Annotation of src/sys/netinet/ip_input.c, Revision 1.348
1.348 ! ozaki-r 1: /* $NetBSD: ip_input.c,v 1.347 2016/12/12 03:55:57 ozaki-r Exp $ */
1.89 itojun 2:
3: /*
4: * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5: * All rights reserved.
1.152 itojun 6: *
1.89 itojun 7: * Redistribution and use in source and binary forms, with or without
8: * modification, are permitted provided that the following conditions
9: * are met:
10: * 1. Redistributions of source code must retain the above copyright
11: * notice, this list of conditions and the following disclaimer.
12: * 2. Redistributions in binary form must reproduce the above copyright
13: * notice, this list of conditions and the following disclaimer in the
14: * documentation and/or other materials provided with the distribution.
15: * 3. Neither the name of the project nor the names of its contributors
16: * may be used to endorse or promote products derived from this software
17: * without specific prior written permission.
1.152 itojun 18: *
1.89 itojun 19: * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22: * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29: * SUCH DAMAGE.
30: */
1.76 thorpej 31:
32: /*-
33: * Copyright (c) 1998 The NetBSD Foundation, Inc.
34: * All rights reserved.
35: *
36: * This code is derived from software contributed to The NetBSD Foundation
37: * by Public Access Networks Corporation ("Panix"). It was developed under
38: * contract to Panix by Eric Haszlakiewicz and Thor Lancelot Simon.
39: *
40: * Redistribution and use in source and binary forms, with or without
41: * modification, are permitted provided that the following conditions
42: * are met:
43: * 1. Redistributions of source code must retain the above copyright
44: * notice, this list of conditions and the following disclaimer.
45: * 2. Redistributions in binary form must reproduce the above copyright
46: * notice, this list of conditions and the following disclaimer in the
47: * documentation and/or other materials provided with the distribution.
48: *
49: * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
50: * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
51: * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
52: * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
53: * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
54: * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
55: * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
56: * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
57: * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
58: * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
59: * POSSIBILITY OF SUCH DAMAGE.
60: */
1.14 cgd 61:
1.1 cgd 62: /*
1.13 mycroft 63: * Copyright (c) 1982, 1986, 1988, 1993
64: * The Regents of the University of California. All rights reserved.
1.1 cgd 65: *
66: * Redistribution and use in source and binary forms, with or without
67: * modification, are permitted provided that the following conditions
68: * are met:
69: * 1. Redistributions of source code must retain the above copyright
70: * notice, this list of conditions and the following disclaimer.
71: * 2. Redistributions in binary form must reproduce the above copyright
72: * notice, this list of conditions and the following disclaimer in the
73: * documentation and/or other materials provided with the distribution.
1.172 agc 74: * 3. Neither the name of the University nor the names of its contributors
1.1 cgd 75: * may be used to endorse or promote products derived from this software
76: * without specific prior written permission.
77: *
78: * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
79: * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
80: * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
81: * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
82: * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
83: * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
84: * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
85: * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
86: * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
87: * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
88: * SUCH DAMAGE.
89: *
1.14 cgd 90: * @(#)ip_input.c 8.2 (Berkeley) 1/4/94
1.1 cgd 91: */
1.141 lukem 92:
93: #include <sys/cdefs.h>
1.348 ! ozaki-r 94: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.347 2016/12/12 03:55:57 ozaki-r Exp $");
1.55 scottr 95:
1.324 pooka 96: #ifdef _KERNEL_OPT
1.184 jonathan 97: #include "opt_inet.h"
1.278 christos 98: #include "opt_compat_netbsd.h"
1.62 matt 99: #include "opt_gateway.h"
1.91 thorpej 100: #include "opt_ipsec.h"
1.55 scottr 101: #include "opt_mrouting.h"
1.167 martin 102: #include "opt_mbuftrace.h"
1.135 thorpej 103: #include "opt_inet_csum.h"
1.344 ozaki-r 104: #include "opt_net_mpsafe.h"
1.324 pooka 105: #endif
1.1 cgd 106:
1.325 roy 107: #include "arp.h"
108:
1.5 mycroft 109: #include <sys/param.h>
110: #include <sys/systm.h>
1.318 rmind 111: #include <sys/cpu.h>
1.5 mycroft 112: #include <sys/mbuf.h>
113: #include <sys/domain.h>
114: #include <sys/protosw.h>
115: #include <sys/socket.h>
1.44 thorpej 116: #include <sys/socketvar.h>
1.5 mycroft 117: #include <sys/errno.h>
118: #include <sys/time.h>
119: #include <sys/kernel.h>
1.72 thorpej 120: #include <sys/pool.h>
1.28 christos 121: #include <sys/sysctl.h>
1.230 elad 122: #include <sys/kauth.h>
1.1 cgd 123:
1.5 mycroft 124: #include <net/if.h>
1.44 thorpej 125: #include <net/if_dl.h>
1.5 mycroft 126: #include <net/route.h>
1.318 rmind 127: #include <net/pktqueue.h>
1.45 mrg 128: #include <net/pfil.h>
1.1 cgd 129:
1.5 mycroft 130: #include <netinet/in.h>
131: #include <netinet/in_systm.h>
132: #include <netinet/ip.h>
133: #include <netinet/in_pcb.h>
1.215 yamt 134: #include <netinet/in_proto.h>
1.5 mycroft 135: #include <netinet/in_var.h>
136: #include <netinet/ip_var.h>
1.266 thorpej 137: #include <netinet/ip_private.h>
1.5 mycroft 138: #include <netinet/ip_icmp.h>
1.89 itojun 139: /* just for gif_ttl */
140: #include <netinet/in_gif.h>
141: #include "gif.h"
1.144 martin 142: #include <net/if_gre.h>
143: #include "gre.h"
1.111 jdolecek 144:
145: #ifdef MROUTING
146: #include <netinet/ip_mroute.h>
147: #endif
1.302 christos 148: #include <netinet/portalgo.h>
1.89 itojun 149:
1.304 christos 150: #ifdef IPSEC
1.173 jonathan 151: #include <netipsec/ipsec.h>
1.305 rmind 152: #endif
1.44 thorpej 153:
1.1 cgd 154: #ifndef IPFORWARDING
155: #ifdef GATEWAY
156: #define IPFORWARDING 1 /* forward IP packets not for us */
157: #else /* GATEWAY */
158: #define IPFORWARDING 0 /* don't forward IP packets not for us */
159: #endif /* GATEWAY */
160: #endif /* IPFORWARDING */
161: #ifndef IPSENDREDIRECTS
162: #define IPSENDREDIRECTS 1
163: #endif
1.26 thorpej 164: #ifndef IPFORWSRCRT
1.47 cjs 165: #define IPFORWSRCRT 1 /* forward source-routed packets */
166: #endif
167: #ifndef IPALLOWSRCRT
1.48 mrg 168: #define IPALLOWSRCRT 1 /* allow source-routed packets */
1.26 thorpej 169: #endif
1.53 kml 170: #ifndef IPMTUDISC
1.153 itojun 171: #define IPMTUDISC 1
1.53 kml 172: #endif
1.60 kml 173: #ifndef IPMTUDISCTIMEOUT
1.61 kml 174: #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */
1.60 kml 175: #endif
1.53 kml 176:
1.278 christos 177: #ifdef COMPAT_50
178: #include <compat/sys/time.h>
179: #include <compat/sys/socket.h>
180: #endif
181:
1.27 thorpej 182: /*
183: * Note: DIRECTED_BROADCAST is handled this way so that previous
184: * configuration using this option will Just Work.
185: */
186: #ifndef IPDIRECTEDBCAST
187: #ifdef DIRECTED_BROADCAST
188: #define IPDIRECTEDBCAST 1
189: #else
190: #define IPDIRECTEDBCAST 0
191: #endif /* DIRECTED_BROADCAST */
192: #endif /* IPDIRECTEDBCAST */
1.1 cgd 193: int ipforwarding = IPFORWARDING;
194: int ipsendredirects = IPSENDREDIRECTS;
1.13 mycroft 195: int ip_defttl = IPDEFTTL;
1.26 thorpej 196: int ip_forwsrcrt = IPFORWSRCRT;
1.27 thorpej 197: int ip_directedbcast = IPDIRECTEDBCAST;
1.47 cjs 198: int ip_allowsrcrt = IPALLOWSRCRT;
1.53 kml 199: int ip_mtudisc = IPMTUDISC;
1.156 itojun 200: int ip_mtudisc_timeout = IPMTUDISCTIMEOUT;
1.1 cgd 201: #ifdef DIAGNOSTIC
202: int ipprintfs = 0;
203: #endif
1.184 jonathan 204:
205: int ip_do_randomid = 0;
206:
1.165 christos 207: /*
208: * XXX - Setting ip_checkinterface mostly implements the receive side of
209: * the Strong ES model described in RFC 1122, but since the routing table
210: * and transmit implementation do not implement the Strong ES model,
211: * setting this to 1 results in an odd hybrid.
212: *
213: * XXX - ip_checkinterface currently must be disabled if you use ipnat
214: * to translate the destination address to another local interface.
215: *
216: * XXX - ip_checkinterface must be disabled if you add IP aliases
217: * to the loopback interface instead of the interface where the
218: * packets for those addresses are received.
