kernel: Make SMP support default (and non-optional).
[dragonfly.git] / sys / netinet / ip_input.c
CommitLineData
984263bc 1/*
66d6c637
JH
2 * Copyright (c) 2003, 2004 Jeffrey M. Hsu. All rights reserved.
3 * Copyright (c) 2003, 2004 The DragonFly Project. All rights reserved.
f23061d4 4 *
66d6c637
JH
5 * This code is derived from software contributed to The DragonFly Project
6 * by Jeffrey M. Hsu.
f23061d4 7 *
66d6c637
JH
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
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.
f23061d4 19 *
66d6c637
JH
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
31 * SUCH DAMAGE.
32 */
33
66d6c637 34/*
984263bc
MD
35 * Copyright (c) 1982, 1986, 1988, 1993
36 * The Regents of the University of California. All rights reserved.
37 *
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
40 * are met:
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.
53 *
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
64 * SUCH DAMAGE.
65 *
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 */
69
70#define _IP_VHL
71
72#include "opt_bootp.h"
984263bc
MD
73#include "opt_ipdn.h"
74#include "opt_ipdivert.h"
984263bc
MD
75#include "opt_ipstealth.h"
76#include "opt_ipsec.h"
fce0e0b4 77#include "opt_rss.h"
984263bc
MD
78
79#include <sys/param.h>
80#include <sys/systm.h>
81#include <sys/mbuf.h>
82#include <sys/malloc.h>
2b89bacf 83#include <sys/mpipe.h>
984263bc
MD
84#include <sys/domain.h>
85#include <sys/protosw.h>
86#include <sys/socket.h>
87#include <sys/time.h>
fc9c7b4f
HP
88#include <sys/globaldata.h>
89#include <sys/thread.h>
984263bc
MD
90#include <sys/kernel.h>
91#include <sys/syslog.h>
92#include <sys/sysctl.h>
3f9db7f8 93#include <sys/in_cksum.h>
8b3db995 94#include <sys/lock.h>
984263bc 95
684a93c4
MD
96#include <sys/mplock2.h>
97
a00138cb
JS
98#include <machine/stdarg.h>
99
984263bc
MD
100#include <net/if.h>
101#include <net/if_types.h>
102#include <net/if_var.h>
103#include <net/if_dl.h>
e7e55f42 104#include <net/pfil.h>
984263bc
MD
105#include <net/route.h>
106#include <net/netisr.h>
984263bc
MD
107
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>
b0cb7c2c 115#include <netinet/ip_divert.h>
4639df5f 116#include <netinet/ip_flow.h>
984263bc 117
4599cf19
MD
118#include <sys/thread2.h>
119#include <sys/msgport2.h>
120#include <net/netmsg2.h>
984263bc
MD
121
122#include <sys/socketvar.h>
123
1f2de5d4
MD
124#include <net/ipfw/ip_fw.h>
125#include <net/dummynet/ip_dummynet.h>
984263bc
MD
126
127#ifdef IPSEC
128#include <netinet6/ipsec.h>
d2438d69 129#include <netproto/key/key.h>
984263bc
MD
130#endif
131
132#ifdef FAST_IPSEC
bf844ffa
JH
133#include <netproto/ipsec/ipsec.h>
134#include <netproto/ipsec/key.h>
984263bc
MD
135#endif
136
137int rsvp_on = 0;
138static int ip_rsvp_on;
139struct socket *ip_rsvpd;
140
dffa46cd 141int ipforwarding = 0;
984263bc
MD
142SYSCTL_INT(_net_inet_ip, IPCTL_FORWARDING, forwarding, CTLFLAG_RW,
143 &ipforwarding, 0, "Enable IP forwarding between interfaces");
144
dffa46cd 145static int ipsendredirects = 1; /* XXX */
984263bc
MD
146SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, CTLFLAG_RW,
147 &ipsendredirects, 0, "Enable sending IP redirects");
148
dffa46cd 149int ip_defttl = IPDEFTTL;
984263bc
MD
150SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW,
151 &ip_defttl, 0, "Maximum TTL on IP packets");
152
dffa46cd 153static int ip_dosourceroute = 0;
984263bc
MD
154SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, CTLFLAG_RW,
155 &ip_dosourceroute, 0, "Enable forwarding source routed IP packets");
156
dffa46cd
JH
157static int ip_acceptsourceroute = 0;
158SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute,
159 CTLFLAG_RW, &ip_acceptsourceroute, 0,
984263bc
MD
160 "Enable accepting source routed IP packets");
161
dffa46cd 162static int ip_keepfaith = 0;
984263bc 163SYSCTL_INT(_net_inet_ip, IPCTL_KEEPFAITH, keepfaith, CTLFLAG_RW,
dffa46cd 164 &ip_keepfaith, 0,
0ca0cd25 165 "Enable packet capture for FAITH IPv4->IPv6 translator daemon");
984263bc 166
dffa46cd
JH
167static int nipq = 0; /* total # of reass queues */
168static int maxnipq;
984263bc 169SYSCTL_INT(_net_inet_ip, OID_AUTO, maxfragpackets, CTLFLAG_RW,
dffa46cd
JH
170 &maxnipq, 0,
171 "Maximum number of IPv4 fragment reassembly queue entries");
984263bc 172
dffa46cd 173static int maxfragsperpacket;
984263bc 174SYSCTL_INT(_net_inet_ip, OID_AUTO, maxfragsperpacket, CTLFLAG_RW,
dffa46cd
JH
175 &maxfragsperpacket, 0,
176 "Maximum number of IPv4 fragments allowed per packet");
984263bc 177
dffa46cd 178static int ip_sendsourcequench = 0;
984263bc 179SYSCTL_INT(_net_inet_ip, OID_AUTO, sendsourcequench, CTLFLAG_RW,
dffa46cd
JH
180 &ip_sendsourcequench, 0,
181 "Enable the transmission of source quench packets");
984263bc 182
b6afe32a 183int ip_do_randomid = 1;
6277137d
MD
184SYSCTL_INT(_net_inet_ip, OID_AUTO, random_id, CTLFLAG_RW,
185 &ip_do_randomid, 0,
186 "Assign random ip_id values");
984263bc
MD
187/*
188 * XXX - Setting ip_checkinterface mostly implements the receive side of
189 * the Strong ES model described in RFC 1122, but since the routing table
190 * and transmit implementation do not implement the Strong ES model,
191 * setting this to 1 results in an odd hybrid.
192 *
193 * XXX - ip_checkinterface currently must be disabled if you use ipnat
194 * to translate the destination address to another local interface.
195 *
196 * XXX - ip_checkinterface must be disabled if you add IP aliases
197 * to the loopback interface instead of the interface where the
198 * packets for those addresses are received.
199 */
dffa46cd 200static int ip_checkinterface = 0;
984263bc
MD
201SYSCTL_INT(_net_inet_ip, OID_AUTO, check_interface, CTLFLAG_RW,
202 &ip_checkinterface, 0, "Verify packet arrives on correct interface");
203
fce0e0b4
SZ
204static u_long ip_hash_count = 0;
205SYSCTL_ULONG(_net_inet_ip, OID_AUTO, hash_count, CTLFLAG_RD,
206 &ip_hash_count, 0, "Number of packets hashed by IP");
207
208#ifdef RSS_DEBUG
209static u_long ip_rehash_count = 0;
210SYSCTL_ULONG(_net_inet_ip, OID_AUTO, rehash_count, CTLFLAG_RD,
211 &ip_rehash_count, 0, "Number of packets rehashed by IP");
212
213static u_long ip_dispatch_fast = 0;
214SYSCTL_ULONG(_net_inet_ip, OID_AUTO, dispatch_fast_count, CTLFLAG_RD,
215 &ip_dispatch_fast, 0, "Number of packets handled on current CPU");
216
217static u_long ip_dispatch_slow = 0;
218SYSCTL_ULONG(_net_inet_ip, OID_AUTO, dispatch_slow_count, CTLFLAG_RD,
219 &ip_dispatch_slow, 0, "Number of packets messaged to another CPU");
220#endif
c3c96e44 221
a3c18566 222static struct lwkt_token ipq_token = LWKT_TOKEN_INITIALIZER(ipq_token);
2d23a8be 223
984263bc 224#ifdef DIAGNOSTIC
dffa46cd 225static int ipprintfs = 0;
984263bc
MD
226#endif
227
984263bc 228extern struct domain inetdomain;
4468b0b4 229extern struct protosw inetsw[];
984263bc 230u_char ip_protox[IPPROTO_MAX];
1b562c24 231struct in_ifaddrhead in_ifaddrheads[MAXCPU]; /* first inet address */
f8983475
SZ
232struct in_ifaddrhashhead *in_ifaddrhashtbls[MAXCPU];
233 /* inet addr hash table */
984263bc 234u_long in_ifaddrhmask; /* mask for hash table */
dffa46cd 235
ae1d3076 236struct ip_stats ipstats_percpu[MAXCPU];
1918fc5c 237
fc9c7b4f
HP
238static int
239sysctl_ipstats(SYSCTL_HANDLER_ARGS)
240{
241 int cpu, error = 0;
242
243 for (cpu = 0; cpu < ncpus; ++cpu) {
ae1d3076 244 if ((error = SYSCTL_OUT(req, &ipstats_percpu[cpu],
fc9c7b4f
HP
245 sizeof(struct ip_stats))))
246 break;
ae1d3076 247 if ((error = SYSCTL_IN(req, &ipstats_percpu[cpu],
fc9c7b4f
HP
248 sizeof(struct ip_stats))))
249 break;
250 }
251
252 return (error);
253}
254SYSCTL_PROC(_net_inet_ip, IPCTL_STATS, stats, (CTLTYPE_OPAQUE | CTLFLAG_RW),
255 0, 0, sysctl_ipstats, "S,ip_stats", "IP statistics");
984263bc
MD
256
257/* Packet reassembly stuff */
dffa46cd
JH
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)
984263bc 263
83c38804 264static TAILQ_HEAD(ipqhead, ipq) ipq[IPREASS_NHASH];
984263bc
MD
265
266#ifdef IPCTL_DEFMTU
267SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW,
268 &ip_mtu, 0, "Default MTU");
269#endif
270
271#ifdef IPSTEALTH
dffa46cd
JH
272static int ipstealth = 0;
273SYSCTL_INT(_net_inet_ip, OID_AUTO, stealth, CTLFLAG_RW, &ipstealth, 0, "");
274#else
275static const int ipstealth = 0;
984263bc
MD
276#endif
277
0030d31a
SZ
278struct mbuf *(*ip_divert_p)(struct mbuf *, int, int);
279
e7e55f42 280struct pfil_head inet_pfil_hook;
984263bc
MD
281
282/*
7e31206a
SZ
283 * struct ip_srcrt_opt is used to store packet state while it travels
284 * through the stack.
