if (fw_enable
&& IPFW_LOADED
) {
615 * If we've been forwarded from the output side, then
616 * skip the firewall a second time
618 if (args
.next_hop
!= NULL
)
622 i
= ip_fw_chk_ptr(&args
);
625 if ((i
& IP_FW_PORT_DENY_FLAG
) || m
== NULL
) { /* drop */
630 ip
= mtod(m
, struct ip
*); /* just in case m changed */
631 if (i
== 0 && args
.next_hop
== NULL
) /* common case */
633 if (DUMMYNET_LOADED
&& (i
& IP_FW_PORT_DYNT_FLAG
)) {
634 /* Send packet to the appropriate pipe */
635 ip_dn_io_ptr(m
, i
&0xffff, DN_TO_IP_IN
, &args
);
639 if (i
!= 0 && !(i
& IP_FW_PORT_DYNT_FLAG
)) {
640 /* Divert or tee packet */
645 if (i
== 0 && args
.next_hop
!= NULL
)
648 * if we get here, the packet must be dropped
656 * Process options and, if not destined for us,
657 * ship it on. ip_dooptions returns 1 when an
658 * error was detected (causing an icmp message
659 * to be sent and the original packet to be freed).
661 ip_nhops
= 0; /* for source routed packets */
662 if (hlen
> sizeof(struct ip
) && ip_dooptions(m
, 0, args
.next_hop
))
665 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no
666 * matter if it is destined to another node, or whether it is
667 * a multicast one, RSVP wants it! and prevents it from being forwarded
668 * anywhere else. Also checks if the rsvp daemon is running before
669 * grabbing the packet.
671 if (rsvp_on
&& ip
->ip_p
== IPPROTO_RSVP
)
675 * Check our list of addresses, to see if the packet is for us.
676 * If we don't have any addresses, assume any unicast packet
677 * we receive might be for us (and let the upper layers deal
680 if (TAILQ_EMPTY(&in_ifaddrhead
) && !(m
->m_flags
& (M_MCAST
| M_BCAST
)))
684 * Cache the destination address of the packet; this may be
685 * changed by use of 'ipfw fwd'.
687 pkt_dst
= args
.next_hop
? args
.next_hop
->sin_addr
: ip
->ip_dst
;
690 * Enable a consistency check between the destination address
691 * and the arrival interface for a unicast packet (the RFC 1122
692 * strong ES model) if IP forwarding is disabled and the packet
693 * is not locally generated and the packet is not subject to
696 * XXX - Checking also should be disabled if the destination
697 * address is ipnat'ed to a different interface.
699 * XXX - Checking is incompatible with IP aliases added
700 * to the loopback interface instead of the interface where
701 * the packets are received.
703 checkif
= ip_checkinterface
&&
705 m
->m_pkthdr
.rcvif
!= NULL
&&
706 !(m
->m_pkthdr
.rcvif
->if_flags
& IFF_LOOPBACK
) &&
707 (args
.next_hop
== NULL
);
710 * Check for exact addresses in the hash bucket.
712 LIST_FOREACH(ia
, INADDR_HASH(pkt_dst
.s_addr
), ia_hash
) {
714 * If the address matches, verify that the packet
715 * arrived via the correct interface if checking is
718 if (IA_SIN(ia
)->sin_addr
.s_addr
== pkt_dst
.s_addr
&&
719 (!checkif
|| ia
->ia_ifp
== m
->m_pkthdr
.rcvif
))
723 * Check for broadcast addresses.
725 * Only accept broadcast packets that arrive via the matching
726 * interface. Reception of forwarded directed broadcasts would
727 * be handled via ip_forward() and ether_output() with the loopback
728 * into the stack for SIMPLEX interfaces handled by ether_output().
730 if (m
->m_pkthdr
.rcvif
->if_flags
& IFF_BROADCAST
) {
731 TAILQ_FOREACH(ifa
, &m
->m_pkthdr
.rcvif
->if_addrhead
, ifa_link
) {
732 if (ifa
->ifa_addr
== NULL
) /* shutdown/startup race */
734 if (ifa
->ifa_addr
->sa_family
!= AF_INET
)
737 if (satosin(&ia
->ia_broadaddr
)->sin_addr
.s_addr
==
740 if (ia
->ia_netbroadcast
.s_addr
== pkt_dst
.s_addr
)
743 if (IA_SIN(ia
)->sin_addr
.s_addr
== INADDR_ANY
)
748 if (IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
))) {
749 struct in_multi
*inm
;
751 if (ip_mrouter
!= NULL
) {
753 * If we are acting as a multicast router, all
754 * incoming multicast packets are passed to the
755 * kernel-level multicast forwarding function.
756 * The packet is returned (relatively) intact; if
757 * ip_mforward() returns a non-zero value, the packet
758 * must be discarded, else it may be accepted below.
760 if (ip_mforward
!= NULL
&&
761 ip_mforward(ip
, m
->m_pkthdr
.rcvif
, m
, NULL
) != 0) {
762 ipstat
.ips_cantforward
++;
768 * The process-level routing daemon needs to receive
769 * all multicast IGMP packets, whether or not this
770 * host belongs to their destination groups.
772 if (ip
->ip_p
== IPPROTO_IGMP
)
774 ipstat
.ips_forward
++;
777 * See if we belong to the destination multicast group on the
780 IN_LOOKUP_MULTI(ip
->ip_dst
, m
->m_pkthdr
.rcvif
, inm
);
782 ipstat
.ips_notmember
++;
788 if (ip
->ip_dst
.s_addr
== INADDR_BROADCAST
)
790 if (ip
->ip_dst
.s_addr
== INADDR_ANY
)
794 * FAITH(Firewall Aided Internet Translator)
796 if (m
->m_pkthdr
.rcvif
&& m
->m_pkthdr
.rcvif
->if_type
== IFT_FAITH
) {
798 if (ip
->ip_p
== IPPROTO_TCP
|| ip
->ip_p
== IPPROTO_ICMP
)
806 * Not for us; forward if possible and desirable.
809 ipstat
.ips_cantforward
++;
814 * Enforce inbound IPsec SPD.