219: */
1.318 rmind 220: static int ip_checkinterface __read_mostly = 0;
1.165 christos 221:
1.60 kml 222: struct rttimer_queue *ip_mtudisc_timeout_q = NULL;
223:
1.318 rmind 224: pktqueue_t * ip_pktq __read_mostly;
225: pfil_head_t * inet_pfil_hook __read_mostly;
226: ipid_state_t * ip_ids __read_mostly;
227: percpu_t * ipstat_percpu __read_mostly;
228:
1.340 ozaki-r 229: static percpu_t *ipforward_rt_percpu __cacheline_aligned;
1.286 tls 230:
1.183 jonathan 231: uint16_t ip_id;
1.75 thorpej 232:
1.135 thorpej 233: #ifdef INET_CSUM_COUNTERS
234: #include <sys/device.h>
235:
236: struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
237: NULL, "inet", "hwcsum bad");
238: struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
239: NULL, "inet", "hwcsum ok");
240: struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC,
241: NULL, "inet", "swcsum");
242:
243: #define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++
244:
1.201 matt 245: EVCNT_ATTACH_STATIC(ip_hwcsum_bad);
246: EVCNT_ATTACH_STATIC(ip_hwcsum_ok);
247: EVCNT_ATTACH_STATIC(ip_swcsum);
248:
1.135 thorpej 249: #else
250:
251: #define INET_CSUM_COUNTER_INCR(ev) /* nothing */
252:
253: #endif /* INET_CSUM_COUNTERS */
254:
1.1 cgd 255: /*
256: * We need to save the IP options in case a protocol wants to respond
257: * to an incoming packet over the same route if the packet got here
258: * using IP source routing. This allows connection establishment and
259: * maintenance when the remote end is on a network that is not known
260: * to us.
261: */
1.311 rmind 262:
263: static int ip_nhops = 0;
264:
1.1 cgd 265: static struct ip_srcrt {
266: struct in_addr dst; /* final destination */
267: char nop; /* one NOP to align */
268: char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */
269: struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)];
270: } ip_srcrt;
271:
1.295 dyoung 272: static int ip_drainwanted;
273:
1.318 rmind 274: struct sockaddr_in ipaddr = {
275: .sin_len = sizeof(ipaddr),
276: .sin_family = AF_INET,
277: };
278:
1.210 perry 279: static void save_rte(u_char *, struct in_addr);
1.35 mycroft 280:
1.164 matt 281: #ifdef MBUFTRACE
1.234 dogcow 282: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
283: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 284: #endif
285:
1.318 rmind 286: static void ipintr(void *);
1.312 rmind 287: static void ip_input(struct mbuf *);
1.329 ozaki-r 288: static void ip_forward(struct mbuf *, int, struct ifnet *);
1.311 rmind 289: static bool ip_dooptions(struct mbuf *);
1.346 ozaki-r 290: static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *);
1.311 rmind 291: static void sysctl_net_inet_ip_setup(struct sysctllog **);
1.284 pooka 292:
1.333 ozaki-r 293: static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *,
294: int *);
295: static struct in_ifaddr *ip_match_our_address_broadcast(struct ifnet *,
296: struct ip *);
297:
1.344 ozaki-r 298: #ifdef NET_MPSAFE
299: #define SOFTNET_LOCK() mutex_enter(softnet_lock)
300: #define SOFTNET_UNLOCK() mutex_exit(softnet_lock)
301: #else
1.318 rmind 302: #define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock))
303: #define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock))
1.344 ozaki-r 304: #endif
1.318 rmind 305:
1.1 cgd 306: /*
307: * IP initialization: fill in IP protocol switch table.
308: * All protocols not implemented in kernel go to raw IP protocol handler.
309: */
1.8 mycroft 310: void
1.211 perry 311: ip_init(void)
1.1 cgd 312: {
1.199 matt 313: const struct protosw *pr;
1.1 cgd 314:
1.311 rmind 315: in_init();
1.284 pooka 316: sysctl_net_inet_ip_setup(NULL);
317:
1.311 rmind 318: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
319: KASSERT(pr != NULL);
1.275 pooka 320:
1.319 ozaki-r 321: ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL);
1.318 rmind 322: KASSERT(ip_pktq != NULL);
323:
1.328 riastrad 324: for (u_int i = 0; i < IPPROTO_MAX; i++) {
325: ip_protox[i] = pr - inetsw;
1.311 rmind 326: }
1.328 riastrad 327: for (pr = inetdomain.dom_protosw;
328: pr < inetdomain.dom_protoswNPROTOSW; pr++)
1.1 cgd 329: if (pr->pr_domain->dom_family == PF_INET &&
330: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
1.328 riastrad 331: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 332:
1.288 rmind 333: ip_reass_init();
1.190 jonathan 334:
1.291 rmind 335: ip_ids = ip_id_init();
1.323 ozaki-r 336: ip_id = time_uptime & 0xfffff;
1.194 jonathan 337:
1.73 thorpej 338: #ifdef GATEWAY
1.311 rmind 339: ipflow_init();
1.73 thorpej 340: #endif
1.121 thorpej 341:
342: /* Register our Packet Filter hook. */
1.308 rmind 343: inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET);
344: KASSERT(inet_pfil_hook != NULL);
1.135 thorpej 345:
1.164 matt 346: #ifdef MBUFTRACE
347: MOWNER_ATTACH(&ip_tx_mowner);
348: MOWNER_ATTACH(&ip_rx_mowner);
349: #endif /* MBUFTRACE */
1.266 thorpej 350:
351: ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
1.340 ozaki-r 352:
353: ipforward_rt_percpu = percpu_alloc(sizeof(struct route));
354: if (ipforward_rt_percpu == NULL)
355: panic("failed to allocate ipforward_rt_percpu");
1.1 cgd 356: }
357:
1.333 ozaki-r 358: static struct in_ifaddr *
359: ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch)
360: {
361: struct in_ifaddr *ia = NULL;
362: int checkif;
363:
364: /*
365: * Enable a consistency check between the destination address
366: * and the arrival interface for a unicast packet (the RFC 1122
367: * strong ES model) if IP forwarding is disabled and the packet
368: * is not locally generated.
369: *
370: * XXX - Checking also should be disabled if the destination
371: * address is ipnat'ed to a different interface.
372: *
373: * XXX - Checking is incompatible with IP aliases added
374: * to the loopback interface instead of the interface where
375: * the packets are received.
376: *
377: * XXX - We need to add a per ifaddr flag for this so that
378: * we get finer grain control.
379: */
380: checkif = ip_checkinterface && (ipforwarding == 0) &&
381: (ifp->if_flags & IFF_LOOPBACK) == 0;
382:
1.334 ozaki-r 383: IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) {
1.333 ozaki-r 384: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
385: if (ia->ia4_flags & IN_IFF_NOTREADY)
386: continue;
387: if (checkif && ia->ia_ifp != ifp)
388: continue;
1.341 roy 389: if ((ia->ia_ifp->if_flags & IFF_UP) != 0 &&
390: (ia->ia4_flags & IN_IFF_DETACHED) == 0)
1.333 ozaki-r 391: break;
392: else
1.338 ozaki-r 393: (*downmatch)++;
1.333 ozaki-r 394: }
395: }
396:
397: return ia;
398: }
399:
400: static struct in_ifaddr *
401: ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip)
402: {
403: struct in_ifaddr *ia = NULL;
404: struct ifaddr *ifa;
405:
1.336 ozaki-r 406: IFADDR_READER_FOREACH(ifa, ifp) {
1.333 ozaki-r 407: if (ifa->ifa_addr->sa_family != AF_INET)
408: continue;
409: ia = ifatoia(ifa);
1.341 roy 410: if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED))
1.333 ozaki-r 411: continue;
412: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
413: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
414: /*
415: * Look for all-0's host part (old broadcast addr),
416: * either for subnet or net.
417: */
418: ip->ip_dst.s_addr == ia->ia_subnet ||
419: ip->ip_dst.s_addr == ia->ia_net)
420: goto matched;
421: /*
422: * An interface with IP address zero accepts
423: * all packets that arrive on that interface.
424: */
425: if (in_nullhost(ia->ia_addr.sin_addr))
426: goto matched;
427: }
428: ia = NULL;
429:
430: matched:
431: return ia;
432: }
433:
1.1 cgd 434: /*
1.318 rmind 435: * IP software interrupt routine.
1.89 itojun 436: */
1.318 rmind 437: static void
438: ipintr(void *arg __unused)
1.89 itojun 439: {
440: struct mbuf *m;
1.286 tls 441:
1.318 rmind 442: KASSERT(cpu_softintr_p());
1.89 itojun 443:
1.344 ozaki-r 444: #ifndef NET_MPSAFE
1.268 ad 445: mutex_enter(softnet_lock);
1.344 ozaki-r 446: #endif
1.318 rmind 447: while ((m = pktq_dequeue(ip_pktq)) != NULL) {
1.89 itojun 448: ip_input(m);
449: }
1.344 ozaki-r 450: #ifndef NET_MPSAFE
1.268 ad 451: mutex_exit(softnet_lock);
1.344 ozaki-r 452: #endif
1.89 itojun 453: }
454:
455: /*
1.318 rmind 456: * IP input routine. Checksum and byte swap header. If fragmented
1.1 cgd 457: * try to reassemble. Process options. Pass to next level.