285 *
286 * XXX Note that the code even makes assumptions on the size and
984263bc 287 * alignment of fields inside struct ip_srcrt so e.g. adding some
7e31206a 288 * fields will break the code. This needs to be fixed.
984263bc
MD
289 *
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
294 * to us.
295 */
7e31206a 296struct ip_srcrt {
984263bc
MD
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)];
7e31206a
SZ
301};
302
303struct ip_srcrt_opt {
304 int ip_nhops;
305 struct ip_srcrt ip_srcrt;
306};
984263bc 307
2b89bacf
MD
308static MALLOC_DEFINE(M_IPQ, "ipq", "IP Fragment Management");
309static struct malloc_pipe ipq_mpipe;
310
7e31206a 311static void save_rte(struct mbuf *, u_char *, struct in_addr);
bb3b17df 312static int ip_dooptions(struct mbuf *m, int, struct sockaddr_in *);
83c38804 313static void ip_freef(struct ipqhead *, struct ipq *);
002c1265 314static void ip_input_handler(netmsg_t);
984263bc
MD
315
316/*
317 * IP initialization: fill in IP protocol switch table.
318 * All protocols not implemented in kernel go to raw IP protocol handler.
319 */
320void
8a3125c6 321ip_init(void)
984263bc 322{
4468b0b4 323 struct protosw *pr;
2256ba69 324 int i;
fc9c7b4f 325 int cpu;
984263bc 326
2b89bacf
MD
327 /*
328 * Make sure we can handle a reasonable number of fragments but
329 * cap it at 4000 (XXX).
330 */
331 mpipe_init(&ipq_mpipe, M_IPQ, sizeof(struct ipq),
fdec03d6 332 IFQ_MAXLEN, 4000, 0, NULL, NULL, NULL);
f8983475 333 for (i = 0; i < ncpus; ++i) {
1b562c24 334 TAILQ_INIT(&in_ifaddrheads[i]);
f8983475
SZ
335 in_ifaddrhashtbls[i] =
336 hashinit(INADDR_NHASH, M_IFADDR, &in_ifaddrhmask);
337 }
4468b0b4 338 pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
dffa46cd 339 if (pr == NULL)
984263bc
MD
340 panic("ip_init");
341 for (i = 0; i < IPPROTO_MAX; i++)
342 ip_protox[i] = pr - inetsw;
4468b0b4 343 for (pr = inetdomain.dom_protosw;
92db3805
SZ
344 pr < inetdomain.dom_protoswNPROTOSW; pr++) {
345 if (pr->pr_domain->dom_family == PF_INET && pr->pr_protocol) {
346 if (pr->pr_protocol != IPPROTO_RAW)
347 ip_protox[pr->pr_protocol] = pr - inetsw;
92db3805
SZ
348 }
349 }
984263bc 350
e7e55f42
JR
351 inet_pfil_hook.ph_type = PFIL_TYPE_AF;
352 inet_pfil_hook.ph_af = AF_INET;
5e3f3b7a 353 if ((i = pfil_head_register(&inet_pfil_hook)) != 0) {
a6ec04bc 354 kprintf("%s: WARNING: unable to register pfil hook, "
e7e55f42 355 "error %d\n", __func__, i);
5e3f3b7a 356 }
e7e55f42 357
984263bc 358 for (i = 0; i < IPREASS_NHASH; i++)
83c38804 359 TAILQ_INIT(&ipq[i]);
984263bc
MD
360
361 maxnipq = nmbclusters / 32;
362 maxfragsperpacket = 16;
363
984263bc 364 ip_id = time_second & 0xffff;
984263bc 365
fc9c7b4f 366 /*
ae1d3076 367 * Initialize IP statistics counters for each CPU.
fc9c7b4f 368 *
fc9c7b4f 369 */
fc9c7b4f 370 for (cpu = 0; cpu < ncpus; ++cpu) {
ae1d3076 371 bzero(&ipstats_percpu[cpu], sizeof(struct ip_stats));
fc9c7b4f 372 }
fc9c7b4f 373
c3c96e44 374 netisr_register(NETISR_IP, ip_input_handler, ip_cpufn_in);
e6f77b88 375 netisr_register_hashcheck(NETISR_IP, ip_hashcheck);
984263bc
MD
376}
377
0c7ac0cd
JH
378/* Do transport protocol processing. */
379static void
5de23090 380transport_processing_oncpu(struct mbuf *m, int hlen, struct ip *ip)
0c7ac0cd 381{
92db3805
SZ
382 const struct protosw *pr = &inetsw[ip_protox[ip->ip_p]];
383
0c7ac0cd
JH
384 /*
385 * Switch out to protocol's input routine.
386 */
92db3805 387 PR_GET_MPLOCK(pr);
002c1265 388 pr->pr_input(&m, &hlen, ip->ip_p);
92db3805 389 PR_REL_MPLOCK(pr);
0c7ac0cd
JH
390}
391
4599cf19 392static void
002c1265 393transport_processing_handler(netmsg_t msg)
0c7ac0cd 394{
002c1265 395 struct netmsg_packet *pmsg = &msg->packet;
0c7ac0cd 396 struct ip *ip;
3d6057fe
SZ
397 int hlen;
398
399 ip = mtod(pmsg->nm_packet, struct ip *);
002c1265 400 hlen = pmsg->base.lmsg.u.ms_result;
0c7ac0cd 401
3d6057fe 402 transport_processing_oncpu(pmsg->nm_packet, hlen, ip);
002c1265 403 /* msg was embedded in the mbuf, do not reply! */
0c7ac0cd
JH
404}
405
4599cf19 406static void
002c1265 407ip_input_handler(netmsg_t msg)
4f277347 408{
002c1265
MD
409 ip_input(msg->packet.nm_packet);
410 /* msg was embedded in the mbuf, do not reply! */
4f277347
JH
411}
412
984263bc 413/*
590b8cd4 414 * IP input routine. Checksum and byte swap header. If fragmented
984263bc
MD
415 * try to reassemble. Process options. Pass to next level.
416 */
417void
4f277347 418ip_input(struct mbuf *m)
984263bc
MD
419{
420 struct ip *ip;
984263bc 421 struct in_ifaddr *ia = NULL;
f8983475 422 struct in_ifaddr_container *iac;
8c6081b9 423 int hlen, checkif;
984263bc
MD
424 u_short sum;
425 struct in_addr pkt_dst;
dffa46cd 426 boolean_t using_srcrt = FALSE; /* forward (by PFIL_HOOKS) */
e7e55f42 427 struct in_addr odst; /* original dst address(NAT) */
984263bc 428 struct m_tag *mtag;
5de23090 429 struct sockaddr_in *next_hop = NULL;
6a704092 430 lwkt_port_t port;
e5ecc832 431#ifdef FAST_IPSEC
984263bc
MD
432 struct tdb_ident *tdbi;
433 struct secpolicy *sp;
1cae611f 434 int error;
dffa46cd 435#endif
984263bc 436
814907cb 437 M_ASSERTPKTHDR(m);
984263bc 438
6a704092 439 /*
c3c96e44
MD
440 * This routine is called from numerous places which may not have
441 * characterized the packet.
6a704092 442 */
c3c96e44 443 if ((m->m_flags & M_HASH) == 0) {
fce0e0b4 444 atomic_add_long(&ip_hash_count, 1);
c3c96e44
MD
445 ip_cpufn(&m, 0, IP_MPORT_IN);
446 if (m == NULL)
447 return;
448 KKASSERT(m->m_flags & M_HASH);
449 }
6a704092
MD
450 ip = mtod(m, struct ip *);
451
452 /*
453 * Pull out certain tags
454 */
5de23090
SZ
455 if (m->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) {
456 /* Next hop */
457 mtag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
458 KKASSERT(mtag != NULL);
459 next_hop = m_tag_data(mtag);
460 }
461
eb241549 462 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
eb241549 463 /* dummynet already filtered us */
984263bc
MD
464 ip = mtod(m, struct ip *);
465 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
dffa46cd 466 goto iphack;
984263bc
MD
467 }
468
469 ipstat.ips_total++;
470
c3c96e44 471 /* length checks already done in ip_cpufn() */
47d96de7 472 KASSERT(m->m_len >= sizeof(struct ip), ("IP header not in one mbuf"));
984263bc
MD
473
474 if (IP_VHL_V(ip->ip_vhl) != IPVERSION) {
475 ipstat.ips_badvers++;
476 goto bad;
477 }
478
479 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
c3c96e44 480 /* length checks already done in ip_cpufn() */
9babcab8 481 KASSERT(hlen >= sizeof(struct ip), ("IP header len too small"));
9b161cc2 482 KASSERT(m->m_len >= hlen, ("complete IP header not in one mbuf"));
984263bc
MD
483
484 /* 127/8 must not appear on wire - RFC1122 */
485 if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
486 (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
dffa46cd 487 if (!(m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK)) {
984263bc
MD
488 ipstat.ips_badaddr++;
489 goto bad;
490 }
491 }
492
493 if (m->m_pkthdr.csum_flags & CSUM_IP_CHECKED) {
494 sum = !(m->m_pkthdr.csum_flags & CSUM_IP_VALID);
495 } else {
459837b1 496 if (hlen == sizeof(struct ip))
984263bc 497 sum = in_cksum_hdr(ip);
459837b1 498 else
984263bc 499 sum = in_cksum(m, hlen);
984263bc 500 }
5fe66e68 501 if (sum != 0) {
984263bc
MD
502 ipstat.ips_badsum++;
503 goto bad;
504 }
505
4d723e5a
JS
506#ifdef ALTQ
507 if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) {
508 /* packet is dropped by traffic conditioner */
509 return;
510 }
511#endif
984263bc
MD
512 /*
513 * Convert fields to host representation.
514 */
515 ip->ip_len = ntohs(ip->ip_len);
984263bc
MD
516 ip->ip_off = ntohs(ip->ip_off);
517
c3c96e44 518 /* length checks already done in ip_cpufn() */
9b161cc2
SZ
519 KASSERT(ip->ip_len >= hlen, ("total length less then header length"));
520 KASSERT(m->m_pkthdr.len >= ip->ip_len, ("mbuf too short"));
521
984263bc 522 /*
9b161cc2 523 * Trim mbufs if longer than the IP header would have us expect.