816 if (ipsec4_in_reject(m
, NULL
)) {
817 ipsecstat
.in_polvio
++;
822 mtag
= m_tag_find(m
, PACKET_TAG_IPSEC_IN_DONE
, NULL
);
825 tdbi
= (struct tdb_ident
*)m_tag_data(mtag
);
826 sp
= ipsec_getpolicy(tdbi
, IPSEC_DIR_INBOUND
);
828 sp
= ipsec_getpolicybyaddr(m
, IPSEC_DIR_INBOUND
,
829 IP_FORWARDING
, &error
);
831 if (sp
== NULL
) { /* NB: can happen if error */
833 /*XXX error stat???*/
834 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
839 * Check security policy against packet attributes.
841 error
= ipsec_in_reject(sp
, m
);
845 ipstat
.ips_cantforward
++;
849 ip_forward(m
, using_srcrt
, args
.next_hop
);
856 * IPSTEALTH: Process non-routing options only
857 * if the packet is destined for us.
860 hlen
> sizeof(struct ip
) &&
861 ip_dooptions(m
, 1, args
.next_hop
))
864 /* Count the packet in the ip address stats */
866 ia
->ia_ifa
.if_ipackets
++;
867 ia
->ia_ifa
.if_ibytes
+= m
->m_pkthdr
.len
;
871 * If offset or IP_MF are set, must reassemble.
872 * Otherwise, nothing need be done.
873 * (We could look in the reassembly queue to see
874 * if the packet was previously fragmented,
875 * but it's not worth the time; just let them time out.)
877 if (ip
->ip_off
& (IP_MF
| IP_OFFMASK
)) {
879 /* If maxnipq is 0, never accept fragments. */
881 ipstat
.ips_fragments
++;
882 ipstat
.ips_fragdropped
++;
886 sum
= IPREASS_HASH(ip
->ip_src
.s_addr
, ip
->ip_id
);
888 * Look for queue of fragments
891 for (fp
= ipq
[sum
].next
; fp
!= &ipq
[sum
]; fp
= fp
->next
)
892 if (ip
->ip_id
== fp
->ipq_id
&&
893 ip
->ip_src
.s_addr
== fp
->ipq_src
.s_addr
&&
894 ip
->ip_dst
.s_addr
== fp
->ipq_dst
.s_addr
&&
895 ip
->ip_p
== fp
->ipq_p
)
901 * Enforce upper bound on number of fragmented packets
902 * for which we attempt reassembly;
903 * If maxnipq is -1, accept all fragments without limitation.
905 if ((nipq
> maxnipq
) && (maxnipq
> 0)) {
907 * drop something from the tail of the current queue
908 * before proceeding further
910 if (ipq
[sum
].prev
== &ipq
[sum
]) { /* gak */
911 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
912 if (ipq
[i
].prev
!= &ipq
[i
]) {
913 ipstat
.ips_fragtimeout
+=
914 ipq
[i
].prev
->ipq_nfrags
;
915 ip_freef(ipq
[i
].prev
);
920 ipstat
.ips_fragtimeout
+=
921 ipq
[sum
].prev
->ipq_nfrags
;
922 ip_freef(ipq
[sum
].prev
);
927 * Adjust ip_len to not reflect header,
928 * convert offset of this to bytes.
931 if (ip
->ip_off
& IP_MF
) {
933 * Make sure that fragments have a data length
934 * that's a non-zero multiple of 8 bytes.
936 if (ip
->ip_len
== 0 || (ip
->ip_len
& 0x7) != 0) {
937 ipstat
.ips_toosmall
++; /* XXX */
940 m
->m_flags
|= M_FRAG
;
942 m
->m_flags
&= ~M_FRAG
;
946 * Attempt reassembly; if it succeeds, proceed.
947 * ip_reass() will return a different mbuf, and update
948 * the divert info in divert_info.
950 ipstat
.ips_fragments
++;
951 m
->m_pkthdr
.header
= ip
;
952 m
= ip_reass(m
, fp
, &ipq
[sum
], &divert_info
);
955 ipstat
.ips_reassembled
++;
956 needredispatch
= TRUE
;
957 ip
= mtod(m
, struct ip
*);
958 /* Get the header length of the reassembled packet */
959 hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
961 /* Restore original checksum before diverting packet */
962 if (divert_info
!= 0) {
964 ip
->ip_len
= htons(ip
->ip_len
);
965 ip
->ip_off
= htons(ip
->ip_off
);
967 if (hlen
== sizeof(struct ip
))
968 ip
->ip_sum
= in_cksum_hdr(ip
);
970 ip
->ip_sum
= in_cksum(m
, hlen
);
971 ip
->ip_off
= ntohs(ip
->ip_off
);
972 ip
->ip_len
= ntohs(ip
->ip_len
);
982 * Divert or tee packet to the divert protocol if required.
984 if (divert_info
!= 0) {
985 struct mbuf
*clone
= NULL
;
987 /* Clone packet if we're doing a 'tee' */
988 if ((divert_info
& IP_FW_PORT_TEE_FLAG
) != 0)
989 clone
= m_dup(m
, MB_DONTWAIT
);
991 /* Restore packet header fields to original values */
993 ip
->ip_len
= htons(ip
->ip_len
);
994 ip
->ip_off
= htons(ip
->ip_off
);
996 /* Deliver packet to divert input routine */
997 divert_packet(m
, 1, divert_info
& 0xffff);
998 ipstat
.ips_delivered
++;
1000 /* If 'tee', continue with original packet */
1004 ip
= mtod(m
, struct ip
*);
1007 * Jump backwards to complete processing of the
1008 * packet. But first clear divert_info to avoid
1009 * entering this block again.
1010 * We do not need to clear args.divert_rule
1011 * or args.next_hop as they will not be used.
1013 * XXX Better safe than sorry, remove the DIVERT tag.
1015 mtag
= m_tag_find(m
, PACKET_TAG_IPFW_DIVERT
, NULL
);
1017 m_tag_delete(m
, mtag
);
1026 * enforce IPsec policy checking if we are seeing last header.
1027 * note that we do not visit this with protocols with pcb layer
1028 * code - like udp/tcp/raw ip.
1030 if ((inetsw
[ip_protox
[ip
->ip_p
]].pr_flags
& PR_LASTHDR
) &&
1031 ipsec4_in_reject(m
, NULL
)) {
1032 ipsecstat
.in_polvio
++;
1038 * enforce IPsec policy checking if we are seeing last header.
1039 * note that we do not visit this with protocols with pcb layer
1040 * code - like udp/tcp/raw ip.