458: */
1.312 rmind 459: static void
1.89 itojun 460: ip_input(struct mbuf *m)
1.1 cgd 461: {
1.109 augustss 462: struct ip *ip = NULL;
1.339 ozaki-r 463: struct in_ifaddr *ia = NULL;
1.288 rmind 464: int hlen = 0, len;
1.100 itojun 465: int downmatch;
1.169 itojun 466: int srcrt = 0;
1.314 rmind 467: ifnet_t *ifp;
1.329 ozaki-r 468: struct psref psref;
1.339 ozaki-r 469: int s;
1.1 cgd 470:
1.318 rmind 471: KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software "
472: "interrupt handler; synchronization assumptions violated");
473:
1.164 matt 474: MCLAIM(m, &ip_rx_mowner);
1.289 rmind 475: KASSERT((m->m_flags & M_PKTHDR) != 0);
1.329 ozaki-r 476:
477: ifp = m_get_rcvif_psref(m, &psref);
1.330 ozaki-r 478: if (__predict_false(ifp == NULL))
1.332 ozaki-r 479: goto out;
1.164 matt 480:
1.1 cgd 481: /*
482: * If no IP addresses have been set yet but the interfaces
483: * are receiving, can't do anything with incoming packets yet.
1.318 rmind 484: * Note: we pre-check without locks held.
1.1 cgd 485: */
1.335 ozaki-r 486: if (IN_ADDRLIST_READER_EMPTY())
1.332 ozaki-r 487: goto out;
1.266 thorpej 488: IP_STATINC(IP_STAT_TOTAL);
1.318 rmind 489:
1.154 thorpej 490: /*
491: * If the IP header is not aligned, slurp it up into a new
492: * mbuf with space for link headers, in the event we forward
493: * it. Otherwise, if it is aligned, make sure the entire
494: * base IP header is in the first mbuf of the chain.
495: */
1.244 christos 496: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 497: if ((m = m_copyup(m, sizeof(struct ip),
498: (max_linkhdr + 3) & ~3)) == NULL) {
499: /* XXXJRT new stat, please */
1.266 thorpej 500: IP_STATINC(IP_STAT_TOOSMALL);
1.329 ozaki-r 501: goto out;
1.154 thorpej 502: }
503: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
504: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.266 thorpej 505: IP_STATINC(IP_STAT_TOOSMALL);
1.329 ozaki-r 506: goto out;
1.154 thorpej 507: }
1.1 cgd 508: }
509: ip = mtod(m, struct ip *);
1.13 mycroft 510: if (ip->ip_v != IPVERSION) {
1.266 thorpej 511: IP_STATINC(IP_STAT_BADVERS);
1.332 ozaki-r 512: goto out;
1.13 mycroft 513: }
1.1 cgd 514: hlen = ip->ip_hl << 2;
515: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.266 thorpej 516: IP_STATINC(IP_STAT_BADHLEN);
1.332 ozaki-r 517: goto out;
1.1 cgd 518: }
519: if (hlen > m->m_len) {
1.298 liamjfoy 520: if ((m = m_pullup(m, hlen)) == NULL) {
1.266 thorpej 521: IP_STATINC(IP_STAT_BADHLEN);
1.329 ozaki-r 522: goto out;
1.1 cgd 523: }
524: ip = mtod(m, struct ip *);
525: }
1.98 thorpej 526:
1.85 hwr 527: /*
1.99 thorpej 528: * RFC1122: packets with a multicast source address are
1.98 thorpej 529: * not allowed.
1.85 hwr 530: */
531: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.266 thorpej 532: IP_STATINC(IP_STAT_BADADDR);
1.332 ozaki-r 533: goto out;
1.129 itojun 534: }
535:
536: /* 127/8 must not appear on wire - RFC1122 */
537: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
538: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.314 rmind 539: if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
1.266 thorpej 540: IP_STATINC(IP_STAT_BADADDR);
1.332 ozaki-r 541: goto out;
1.130 itojun 542: }
1.85 hwr 543: }
544:
1.135 thorpej 545: switch (m->m_pkthdr.csum_flags &
1.314 rmind 546: ((ifp->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 547: M_CSUM_IPv4_BAD)) {
548: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
549: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
1.332 ozaki-r 550: IP_STATINC(IP_STAT_BADSUM);
551: goto out;
1.135 thorpej 552:
553: case M_CSUM_IPv4:
554: /* Checksum was okay. */
555: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
556: break;
557:
558: default:
1.206 thorpej 559: /*
560: * Must compute it ourselves. Maybe skip checksum on
561: * loopback interfaces.
562: */
1.314 rmind 563: if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) ||
564: ip_do_loopback_cksum)) {
1.206 thorpej 565: INET_CSUM_COUNTER_INCR(&ip_swcsum);
1.332 ozaki-r 566: if (in_cksum(m, hlen) != 0) {
567: IP_STATINC(IP_STAT_BADSUM);
568: goto out;
569: }
1.206 thorpej 570: }
1.135 thorpej 571: break;
1.1 cgd 572: }
573:
1.121 thorpej 574: /* Retrieve the packet length. */
575: len = ntohs(ip->ip_len);
1.81 proff 576:
577: /*
578: * Check for additional length bogosity
579: */
1.84 proff 580: if (len < hlen) {
1.266 thorpej 581: IP_STATINC(IP_STAT_BADLEN);
1.332 ozaki-r 582: goto out;
1.81 proff 583: }
1.1 cgd 584:
585: /*
586: * Check that the amount of data in the buffers
587: * is as at least much as the IP header would have us expect.
588: * Trim mbufs if longer than we expect.
589: * Drop packet if shorter than we expect.
590: */
1.35 mycroft 591: if (m->m_pkthdr.len < len) {
1.266 thorpej 592: IP_STATINC(IP_STAT_TOOSHORT);
1.332 ozaki-r 593: goto out;
1.1 cgd 594: }
1.35 mycroft 595: if (m->m_pkthdr.len > len) {
1.1 cgd 596: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 597: m->m_len = len;
598: m->m_pkthdr.len = len;
1.1 cgd 599: } else
1.35 mycroft 600: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 601: }
602:
1.64 thorpej 603: /*
604: * Assume that we can create a fast-forward IP flow entry
605: * based on this packet.
606: */
607: m->m_flags |= M_CANFASTFWD;
608:
1.33 mrg 609: /*
1.64 thorpej 610: * Run through list of hooks for input packets. If there are any
611: * filters which require that additional packets in the flow are
612: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
613: * Note that filters must _never_ set this flag, as another filter
614: * in the list may have previously cleared it.
1.33 mrg 615: */
1.304 christos 616: #if defined(IPSEC)
1.317 christos 617: if (!ipsec_used || !ipsec_indone(m))
1.127 itojun 618: #else
619: if (1)
620: #endif
621: {
1.314 rmind 622: struct in_addr odst = ip->ip_dst;
1.318 rmind 623: bool freed;
1.169 itojun 624:
1.318 rmind 625: freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0;
626: if (freed || m == NULL) {
1.343 ozaki-r 627: m = NULL;
1.329 ozaki-r 628: goto out;
1.318 rmind 629: }
1.127 itojun 630: ip = mtod(m, struct ip *);
1.142 darrenr 631: hlen = ip->ip_hl << 2;
1.318 rmind 632:
1.205 darrenr 633: /*
634: * XXX The setting of "srcrt" here is to prevent ip_forward()
635: * from generating ICMP redirects for packets that have
636: * been redirected by a hook back out on to the same LAN that
637: * they came from and is not an indication that the packet
638: * is being inffluenced by source routing options. This
639: * allows things like
640: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
641: * where tlp0 is both on the 1.1.1.0/24 network and is the
642: * default route for hosts on 1.1.1.0/24. Of course this
643: * also requires a "map tlp0 ..." to complete the story.
644: * One might argue whether or not this kind of network config.
1.212 perry 645: * should be supported in this manner...
1.205 darrenr 646: */
1.169 itojun 647: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 648: }
1.123 thorpej 649:
650: #ifdef ALTQ
651: /* XXX Temporary until ALTQ is changed to use a pfil hook */
1.318 rmind 652: if (altq_input) {
653: SOFTNET_LOCK();
654: if ((*altq_input)(m, AF_INET) == 0) {
655: /* Packet dropped by traffic conditioner. */
656: SOFTNET_UNLOCK();
1.343 ozaki-r 657: m = NULL;
1.329 ozaki-r 658: goto out;
1.318 rmind 659: }
660: SOFTNET_UNLOCK();
1.123 thorpej 661: }
662: #endif
1.121 thorpej 663:
664: /*
1.1 cgd 665: * Process options and, if not destined for us,
666: * ship it on. ip_dooptions returns 1 when an
667: * error was detected (causing an icmp message
668: * to be sent and the original packet to be freed).