984263bc 524 */
984263bc
MD
525 if (m->m_pkthdr.len > ip->ip_len) {
526 if (m->m_len == m->m_pkthdr.len) {
527 m->m_len = ip->ip_len;
528 m->m_pkthdr.len = ip->ip_len;
459837b1 529 } else {
984263bc 530 m_adj(m, ip->ip_len - m->m_pkthdr.len);
459837b1 531 }
984263bc
MD
532 }
533#if defined(IPSEC) && !defined(IPSEC_FILTERGIF)
534 /*
535 * Bypass packet filtering for packets from a tunnel (gif).
536 */
537 if (ipsec_gethist(m, NULL))
538 goto pass;
539#endif
540
541 /*
542 * IpHack's section.
543 * Right now when no processing on packet has done
544 * and it is still fresh out of network we do our black
545 * deals with it.
546 * - Firewall: deny/allow/divert
547 * - Xlate: translate packet's addr/port (NAT).
548 * - Pipe: pass pkt through dummynet.
549 * - Wrap: fake packet's addr/port <unimpl.>
550 * - Encapsulate: put it in another IP and send out. <unimp.>
dffa46cd 551 */
984263bc
MD
552
553iphack:
a5d1fda3
SZ
554 /*
555 * If we've been forwarded from the output side, then
556 * skip the firewall a second time
557 */
558 if (next_hop != NULL)
559 goto ours;
dffa46cd 560
a93c9c2f
SZ
561 /* No pfil hooks */
562 if (!pfil_has_hooks(&inet_pfil_hook)) {
563 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
564 /*
565 * Strip dummynet tags from stranded packets
566 */
567 mtag = m_tag_find(m, PACKET_TAG_DUMMYNET, NULL);
568 KKASSERT(mtag != NULL);
569 m_tag_delete(m, mtag);
570 m->m_pkthdr.fw_flags &= ~DUMMYNET_MBUF_TAGGED;
571 }
572 goto pass;
573 }
574
984263bc 575 /*
e7e55f42
JR
576 * Run through list of hooks for input packets.
577 *
c3c96e44
MD
578 * NOTE! If the packet is rewritten pf/ipfw/whoever must
579 * clear M_HASH.
984263bc 580 */
a93c9c2f
SZ
581 odst = ip->ip_dst;
582 if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, PFIL_IN))
583 return;
584 if (m == NULL) /* consumed by filter */
585 return;
586 ip = mtod(m, struct ip *);
587 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
588 using_srcrt = (odst.s_addr != ip->ip_dst.s_addr);
dffa46cd 589
a5d1fda3
SZ
590 if (m->m_pkthdr.fw_flags & IPFORWARD_MBUF_TAGGED) {
591 mtag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
592 KKASSERT(mtag != NULL);
593 next_hop = m_tag_data(mtag);
594 }
e4d4f9c3
SZ
595 if (m->m_pkthdr.fw_flags & DUMMYNET_MBUF_TAGGED) {
596 ip_dn_queue(m);
597 return;
598 }
012d335d 599 if (m->m_pkthdr.fw_flags & FW_MBUF_REDISPATCH) {
012d335d
SZ
600 m->m_pkthdr.fw_flags &= ~FW_MBUF_REDISPATCH;
601 }
8c6081b9 602pass:
984263bc
MD
603 /*
604 * Process options and, if not destined for us,
605 * ship it on. ip_dooptions returns 1 when an
606 * error was detected (causing an icmp message
607 * to be sent and the original packet to be freed).
608 */
5de23090 609 if (hlen > sizeof(struct ip) && ip_dooptions(m, 0, next_hop))
984263bc
MD
610 return;
611
dffa46cd
JH
612 /* greedy RSVP, snatches any PATH packet of the RSVP protocol and no
613 * matter if it is destined to another node, or whether it is
614 * a multicast one, RSVP wants it! and prevents it from being forwarded
615 * anywhere else. Also checks if the rsvp daemon is running before
984263bc 616 * grabbing the packet.
dffa46cd
JH
617 */
618 if (rsvp_on && ip->ip_p == IPPROTO_RSVP)
984263bc
MD
619 goto ours;
620
621 /*
622 * Check our list of addresses, to see if the packet is for us.
623 * If we don't have any addresses, assume any unicast packet
624 * we receive might be for us (and let the upper layers deal
625 * with it).
626 */
1b562c24
SZ
627 if (TAILQ_EMPTY(&in_ifaddrheads[mycpuid]) &&
628 !(m->m_flags & (M_MCAST | M_BCAST)))
984263bc
MD
629 goto ours;
630
631 /*
632 * Cache the destination address of the packet; this may be
633 * changed by use of 'ipfw fwd'.
634 */
5de23090 635 pkt_dst = next_hop ? next_hop->sin_addr : ip->ip_dst;
984263bc
MD
636
637 /*
638 * Enable a consistency check between the destination address
639 * and the arrival interface for a unicast packet (the RFC 1122
640 * strong ES model) if IP forwarding is disabled and the packet
641 * is not locally generated and the packet is not subject to
642 * 'ipfw fwd'.
643 *
644 * XXX - Checking also should be disabled if the destination
645 * address is ipnat'ed to a different interface.
646 *
647 * XXX - Checking is incompatible with IP aliases added
648 * to the loopback interface instead of the interface where
649 * the packets are received.
650 */
dffa46cd
JH
651 checkif = ip_checkinterface &&
652 !ipforwarding &&
653 m->m_pkthdr.rcvif != NULL &&
654 !(m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) &&
5de23090 655 next_hop == NULL;
984263bc
MD
656
657 /*
658 * Check for exact addresses in the hash bucket.
659 */
f8983475
SZ
660 LIST_FOREACH(iac, INADDR_HASH(pkt_dst.s_addr), ia_hash) {
661 ia = iac->ia;
662
984263bc
MD
663 /*
664 * If the address matches, verify that the packet
665 * arrived via the correct interface if checking is
666 * enabled.
667 */
dffa46cd 668 if (IA_SIN(ia)->sin_addr.s_addr == pkt_dst.s_addr &&
984263bc
MD
669 (!checkif || ia->ia_ifp == m->m_pkthdr.rcvif))
670 goto ours;
671 }
f8983475
SZ
672 ia = NULL;
673
984263bc
MD
674 /*
675 * Check for broadcast addresses.
676 *
677 * Only accept broadcast packets that arrive via the matching
678 * interface. Reception of forwarded directed broadcasts would
679 * be handled via ip_forward() and ether_output() with the loopback
680 * into the stack for SIMPLEX interfaces handled by ether_output().
681 */
682 if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) {
b2632176
SZ
683 struct ifaddr_container *ifac;
684
685 TAILQ_FOREACH(ifac, &m->m_pkthdr.rcvif->if_addrheads[mycpuid],
686 ifa_link) {
687 struct ifaddr *ifa = ifac->ifa;
688
83a415f4
MD
689 if (ifa->ifa_addr == NULL) /* shutdown/startup race */
690 continue;
984263bc
MD
691 if (ifa->ifa_addr->sa_family != AF_INET)
692 continue;
693 ia = ifatoia(ifa);
694 if (satosin(&ia->ia_broadaddr)->sin_addr.s_addr ==
dffa46cd 695 pkt_dst.s_addr)
984263bc
MD
696 goto ours;
697 if (ia->ia_netbroadcast.s_addr == pkt_dst.s_addr)
698 goto ours;
699#ifdef BOOTP_COMPAT
700 if (IA_SIN(ia)->sin_addr.s_addr == INADDR_ANY)
701 goto ours;
702#endif
703 }
704 }
705 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
706 struct in_multi *inm;
5fe66e68 707
87b66be9
SZ
708 /* XXX Multicast is not MPSAFE yet */
709 get_mplock();
710
5fe66e68 711 if (ip_mrouter != NULL) {
984263bc
MD
712 /*
713 * If we are acting as a multicast router, all
714 * incoming multicast packets are passed to the
715 * kernel-level multicast forwarding function.
716 * The packet is returned (relatively) intact; if
717 * ip_mforward() returns a non-zero value, the packet
718 * must be discarded, else it may be accepted below.
719 */
5fe66e68 720 if (ip_mforward != NULL &&
dffa46cd 721 ip_mforward(ip, m->m_pkthdr.rcvif, m, NULL) != 0) {
87b66be9 722 rel_mplock();
984263bc
MD
723 ipstat.ips_cantforward++;
724 m_freem(m);
725 return;
726 }
727
728 /*
729 * The process-level routing daemon needs to receive
730 * all multicast IGMP packets, whether or not this
731 * host belongs to their destination groups.
732 */
87b66be9
SZ
733 if (ip->ip_p == IPPROTO_IGMP) {
734 rel_mplock();
984263bc 735 goto ours;
87b66be9 736 }
984263bc
MD
737 ipstat.ips_forward++;
738 }
739 /*
740 * See if we belong to the destination multicast group on the
741 * arrival interface.
742 */
743 IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm);
744 if (inm == NULL) {
87b66be9 745 rel_mplock();
984263bc
MD
746 ipstat.ips_notmember++;
747 m_freem(m);
748 return;
749 }
87b66be9
SZ
750
751 rel_mplock();
984263bc
MD
752 goto ours;
753 }
dffa46cd 754 if (ip->ip_dst.s_addr == INADDR_BROADCAST)
984263bc
MD
755 goto ours;
756 if (ip->ip_dst.s_addr == INADDR_ANY)
757 goto ours;
758
759 /*
760 * FAITH(Firewall Aided Internet Translator)
761 */
762 if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type == IFT_FAITH) {
763 if (ip_keepfaith) {
dffa46cd 764 if (ip->ip_p == IPPROTO_TCP || ip->ip_p == IPPROTO_ICMP)
984263bc
MD
765 goto ours;
766 }
767 m_freem(m);
768 return;
769 }
770
771 /*
772 * Not for us; forward if possible and desirable.
773 */
dffa46cd 774 if (!ipforwarding) {
984263bc
MD
775 ipstat.ips_cantforward++;
776 m_freem(m);
777 } else {
778#ifdef IPSEC
779 /*
780 * Enforce inbound IPsec SPD.