1042 if (inetsw
[ip_protox
[ip
->ip_p
]].pr_flags
& PR_LASTHDR
) {
1044 * Check if the packet has already had IPsec processing
1045 * done. If so, then just pass it along. This tag gets
1046 * set during AH, ESP, etc. input handling, before the
1047 * packet is returned to the ip input queue for delivery.
1049 mtag
= m_tag_find(m
, PACKET_TAG_IPSEC_IN_DONE
, NULL
);
1052 tdbi
= (struct tdb_ident
*)m_tag_data(mtag
);
1053 sp
= ipsec_getpolicy(tdbi
, IPSEC_DIR_INBOUND
);
1055 sp
= ipsec_getpolicybyaddr(m
, IPSEC_DIR_INBOUND
,
1056 IP_FORWARDING
, &error
);
1060 * Check security policy against packet attributes.
1062 error
= ipsec_in_reject(sp
, m
);
1065 /* XXX error stat??? */
1067 DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
1074 #endif /* FAST_IPSEC */
1076 ipstat
.ips_delivered
++;
1077 if (needredispatch
) {
1078 struct netmsg_transport_packet
*msg
;
1081 ip
->ip_off
= htons(ip
->ip_off
);
1082 ip
->ip_len
= htons(ip
->ip_len
);
1083 port
= ip_mport(&m
);
1087 msg
= kmalloc(sizeof(struct netmsg_transport_packet
), M_LWKTMSG
,
1088 M_INTWAIT
| M_NULLOK
);
1092 netmsg_init(&msg
->nm_netmsg
, &netisr_afree_rport
, 0,
1093 transport_processing_handler
);
1094 msg
->nm_hlen
= hlen
;
1095 msg
->nm_hasnexthop
= (args
.next_hop
!= NULL
);
1096 if (msg
->nm_hasnexthop
)
1097 msg
->nm_nexthop
= *args
.next_hop
; /* structure copy */
1100 ip
= mtod(m
, struct ip
*);
1101 ip
->ip_len
= ntohs(ip
->ip_len
);
1102 ip
->ip_off
= ntohs(ip
->ip_off
);
1103 lwkt_sendmsg(port
, &msg
->nm_netmsg
.nm_lmsg
);
1105 transport_processing_oncpu(m
, hlen
, ip
, args
.next_hop
);
1114 * Take incoming datagram fragment and try to reassemble it into
1115 * whole datagram. If a chain for reassembly of this datagram already
1116 * exists, then it is given as fp; otherwise have to make a chain.
1118 * When IPDIVERT enabled, keep additional state with each packet that
1119 * tells us if we need to divert or tee the packet we're building.
1120 * In particular, *divinfo includes the port and TEE flag.
1123 static struct mbuf
*
1124 ip_reass(struct mbuf
*m
, struct ipq
*fp
, struct ipq
*where
,
1127 struct ip
*ip
= mtod(m
, struct ip
*);
1128 struct mbuf
*p
= NULL
, *q
, *nq
;
1130 int hlen
= IP_VHL_HL(ip
->ip_vhl
) << 2;
1137 * If the hardware has not done csum over this fragment
1138 * then csum_data is not valid at all.
1140 if ((m
->m_pkthdr
.csum_flags
& (CSUM_FRAG_NOT_CHECKED
| CSUM_DATA_VALID
))
1141 == (CSUM_FRAG_NOT_CHECKED
| CSUM_DATA_VALID
)) {
1142 m
->m_pkthdr
.csum_data
= 0;
1143 m
->m_pkthdr
.csum_flags
&= ~(CSUM_DATA_VALID
| CSUM_PSEUDO_HDR
);
1147 * Presence of header sizes in mbufs
1148 * would confuse code below.
1154 * If first fragment to arrive, create a reassembly queue.
1157 if ((fp
= mpipe_alloc_nowait(&ipq_mpipe
)) == NULL
)
1162 fp
->ipq_ttl
= IPFRAGTTL
;
1163 fp
->ipq_p
= ip
->ip_p
;
1164 fp
->ipq_id
= ip
->ip_id
;
1165 fp
->ipq_src
= ip
->ip_src
;
1166 fp
->ipq_dst
= ip
->ip_dst
;
1168 m
->m_nextpkt
= NULL
;
1170 fp
->ipq_div_info
= 0;
1177 #define GETIP(m) ((struct ip*)((m)->m_pkthdr.header))
1180 * Find a segment which begins after this one does.
1182 for (p
= NULL
, q
= fp
->ipq_frags
; q
; p
= q
, q
= q
->m_nextpkt
)
1183 if (GETIP(q
)->ip_off
> ip
->ip_off
)
1187 * If there is a preceding segment, it may provide some of
1188 * our data already. If so, drop the data from the incoming
1189 * segment. If it provides all of our data, drop us, otherwise
1190 * stick new segment in the proper place.
1192 * If some of the data is dropped from the the preceding
1193 * segment, then it's checksum is invalidated.
1196 i
= GETIP(p
)->ip_off
+ GETIP(p
)->ip_len
- ip
->ip_off
;
1198 if (i
>= ip
->ip_len
)
1201 m
->m_pkthdr
.csum_flags
= 0;
1205 m
->m_nextpkt
= p
->m_nextpkt
;
1208 m
->m_nextpkt
= fp
->ipq_frags
;
1213 * While we overlap succeeding segments trim them or,
1214 * if they are completely covered, dequeue them.
1216 for (; q
!= NULL
&& ip
->ip_off
+ ip
->ip_len
> GETIP(q
)->ip_off
;
1218 i
= (ip
->ip_off
+ ip
->ip_len
) - GETIP(q
)->ip_off
;
1219 if (i
< GETIP(q
)->ip_len
) {
1220 GETIP(q
)->ip_len
-= i
;
1221 GETIP(q
)->ip_off
+= i
;
1223 q
->m_pkthdr
.csum_flags
= 0;
1228 ipstat
.ips_fragdropped
++;
1230 q
->m_nextpkt
= NULL
;
1238 * Transfer firewall instructions to the fragment structure.
1239 * Only trust info in the fragment at offset 0.
1241 if (ip
->ip_off
== 0) {
1242 fp
->ipq_div_info
= *divinfo
;
1244 mtag
= m_tag_find(m
, PACKET_TAG_IPFW_DIVERT
, NULL
);
1246 m_tag_delete(m
, mtag
);
1252 * Check for complete reassembly and perform frag per packet
1255 * Frag limiting is performed here so that the nth frag has
1256 * a chance to complete the packet before we drop the packet.