669: */
670: ip_nhops = 0; /* for source routed packets */
1.343 ozaki-r 671: if (hlen > sizeof (struct ip) && ip_dooptions(m)) {
672: m = NULL;
1.329 ozaki-r 673: goto out;
1.343 ozaki-r 674: }
1.1 cgd 675:
676: /*
677: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 678: *
679: * Traditional 4.4BSD did not consult IFF_UP at all.
680: * The behavior here is to treat addresses on !IFF_UP interface
1.321 roy 681: * or IN_IFF_NOTREADY addresses as not mine.
1.1 cgd 682: */
1.100 itojun 683: downmatch = 0;
1.339 ozaki-r 684: s = pserialize_read_enter();
1.333 ozaki-r 685: ia = ip_match_our_address(ifp, ip, &downmatch);
1.339 ozaki-r 686: if (ia != NULL) {
687: pserialize_read_exit(s);
1.86 thorpej 688: goto ours;
1.339 ozaki-r 689: }
1.333 ozaki-r 690:
1.315 christos 691: if (ifp->if_flags & IFF_BROADCAST) {
1.333 ozaki-r 692: ia = ip_match_our_address_broadcast(ifp, ip);
1.339 ozaki-r 693: if (ia != NULL) {
694: pserialize_read_exit(s);
1.333 ozaki-r 695: goto ours;
1.339 ozaki-r 696: }
1.1 cgd 697: }
1.339 ozaki-r 698: pserialize_read_exit(s);
1.333 ozaki-r 699:
1.18 mycroft 700: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 701: #ifdef MROUTING
702: extern struct socket *ip_mrouter;
1.10 brezak 703:
1.4 hpeyerl 704: if (ip_mrouter) {
705: /*
706: * If we are acting as a multicast router, all
707: * incoming multicast packets are passed to the
708: * kernel-level multicast forwarding function.
709: * The packet is returned (relatively) intact; if
710: * ip_mforward() returns a non-zero value, the packet
711: * must be discarded, else it may be accepted below.
712: *
713: * (The IP ident field is put in the same byte order
714: * as expected when ip_mforward() is called from
715: * ip_output().)
716: */
1.318 rmind 717: SOFTNET_LOCK();
1.314 rmind 718: if (ip_mforward(m, ifp) != 0) {
1.318 rmind 719: SOFTNET_UNLOCK();
1.266 thorpej 720: IP_STATINC(IP_STAT_CANTFORWARD);
1.332 ozaki-r 721: goto out;
1.4 hpeyerl 722: }
1.318 rmind 723: SOFTNET_UNLOCK();
1.4 hpeyerl 724:
725: /*
726: * The process-level routing demon needs to receive
727: * all multicast IGMP packets, whether or not this
728: * host belongs to their destination groups.
729: */
1.318 rmind 730: if (ip->ip_p == IPPROTO_IGMP) {
1.4 hpeyerl 731: goto ours;
1.318 rmind 732: }
1.266 thorpej 733: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 734: }
735: #endif
736: /*
737: * See if we belong to the destination multicast group on the
738: * arrival interface.
739: */
1.316 rmind 740: if (!in_multi_group(ip->ip_dst, ifp, 0)) {
1.266 thorpej 741: IP_STATINC(IP_STAT_CANTFORWARD);
1.332 ozaki-r 742: goto out;
1.4 hpeyerl 743: }
744: goto ours;
745: }
1.19 mycroft 746: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 747: in_nullhost(ip->ip_dst))
1.1 cgd 748: goto ours;
749:
750: /*
751: * Not for us; forward if possible and desirable.
752: */
753: if (ipforwarding == 0) {
1.329 ozaki-r 754: m_put_rcvif_psref(ifp, &psref);
1.266 thorpej 755: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 756: m_freem(m);
1.100 itojun 757: } else {
758: /*
759: * If ip_dst matched any of my address on !IFF_UP interface,
760: * and there's no IFF_UP interface that matches ip_dst,
761: * send icmp unreach. Forwarding it will result in in-kernel
762: * forwarding loop till TTL goes to 0.
763: */
764: if (downmatch) {
1.329 ozaki-r 765: m_put_rcvif_psref(ifp, &psref);
1.100 itojun 766: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.266 thorpej 767: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 768: return;
769: }
1.304 christos 770: #ifdef IPSEC
1.305 rmind 771: /* Perform IPsec, if any. */
1.318 rmind 772: if (ipsec_used) {
773: SOFTNET_LOCK();
774: if (ipsec4_input(m, IP_FORWARDING |
775: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) {
776: SOFTNET_UNLOCK();
1.332 ozaki-r 777: goto out;
1.318 rmind 778: }
779: SOFTNET_UNLOCK();
1.173 jonathan 780: }
1.305 rmind 781: #endif
1.329 ozaki-r 782: ip_forward(m, srcrt, ifp);
783: m_put_rcvif_psref(ifp, &psref);
1.100 itojun 784: }
1.89 itojun 785: return;
1.1 cgd 786:
787: ours:
1.329 ozaki-r 788: m_put_rcvif_psref(ifp, &psref);
1.331 ozaki-r 789: ifp = NULL;
1.329 ozaki-r 790:
1.1 cgd 791: /*
792: * If offset or IP_MF are set, must reassemble.
793: */
1.155 itojun 794: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.289 rmind 795: /*
796: * Pass to IP reassembly mechanism.
797: */
1.290 rmind 798: if (ip_reass_packet(&m, ip) != 0) {
1.289 rmind 799: /* Failed; invalid fragment(s) or packet. */
1.332 ozaki-r 800: goto out;
1.16 cgd 801: }
1.290 rmind 802: if (m == NULL) {
1.289 rmind 803: /* More fragments should come; silently return. */
1.332 ozaki-r 804: goto out;
1.289 rmind 805: }
1.290 rmind 806: /*
807: * Reassembly is done, we have the final packet.
808: * Updated cached data in local variable(s).
809: */
1.289 rmind 810: ip = mtod(m, struct ip *);
811: hlen = ip->ip_hl << 2;
1.79 mycroft 812: }
1.128 itojun 813:
1.304 christos 814: #ifdef IPSEC
1.173 jonathan 815: /*
1.305 rmind 816: * Enforce IPsec policy checking if we are seeing last header.
817: * Note that we do not visit this with protocols with PCB layer
818: * code - like UDP/TCP/raw IP.
1.173 jonathan 819: */
1.317 christos 820: if (ipsec_used &&
1.328 riastrad 821: (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
1.318 rmind 822: SOFTNET_LOCK();
1.305 rmind 823: if (ipsec4_input(m, 0) != 0) {
1.318 rmind 824: SOFTNET_UNLOCK();
1.332 ozaki-r 825: goto out;
1.173 jonathan 826: }
1.318 rmind 827: SOFTNET_UNLOCK();
1.173 jonathan 828: }
1.305 rmind 829: #endif
1.1 cgd 830:
831: /*
832: * Switch out to protocol's input routine.
833: */
1.82 aidan 834: #if IFA_STATS
1.339 ozaki-r 835: if (ia && ip) {
836: struct in_ifaddr *_ia;
837: /*
838: * Keep a reference from ip_match_our_address with psref
839: * is expensive, so explore ia here again.
840: */
841: s = pserialize_read_enter();
1.342 ozaki-r 842: _ia = in_get_ia(ip->ip_dst);
1.339 ozaki-r 843: _ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
844: pserialize_read_exit(s);
845: }
1.82 aidan 846: #endif
1.266 thorpej 847: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 848:
1.318 rmind 849: const int off = hlen, nh = ip->ip_p;
850:
851: SOFTNET_LOCK();
1.328 riastrad 852: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1.318 rmind 853: SOFTNET_UNLOCK();
1.89 itojun 854: return;
1.135 thorpej 855:
1.329 ozaki-r 856: out:
857: m_put_rcvif_psref(ifp, &psref);
1.332 ozaki-r 858: if (m != NULL)
859: m_freem(m);
1.1 cgd 860: }
861:
862: /*
1.288 rmind 863: * IP timer processing.
1.1 cgd 864: */
1.8 mycroft 865: void
1.211 perry 866: ip_slowtimo(void)
1.1 cgd 867: {
1.268 ad 868:
1.348 ! ozaki-r 869: #ifndef NET_MPSAFE
1.268 ad 870: mutex_enter(softnet_lock);
871: KERNEL_LOCK(1, NULL);
1.348 ! ozaki-r 872: #endif
1.1 cgd 873:
1.288 rmind 874: ip_reass_slowtimo();
1.268 ad 875:
1.348 ! ozaki-r 876: #ifndef NET_MPSAFE
1.268 ad 877: KERNEL_UNLOCK_ONE(NULL);
878: mutex_exit(softnet_lock);
1.348 ! ozaki-r 879: #endif
1.1 cgd 880: }
881:
882: /*
1.288 rmind 883: * IP drain processing.
1.1 cgd 884: */
1.8 mycroft 885: void
1.211 perry 886: ip_drain(void)
1.1 cgd 887: {
888:
1.268 ad 889: KERNEL_LOCK(1, NULL);
1.288 rmind 890: ip_reass_drain();
1.268 ad 891: KERNEL_UNLOCK_ONE(NULL);
1.1 cgd 892: }
893:
894: /*
1.311 rmind 895: * ip_dooptions: perform option processing on a datagram, possibly discarding
896: * it if bad options are encountered, or forwarding it if source-routed.