781 */
782 if (ipsec4_in_reject(m, NULL)) {
783 ipsecstat.in_polvio++;
784 goto bad;
785 }
dffa46cd 786#endif
984263bc
MD
787#ifdef FAST_IPSEC
788 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1cae611f 789 crit_enter();
984263bc 790 if (mtag != NULL) {
d031aa80 791 tdbi = (struct tdb_ident *)m_tag_data(mtag);
984263bc
MD
792 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
793 } else {
794 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
dffa46cd 795 IP_FORWARDING, &error);
984263bc
MD
796 }
797 if (sp == NULL) { /* NB: can happen if error */
1cae611f 798 crit_exit();
984263bc
MD
799 /*XXX error stat???*/
800 DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/
801 goto bad;
802 }
803
804 /*
805 * Check security policy against packet attributes.
806 */
807 error = ipsec_in_reject(sp, m);
808 KEY_FREESP(&sp);
1cae611f 809 crit_exit();
984263bc
MD
810 if (error) {
811 ipstat.ips_cantforward++;
812 goto bad;
813 }
dffa46cd 814#endif
5de23090 815 ip_forward(m, using_srcrt, next_hop);
984263bc
MD
816 }
817 return;
818
819ours:
dffa46cd 820
984263bc
MD
821 /*
822 * IPSTEALTH: Process non-routing options only
823 * if the packet is destined for us.
824 */
dffa46cd
JH
825 if (ipstealth &&
826 hlen > sizeof(struct ip) &&
5de23090 827 ip_dooptions(m, 1, next_hop))
984263bc 828 return;
984263bc
MD
829
830 /* Count the packet in the ip address stats */
831 if (ia != NULL) {
832 ia->ia_ifa.if_ipackets++;
833 ia->ia_ifa.if_ibytes += m->m_pkthdr.len;
834 }
835
836 /*
837 * If offset or IP_MF are set, must reassemble.
838 * Otherwise, nothing need be done.
839 * (We could look in the reassembly queue to see
840 * if the packet was previously fragmented,
841 * but it's not worth the time; just let them time out.)
842 */
843 if (ip->ip_off & (IP_MF | IP_OFFMASK)) {
984263bc 844 /*
c3c96e44
MD
845 * Attempt reassembly; if it succeeds, proceed. ip_reass()
846 * will return a different mbuf.
847 *
848 * NOTE: ip_reass() returns m with M_HASH cleared to force
849 * us to recharacterize the packet.
984263bc 850 */
b0cb7c2c 851 m = ip_reass(m);
dffa46cd 852 if (m == NULL)
984263bc 853 return;
984263bc 854 ip = mtod(m, struct ip *);
b0cb7c2c 855
984263bc
MD
856 /* Get the header length of the reassembled packet */
857 hlen = IP_VHL_HL(ip->ip_vhl) << 2;
bf82f9b7 858 } else {
984263bc 859 ip->ip_len -= hlen;
bf82f9b7 860 }
984263bc 861
984263bc
MD
862#ifdef IPSEC
863 /*
864 * enforce IPsec policy checking if we are seeing last header.
865 * note that we do not visit this with protocols with pcb layer
866 * code - like udp/tcp/raw ip.
867 */
dffa46cd 868 if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) &&
984263bc
MD
869 ipsec4_in_reject(m, NULL)) {
870 ipsecstat.in_polvio++;
871 goto bad;
872 }
873#endif
874#if FAST_IPSEC
875 /*
876 * enforce IPsec policy checking if we are seeing last header.
877 * note that we do not visit this with protocols with pcb layer
878 * code - like udp/tcp/raw ip.
879 */
dffa46cd 880 if (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) {
984263bc
MD
881 /*
882 * Check if the packet has already had IPsec processing
883 * done. If so, then just pass it along. This tag gets
884 * set during AH, ESP, etc. input handling, before the
885 * packet is returned to the ip input queue for delivery.
dffa46cd 886 */
984263bc 887 mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL);
1cae611f 888 crit_enter();
984263bc 889 if (mtag != NULL) {
d031aa80 890 tdbi = (struct tdb_ident *)m_tag_data(mtag);
984263bc
MD
891 sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND);
892 } else {
893 sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND,
dffa46cd 894 IP_FORWARDING, &error);
984263bc
MD
895 }
896 if (sp != NULL) {
897 /*
898 * Check security policy against packet attributes.
899 */
900 error = ipsec_in_reject(sp, m);
901 KEY_FREESP(&sp);
902 } else {
903 /* XXX error stat??? */
904 error = EINVAL;
905DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/
906 goto bad;
907 }
1cae611f 908 crit_exit();
984263bc
MD
909 if (error)
910 goto bad;
911 }
912#endif /* FAST_IPSEC */
913
a40c0023 914 /*
6a704092
MD
915 * We must forward the packet to the correct protocol thread if
916 * we are not already in it.
c3c96e44
MD
917 *
918 * NOTE: ip_len is now in host form. ip_len is not adjusted
919 * further for protocol processing, instead we pass hlen
920 * to the protosw and let it deal with it.
a40c0023 921 */
984263bc 922 ipstat.ips_delivered++;
984263bc 923
c3c96e44 924 if ((m->m_flags & M_HASH) == 0) {
fce0e0b4
SZ
925#ifdef RSS_DEBUG
926 atomic_add_long(&ip_rehash_count, 1);
927#endif
76c4458a 928 ip->ip_len = htons(ip->ip_len + hlen);
c3c96e44
MD
929 ip->ip_off = htons(ip->ip_off);
930
931 ip_cpufn(&m, 0, IP_MPORT_IN);
932 if (m == NULL)
933 return;
934
935 ip = mtod(m, struct ip *);
76c4458a 936 ip->ip_len = ntohs(ip->ip_len) - hlen;
c3c96e44
MD
937 ip->ip_off = ntohs(ip->ip_off);
938 KKASSERT(m->m_flags & M_HASH);
939 }
3abced87 940 port = netisr_portfn(m->m_pkthdr.hash);
c3c96e44 941
6a704092
MD
942 if (port != &curthread->td_msgport) {
943 struct netmsg_packet *pmsg;
35658e20 944
fce0e0b4
SZ
945#ifdef RSS_DEBUG
946 atomic_add_long(&ip_dispatch_slow, 1);
947#endif
c3c96e44 948
3d6057fe 949 pmsg = &m->m_hdr.mh_netmsg;
002c1265 950 netmsg_init(&pmsg->base, NULL, &netisr_apanic_rport,
c3c96e44 951 0, transport_processing_handler);
3d6057fe 952 pmsg->nm_packet = m;
002c1265
MD
953 pmsg->base.lmsg.u.ms_result = hlen;
954 lwkt_sendmsg(port, &pmsg->base.lmsg);
0c7ac0cd 955 } else {
fce0e0b4
SZ
956#ifdef RSS_DEBUG
957 atomic_add_long(&ip_dispatch_fast, 1);
958#endif
5de23090 959 transport_processing_oncpu(m, hlen, ip);
bf82f9b7 960 }
984263bc 961 return;
a80cf23b 962
984263bc
MD
963bad:
964 m_freem(m);
965}
966
984263bc
MD
967/*
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.
984263bc 971 */
0030d31a 972struct mbuf *
b0cb7c2c 973ip_reass(struct mbuf *m)
984263bc
MD
974{
975 struct ip *ip = mtod(m, struct ip *);
dffa46cd 976 struct mbuf *p = NULL, *q, *nq;
2b89bacf 977 struct mbuf *n;
9691feb6 978 struct ipq *fp = NULL;
83c38804 979 struct ipqhead *head;
984263bc
MD
980 int hlen = IP_VHL_HL(ip->ip_vhl) << 2;
981 int i, next;
9691feb6 982 u_short sum;
984263bc 983
9691feb6
SZ
984 /* If maxnipq is 0, never accept fragments. */
985 if (maxnipq == 0) {
986 ipstat.ips_fragments++;
987 ipstat.ips_fragdropped++;
988 m_freem(m);
989 return NULL;
990 }
991
992 sum = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id);
993 /*
994 * Look for queue of fragments of this datagram.
995 */
2d23a8be 996 lwkt_gettoken(&ipq_token);
83c38804
PA
997 head = &ipq[sum];
998 TAILQ_FOREACH(fp, head, ipq_list) {
9691feb6
SZ
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)
1003 goto found;
2d23a8be 1004 }
9691feb6
SZ
1005
1006 fp = NULL;
1007
1008 /*
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.
1012 */
1013 if (nipq > maxnipq && maxnipq > 0) {
1014 /*
1015 * drop something from the tail of the current queue
1016 * before proceeding further
1017 */
83c38804
PA
1018 struct ipq *q = TAILQ_LAST(head, ipqhead);
1019 if (q == NULL) {
1020 /*
1021 * The current queue is empty,
1022 * so drop from one of the others.
1023 */
9691feb6 1024 for (i = 0; i < IPREASS_NHASH; i++) {
83c38804
PA
1025 struct ipq *r = TAILQ_LAST(&ipq[i], ipqhead);
1026 if (r) {
1027 ipstat.ips_fragtimeout += r->ipq_nfrags;
1028 ip_freef(&ipq[i], r);
9691feb6
SZ
1029 break;
1030 }
1031 }
1032 } else {
83c38804
PA
1033 ipstat.ips_fragtimeout += q->ipq_nfrags;
1034 ip_freef(head, q);
9691feb6
SZ
1035 }
1036 }
1037found:
1038 /*
1039 * Adjust ip_len to not reflect header,
1040 * convert offset of this to bytes.
1041 */
1042 ip->ip_len -= hlen;
1043 if (ip->ip_off & IP_MF) {
1044 /*
1045 * Make sure that fragments have a data length
1046 * that's a non-zero multiple of 8 bytes.
1047 */
1048 if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) {
1049 ipstat.ips_toosmall++; /* XXX */
1050 m_freem(m);
2d23a8be 1051 goto done;
9691feb6
SZ
1052 }
1053 m->m_flags |= M_FRAG;
2d23a8be 1054 } else {
9691feb6 1055 m->m_flags &= ~M_FRAG;
2d23a8be 1056 }
9691feb6
SZ
1057 ip->ip_off <<= 3;
1058
1059 ipstat.ips_fragments++;
1060 m->m_pkthdr.header = ip;
1061
fbb35ef0
SZ
1062 /*
1063 * If the hardware has not done csum over this fragment
1064 * then csum_data is not valid at all.
1065 */
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);
1070 }
1071
984263bc
MD
1072 /*
1073 * Presence of header sizes in mbufs
1074 * would confuse code below.
1075 */
1076 m->m_data += hlen;
1077 m->m_len -= hlen;
1078
1079 /*
1080 * If first fragment to arrive, create a reassembly queue.