1257 * As a result, n+1 frags are actually allowed per packet, but
1258 * only n will ever be stored. (n = maxfragsperpacket.)
1262 for (p
= NULL
, q
= fp
->ipq_frags
; q
; p
= q
, q
= q
->m_nextpkt
) {
1263 if (GETIP(q
)->ip_off
!= next
) {
1264 if (fp
->ipq_nfrags
> maxfragsperpacket
) {
1265 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1270 next
+= GETIP(q
)->ip_len
;
1272 /* Make sure the last packet didn't have the IP_MF flag */
1273 if (p
->m_flags
& M_FRAG
) {
1274 if (fp
->ipq_nfrags
> maxfragsperpacket
) {
1275 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1282 * Reassembly is complete. Make sure the packet is a sane size.
1286 if (next
+ (IP_VHL_HL(ip
->ip_vhl
) << 2) > IP_MAXPACKET
) {
1287 ipstat
.ips_toolong
++;
1288 ipstat
.ips_fragdropped
+= fp
->ipq_nfrags
;
1294 * Concatenate fragments.
1301 q
->m_nextpkt
= NULL
;
1302 for (q
= nq
; q
!= NULL
; q
= nq
) {
1304 q
->m_nextpkt
= NULL
;
1305 m
->m_pkthdr
.csum_flags
&= q
->m_pkthdr
.csum_flags
;
1306 m
->m_pkthdr
.csum_data
+= q
->m_pkthdr
.csum_data
;
1311 * Clean up the 1's complement checksum. Carry over 16 bits must
1312 * be added back. This assumes no more then 65535 packet fragments
1313 * were reassembled. A second carry can also occur (but not a third).
1315 m
->m_pkthdr
.csum_data
= (m
->m_pkthdr
.csum_data
& 0xffff) +
1316 (m
->m_pkthdr
.csum_data
>> 16);
1317 if (m
->m_pkthdr
.csum_data
> 0xFFFF)
1318 m
->m_pkthdr
.csum_data
-= 0xFFFF;
1323 * Extract firewall instructions from the fragment structure.
1325 *divinfo
= fp
->ipq_div_info
;
1329 * Create header for new ip packet by
1330 * modifying header of first packet;
1331 * dequeue and discard fragment reassembly header.
1332 * Make header visible.
1335 ip
->ip_src
= fp
->ipq_src
;
1336 ip
->ip_dst
= fp
->ipq_dst
;
1339 mpipe_free(&ipq_mpipe
, fp
);
1340 m
->m_len
+= (IP_VHL_HL(ip
->ip_vhl
) << 2);
1341 m
->m_data
-= (IP_VHL_HL(ip
->ip_vhl
) << 2);
1342 /* some debugging cruft by sklower, below, will go away soon */
1343 if (m
->m_flags
& M_PKTHDR
) { /* XXX this should be done elsewhere */
1346 for (n
= m
; n
; n
= n
->m_next
)
1348 m
->m_pkthdr
.len
= plen
;
1356 ipstat
.ips_fragdropped
++;
1366 * Free a fragment reassembly header and all
1367 * associated datagrams.
1370 ip_freef(struct ipq
*fp
)
1374 while (fp
->ipq_frags
) {
1376 fp
->ipq_frags
= q
->m_nextpkt
;
1377 q
->m_nextpkt
= NULL
;
1381 mpipe_free(&ipq_mpipe
, fp
);
1386 * IP timer processing;
1387 * if a timer expires on a reassembly
1388 * queue, discard it.
1397 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1401 while (fp
!= &ipq
[i
]) {
1404 if (fp
->prev
->ipq_ttl
== 0) {
1405 ipstat
.ips_fragtimeout
+= fp
->prev
->ipq_nfrags
;
1411 * If we are over the maximum number of fragments
1412 * (due to the limit being lowered), drain off
1413 * enough to get down to the new limit.
1415 if (maxnipq
>= 0 && nipq
> maxnipq
) {
1416 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1417 while (nipq
> maxnipq
&&
1418 (ipq
[i
].next
!= &ipq
[i
])) {
1419 ipstat
.ips_fragdropped
+=
1420 ipq
[i
].next
->ipq_nfrags
;
1421 ip_freef(ipq
[i
].next
);
1430 * Drain off all datagram fragments.
1437 for (i
= 0; i
< IPREASS_NHASH
; i
++) {
1438 while (ipq
[i
].next
!= &ipq
[i
]) {
1439 ipstat
.ips_fragdropped
+= ipq
[i
].next
->ipq_nfrags
;
1440 ip_freef(ipq
[i
].next
);
1447 * Do option processing on a datagram,
1448 * possibly discarding it if bad options are encountered,
1449 * or forwarding it if source-routed.
1450 * The pass argument is used when operating in the IPSTEALTH
1451 * mode to tell what options to process:
1452 * [LS]SRR (pass 0) or the others (pass 1).
1453 * The reason for as many as two passes is that when doing IPSTEALTH,
1454 * non-routing options should be processed only if the packet is for us.
1455 * Returns 1 if packet has been forwarded/freed,
1456 * 0 if the packet should be processed further.
1459 ip_dooptions(struct mbuf
*m
, int pass
, struct sockaddr_in
*next_hop
)
1461 struct sockaddr_in ipaddr
= { sizeof ipaddr
, AF_INET
};
1462 struct ip
*ip
= mtod(m
, struct ip
*);
1464 struct in_ifaddr
*ia
;
1465 int opt
, optlen
, cnt
, off
, code
, type
= ICMP_PARAMPROB
;
1466 boolean_t forward
= FALSE
;
1467 struct in_addr
*sin
, dst
;
1471 cp
= (u_char
*)(ip
+ 1);
1472 cnt
= (IP_VHL_HL(ip
->ip_vhl
) << 2) - sizeof(struct ip
);
1473 for (; cnt
> 0; cnt
-= optlen
, cp
+= optlen
) {
1474 opt
= cp
[IPOPT_OPTVAL
];
1475 if (opt
== IPOPT_EOL
)
1477 if (opt
== IPOPT_NOP
)
1480 if (cnt
< IPOPT_OLEN
+ sizeof(*cp
)) {
1481 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1484 optlen
= cp
[IPOPT_OLEN
];
1485 if (optlen
< IPOPT_OLEN
+ sizeof(*cp
) || optlen
> cnt
) {
1486 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1496 * Source routing with record.
1497 * Find interface with current destination address.