897: *
898: * => Returns true if packet has been forwarded/freed.
899: * => Returns false if the packet should be processed further.
1.1 cgd 900: */
1.311 rmind 901: static bool
1.211 perry 902: ip_dooptions(struct mbuf *m)
1.1 cgd 903: {
1.109 augustss 904: struct ip *ip = mtod(m, struct ip *);
905: u_char *cp, *cp0;
906: struct ip_timestamp *ipt;
907: struct in_ifaddr *ia;
1.1 cgd 908: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 909: struct in_addr dst;
1.1 cgd 910: n_time ntime;
1.339 ozaki-r 911: struct ifaddr *ifa;
912: int s;
1.1 cgd 913:
1.13 mycroft 914: dst = ip->ip_dst;
1.1 cgd 915: cp = (u_char *)(ip + 1);
916: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
917: for (; cnt > 0; cnt -= optlen, cp += optlen) {
918: opt = cp[IPOPT_OPTVAL];
919: if (opt == IPOPT_EOL)
920: break;
921: if (opt == IPOPT_NOP)
922: optlen = 1;
923: else {
1.113 itojun 924: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
925: code = &cp[IPOPT_OLEN] - (u_char *)ip;
926: goto bad;
927: }
1.1 cgd 928: optlen = cp[IPOPT_OLEN];
1.114 itojun 929: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 930: code = &cp[IPOPT_OLEN] - (u_char *)ip;
931: goto bad;
932: }
933: }
934: switch (opt) {
935:
936: default:
937: break;
938:
939: /*
940: * Source routing with record.
941: * Find interface with current destination address.
942: * If none on this machine then drop if strictly routed,
943: * or do nothing if loosely routed.
944: * Record interface address and bring up next address
945: * component. If strictly routed make sure next
946: * address is on directly accessible net.
947: */
948: case IPOPT_LSRR:
1.346 ozaki-r 949: case IPOPT_SSRR: {
950: struct psref psref;
1.47 cjs 951: if (ip_allowsrcrt == 0) {
952: type = ICMP_UNREACH;
953: code = ICMP_UNREACH_NET_PROHIB;
954: goto bad;
955: }
1.114 itojun 956: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
957: code = &cp[IPOPT_OLEN] - (u_char *)ip;
958: goto bad;
959: }
1.1 cgd 960: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
961: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
962: goto bad;
963: }
964: ipaddr.sin_addr = ip->ip_dst;
1.339 ozaki-r 965:
966: s = pserialize_read_enter();
967: ifa = ifa_ifwithaddr(sintosa(&ipaddr));
968: if (ifa == NULL) {
969: pserialize_read_exit(s);
1.1 cgd 970: if (opt == IPOPT_SSRR) {
971: type = ICMP_UNREACH;
972: code = ICMP_UNREACH_SRCFAIL;
973: goto bad;
974: }
975: /*
976: * Loose routing, and not at next destination
977: * yet; nothing to do except forward.
978: */
979: break;
980: }
1.339 ozaki-r 981: pserialize_read_exit(s);
982:
1.1 cgd 983: off--; /* 0 origin */
1.112 sommerfe 984: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 985: /*
986: * End of source route. Should be for us.
987: */
988: save_rte(cp, ip->ip_src);
989: break;
990: }
991: /*
992: * locate outgoing interface
993: */
1.281 tsutsui 994: memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
1.1 cgd 995: sizeof(ipaddr.sin_addr));
1.339 ozaki-r 996: if (opt == IPOPT_SSRR) {
1.346 ozaki-r 997: ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr),
998: &psref);
1.339 ozaki-r 999: if (ifa != NULL)
1000: ia = ifatoia(ifa);
1001: else
1002: ia = NULL;
1003: } else {
1.346 ozaki-r 1004: ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1.339 ozaki-r 1005: }
1006: if (ia == NULL) {
1.1 cgd 1007: type = ICMP_UNREACH;
1008: code = ICMP_UNREACH_SRCFAIL;
1009: goto bad;
1010: }
1011: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1012: bcopy((void *)&ia->ia_addr.sin_addr,
1013: (void *)(cp + off), sizeof(struct in_addr));
1.346 ozaki-r 1014: ia4_release(ia, &psref);
1.1 cgd 1015: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1016: /*
1017: * Let ip_intr's mcast routing check handle mcast pkts
1018: */
1.18 mycroft 1019: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1020: break;
1.346 ozaki-r 1021: }
1.1 cgd 1022:
1.346 ozaki-r 1023: case IPOPT_RR: {
1024: struct psref psref;
1.114 itojun 1025: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1026: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1027: goto bad;
1028: }
1.1 cgd 1029: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1030: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1031: goto bad;
1032: }
1033: /*
1034: * If no space remains, ignore.
1035: */
1036: off--; /* 0 origin */
1.112 sommerfe 1037: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1038: break;
1.281 tsutsui 1039: memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1.1 cgd 1040: sizeof(ipaddr.sin_addr));
1041: /*
1042: * locate outgoing interface; if we're the destination,
1043: * use the incoming interface (should be same).
1044: */
1.346 ozaki-r 1045: ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref);
1.339 ozaki-r 1046: if (ifa == NULL) {
1.346 ozaki-r 1047: ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1.339 ozaki-r 1048: if (ia == NULL) {
1049: type = ICMP_UNREACH;
1050: code = ICMP_UNREACH_HOST;
1051: goto bad;
1052: }
1053: } else {
1054: ia = ifatoia(ifa);
1.1 cgd 1055: }
1.244 christos 1056: bcopy((void *)&ia->ia_addr.sin_addr,
1057: (void *)(cp + off), sizeof(struct in_addr));
1.346 ozaki-r 1058: ia4_release(ia, &psref);
1.1 cgd 1059: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1060: break;
1.346 ozaki-r 1061: }
1.1 cgd 1062:
1063: case IPOPT_TS:
1064: code = cp - (u_char *)ip;
1065: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1066: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1067: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1068: goto bad;
1.114 itojun 1069: }
1070: if (ipt->ipt_ptr < 5) {
1071: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1072: goto bad;
1073: }
1.15 cgd 1074: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1075: if (++ipt->ipt_oflw == 0) {
1076: code = (u_char *)&ipt->ipt_ptr -
1077: (u_char *)ip;
1.1 cgd 1078: goto bad;
1.114 itojun 1079: }
1.1 cgd 1080: break;
1081: }
1.104 thorpej 1082: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1083: switch (ipt->ipt_flg) {
1084:
1085: case IPOPT_TS_TSONLY:
1086: break;
1087:
1.329 ozaki-r 1088: case IPOPT_TS_TSANDADDR: {
1089: struct ifnet *rcvif;
1.339 ozaki-r 1090: int _s, _ss;
1.329 ozaki-r 1091:
1.66 thorpej 1092: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1093: sizeof(struct in_addr) > ipt->ipt_len) {
1094: code = (u_char *)&ipt->ipt_ptr -
1095: (u_char *)ip;
1.1 cgd 1096: goto bad;
1.114 itojun 1097: }
1.13 mycroft 1098: ipaddr.sin_addr = dst;
1.339 ozaki-r 1099: _ss = pserialize_read_enter();
1100: rcvif = m_get_rcvif(m, &_s);
1101: ifa = ifaof_ifpforaddr(sintosa(&ipaddr), rcvif);
1102: m_put_rcvif(rcvif, &_s);
1103: if (ifa == NULL) {
1104: pserialize_read_exit(_ss);
1105: break;
1106: }
1107: ia = ifatoia(ifa);
1.104 thorpej 1108: bcopy(&ia->ia_addr.sin_addr,
1109: cp0, sizeof(struct in_addr));
1.339 ozaki-r 1110: pserialize_read_exit(_ss);
1.1 cgd 1111: ipt->ipt_ptr += sizeof(struct in_addr);
1112: break;
1.329 ozaki-r 1113: }
1.1 cgd 1114:
1115: case IPOPT_TS_PRESPEC:
1.66 thorpej 1116: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1117: sizeof(struct in_addr) > ipt->ipt_len) {
1118: code = (u_char *)&ipt->ipt_ptr -
1119: (u_char *)ip;
1.1 cgd 1120: goto bad;
1.114 itojun 1121: }
1.281 tsutsui 1122: memcpy(&ipaddr.sin_addr, cp0,
1.1 cgd 1123: sizeof(struct in_addr));
1.339 ozaki-r 1124: s = pserialize_read_enter();
1125: ifa = ifa_ifwithaddr(sintosa(&ipaddr));
1126: if (ifa == NULL) {
1127: pserialize_read_exit(s);
1.1 cgd 1128: continue;
1.339 ozaki-r 1129: }
1130: pserialize_read_exit(s);
1.1 cgd 1131: ipt->ipt_ptr += sizeof(struct in_addr);
1132: break;
1133:
1134: default:
1.114 itojun 1135: /* XXX can't take &ipt->ipt_flg */
1136: code = (u_char *)&ipt->ipt_ptr -
1137: (u_char *)ip + 1;
1.1 cgd 1138: goto bad;
1139: }
1140: ntime = iptime();
1.107 thorpej 1141: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1142: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1143: sizeof(n_time));
1144: ipt->ipt_ptr += sizeof(n_time);
1145: }
1146: }
1147: if (forward) {
1.329 ozaki-r 1148: struct ifnet *rcvif;
1.339 ozaki-r 1149: struct psref _psref;
1.329 ozaki-r 1150:
1.26 thorpej 1151: if (ip_forwsrcrt == 0) {
1152: type = ICMP_UNREACH;
1153: code = ICMP_UNREACH_SRCFAIL;
1154: goto bad;
1155: }
1.329 ozaki-r 1156:
1.339 ozaki-r 1157: rcvif = m_get_rcvif_psref(m, &_psref);
1.330 ozaki-r 1158: if (__predict_false(rcvif == NULL)) {
1159: type = ICMP_UNREACH;
1160: code = ICMP_UNREACH_HOST;
1161: goto bad;
1162: }
1.329 ozaki-r 1163: ip_forward(m, 1, rcvif);
1.339 ozaki-r 1164: m_put_rcvif_psref(rcvif, &_psref);
1.311 rmind 1165: return true;
1.13 mycroft 1166: }
1.311 rmind 1167: return false;
1.1 cgd 1168: bad:
1.13 mycroft 1169: icmp_error(m, type, code, 0, 0);
1.266 thorpej 1170: IP_STATINC(IP_STAT_BADOPTIONS);
1.311 rmind 1171: return true;
1.1 cgd 1172: }
1173:
1174: /*
1.311 rmind 1175: * ip_rtaddr: given address of next destination (final or next hop),
1.1 cgd 1176: * return internet address info of interface to be used to get there.