1081 */
dffa46cd 1082 if (fp == NULL) {
2b89bacf 1083 if ((fp = mpipe_alloc_nowait(&ipq_mpipe)) == NULL)
984263bc 1084 goto dropfrag;
83c38804 1085 TAILQ_INSERT_HEAD(head, fp, ipq_list);
984263bc
MD
1086 nipq++;
1087 fp->ipq_nfrags = 1;
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;
1093 fp->ipq_frags = m;
1094 m->m_nextpkt = NULL;
984263bc
MD
1095 goto inserted;
1096 } else {
1097 fp->ipq_nfrags++;
1098 }
1099
dffa46cd 1100#define GETIP(m) ((struct ip*)((m)->m_pkthdr.header))
984263bc
MD
1101
1102 /*
1103 * Find a segment which begins after this one does.
1104 */
2d23a8be 1105 for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt) {
984263bc
MD
1106 if (GETIP(q)->ip_off > ip->ip_off)
1107 break;
2d23a8be 1108 }
984263bc
MD
1109
1110 /*
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.
1115 *
1116 * If some of the data is dropped from the the preceding
1117 * segment, then it's checksum is invalidated.
1118 */
1119 if (p) {
1120 i = GETIP(p)->ip_off + GETIP(p)->ip_len - ip->ip_off;
1121 if (i > 0) {
1122 if (i >= ip->ip_len)
1123 goto dropfrag;
1124 m_adj(m, i);
1125 m->m_pkthdr.csum_flags = 0;
1126 ip->ip_off += i;
1127 ip->ip_len -= i;
1128 }
1129 m->m_nextpkt = p->m_nextpkt;
1130 p->m_nextpkt = m;
1131 } else {
1132 m->m_nextpkt = fp->ipq_frags;
1133 fp->ipq_frags = m;
1134 }
1135
1136 /*
1137 * While we overlap succeeding segments trim them or,
1138 * if they are completely covered, dequeue them.
1139 */
1140 for (; q != NULL && ip->ip_off + ip->ip_len > GETIP(q)->ip_off;
1141 q = nq) {
dffa46cd 1142 i = (ip->ip_off + ip->ip_len) - GETIP(q)->ip_off;
984263bc
MD
1143 if (i < GETIP(q)->ip_len) {
1144 GETIP(q)->ip_len -= i;
1145 GETIP(q)->ip_off += i;
1146 m_adj(q, i);
1147 q->m_pkthdr.csum_flags = 0;
1148 break;
1149 }
1150 nq = q->m_nextpkt;
1151 m->m_nextpkt = nq;
1152 ipstat.ips_fragdropped++;
1153 fp->ipq_nfrags--;
c4933956 1154 q->m_nextpkt = NULL;
984263bc
MD
1155 m_freem(q);
1156 }
1157
1158inserted:
984263bc
MD
1159 /*
1160 * Check for complete reassembly and perform frag per packet
1161 * limiting.
1162 *
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.)
1167 *
1168 */
1169 next = 0;
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;
83c38804 1174 ip_freef(head, fp);
984263bc 1175 }
2d23a8be 1176 goto done;
984263bc
MD
1177 }
1178 next += GETIP(q)->ip_len;
1179 }
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;
83c38804 1184 ip_freef(head, fp);
984263bc 1185 }
2d23a8be 1186 goto done;
984263bc
MD
1187 }
1188
1189 /*
1190 * Reassembly is complete. Make sure the packet is a sane size.
1191 */
1192 q = fp->ipq_frags;
1193 ip = GETIP(q);
1194 if (next + (IP_VHL_HL(ip->ip_vhl) << 2) > IP_MAXPACKET) {
1195 ipstat.ips_toolong++;
1196 ipstat.ips_fragdropped += fp->ipq_nfrags;
83c38804 1197 ip_freef(head, fp);
2d23a8be 1198 goto done;
984263bc
MD
1199 }
1200
1201 /*
1202 * Concatenate fragments.
1203 */
1204 m = q;
2b89bacf 1205 n = m->m_next;
dffa46cd 1206 m->m_next = NULL;
2b89bacf 1207 m_cat(m, n);
984263bc 1208 nq = q->m_nextpkt;
dffa46cd 1209 q->m_nextpkt = NULL;
984263bc
MD
1210 for (q = nq; q != NULL; q = nq) {
1211 nq = q->m_nextpkt;
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;
1215 m_cat(m, q);
1216 }
1217
1e7aea7f
MD
1218 /*
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
37ed010a 1221 * were reassembled. A second carry can also occur (but not a third).
1e7aea7f
MD
1222 */
1223 m->m_pkthdr.csum_data = (m->m_pkthdr.csum_data & 0xffff) +
1224 (m->m_pkthdr.csum_data >> 16);
37ed010a
MD
1225 if (m->m_pkthdr.csum_data > 0xFFFF)
1226 m->m_pkthdr.csum_data -= 0xFFFF;
1e7aea7f 1227
984263bc
MD
1228 /*
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.
1233 */
1234 ip->ip_len = next;
1235 ip->ip_src = fp->ipq_src;
1236 ip->ip_dst = fp->ipq_dst;
83c38804 1237 TAILQ_REMOVE(head, fp, ipq_list);
984263bc 1238 nipq--;
2b89bacf 1239 mpipe_free(&ipq_mpipe, fp);
984263bc
MD
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 */
2256ba69 1244 int plen = 0;
dffa46cd 1245
2b89bacf
MD
1246 for (n = m; n; n = n->m_next)
1247 plen += n->m_len;
984263bc
MD
1248 m->m_pkthdr.len = plen;
1249 }
d2f6f5b2 1250
c3c96e44
MD
1251 /*
1252 * Reassembly complete, return the next protocol.
1253 *
1254 * Be sure to clear M_HASH to force the packet
1255 * to be re-characterized.
1256 *
1257 * Clear M_FRAG, we are no longer a fragment.
1258 */
1259 m->m_flags &= ~(M_HASH | M_FRAG);
1260
d2f6f5b2 1261 ipstat.ips_reassembled++;
2d23a8be 1262 lwkt_reltoken(&ipq_token);
984263bc
MD
1263 return (m);
1264
1265dropfrag:
984263bc 1266 ipstat.ips_fragdropped++;
dffa46cd 1267 if (fp != NULL)
984263bc
MD
1268 fp->ipq_nfrags--;
1269 m_freem(m);
2d23a8be
MD
1270done:
1271 lwkt_reltoken(&ipq_token);
dffa46cd 1272 return (NULL);
984263bc
MD
1273
1274#undef GETIP
1275}
1276
1277/*
1278 * Free a fragment reassembly header and all
1279 * associated datagrams.
2d23a8be
MD
1280 *
1281 * Called with ipq_token held.
984263bc
MD
1282 */
1283static void
83c38804 1284ip_freef(struct ipqhead *fhp, struct ipq *fp)
984263bc 1285{
2256ba69 1286 struct mbuf *q;
984263bc 1287
2d23a8be
MD
1288 /*
1289 * Remove first to protect against blocking
1290 */
83c38804 1291 TAILQ_REMOVE(fhp, fp, ipq_list);
2d23a8be
MD
1292
1293 /*
1294 * Clean out at our leisure
1295 */
984263bc
MD
1296 while (fp->ipq_frags) {
1297 q = fp->ipq_frags;
1298 fp->ipq_frags = q->m_nextpkt;
c4933956 1299 q->m_nextpkt = NULL;
984263bc
MD
1300 m_freem(q);
1301 }
2b89bacf 1302 mpipe_free(&ipq_mpipe, fp);
984263bc
MD
1303 nipq--;
1304}
1305
1306/*
1307 * IP timer processing;
1308 * if a timer expires on a reassembly
1309 * queue, discard it.
1310 */
1311void
8a3125c6 1312ip_slowtimo(void)
984263bc 1313{
83c38804
PA
1314 struct ipq *fp, *fp_temp;
1315 struct ipqhead *head;
984263bc
MD
1316 int i;
1317
2d23a8be 1318 lwkt_gettoken(&ipq_token);
984263bc 1319 for (i = 0; i < IPREASS_NHASH; i++) {
83c38804
PA
1320 head = &ipq[i];
1321 TAILQ_FOREACH_MUTABLE(fp, head, ipq_list, fp_temp) {
1322 if (--fp->ipq_ttl == 0) {
1323 ipstat.ips_fragtimeout += fp->ipq_nfrags;
1324 ip_freef(head, fp);
984263bc
MD
1325 }
1326 }
1327 }
1328 /*
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.
1332 */
1333 if (maxnipq >= 0 && nipq > maxnipq) {
1334 for (i = 0; i < IPREASS_NHASH; i++) {
83c38804
PA
1335 head = &ipq[i];
1336 while (nipq > maxnipq && !TAILQ_EMPTY(head)) {
984263bc 1337 ipstat.ips_fragdropped +=
83c38804
PA
1338 TAILQ_FIRST(head)->ipq_nfrags;
1339 ip_freef(head, TAILQ_FIRST(head));
984263bc
MD
1340 }
1341 }
1342 }
2d23a8be 1343 lwkt_reltoken(&ipq_token);
984263bc 1344 ipflow_slowtimo();
984263bc
MD
1345}
1346
1347/*
1348 * Drain off all datagram fragments.
1349 */
1350void
8a3125c6 1351ip_drain(void)
984263bc 1352{
83c38804 1353 struct ipqhead *head;
dffa46cd 1354 int i;
984263bc 1355
2d23a8be 1356 lwkt_gettoken(&ipq_token);
984263bc 1357 for (i = 0; i < IPREASS_NHASH; i++) {
83c38804
PA
1358 head = &ipq[i];
1359 while (!TAILQ_EMPTY(head)) {
1360 ipstat.ips_fragdropped += TAILQ_FIRST(head)->ipq_nfrags;
1361 ip_freef(head, TAILQ_FIRST(head));
984263bc
MD
1362 }
1363 }
2d23a8be 1364 lwkt_reltoken(&ipq_token);
984263bc
MD
1365 in_rtqdrain();
1366}
1367
1368/*
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.