1498 * If none on this machine then drop if strictly routed,
1499 * or do nothing if loosely routed.
1500 * Record interface address and bring up next address
1501 * component. If strictly routed make sure next
1502 * address is on directly accessible net.
1506 if (ipstealth
&& pass
> 0)
1508 if (optlen
< IPOPT_OFFSET
+ sizeof(*cp
)) {
1509 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1512 if ((off
= cp
[IPOPT_OFFSET
]) < IPOPT_MINOFF
) {
1513 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1516 ipaddr
.sin_addr
= ip
->ip_dst
;
1517 ia
= (struct in_ifaddr
*)
1518 ifa_ifwithaddr((struct sockaddr
*)&ipaddr
);
1520 if (opt
== IPOPT_SSRR
) {
1521 type
= ICMP_UNREACH
;
1522 code
= ICMP_UNREACH_SRCFAIL
;
1525 if (!ip_dosourceroute
)
1526 goto nosourcerouting
;
1528 * Loose routing, and not at next destination
1529 * yet; nothing to do except forward.
1533 off
--; /* 0 origin */
1534 if (off
> optlen
- (int)sizeof(struct in_addr
)) {
1536 * End of source route. Should be for us.
1538 if (!ip_acceptsourceroute
)
1539 goto nosourcerouting
;
1540 save_rte(cp
, ip
->ip_src
);
1545 if (!ip_dosourceroute
) {
1547 char buf
[sizeof "aaa.bbb.ccc.ddd"];
1550 * Acting as a router, so generate ICMP
1553 strcpy(buf
, inet_ntoa(ip
->ip_dst
));
1555 "attempted source route from %s to %s\n",
1556 inet_ntoa(ip
->ip_src
), buf
);
1557 type
= ICMP_UNREACH
;
1558 code
= ICMP_UNREACH_SRCFAIL
;
1562 * Not acting as a router,
1566 ipstat
.ips_cantforward
++;
1573 * locate outgoing interface
1575 memcpy(&ipaddr
.sin_addr
, cp
+ off
,
1576 sizeof ipaddr
.sin_addr
);
1578 if (opt
== IPOPT_SSRR
) {
1579 #define INA struct in_ifaddr *
1580 #define SA struct sockaddr *
1581 if ((ia
= (INA
)ifa_ifwithdstaddr((SA
)&ipaddr
))
1583 ia
= (INA
)ifa_ifwithnet((SA
)&ipaddr
);
1585 ia
= ip_rtaddr(ipaddr
.sin_addr
,
1586 &ipforward_rt
[mycpuid
]);
1588 type
= ICMP_UNREACH
;
1589 code
= ICMP_UNREACH_SRCFAIL
;
1592 ip
->ip_dst
= ipaddr
.sin_addr
;
1593 memcpy(cp
+ off
, &IA_SIN(ia
)->sin_addr
,
1594 sizeof(struct in_addr
));
1595 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1597 * Let ip_intr's mcast routing check handle mcast pkts
1599 forward
= !IN_MULTICAST(ntohl(ip
->ip_dst
.s_addr
));
1603 if (ipstealth
&& pass
== 0)
1605 if (optlen
< IPOPT_OFFSET
+ sizeof(*cp
)) {
1606 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1609 if ((off
= cp
[IPOPT_OFFSET
]) < IPOPT_MINOFF
) {
1610 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1614 * If no space remains, ignore.
1616 off
--; /* 0 origin */
1617 if (off
> optlen
- (int)sizeof(struct in_addr
))
1619 memcpy(&ipaddr
.sin_addr
, &ip
->ip_dst
,
1620 sizeof ipaddr
.sin_addr
);
1622 * locate outgoing interface; if we're the destination,
1623 * use the incoming interface (should be same).
1625 if ((ia
= (INA
)ifa_ifwithaddr((SA
)&ipaddr
)) == NULL
&&
1626 (ia
= ip_rtaddr(ipaddr
.sin_addr
,
1627 &ipforward_rt
[mycpuid
]))
1629 type
= ICMP_UNREACH
;
1630 code
= ICMP_UNREACH_HOST
;
1633 memcpy(cp
+ off
, &IA_SIN(ia
)->sin_addr
,
1634 sizeof(struct in_addr
));
1635 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1639 if (ipstealth
&& pass
== 0)
1641 code
= cp
- (u_char
*)ip
;
1642 if (optlen
< 4 || optlen
> 40) {
1643 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1646 if ((off
= cp
[IPOPT_OFFSET
]) < 5) {
1647 code
= &cp
[IPOPT_OLEN
] - (u_char
*)ip
;
1650 if (off
> optlen
- (int)sizeof(int32_t)) {
1651 cp
[IPOPT_OFFSET
+ 1] += (1 << 4);
1652 if ((cp
[IPOPT_OFFSET
+ 1] & 0xf0) == 0) {
1653 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1658 off
--; /* 0 origin */
1659 sin
= (struct in_addr
*)(cp
+ off
);
1660 switch (cp
[IPOPT_OFFSET
+ 1] & 0x0f) {
1662 case IPOPT_TS_TSONLY
:
1665 case IPOPT_TS_TSANDADDR
:
1666 if (off
+ sizeof(n_time
) +
1667 sizeof(struct in_addr
) > optlen
) {
1668 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1671 ipaddr
.sin_addr
= dst
;
1672 ia
= (INA
)ifaof_ifpforaddr((SA
)&ipaddr
,
1676 memcpy(sin
, &IA_SIN(ia
)->sin_addr
,
1677 sizeof(struct in_addr
));
1678 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1679 off
+= sizeof(struct in_addr
);
1682 case IPOPT_TS_PRESPEC
:
1683 if (off
+ sizeof(n_time
) +
1684 sizeof(struct in_addr
) > optlen
) {
1685 code
= &cp
[IPOPT_OFFSET
] - (u_char
*)ip
;
1688 memcpy(&ipaddr
.sin_addr
, sin
,
1689 sizeof(struct in_addr
));
1690 if (ifa_ifwithaddr((SA
)&ipaddr
) == NULL
)
1692 cp
[IPOPT_OFFSET
] += sizeof(struct in_addr
);
1693 off
+= sizeof(struct in_addr
);
1697 code
= &cp
[IPOPT_OFFSET
+ 1] - (u_char
*)ip
;
1701 memcpy(cp
+ off
, &ntime
, sizeof(n_time
));
1702 cp
[IPOPT_OFFSET
] += sizeof(n_time
);
1705 if (forward
&& ipforwarding
) {
1706 ip_forward(m
, TRUE
, next_hop
);
1711 icmp_error(m
, type
, code
, 0, 0);
1712 ipstat
.ips_badoptions
++;
1717 * Given address of next destination (final or next hop),
1718 * return internet address info of interface to be used to get there.