1177: */
1.311 rmind 1178: static struct in_ifaddr *
1.346 ozaki-r 1179: ip_rtaddr(struct in_addr dst, struct psref *psref)
1.1 cgd 1180: {
1.249 dyoung 1181: struct rtentry *rt;
1182: union {
1183: struct sockaddr dst;
1184: struct sockaddr_in dst4;
1185: } u;
1.340 ozaki-r 1186: struct route *ro;
1.249 dyoung 1187:
1188: sockaddr_in_init(&u.dst4, &dst, 0);
1189:
1.340 ozaki-r 1190: ro = percpu_getref(ipforward_rt_percpu);
1191: rt = rtcache_lookup(ro, &u.dst);
1.345 ozaki-r 1192: if (rt == NULL) {
1193: percpu_putref(ipforward_rt_percpu);
1194: return NULL;
1195: }
1196:
1.346 ozaki-r 1197: ia4_acquire(ifatoia(rt->rt_ifa), psref);
1.345 ozaki-r 1198: rtcache_unref(rt, ro);
1.340 ozaki-r 1199: percpu_putref(ipforward_rt_percpu);
1.249 dyoung 1200:
1201: return ifatoia(rt->rt_ifa);
1.1 cgd 1202: }
1203:
1204: /*
1.311 rmind 1205: * save_rte: save incoming source route for use in replies, to be picked
1206: * up later by ip_srcroute if the receiver is interested.
1.1 cgd 1207: */
1.311 rmind 1208: static void
1.211 perry 1209: save_rte(u_char *option, struct in_addr dst)
1.1 cgd 1210: {
1211: unsigned olen;
1212:
1213: olen = option[IPOPT_OLEN];
1214: if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst)))
1215: return;
1.281 tsutsui 1216: memcpy((void *)ip_srcrt.srcopt, (void *)option, olen);
1.1 cgd 1217: ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
1218: ip_srcrt.dst = dst;
1219: }
1220:
1221: /*
1222: * Retrieve incoming source route for use in replies,
1223: * in the same form used by setsockopt.
1224: * The first hop is placed before the options, will be removed later.
1225: */
1226: struct mbuf *
1.211 perry 1227: ip_srcroute(void)
1.1 cgd 1228: {
1.109 augustss 1229: struct in_addr *p, *q;
1230: struct mbuf *m;
1.1 cgd 1231:
1232: if (ip_nhops == 0)
1.237 dyoung 1233: return NULL;
1.1 cgd 1234: m = m_get(M_DONTWAIT, MT_SOOPTS);
1235: if (m == 0)
1.237 dyoung 1236: return NULL;
1.1 cgd 1237:
1.164 matt 1238: MCLAIM(m, &inetdomain.dom_mowner);
1.13 mycroft 1239: #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt))
1.1 cgd 1240:
1241: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */
1242: m->m_len = ip_nhops * sizeof(struct in_addr) + sizeof(struct in_addr) +
1243: OPTSIZ;
1244:
1245: /*
1246: * First save first hop for return route
1247: */
1248: p = &ip_srcrt.route[ip_nhops - 1];
1249: *(mtod(m, struct in_addr *)) = *p--;
1250:
1251: /*
1252: * Copy option fields and padding (nop) to mbuf.
1253: */
1254: ip_srcrt.nop = IPOPT_NOP;
1255: ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF;
1.244 christos 1256: memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop,
1257: OPTSIZ);
1258: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1259: sizeof(struct in_addr) + OPTSIZ);
1260: #undef OPTSIZ
1261: /*
1262: * Record return path as an IP source route,
1263: * reversing the path (pointers are now aligned).
1264: */
1265: while (p >= ip_srcrt.route) {
1266: *q++ = *p--;
1267: }
1268: /*
1269: * Last hop goes to final destination.
1270: */
1271: *q = ip_srcrt.dst;
1272: return (m);
1273: }
1274:
1.139 matt 1275: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1276: [PRC_MSGSIZE] = EMSGSIZE,
1277: [PRC_HOSTDEAD] = EHOSTDOWN,
1278: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1279: [PRC_UNREACH_NET] = EHOSTUNREACH,
1280: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1281: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1282: [PRC_UNREACH_PORT] = ECONNREFUSED,
1283: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1284: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1285: };
1286:
1.295 dyoung 1287: void
1288: ip_fasttimo(void)
1289: {
1290: if (ip_drainwanted) {
1291: ip_drain();
1292: ip_drainwanted = 0;
1293: }
1294: }
1295:
1296: void
1297: ip_drainstub(void)
1298: {
1299: ip_drainwanted = 1;
1300: }
1301:
1.1 cgd 1302: /*
1303: * Forward a packet. If some error occurs return the sender
1304: * an icmp packet. Note we can't always generate a meaningful
1305: * icmp message because icmp doesn't have a large enough repertoire
1306: * of codes and types.
1307: *
1308: * If not forwarding, just drop the packet. This could be confusing
1309: * if ipforwarding was zero but some routing protocol was advancing
1310: * us as a gateway to somewhere. However, we must let the routing
1311: * protocol deal with that.
1312: *
1313: * The srcrt parameter indicates whether the packet is being forwarded
1314: * via a source route.
1315: */
1.313 rmind 1316: static void
1.329 ozaki-r 1317: ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif)
1.1 cgd 1318: {
1.109 augustss 1319: struct ip *ip = mtod(m, struct ip *);
1320: struct rtentry *rt;
1.220 christos 1321: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1322: struct mbuf *mcopy;
1.13 mycroft 1323: n_long dest;
1.249 dyoung 1324: union {
1325: struct sockaddr dst;
1326: struct sockaddr_in dst4;
1327: } u;
1.320 ozaki-r 1328: uint64_t *ips;
1.340 ozaki-r 1329: struct route *ro;
1.164 matt 1330:
1.318 rmind 1331: KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software "
1332: "interrupt handler; synchronization assumptions violated");
1333:
1.164 matt 1334: /*
1335: * We are now in the output path.
1336: */
1337: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1338:
1339: /*
1340: * Clear any in-bound checksum flags for this packet.
1341: */
1342: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1343:
1.13 mycroft 1344: dest = 0;
1.93 sommerfe 1345: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.266 thorpej 1346: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1347: m_freem(m);
1348: return;
1349: }
1.318 rmind 1350:
1.1 cgd 1351: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1352: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1353: return;
1354: }
1355:
1.249 dyoung 1356: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1.318 rmind 1357:
1.340 ozaki-r 1358: ro = percpu_getref(ipforward_rt_percpu);
1.345 ozaki-r 1359: rt = rtcache_lookup(ro, &u.dst);
1360: if (rt == NULL) {
1.340 ozaki-r 1361: percpu_putref(ipforward_rt_percpu);
1.249 dyoung 1362: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1363: return;
1.1 cgd 1364: }
1365:
1366: /*
1.34 mycroft 1367: * Save at most 68 bytes of the packet in case
1.1 cgd 1368: * we need to generate an ICMP message to the src.
1.119 itojun 1369: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1370: */
1.155 itojun 1371: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1372: if (mcopy)
1373: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1374:
1.221 christos 1375: ip->ip_ttl -= IPTTLDEC;
1376:
1.1 cgd 1377: /*
1378: * If forwarding packet using same interface that it came in on,
1379: * perhaps should send a redirect to sender to shortcut a hop.
1380: * Only send redirect if source is sending directly to us,
1381: * and if packet was not source routed (or has any options).
1382: * Also, don't send redirect if forwarding using a default route
1383: * or a route modified by a redirect.