1379 */
1380static int
1381ip_dooptions(struct mbuf *m, int pass, struct sockaddr_in *next_hop)
1382{
407e896e 1383 struct sockaddr_in ipaddr = { sizeof ipaddr, AF_INET };
984263bc
MD
1384 struct ip *ip = mtod(m, struct ip *);
1385 u_char *cp;
1386 struct in_ifaddr *ia;
dffa46cd
JH
1387 int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB;
1388 boolean_t forward = FALSE;
984263bc
MD
1389 struct in_addr *sin, dst;
1390 n_time ntime;
1391
1392 dst = ip->ip_dst;
1393 cp = (u_char *)(ip + 1);
dffa46cd 1394 cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
984263bc
MD
1395 for (; cnt > 0; cnt -= optlen, cp += optlen) {
1396 opt = cp[IPOPT_OPTVAL];
1397 if (opt == IPOPT_EOL)
1398 break;
1399 if (opt == IPOPT_NOP)
1400 optlen = 1;
1401 else {
1402 if (cnt < IPOPT_OLEN + sizeof(*cp)) {
1403 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1404 goto bad;
1405 }
1406 optlen = cp[IPOPT_OLEN];
1407 if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1408 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1409 goto bad;
1410 }
1411 }
1412 switch (opt) {
1413
1414 default:
1415 break;
1416
1417 /*
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.
1425 */
1426 case IPOPT_LSRR:
1427 case IPOPT_SSRR:
984263bc
MD
1428 if (ipstealth && pass > 0)
1429 break;
984263bc
MD
1430 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1431 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1432 goto bad;
1433 }
1434 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1435 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1436 goto bad;
1437 }
1438 ipaddr.sin_addr = ip->ip_dst;
1439 ia = (struct in_ifaddr *)
1440 ifa_ifwithaddr((struct sockaddr *)&ipaddr);
dffa46cd 1441 if (ia == NULL) {
984263bc
MD
1442 if (opt == IPOPT_SSRR) {
1443 type = ICMP_UNREACH;
1444 code = ICMP_UNREACH_SRCFAIL;
1445 goto bad;
1446 }
1447 if (!ip_dosourceroute)
1448 goto nosourcerouting;
1449 /*
1450 * Loose routing, and not at next destination
1451 * yet; nothing to do except forward.
1452 */
1453 break;
1454 }
1455 off--; /* 0 origin */
1456 if (off > optlen - (int)sizeof(struct in_addr)) {
1457 /*
1458 * End of source route. Should be for us.
1459 */
1460 if (!ip_acceptsourceroute)
1461 goto nosourcerouting;
7e31206a 1462 save_rte(m, cp, ip->ip_src);
984263bc
MD
1463 break;
1464 }
984263bc
MD
1465 if (ipstealth)
1466 goto dropit;
984263bc
MD
1467 if (!ip_dosourceroute) {
1468 if (ipforwarding) {
1141eb20
JH
1469 char buf[sizeof "aaa.bbb.ccc.ddd"];
1470
984263bc
MD
1471 /*
1472 * Acting as a router, so generate ICMP
1473 */
1474nosourcerouting:
1475 strcpy(buf, inet_ntoa(ip->ip_dst));
dffa46cd 1476 log(LOG_WARNING,
984263bc
MD
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;
1481 goto bad;
1482 } else {
1483 /*
dffa46cd
JH
1484 * Not acting as a router,
1485 * so silently drop.
984263bc 1486 */
984263bc 1487dropit:
984263bc
MD
1488 ipstat.ips_cantforward++;
1489 m_freem(m);
1490 return (1);
1491 }
1492 }
1493
1494 /*
1495 * locate outgoing interface
1496 */
f23061d4
JH
1497 memcpy(&ipaddr.sin_addr, cp + off,
1498 sizeof ipaddr.sin_addr);
984263bc
MD
1499
1500 if (opt == IPOPT_SSRR) {
1501#define INA struct in_ifaddr *
1502#define SA struct sockaddr *
dffa46cd
JH
1503 if ((ia = (INA)ifa_ifwithdstaddr((SA)&ipaddr))
1504 == NULL)
1505 ia = (INA)ifa_ifwithnet((SA)&ipaddr);
d27ef928
SZ
1506 } else {
1507 ia = ip_rtaddr(ipaddr.sin_addr, NULL);
1508 }
dffa46cd 1509 if (ia == NULL) {
984263bc
MD
1510 type = ICMP_UNREACH;
1511 code = ICMP_UNREACH_SRCFAIL;
1512 goto bad;
1513 }
1514 ip->ip_dst = ipaddr.sin_addr;
f23061d4 1515 memcpy(cp + off, &IA_SIN(ia)->sin_addr,
984263bc
MD
1516 sizeof(struct in_addr));
1517 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1518 /*
1519 * Let ip_intr's mcast routing check handle mcast pkts
1520 */
1521 forward = !IN_MULTICAST(ntohl(ip->ip_dst.s_addr));
1522 break;
1523
1524 case IPOPT_RR:
984263bc
MD
1525 if (ipstealth && pass == 0)
1526 break;
984263bc
MD
1527 if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1528 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1529 goto bad;
1530 }
1531 if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1532 code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1533 goto bad;
1534 }
1535 /*
1536 * If no space remains, ignore.
1537 */
1538 off--; /* 0 origin */
1539 if (off > optlen - (int)sizeof(struct in_addr))
1540 break;
f23061d4
JH
1541 memcpy(&ipaddr.sin_addr, &ip->ip_dst,
1542 sizeof ipaddr.sin_addr);
984263bc
MD
1543 /*
1544 * locate outgoing interface; if we're the destination,
1545 * use the incoming interface (should be same).
1546 */
dffa46cd 1547 if ((ia = (INA)ifa_ifwithaddr((SA)&ipaddr)) == NULL &&
d27ef928 1548 (ia = ip_rtaddr(ipaddr.sin_addr, NULL)) == NULL) {
984263bc
MD
1549 type = ICMP_UNREACH;
1550 code = ICMP_UNREACH_HOST;
1551 goto bad;
1552 }
f23061d4 1553 memcpy(cp + off, &IA_SIN(ia)->sin_addr,
984263bc
MD
1554 sizeof(struct in_addr));
1555 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1556 break;
1557
1558 case IPOPT_TS:
984263bc
MD
1559 if (ipstealth && pass == 0)
1560 break;
984263bc
MD
1561 code = cp - (u_char *)ip;
1562 if (optlen < 4 || optlen > 40) {
1563 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1564 goto bad;
1565 }
1566 if ((off = cp[IPOPT_OFFSET]) < 5) {
1567 code = &cp[IPOPT_OLEN] - (u_char *)ip;
1568 goto bad;
1569 }
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;
1574 goto bad;
1575 }
1576 break;
1577 }
1578 off--; /* 0 origin */
1579 sin = (struct in_addr *)(cp + off);
1580 switch (cp[IPOPT_OFFSET + 1] & 0x0f) {
1581
1582 case IPOPT_TS_TSONLY:
1583 break;
1584
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;
1589 goto bad;
1590 }
1591 ipaddr.sin_addr = dst;
1592 ia = (INA)ifaof_ifpforaddr((SA)&ipaddr,
1593 m->m_pkthdr.rcvif);
dffa46cd 1594 if (ia == NULL)
984263bc 1595 continue;
f23061d4 1596 memcpy(sin, &IA_SIN(ia)->sin_addr,
984263bc
MD
1597 sizeof(struct in_addr));
1598 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1599 off += sizeof(struct in_addr);
1600 break;
1601
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;
1606 goto bad;
1607 }
f23061d4 1608 memcpy(&ipaddr.sin_addr, sin,
984263bc 1609 sizeof(struct in_addr));
dffa46cd 1610 if (ifa_ifwithaddr((SA)&ipaddr) == NULL)
984263bc
MD
1611 continue;
1612 cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1613 off += sizeof(struct in_addr);
1614 break;
1615
1616 default:
1617 code = &cp[IPOPT_OFFSET + 1] - (u_char *)ip;
1618 goto bad;
1619 }
1620 ntime = iptime();
f23061d4 1621 memcpy(cp + off, &ntime, sizeof(n_time));
984263bc
MD
1622 cp[IPOPT_OFFSET] += sizeof(n_time);
1623 }
1624 }
1625 if (forward && ipforwarding) {
41332872 1626 ip_forward(m, TRUE, next_hop);
984263bc
MD
1627 return (1);
1628 }
1629 return (0);
1630bad:
981773b8 1631 icmp_error(m, type, code, 0, 0);
984263bc
MD
1632 ipstat.ips_badoptions++;
1633 return (1);
1634}
1635
1636/*
1637 * Given address of next destination (final or next hop),
1638 * return internet address info of interface to be used to get there.
1639 */
1640struct in_ifaddr *
d27ef928 1641ip_rtaddr(struct in_addr dst, struct route *ro0)
984263bc 1642{
d27ef928 1643 struct route sro, *ro;
2256ba69 1644 struct sockaddr_in *sin;
d27ef928
SZ
1645 struct in_ifaddr *ia;
1646
1647 if (ro0 != NULL) {
1648 ro = ro0;
1649 } else {
1650 bzero(&sro, sizeof(sro));
1651 ro = &sro;
1652 }
984263bc 1653
41332872 1654 sin = (struct sockaddr_in *)&ro->ro_dst;
984263bc 1655
41332872
JH
1656 if (ro->ro_rt == NULL || dst.s_addr != sin->sin_addr.s_addr) {
1657 if (ro->ro_rt != NULL) {
1658 RTFREE(ro->ro_rt);
1659 ro->ro_rt = NULL;
984263bc
MD
1660 }
1661 sin->sin_family = AF_INET;
5fe66e68 1662 sin->sin_len = sizeof *sin;
984263bc 1663 sin->sin_addr = dst;
41332872 1664 rtalloc_ign(ro, RTF_PRCLONING);
984263bc 1665 }
dffa46cd 1666
41332872 1667 if (ro->ro_rt == NULL)
dffa46cd
JH
1668 return (NULL);
1669
d27ef928
SZ
1670 ia = ifatoia(ro->ro_rt->rt_ifa);
1671
1672 if (ro == &sro)
1673 RTFREE(ro->ro_rt);
1674 return ia;
984263bc
MD
1675}
1676
1677/*
1678 * Save incoming source route for use in replies,
1679 * to be picked up later by ip_srcroute if the receiver is interested.
1680 */
102cfd56 1681static void
7e31206a 1682save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
984263bc 1683{
7e31206a
SZ
1684 struct m_tag *mtag;
1685 struct ip_srcrt_opt *opt;
984263bc
MD
1686 unsigned olen;
1687
7e31206a
SZ
1688 mtag = m_tag_get(PACKET_TAG_IPSRCRT, sizeof(*opt), MB_DONTWAIT);
1689 if (mtag == NULL)
1690 return;
1691 opt = m_tag_data(mtag);
1692
984263bc
MD
1693 olen = option[IPOPT_OLEN];
1694#ifdef DIAGNOSTIC
1695 if (ipprintfs)
a6ec04bc 1696 kprintf("save_rte: olen %d\n", olen);
984263bc 1697#endif
7e31206a
SZ
1698 if (olen > sizeof(opt->ip_srcrt) - (1 + sizeof(dst))) {
1699 m_tag_free(mtag);
984263bc 1700 return;
7e31206a
SZ
1701 }
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);
984263bc
MD
1706}
1707
1708/*
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.