1721 ip_rtaddr(struct in_addr dst
, struct route
*ro
)
1723 struct sockaddr_in
*sin
;
1725 sin
= (struct sockaddr_in
*)&ro
->ro_dst
;
1727 if (ro
->ro_rt
== NULL
|| dst
.s_addr
!= sin
->sin_addr
.s_addr
) {
1728 if (ro
->ro_rt
!= NULL
) {
1732 sin
->sin_family
= AF_INET
;
1733 sin
->sin_len
= sizeof *sin
;
1734 sin
->sin_addr
= dst
;
1735 rtalloc_ign(ro
, RTF_PRCLONING
);
1738 if (ro
->ro_rt
== NULL
)
1741 return (ifatoia(ro
->ro_rt
->rt_ifa
));
1745 * Save incoming source route for use in replies,
1746 * to be picked up later by ip_srcroute if the receiver is interested.
1749 save_rte(u_char
*option
, struct in_addr dst
)
1753 olen
= option
[IPOPT_OLEN
];
1756 kprintf("save_rte: olen %d\n", olen
);
1758 if (olen
> sizeof(ip_srcrt
) - (1 + sizeof(dst
)))
1760 bcopy(option
, ip_srcrt
.srcopt
, olen
);
1761 ip_nhops
= (olen
- IPOPT_OFFSET
- 1) / sizeof(struct in_addr
);
1766 * Retrieve incoming source route for use in replies,
1767 * in the same form used by setsockopt.
1768 * The first hop is placed before the options, will be removed later.
1773 struct in_addr
*p
, *q
;
1778 m
= m_get(MB_DONTWAIT
, MT_HEADER
);
1782 #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1784 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1785 m
->m_len
= ip_nhops
* sizeof(struct in_addr
) + sizeof(struct in_addr
) +
1789 kprintf("ip_srcroute: nhops %d mlen %d", ip_nhops
, m
->m_len
);
1793 * First save first hop for return route
1795 p
= &ip_srcrt
.route
[ip_nhops
- 1];
1796 *(mtod(m
, struct in_addr
*)) = *p
--;
1799 kprintf(" hops %x", ntohl(mtod(m
, struct in_addr
*)->s_addr
));
1803 * Copy option fields and padding (nop) to mbuf.
1805 ip_srcrt
.nop
= IPOPT_NOP
;
1806 ip_srcrt
.srcopt
[IPOPT_OFFSET
] = IPOPT_MINOFF
;
1807 memcpy(mtod(m
, caddr_t
) + sizeof(struct in_addr
), &ip_srcrt
.nop
,
1809 q
= (struct in_addr
*)(mtod(m
, caddr_t
) +
1810 sizeof(struct in_addr
) + OPTSIZ
);
1813 * Record return path as an IP source route,
1814 * reversing the path (pointers are now aligned).
1816 while (p
>= ip_srcrt
.route
) {
1819 kprintf(" %x", ntohl(q
->s_addr
));
1824 * Last hop goes to final destination.
1829 kprintf(" %x\n", ntohl(q
->s_addr
));
1835 * Strip out IP options.
1838 ip_stripoptions(struct mbuf
*m
)
1841 struct ip
*ip
= mtod(m
, struct ip
*);
1845 optlen
= (IP_VHL_HL(ip
->ip_vhl
) << 2) - sizeof(struct ip
);
1846 opts
= (caddr_t
)(ip
+ 1);
1847 datalen
= m
->m_len
- (sizeof(struct ip
) + optlen
);
1848 bcopy(opts
+ optlen
, opts
, datalen
);
1850 if (m
->m_flags
& M_PKTHDR
)
1851 m
->m_pkthdr
.len
-= optlen
;
1852 ip
->ip_vhl
= IP_MAKE_VHL(IPVERSION
, sizeof(struct ip
) >> 2);
1855 u_char inetctlerrmap
[PRC_NCMDS
] = {
1857 0, EMSGSIZE
, EHOSTDOWN
, EHOSTUNREACH
,
1858 EHOSTUNREACH
, EHOSTUNREACH
, ECONNREFUSED
, ECONNREFUSED
,
1859 EMSGSIZE
, EHOSTUNREACH
, 0, 0,
1861 ENOPROTOOPT
, ECONNREFUSED
1865 * Forward a packet. If some error occurs return the sender
1866 * an icmp packet. Note we can't always generate a meaningful
1867 * icmp message because icmp doesn't have a large enough repertoire
1868 * of codes and types.
1870 * If not forwarding, just drop the packet. This could be confusing
1871 * if ipforwarding was zero but some routing protocol was advancing
1872 * us as a gateway to somewhere. However, we must let the routing
1873 * protocol deal with that.
1875 * The using_srcrt parameter indicates whether the packet is being forwarded
1876 * via a source route.
1879 ip_forward(struct mbuf
*m
, boolean_t using_srcrt
, struct sockaddr_in
*next_hop
)
1881 struct ip
*ip
= mtod(m
, struct ip
*);
1882 struct sockaddr_in
*ipforward_rtaddr
;
1884 int error
, type
= 0, code
= 0, destmtu
= 0;
1887 struct in_addr pkt_dst
;
1889 struct route
*cache_rt
= &ipforward_rt
[mycpuid
];
1893 * Cache the destination address of the packet; this may be
1894 * changed by use of 'ipfw fwd'.