1384: */
1.329 ozaki-r 1385: if (rt->rt_ifp == rcvif &&
1.1 cgd 1386: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1387: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1388: ipsendredirects && !srcrt) {
1.19 mycroft 1389: if (rt->rt_ifa &&
1390: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1391: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1392: if (rt->rt_flags & RTF_GATEWAY)
1393: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1394: else
1395: dest = ip->ip_dst.s_addr;
1396: /*
1397: * Router requirements says to only send host
1398: * redirects.
1399: */
1400: type = ICMP_REDIRECT;
1401: code = ICMP_REDIRECT_HOST;
1.1 cgd 1402: }
1403: }
1.345 ozaki-r 1404: rtcache_unref(rt, ro);
1.1 cgd 1405:
1.340 ozaki-r 1406: error = ip_output(m, NULL, ro,
1.173 jonathan 1407: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.296 plunky 1408: NULL, NULL);
1.173 jonathan 1409:
1.320 ozaki-r 1410: if (error) {
1.266 thorpej 1411: IP_STATINC(IP_STAT_CANTFORWARD);
1.320 ozaki-r 1412: goto error;
1413: }
1414:
1415: ips = IP_STAT_GETREF();
1416: ips[IP_STAT_FORWARD]++;
1417:
1418: if (type) {
1419: ips[IP_STAT_REDIRECTSENT]++;
1420: IP_STAT_PUTREF();
1421: goto redirect;
1422: }
1423:
1424: IP_STAT_PUTREF();
1425: if (mcopy) {
1.63 matt 1426: #ifdef GATEWAY
1.320 ozaki-r 1427: if (mcopy->m_flags & M_CANFASTFWD)
1.340 ozaki-r 1428: ipflow_create(ro, mcopy);
1.63 matt 1429: #endif
1.320 ozaki-r 1430: m_freem(mcopy);
1.1 cgd 1431: }
1.320 ozaki-r 1432:
1.340 ozaki-r 1433: percpu_putref(ipforward_rt_percpu);
1.320 ozaki-r 1434: return;
1435:
1436: redirect:
1437: error:
1.318 rmind 1438: if (mcopy == NULL) {
1.340 ozaki-r 1439: percpu_putref(ipforward_rt_percpu);
1.1 cgd 1440: return;
1.318 rmind 1441: }
1.13 mycroft 1442:
1.1 cgd 1443: switch (error) {
1444:
1445: case 0: /* forwarded, but need redirect */
1446: /* type, code set above */
1447: break;
1448:
1449: case ENETUNREACH: /* shouldn't happen, checked above */
1450: case EHOSTUNREACH:
1451: case ENETDOWN:
1452: case EHOSTDOWN:
1453: default:
1454: type = ICMP_UNREACH;
1455: code = ICMP_UNREACH_HOST;
1456: break;
1457:
1458: case EMSGSIZE:
1459: type = ICMP_UNREACH;
1460: code = ICMP_UNREACH_NEEDFRAG;
1.263 cube 1461:
1.345 ozaki-r 1462: if ((rt = rtcache_validate(ro)) != NULL) {
1.274 seanb 1463: destmtu = rt->rt_ifp->if_mtu;
1.345 ozaki-r 1464: rtcache_unref(rt, ro);
1465: }
1.305 rmind 1466: #ifdef IPSEC
1.317 christos 1467: if (ipsec_used)
1468: (void)ipsec4_forward(mcopy, &destmtu);
1.305 rmind 1469: #endif
1.266 thorpej 1470: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 1471: break;
1472:
1473: case ENOBUFS:
1.143 itojun 1474: /*
1.311 rmind 1475: * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812,
1476: * Requirements for IP Version 4 Routers. Source quench can
1477: * big problem under DoS attacks or if the underlying
1478: * interface is rate-limited.
1.143 itojun 1479: */
1480: if (mcopy)
1481: m_freem(mcopy);
1.340 ozaki-r 1482: percpu_putref(ipforward_rt_percpu);
1.143 itojun 1483: return;
1.1 cgd 1484: }
1.220 christos 1485: icmp_error(mcopy, type, code, dest, destmtu);
1.340 ozaki-r 1486: percpu_putref(ipforward_rt_percpu);
1.44 thorpej 1487: }
1488:
1489: void
1.211 perry 1490: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
1491: struct mbuf *m)
1.44 thorpej 1492: {
1.311 rmind 1493: struct socket *so = inp->inp_socket;
1.329 ozaki-r 1494: ifnet_t *ifp;
1.311 rmind 1495: int inpflags = inp->inp_flags;
1.329 ozaki-r 1496: struct psref psref;
1.44 thorpej 1497:
1.329 ozaki-r 1498: ifp = m_get_rcvif_psref(m, &psref);
1.330 ozaki-r 1499: if (__predict_false(ifp == NULL))
1500: return; /* XXX should report error? */
1501:
1.311 rmind 1502: if (so->so_options & SO_TIMESTAMP
1.278 christos 1503: #ifdef SO_OTIMESTAMP
1.311 rmind 1504: || so->so_options & SO_OTIMESTAMP
1.278 christos 1505: #endif
1506: ) {
1.44 thorpej 1507: struct timeval tv;
1508:
1509: microtime(&tv);
1.278 christos 1510: #ifdef SO_OTIMESTAMP
1.311 rmind 1511: if (so->so_options & SO_OTIMESTAMP) {
1.278 christos 1512: struct timeval50 tv50;
1513: timeval_to_timeval50(&tv, &tv50);
1514: *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1515: SCM_OTIMESTAMP, SOL_SOCKET);
1516: } else
1517: #endif
1.244 christos 1518: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 1519: SCM_TIMESTAMP, SOL_SOCKET);
1520: if (*mp)
1521: mp = &(*mp)->m_next;
1522: }
1.311 rmind 1523: if (inpflags & INP_RECVDSTADDR) {
1.244 christos 1524: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 1525: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1526: if (*mp)
1527: mp = &(*mp)->m_next;
1528: }
1.311 rmind 1529: if (inpflags & INP_RECVPKTINFO) {
1.306 christos 1530: struct in_pktinfo ipi;
1.307 christos 1531: ipi.ipi_addr = ip->ip_src;
1.314 rmind 1532: ipi.ipi_ifindex = ifp->if_index;
1.306 christos 1533: *mp = sbcreatecontrol((void *) &ipi,
1534: sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP);
1535: if (*mp)
1536: mp = &(*mp)->m_next;
1537: }
1.311 rmind 1538: if (inpflags & INP_PKTINFO) {
1.306 christos 1539: struct in_pktinfo ipi;
1.307 christos 1540: ipi.ipi_addr = ip->ip_dst;
1.314 rmind 1541: ipi.ipi_ifindex = ifp->if_index;
1.306 christos 1542: *mp = sbcreatecontrol((void *) &ipi,
1543: sizeof(ipi), IP_PKTINFO, IPPROTO_IP);
1544: if (*mp)
1545: mp = &(*mp)->m_next;
1546: }
1.311 rmind 1547: if (inpflags & INP_RECVIF) {
1.44 thorpej 1548: struct sockaddr_dl sdl;
1549:
1.337 ozaki-r 1550: sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0,
1551: NULL, 0);
1.251 dyoung 1552: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 1553: if (*mp)
1554: mp = &(*mp)->m_next;
1555: }
1.311 rmind 1556: if (inpflags & INP_RECVTTL) {
1.282 minskim 1557: *mp = sbcreatecontrol((void *) &ip->ip_ttl,
1558: sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
1559: if (*mp)
1560: mp = &(*mp)->m_next;
1561: }
1.329 ozaki-r 1562: m_put_rcvif_psref(ifp, &psref);
1.13 mycroft 1563: }
1564:
1.189 atatat 1565: /*
1.228 elad 1566: * sysctl helper routine for net.inet.ip.forwsrcrt.
1567: */
1568: static int
1569: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
1570: {
1571: int error, tmp;
1572: struct sysctlnode node;
1573:
1574: node = *rnode;
1575: tmp = ip_forwsrcrt;
1576: node.sysctl_data = &tmp;
1577: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1578: if (error || newp == NULL)
1579: return (error);
1580:
1.280 elad 1581: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1582: 0, NULL, NULL, NULL);
1583: if (error)
1584: return (error);
1.228 elad 1585:
1586: ip_forwsrcrt = tmp;
1587:
1588: return (0);
1589: }
1590:
1591: /*
1.189 atatat 1592: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
1593: * range of the new value and tweaks timers if it changes.