1712 */
1713struct mbuf *
7e31206a 1714ip_srcroute(struct mbuf *m0)
984263bc 1715{
2256ba69
RG
1716 struct in_addr *p, *q;
1717 struct mbuf *m;
7e31206a
SZ
1718 struct m_tag *mtag;
1719 struct ip_srcrt_opt *opt;
1720
1721 if (m0 == NULL)
1722 return NULL;
984263bc 1723
7e31206a
SZ
1724 mtag = m_tag_find(m0, PACKET_TAG_IPSRCRT, NULL);
1725 if (mtag == NULL)
1726 return NULL;
1727 opt = m_tag_data(mtag);
1728
1729 if (opt->ip_nhops == 0)
dffa46cd 1730 return (NULL);
74f1caca 1731 m = m_get(MB_DONTWAIT, MT_HEADER);
dffa46cd
JH
1732 if (m == NULL)
1733 return (NULL);
984263bc 1734
7e31206a 1735#define OPTSIZ (sizeof(opt->ip_srcrt.nop) + sizeof(opt->ip_srcrt.srcopt))
984263bc
MD
1736
1737 /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
7e31206a
SZ
1738 m->m_len = opt->ip_nhops * sizeof(struct in_addr) +
1739 sizeof(struct in_addr) + OPTSIZ;
984263bc 1740#ifdef DIAGNOSTIC
7e31206a
SZ
1741 if (ipprintfs) {
1742 kprintf("ip_srcroute: nhops %d mlen %d",
1743 opt->ip_nhops, m->m_len);
1744 }
984263bc
MD
1745#endif
1746
1747 /*
1748 * First save first hop for return route
1749 */
7e31206a 1750 p = &opt->ip_srcrt.route[opt->ip_nhops - 1];
984263bc
MD
1751 *(mtod(m, struct in_addr *)) = *p--;
1752#ifdef DIAGNOSTIC
1753 if (ipprintfs)
a6ec04bc 1754 kprintf(" hops %x", ntohl(mtod(m, struct in_addr *)->s_addr));
984263bc
MD
1755#endif
1756
1757 /*
1758 * Copy option fields and padding (nop) to mbuf.
1759 */
7e31206a
SZ
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,
dffa46cd 1763 OPTSIZ);
984263bc
MD
1764 q = (struct in_addr *)(mtod(m, caddr_t) +
1765 sizeof(struct in_addr) + OPTSIZ);
1766#undef OPTSIZ
1767 /*
1768 * Record return path as an IP source route,
1769 * reversing the path (pointers are now aligned).
1770 */
7e31206a 1771 while (p >= opt->ip_srcrt.route) {
984263bc
MD
1772#ifdef DIAGNOSTIC
1773 if (ipprintfs)
a6ec04bc 1774 kprintf(" %x", ntohl(q->s_addr));
984263bc
MD
1775#endif
1776 *q++ = *p--;
1777 }
1778 /*
1779 * Last hop goes to final destination.
1780 */
7e31206a
SZ
1781 *q = opt->ip_srcrt.dst;
1782 m_tag_delete(m0, mtag);
984263bc
MD
1783#ifdef DIAGNOSTIC
1784 if (ipprintfs)
a6ec04bc 1785 kprintf(" %x\n", ntohl(q->s_addr));
984263bc
MD
1786#endif
1787 return (m);
1788}
1789
1790/*
bddf0751 1791 * Strip out IP options.
984263bc
MD
1792 */
1793void
bddf0751 1794ip_stripoptions(struct mbuf *m)
984263bc 1795{
bddf0751 1796 int datalen;
984263bc 1797 struct ip *ip = mtod(m, struct ip *);
2256ba69 1798 caddr_t opts;
bddf0751 1799 int optlen;
984263bc 1800
bddf0751 1801 optlen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip);
984263bc 1802 opts = (caddr_t)(ip + 1);
bddf0751
JH
1803 datalen = m->m_len - (sizeof(struct ip) + optlen);
1804 bcopy(opts + optlen, opts, datalen);
1805 m->m_len -= optlen;
984263bc 1806 if (m->m_flags & M_PKTHDR)
bddf0751 1807 m->m_pkthdr.len -= optlen;
984263bc
MD
1808 ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2);
1809}
1810
1811u_char inetctlerrmap[PRC_NCMDS] = {
1812 0, 0, 0, 0,
1813 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH,
1814 EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED,
1815 EMSGSIZE, EHOSTUNREACH, 0, 0,
1816 0, 0, 0, 0,
1817 ENOPROTOOPT, ECONNREFUSED
1818};
1819
1820/*
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.
1825 *
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.
1830 *
dffa46cd 1831 * The using_srcrt parameter indicates whether the packet is being forwarded
984263bc
MD
1832 * via a source route.
1833 */
33cc7bb1 1834void
41332872 1835ip_forward(struct mbuf *m, boolean_t using_srcrt, struct sockaddr_in *next_hop)
984263bc
MD
1836{
1837 struct ip *ip = mtod(m, struct ip *);
984263bc 1838 struct rtentry *rt;
6164b468 1839 struct route fwd_ro;
981773b8 1840 int error, type = 0, code = 0, destmtu = 0;
984263bc
MD
1841 struct mbuf *mcopy;
1842 n_long dest;
1843 struct in_addr pkt_dst;
984263bc 1844
be23faf1 1845 dest = INADDR_ANY;
984263bc
MD
1846 /*
1847 * Cache the destination address of the packet; this may be
1848 * changed by use of 'ipfw fwd'.
1849 */
5fe66e68 1850 pkt_dst = (next_hop != NULL) ? next_hop->sin_addr : ip->ip_dst;
984263bc
MD
1851
1852#ifdef DIAGNOSTIC
1853 if (ipprintfs)
a6ec04bc 1854 kprintf("forward: src %x dst %x ttl %x\n",
dffa46cd 1855 ip->ip_src.s_addr, pkt_dst.s_addr, ip->ip_ttl);
984263bc
MD
1856#endif
1857
dffa46cd 1858 if (m->m_flags & (M_BCAST | M_MCAST) || !in_canforward(pkt_dst)) {
984263bc
MD
1859 ipstat.ips_cantforward++;
1860 m_freem(m);
1861 return;
1862 }
dffa46cd 1863 if (!ipstealth && ip->ip_ttl <= IPTTLDEC) {
981773b8 1864 icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
dffa46cd 1865 return;
984263bc 1866 }
984263bc 1867
6164b468
MD
1868 bzero(&fwd_ro, sizeof(fwd_ro));
1869 ip_rtaddr(pkt_dst, &fwd_ro);
1870 if (fwd_ro.ro_rt == NULL) {
984322d1
SZ
1871 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0);
1872 return;
984263bc 1873 }
6164b468 1874 rt = fwd_ro.ro_rt;
984263bc
MD
1875
1876 /*
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.
1879 *
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.
1886 *
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().
1891 */
85c8d4e6 1892 MGETHDR(mcopy, MB_DONTWAIT, m->m_type);
74f1caca 1893 if (mcopy != NULL && !m_dup_pkthdr(mcopy, m, MB_DONTWAIT)) {
984263bc
MD
1894 /*
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.
1899 */
1900 m_free(mcopy);
1901 mcopy = NULL;
1902 }
1903 if (mcopy != NULL) {
1904 mcopy->m_len = imin((IP_VHL_HL(ip->ip_vhl) << 2) + 8,
1905 (int)ip->ip_len);
52560a13 1906 mcopy->m_pkthdr.len = mcopy->m_len;
984263bc
MD
1907 m_copydata(m, 0, mcopy->m_len, mtod(mcopy, caddr_t));
1908 }
1909
dffa46cd 1910 if (!ipstealth)
984263bc 1911 ip->ip_ttl -= IPTTLDEC;
984263bc
MD
1912
1913 /*
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.
1920 */
1921 if (rt->rt_ifp == m->m_pkthdr.rcvif &&
dffa46cd
JH
1922 !(rt->rt_flags & (RTF_DYNAMIC | RTF_MODIFIED)) &&
1923 satosin(rt_key(rt))->sin_addr.s_addr != INADDR_ANY &&
983015c8 1924 ipsendredirects && !using_srcrt && next_hop == NULL) {
984263bc 1925 u_long src = ntohl(ip->ip_src.s_addr);
41332872 1926 struct in_ifaddr *rt_ifa = (struct in_ifaddr *)rt->rt_ifa;
984263bc 1927
41332872
JH
1928 if (rt_ifa != NULL &&
1929 (src & rt_ifa->ia_subnetmask) == rt_ifa->ia_subnet) {
dffa46cd
JH
1930 if (rt->rt_flags & RTF_GATEWAY)
1931 dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1932 else
1933 dest = pkt_dst.s_addr;
1934 /*
1935 * Router requirements says to only send
1936 * host redirects.
1937 */
1938 type = ICMP_REDIRECT;
1939 code = ICMP_REDIRECT_HOST;
984263bc 1940#ifdef DIAGNOSTIC
dffa46cd 1941 if (ipprintfs)
a6ec04bc 1942 kprintf("redirect (%d) to %x\n", code, dest);
984263bc
MD
1943#endif
1944 }
1945 }
1946
6164b468 1947 error = ip_output(m, NULL, &fwd_ro, IP_FORWARDING, NULL, NULL);
41332872 1948 if (error == 0) {
984263bc 1949 ipstat.ips_forward++;
41332872 1950 if (type == 0) {
984263bc 1951 if (mcopy) {
6164b468 1952 ipflow_create(&fwd_ro, mcopy);
297c8124 1953 m_freem(mcopy);
984263bc 1954 }
984322d1 1955 goto done;
41332872
JH
1956 } else {
1957 ipstat.ips_redirectsent++;
984263bc 1958 }
41332872
JH
1959 } else {
1960 ipstat.ips_cantforward++;
984263bc 1961 }
41332872 1962
984263bc 1963 if (mcopy == NULL)
984322d1 1964 goto done;
41332872
JH
1965
1966 /*
1967 * Send ICMP message.