1896 pkt_dst
= (next_hop
!= NULL
) ? next_hop
->sin_addr
: ip
->ip_dst
;
1900 kprintf("forward: src %x dst %x ttl %x\n",
1901 ip
->ip_src
.s_addr
, pkt_dst
.s_addr
, ip
->ip_ttl
);
1904 if (m
->m_flags
& (M_BCAST
| M_MCAST
) || !in_canforward(pkt_dst
)) {
1905 ipstat
.ips_cantforward
++;
1909 if (!ipstealth
&& ip
->ip_ttl
<= IPTTLDEC
) {
1910 icmp_error(m
, ICMP_TIMXCEED
, ICMP_TIMXCEED_INTRANS
, dest
, 0);
1914 ipforward_rtaddr
= (struct sockaddr_in
*) &cache_rt
->ro_dst
;
1915 if (cache_rt
->ro_rt
== NULL
||
1916 ipforward_rtaddr
->sin_addr
.s_addr
!= pkt_dst
.s_addr
) {
1917 if (cache_rt
->ro_rt
!= NULL
) {
1918 RTFREE(cache_rt
->ro_rt
);
1919 cache_rt
->ro_rt
= NULL
;
1921 ipforward_rtaddr
->sin_family
= AF_INET
;
1922 ipforward_rtaddr
->sin_len
= sizeof(struct sockaddr_in
);
1923 ipforward_rtaddr
->sin_addr
= pkt_dst
;
1924 rtalloc_ign(cache_rt
, RTF_PRCLONING
);
1925 if (cache_rt
->ro_rt
== NULL
) {
1926 icmp_error(m
, ICMP_UNREACH
, ICMP_UNREACH_HOST
, dest
, 0);
1930 rt
= cache_rt
->ro_rt
;
1933 * Save the IP header and at most 8 bytes of the payload,
1934 * in case we need to generate an ICMP message to the src.
1936 * XXX this can be optimized a lot by saving the data in a local
1937 * buffer on the stack (72 bytes at most), and only allocating the
1938 * mbuf if really necessary. The vast majority of the packets
1939 * are forwarded without having to send an ICMP back (either
1940 * because unnecessary, or because rate limited), so we are
1941 * really we are wasting a lot of work here.
1943 * We don't use m_copy() because it might return a reference
1944 * to a shared cluster. Both this function and ip_output()
1945 * assume exclusive access to the IP header in `m', so any
1946 * data in a cluster may change before we reach icmp_error().
1948 MGETHDR(mcopy
, MB_DONTWAIT
, m
->m_type
);
1949 if (mcopy
!= NULL
&& !m_dup_pkthdr(mcopy
, m
, MB_DONTWAIT
)) {
1951 * It's probably ok if the pkthdr dup fails (because
1952 * the deep copy of the tag chain failed), but for now
1953 * be conservative and just discard the copy since
1954 * code below may some day want the tags.
1959 if (mcopy
!= NULL
) {
1960 mcopy
->m_len
= imin((IP_VHL_HL(ip
->ip_vhl
) << 2) + 8,
1962 mcopy
->m_pkthdr
.len
= mcopy
->m_len
;
1963 m_copydata(m
, 0, mcopy
->m_len
, mtod(mcopy
, caddr_t
));
1967 ip
->ip_ttl
-= IPTTLDEC
;
1970 * If forwarding packet using same interface that it came in on,
1971 * perhaps should send a redirect to sender to shortcut a hop.
1972 * Only send redirect if source is sending directly to us,
1973 * and if packet was not source routed (or has any options).
1974 * Also, don't send redirect if forwarding using a default route
1975 * or a route modified by a redirect.
1977 if (rt
->rt_ifp
== m
->m_pkthdr
.rcvif
&&
1978 !(rt
->rt_flags
& (RTF_DYNAMIC
| RTF_MODIFIED
)) &&
1979 satosin(rt_key(rt
))->sin_addr
.s_addr
!= INADDR_ANY
&&
1980 ipsendredirects
&& !using_srcrt
&& next_hop
== NULL
) {
1981 u_long src
= ntohl(ip
->ip_src
.s_addr
);
1982 struct in_ifaddr
*rt_ifa
= (struct in_ifaddr
*)rt
->rt_ifa
;
1984 if (rt_ifa
!= NULL
&&
1985 (src
& rt_ifa
->ia_subnetmask
) == rt_ifa
->ia_subnet
) {
1986 if (rt
->rt_flags
& RTF_GATEWAY
)
1987 dest
= satosin(rt
->rt_gateway
)->sin_addr
.s_addr
;
1989 dest
= pkt_dst
.s_addr
;
1991 * Router requirements says to only send
1994 type
= ICMP_REDIRECT
;
1995 code
= ICMP_REDIRECT_HOST
;
1998 kprintf("redirect (%d) to %x\n", code
, dest
);
2003 if (next_hop
!= NULL
) {
2004 /* Pass IPFORWARD info if available */
2005 tag
.mh_type
= MT_TAG
;
2006 tag
.mh_flags
= PACKET_TAG_IPFORWARD
;
2007 tag
.mh_data
= (caddr_t
)next_hop
;
2009 m
= (struct mbuf
*)&tag
;
2012 error
= ip_output(m
, NULL
, cache_rt
, IP_FORWARDING
, NULL
,
2015 ipstat
.ips_forward
++;
2018 ipflow_create(cache_rt
, mcopy
);
2021 return; /* most common case */
2023 ipstat
.ips_redirectsent
++;
2026 ipstat
.ips_cantforward
++;
2033 * Send ICMP message.
2038 case 0: /* forwarded, but need redirect */
2039 /* type, code set above */
2042 case ENETUNREACH
: /* shouldn't happen, checked above */
2047 type
= ICMP_UNREACH
;
2048 code
= ICMP_UNREACH_HOST
;
2052 type
= ICMP_UNREACH
;
2053 code
= ICMP_UNREACH_NEEDFRAG
;
2056 * If the packet is routed over IPsec tunnel, tell the
2057 * originator the tunnel MTU.
2058 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
2061 if (cache_rt
->ro_rt
!= NULL
) {
2062 struct secpolicy
*sp
= NULL
;
2067 sp
= ipsec4_getpolicybyaddr(mcopy
,
2073 destmtu
= cache_rt
->ro_rt
->rt_ifp
->if_mtu
;
2075 /* count IPsec header size */
2076 ipsechdr
= ipsec4_hdrsiz(mcopy
,
2081 * find the correct route for outer IPv4
2082 * header, compute tunnel MTU.
2085 if (sp
->req
!= NULL
&& sp
->req
->sav
!= NULL
&&
2086 sp
->req
->sav
->sah
!= NULL
) {
2087 ro
= &sp
->req
->sav
->sah
->sa_route
;
2088 if (ro
->ro_rt
!= NULL
&&
2089 ro
->ro_rt
->rt_ifp
!= NULL
) {
2091 ro
->ro_rt
->rt_ifp
->if_mtu
;
2092 destmtu
-= ipsechdr
;
2101 * If the packet is routed over IPsec tunnel, tell the
2102 * originator the tunnel MTU.