1594: */
1595: static int
1596: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 1597: {
1.189 atatat 1598: int error, tmp;
1599: struct sysctlnode node;
1600:
1601: node = *rnode;
1602: tmp = ip_mtudisc_timeout;
1603: node.sysctl_data = &tmp;
1604: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1605: if (error || newp == NULL)
1606: return (error);
1607: if (tmp < 0)
1608: return (EINVAL);
1.52 thorpej 1609:
1.348 ! ozaki-r 1610: /* XXX NOMPSAFE still need softnet_lock */
1.273 matt 1611: mutex_enter(softnet_lock);
1612:
1.189 atatat 1613: ip_mtudisc_timeout = tmp;
1614: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
1615:
1.273 matt 1616: mutex_exit(softnet_lock);
1617:
1.189 atatat 1618: return (0);
1619: }
1.54 lukem 1620:
1.266 thorpej 1621: static int
1622: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
1623: {
1624:
1.271 thorpej 1625: return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1.266 thorpej 1626: }
1.131 itojun 1627:
1.284 pooka 1628: static void
1629: sysctl_net_inet_ip_setup(struct sysctllog **clog)
1.189 atatat 1630: {
1.197 atatat 1631: sysctl_createv(clog, 0, NULL, NULL,
1632: CTLFLAG_PERMANENT,
1.203 atatat 1633: CTLTYPE_NODE, "inet",
1634: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 1635: NULL, 0, NULL, 0,
1636: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 1637: sysctl_createv(clog, 0, NULL, NULL,
1638: CTLFLAG_PERMANENT,
1.203 atatat 1639: CTLTYPE_NODE, "ip",
1640: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 1641: NULL, 0, NULL, 0,
1642: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 1643:
1.197 atatat 1644: sysctl_createv(clog, 0, NULL, NULL,
1645: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1646: CTLTYPE_INT, "forwarding",
1647: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 1648: NULL, 0, &ipforwarding, 0,
1649: CTL_NET, PF_INET, IPPROTO_IP,
1650: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 1651: sysctl_createv(clog, 0, NULL, NULL,
1652: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1653: CTLTYPE_INT, "redirect",
1654: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 1655: NULL, 0, &ipsendredirects, 0,
1656: CTL_NET, PF_INET, IPPROTO_IP,
1657: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 1658: sysctl_createv(clog, 0, NULL, NULL,
1659: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1660: CTLTYPE_INT, "ttl",
1661: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 1662: NULL, 0, &ip_defttl, 0,
1663: CTL_NET, PF_INET, IPPROTO_IP,
1664: IPCTL_DEFTTL, CTL_EOL);
1665: #ifdef IPCTL_DEFMTU
1.197 atatat 1666: sysctl_createv(clog, 0, NULL, NULL,
1667: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 1668: CTLTYPE_INT, "mtu",
1669: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 1670: NULL, 0, &ip_mtu, 0,
1671: CTL_NET, PF_INET, IPPROTO_IP,
1672: IPCTL_DEFMTU, CTL_EOL);
1673: #endif /* IPCTL_DEFMTU */
1.197 atatat 1674: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 1675: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1676: CTLTYPE_INT, "forwsrcrt",
1677: SYSCTL_DESCR("Enable forwarding of source-routed "
1678: "datagrams"),
1.228 elad 1679: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 1680: CTL_NET, PF_INET, IPPROTO_IP,
1681: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 1682: sysctl_createv(clog, 0, NULL, NULL,
1683: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1684: CTLTYPE_INT, "directed-broadcast",
1685: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 1686: NULL, 0, &ip_directedbcast, 0,
1687: CTL_NET, PF_INET, IPPROTO_IP,
1688: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 1689: sysctl_createv(clog, 0, NULL, NULL,
1690: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1691: CTLTYPE_INT, "allowsrcrt",
1692: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 1693: NULL, 0, &ip_allowsrcrt, 0,
1694: CTL_NET, PF_INET, IPPROTO_IP,
1695: IPCTL_ALLOWSRCRT, CTL_EOL);
1.311 rmind 1696:
1.197 atatat 1697: sysctl_createv(clog, 0, NULL, NULL,
1698: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1699: CTLTYPE_INT, "mtudisc",
1700: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 1701: NULL, 0, &ip_mtudisc, 0,
1702: CTL_NET, PF_INET, IPPROTO_IP,
1703: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 1704: sysctl_createv(clog, 0, NULL, NULL,
1705: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1706: CTLTYPE_INT, "anonportmin",
1707: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 1708: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
1709: CTL_NET, PF_INET, IPPROTO_IP,
1710: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 1711: sysctl_createv(clog, 0, NULL, NULL,
1712: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1713: CTLTYPE_INT, "anonportmax",
1714: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 1715: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
1716: CTL_NET, PF_INET, IPPROTO_IP,
1717: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 1718: sysctl_createv(clog, 0, NULL, NULL,
1719: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1720: CTLTYPE_INT, "mtudisctimeout",
1721: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.300 dsl 1722: sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0,
1.189 atatat 1723: CTL_NET, PF_INET, IPPROTO_IP,
1724: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
1725: #ifndef IPNOPRIVPORTS
1.197 atatat 1726: sysctl_createv(clog, 0, NULL, NULL,
1727: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1728: CTLTYPE_INT, "lowportmin",
1729: SYSCTL_DESCR("Lowest privileged ephemeral port number "
1730: "to assign"),
1.189 atatat 1731: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
1732: CTL_NET, PF_INET, IPPROTO_IP,
1733: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 1734: sysctl_createv(clog, 0, NULL, NULL,
1735: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1736: CTLTYPE_INT, "lowportmax",
1737: SYSCTL_DESCR("Highest privileged ephemeral port number "
1738: "to assign"),
1.189 atatat 1739: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
1740: CTL_NET, PF_INET, IPPROTO_IP,
1741: IPCTL_LOWPORTMAX, CTL_EOL);
1742: #endif /* IPNOPRIVPORTS */
1743: #if NGRE > 0
1.197 atatat 1744: sysctl_createv(clog, 0, NULL, NULL,
1745: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1746: CTLTYPE_INT, "grettl",
1747: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 1748: NULL, 0, &ip_gre_ttl, 0,
1749: CTL_NET, PF_INET, IPPROTO_IP,
1750: IPCTL_GRE_TTL, CTL_EOL);
1751: #endif /* NGRE */
1.197 atatat 1752: sysctl_createv(clog, 0, NULL, NULL,
1753: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1754: CTLTYPE_INT, "checkinterface",
1755: SYSCTL_DESCR("Enable receive side of Strong ES model "
1756: "from RFC1122"),
1.189 atatat 1757: NULL, 0, &ip_checkinterface, 0,
1758: CTL_NET, PF_INET, IPPROTO_IP,
1759: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 1760: sysctl_createv(clog, 0, NULL, NULL,
1761: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1762: CTLTYPE_INT, "random_id",
1763: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 1764: NULL, 0, &ip_do_randomid, 0,
1765: CTL_NET, PF_INET, IPPROTO_IP,
1766: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 1767: sysctl_createv(clog, 0, NULL, NULL,
1768: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1769: CTLTYPE_INT, "do_loopback_cksum",
1770: SYSCTL_DESCR("Perform IP checksum on loopback"),
1771: NULL, 0, &ip_do_loopback_cksum, 0,
1772: CTL_NET, PF_INET, IPPROTO_IP,
1773: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 1774: sysctl_createv(clog, 0, NULL, NULL,
1775: CTLFLAG_PERMANENT,
1776: CTLTYPE_STRUCT, "stats",
1777: SYSCTL_DESCR("IP statistics"),
1.266 thorpej 1778: sysctl_net_inet_ip_stats, 0, NULL, 0,
1.219 elad 1779: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
1780: CTL_EOL);
1.322 joerg 1781: #if NARP
1.321 roy 1782: sysctl_createv(clog, 0, NULL, NULL,
1783: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1784: CTLTYPE_INT, "dad_count",
1785: SYSCTL_DESCR("Number of Duplicate Address Detection "
1786: "probes to send"),
1787: NULL, 0, &ip_dad_count, 0,
1788: CTL_NET, PF_INET, IPPROTO_IP,
1789: IPCTL_DAD_COUNT, CTL_EOL);
1.322 joerg 1790: #endif
1.301 christos 1791:
1792: /* anonportalgo RFC6056 subtree */
1.302 christos 1793: const struct sysctlnode *portalgo_node;
1794: sysctl_createv(clog, 0, NULL, &portalgo_node,
1.301 christos 1795: CTLFLAG_PERMANENT,
1796: CTLTYPE_NODE, "anonportalgo",
1797: SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"),
1798: NULL, 0, NULL, 0,
1799: CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL);
1.302 christos 1800: sysctl_createv(clog, 0, &portalgo_node, NULL,
1.301 christos 1801: CTLFLAG_PERMANENT,
1802: CTLTYPE_STRING, "available",
1803: SYSCTL_DESCR("available algorithms"),
1.302 christos 1804: sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1.301 christos 1805: CTL_CREATE, CTL_EOL);
1.302 christos 1806: sysctl_createv(clog, 0, &portalgo_node, NULL,
1.301 christos 1807: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1808: CTLTYPE_STRING, "selected",
1809: SYSCTL_DESCR("selected algorithm"),
1.303 christos 1810: sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN,
1811: CTL_CREATE, CTL_EOL);
1812: sysctl_createv(clog, 0, &portalgo_node, NULL,
1813: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1814: CTLTYPE_STRUCT, "reserve",
1815: SYSCTL_DESCR("bitmap of reserved ports"),
1816: sysctl_portalgo_reserve4, 0, NULL, 0,
1.301 christos 1817: CTL_CREATE, CTL_EOL);
1.1 cgd 1818: }
1.266 thorpej 1819:
1820: void
1821: ip_statinc(u_int stat)
1822: {
1823:
1824: KASSERT(stat < IP_NSTATS);
1825: IP_STATINC(stat);
1826: }
CVSweb <webmaster@jp.NetBSD.org>