1968 */
984263bc
MD
1969
1970 switch (error) {
1971
1972 case 0: /* forwarded, but need redirect */
1973 /* type, code set above */
1974 break;
1975
1976 case ENETUNREACH: /* shouldn't happen, checked above */
1977 case EHOSTUNREACH:
1978 case ENETDOWN:
1979 case EHOSTDOWN:
1980 default:
1981 type = ICMP_UNREACH;
1982 code = ICMP_UNREACH_HOST;
1983 break;
1984
1985 case EMSGSIZE:
1986 type = ICMP_UNREACH;
1987 code = ICMP_UNREACH_NEEDFRAG;
1988#ifdef IPSEC
1989 /*
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
1993 * XXX quickhack!!!
1994 */
6164b468 1995 if (fwd_ro.ro_rt != NULL) {
984263bc
MD
1996 struct secpolicy *sp = NULL;
1997 int ipsecerror;
1998 int ipsechdr;
1999 struct route *ro;
2000
2001 sp = ipsec4_getpolicybyaddr(mcopy,
2002 IPSEC_DIR_OUTBOUND,
dffa46cd
JH
2003 IP_FORWARDING,
2004 &ipsecerror);
984263bc
MD
2005
2006 if (sp == NULL)
6164b468 2007 destmtu = fwd_ro.ro_rt->rt_ifp->if_mtu;
984263bc
MD
2008 else {
2009 /* count IPsec header size */
2010 ipsechdr = ipsec4_hdrsiz(mcopy,
2011 IPSEC_DIR_OUTBOUND,
2012 NULL);
2013
2014 /*
2015 * find the correct route for outer IPv4
2016 * header, compute tunnel MTU.
2017 *
984263bc 2018 */
012d77ea
JH
2019 if (sp->req != NULL && sp->req->sav != NULL &&
2020 sp->req->sav->sah != NULL) {
984263bc 2021 ro = &sp->req->sav->sah->sa_route;
012d77ea
JH
2022 if (ro->ro_rt != NULL &&
2023 ro->ro_rt->rt_ifp != NULL) {
981773b8 2024 destmtu =
984263bc 2025 ro->ro_rt->rt_ifp->if_mtu;
981773b8 2026 destmtu -= ipsechdr;
984263bc
MD
2027 }
2028 }
2029
2030 key_freesp(sp);
2031 }
2032 }
2033#elif FAST_IPSEC
2034 /*
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
2038 * XXX quickhack!!!
2039 */
6164b468 2040 if (fwd_ro.ro_rt != NULL) {
984263bc
MD
2041 struct secpolicy *sp = NULL;
2042 int ipsecerror;
2043 int ipsechdr;
2044 struct route *ro;
2045
2046 sp = ipsec_getpolicybyaddr(mcopy,
2047 IPSEC_DIR_OUTBOUND,
dffa46cd
JH
2048 IP_FORWARDING,
2049 &ipsecerror);
984263bc
MD
2050
2051 if (sp == NULL)
6164b468 2052 destmtu = fwd_ro.ro_rt->rt_ifp->if_mtu;
984263bc
MD
2053 else {
2054 /* count IPsec header size */
2055 ipsechdr = ipsec4_hdrsiz(mcopy,
2056 IPSEC_DIR_OUTBOUND,
2057 NULL);
2058
2059 /*
2060 * find the correct route for outer IPv4
2061 * header, compute tunnel MTU.
984263bc 2062 */
981773b8 2063
012d77ea
JH
2064 if (sp->req != NULL &&
2065 sp->req->sav != NULL &&
2066 sp->req->sav->sah != NULL) {
984263bc 2067 ro = &sp->req->sav->sah->sa_route;
012d77ea
JH
2068 if (ro->ro_rt != NULL &&
2069 ro->ro_rt->rt_ifp != NULL) {
981773b8 2070 destmtu =
984263bc 2071 ro->ro_rt->rt_ifp->if_mtu;
981773b8 2072 destmtu -= ipsechdr;
984263bc
MD
2073 }
2074 }
2075
2076 KEY_FREESP(&sp);
2077 }
2078 }
2079#else /* !IPSEC && !FAST_IPSEC */
6164b468
MD
2080 if (fwd_ro.ro_rt != NULL)
2081 destmtu = fwd_ro.ro_rt->rt_ifp->if_mtu;
984263bc
MD
2082#endif /*IPSEC*/
2083 ipstat.ips_cantfrag++;
2084 break;
2085
2086 case ENOBUFS:
2087 /*
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.
2094 */
dffa46cd 2095 if (!ip_sendsourcequench) {
984263bc 2096 m_freem(mcopy);
984322d1 2097 goto done;
984263bc
MD
2098 } else {
2099 type = ICMP_SOURCEQUENCH;
2100 code = 0;
2101 }
2102 break;
2103
2104 case EACCES: /* ipfw denied packet */
2105 m_freem(mcopy);
984322d1 2106 goto done;
984263bc 2107 }
981773b8 2108 icmp_error(mcopy, type, code, dest, destmtu);
984322d1 2109done:
6164b468
MD
2110 if (fwd_ro.ro_rt != NULL)
2111 RTFREE(fwd_ro.ro_rt);
984263bc
MD
2112}
2113
2114void
dffa46cd
JH
2115ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
2116 struct mbuf *m)
984263bc
MD
2117{
2118 if (inp->inp_socket->so_options & SO_TIMESTAMP) {
2119 struct timeval tv;
2120
2121 microtime(&tv);
2122 *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv),
dffa46cd 2123 SCM_TIMESTAMP, SOL_SOCKET);
984263bc
MD
2124 if (*mp)
2125 mp = &(*mp)->m_next;
2126 }
2127 if (inp->inp_flags & INP_RECVDSTADDR) {
2128 *mp = sbcreatecontrol((caddr_t) &ip->ip_dst,
2129 sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
2130 if (*mp)
2131 mp = &(*mp)->m_next;
2132 }
95926362
MD
2133 if (inp->inp_flags & INP_RECVTTL) {
2134 *mp = sbcreatecontrol((caddr_t) &ip->ip_ttl,
2135 sizeof(u_char), IP_RECVTTL, IPPROTO_IP);
2136 if (*mp)
2137 mp = &(*mp)->m_next;
2138 }
984263bc
MD
2139#ifdef notyet
2140 /* XXX
2141 * Moving these out of udp_input() made them even more broken
2142 * than they already were.
2143 */
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);
2148 if (*mp)
2149 mp = &(*mp)->m_next;
2150 }
2151 /* ip_srcroute doesn't do what we want here, need to fix */
2152 if (inp->inp_flags & INP_RECVRETOPTS) {
7e31206a 2153 *mp = sbcreatecontrol((caddr_t) ip_srcroute(m),
984263bc
MD
2154 sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP);
2155 if (*mp)
2156 mp = &(*mp)->m_next;
2157 }
2158#endif
2159 if (inp->inp_flags & INP_RECVIF) {
2160 struct ifnet *ifp;
2161 struct sdlbuf {
2162 struct sockaddr_dl sdl;
2163 u_char pad[32];
2164 } sdlbuf;
2165 struct sockaddr_dl *sdp;
2166 struct sockaddr_dl *sdl2 = &sdlbuf.sdl;
2167
dffa46cd
JH
2168 if (((ifp = m->m_pkthdr.rcvif)) &&
2169 ((ifp->if_index != 0) && (ifp->if_index <= if_index))) {
f2682cb9 2170 sdp = IF_LLSOCKADDR(ifp);
984263bc
MD
2171 /*
2172 * Change our mind and don't try copy.
2173 */
dffa46cd
JH
2174 if ((sdp->sdl_family != AF_LINK) ||
2175 (sdp->sdl_len > sizeof(sdlbuf))) {
984263bc
MD
2176 goto makedummy;
2177 }
2178 bcopy(sdp, sdl2, sdp->sdl_len);
2179 } else {
dffa46cd
JH
2180makedummy:
2181 sdl2->sdl_len =
2182 offsetof(struct sockaddr_dl, sdl_data[0]);
984263bc
MD
2183 sdl2->sdl_family = AF_LINK;
2184 sdl2->sdl_index = 0;
2185 sdl2->sdl_nlen = sdl2->sdl_alen = sdl2->sdl_slen = 0;
2186 }
2187 *mp = sbcreatecontrol((caddr_t) sdl2, sdl2->sdl_len,
2188 IP_RECVIF, IPPROTO_IP);
2189 if (*mp)
2190 mp = &(*mp)->m_next;
2191 }
2192}
2193
2194/*
2195 * XXX these routines are called from the upper part of the kernel.
2196 *
2197 * They could also be moved to ip_mroute.c, since all the RSVP
2198 * handling is done there already.
2199 */
2200int
2201ip_rsvp_init(struct socket *so)
2202{
2203 if (so->so_type != SOCK_RAW ||
2204 so->so_proto->pr_protocol != IPPROTO_RSVP)
2205 return EOPNOTSUPP;
2206
2207 if (ip_rsvpd != NULL)
2208 return EADDRINUSE;
2209
2210 ip_rsvpd = so;
2211 /*
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.
2214 */
2215 if (!ip_rsvp_on) {
2216 ip_rsvp_on = 1;
2217 rsvp_on++;
2218 }
2219
2220 return 0;
2221}
2222
2223int
2224ip_rsvp_done(void)
2225{
2226 ip_rsvpd = NULL;
2227 /*
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.
2230 */
2231 if (ip_rsvp_on) {
2232 ip_rsvp_on = 0;
2233 rsvp_on--;
2234 }
2235 return 0;
2236}
2237
002c1265
MD
2238int
2239rsvp_input(struct mbuf **mp, int *offp, int proto)
984263bc 2240{
002c1265 2241 struct mbuf *m = *mp;
a00138cb 2242
002c1265 2243 *mp = NULL;
a00138cb 2244
984263bc 2245 if (rsvp_input_p) { /* call the real one if loaded */
002c1265
MD
2246 *mp = m;
2247 rsvp_input_p(mp, offp, proto);
2248 return(IPPROTO_DONE);
984263bc
MD
2249 }
2250
2251 /* Can still get packets with rsvp_on = 0 if there is a local member
2252 * of the group to which the RSVP packet is addressed. But in this
2253 * case we want to throw the packet away.
2254 */
2255
2256 if (!rsvp_on) {
2257 m_freem(m);
002c1265 2258 return(IPPROTO_DONE);
984263bc
MD
2259 }
2260
dffa46cd 2261 if (ip_rsvpd != NULL) {
002c1265
MD
2262 *mp = m;
2263 rip_input(mp, offp, proto);
2264 return(IPPROTO_DONE);
984263bc
MD
2265 }
2266 /* Drop the packet */
2267 m_freem(m);
002c1265 2268 return(IPPROTO_DONE);
984263bc 2269}