2103 * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz
2106 if (cache_rt
->ro_rt
!= NULL
) {
2107 struct secpolicy
*sp
= NULL
;
2112 sp
= ipsec_getpolicybyaddr(mcopy
,
2118 destmtu
= cache_rt
->ro_rt
->rt_ifp
->if_mtu
;
2120 /* count IPsec header size */
2121 ipsechdr
= ipsec4_hdrsiz(mcopy
,
2126 * find the correct route for outer IPv4
2127 * header, compute tunnel MTU.
2130 if (sp
->req
!= NULL
&&
2131 sp
->req
->sav
!= NULL
&&
2132 sp
->req
->sav
->sah
!= NULL
) {
2133 ro
= &sp
->req
->sav
->sah
->sa_route
;
2134 if (ro
->ro_rt
!= NULL
&&
2135 ro
->ro_rt
->rt_ifp
!= NULL
) {
2137 ro
->ro_rt
->rt_ifp
->if_mtu
;
2138 destmtu
-= ipsechdr
;
2145 #else /* !IPSEC && !FAST_IPSEC */
2146 if (cache_rt
->ro_rt
!= NULL
)
2147 destmtu
= cache_rt
->ro_rt
->rt_ifp
->if_mtu
;
2149 ipstat
.ips_cantfrag
++;
2154 * A router should not generate ICMP_SOURCEQUENCH as
2155 * required in RFC1812 Requirements for IP Version 4 Routers.
2156 * Source quench could be a big problem under DoS attacks,
2157 * or if the underlying interface is rate-limited.
2158 * Those who need source quench packets may re-enable them
2159 * via the net.inet.ip.sendsourcequench sysctl.
2161 if (!ip_sendsourcequench
) {
2165 type
= ICMP_SOURCEQUENCH
;
2170 case EACCES
: /* ipfw denied packet */
2174 icmp_error(mcopy
, type
, code
, dest
, destmtu
);
2178 ip_savecontrol(struct inpcb
*inp
, struct mbuf
**mp
, struct ip
*ip
,
2181 if (inp
->inp_socket
->so_options
& SO_TIMESTAMP
) {
2185 *mp
= sbcreatecontrol((caddr_t
) &tv
, sizeof(tv
),
2186 SCM_TIMESTAMP
, SOL_SOCKET
);
2188 mp
= &(*mp
)->m_next
;
2190 if (inp
->inp_flags
& INP_RECVDSTADDR
) {
2191 *mp
= sbcreatecontrol((caddr_t
) &ip
->ip_dst
,
2192 sizeof(struct in_addr
), IP_RECVDSTADDR
, IPPROTO_IP
);
2194 mp
= &(*mp
)->m_next
;
2196 if (inp
->inp_flags
& INP_RECVTTL
) {
2197 *mp
= sbcreatecontrol((caddr_t
) &ip
->ip_ttl
,
2198 sizeof(u_char
), IP_RECVTTL
, IPPROTO_IP
);
2200 mp
= &(*mp
)->m_next
;
2204 * Moving these out of udp_input() made them even more broken
2205 * than they already were.
2207 /* options were tossed already */
2208 if (inp
->inp_flags
& INP_RECVOPTS
) {
2209 *mp
= sbcreatecontrol((caddr_t
) opts_deleted_above
,
2210 sizeof(struct in_addr
), IP_RECVOPTS
, IPPROTO_IP
);
2212 mp
= &(*mp
)->m_next
;
2214 /* ip_srcroute doesn't do what we want here, need to fix */
2215 if (inp
->inp_flags
& INP_RECVRETOPTS
) {
2216 *mp
= sbcreatecontrol((caddr_t
) ip_srcroute(),
2217 sizeof(struct in_addr
), IP_RECVRETOPTS
, IPPROTO_IP
);
2219 mp
= &(*mp
)->m_next
;
2222 if (inp
->inp_flags
& INP_RECVIF
) {
2225 struct sockaddr_dl sdl
;
2228 struct sockaddr_dl
*sdp
;
2229 struct sockaddr_dl
*sdl2
= &sdlbuf
.sdl
;
2231 if (((ifp
= m
->m_pkthdr
.rcvif
)) &&
2232 ((ifp
->if_index
!= 0) && (ifp
->if_index
<= if_index
))) {
2233 sdp
= IF_LLSOCKADDR(ifp
);
2235 * Change our mind and don't try copy.
2237 if ((sdp
->sdl_family
!= AF_LINK
) ||
2238 (sdp
->sdl_len
> sizeof(sdlbuf
))) {
2241 bcopy(sdp
, sdl2
, sdp
->sdl_len
);
2245 offsetof(struct sockaddr_dl
, sdl_data
[0]);
2246 sdl2
->sdl_family
= AF_LINK
;
2247 sdl2
->sdl_index
= 0;
2248 sdl2
->sdl_nlen
= sdl2
->sdl_alen
= sdl2
->sdl_slen
= 0;
2250 *mp
= sbcreatecontrol((caddr_t
) sdl2
, sdl2
->sdl_len
,
2251 IP_RECVIF
, IPPROTO_IP
);
2253 mp
= &(*mp
)->m_next
;
2258 * XXX these routines are called from the upper part of the kernel.
2260 * They could also be moved to ip_mroute.c, since all the RSVP
2261 * handling is done there already.
2264 ip_rsvp_init(struct socket
*so
)
2266 if (so
->so_type
!= SOCK_RAW
||
2267 so
->so_proto
->pr_protocol
!= IPPROTO_RSVP
)
2270 if (ip_rsvpd
!= NULL
)
2275 * This may seem silly, but we need to be sure we don't over-increment
2276 * the RSVP counter, in case something slips up.
2291 * This may seem silly, but we need to be sure we don't over-decrement
2292 * the RSVP counter, in case something slips up.
2302 rsvp_input(struct mbuf
*m
, ...) /* XXX must fixup manually */
2308 off
= __va_arg(ap
, int);
2309 proto
= __va_arg(ap
, int);
2312 if (rsvp_input_p
) { /* call the real one if loaded */
2313 rsvp_input_p(m
, off
, proto
);
2317 /* Can still get packets with rsvp_on = 0 if there is a local member
2318 * of the group to which the RSVP packet is addressed. But in this
2319 * case we want to throw the packet away.
2327 if (ip_rsvpd
!= NULL
) {
2328 rip_input(m
, off
, proto
);
2331 /* Drop the packet */