Annotation of src/sys/netinet/ip_input.c, Revision 1.354
1.354 ! ozaki-r 1: /* $NetBSD: ip_input.c,v 1.353 2017/03/31 06:42:19 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>
1.354 ! ozaki-r 94: __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.353 2017/03/31 06:42:19 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: /*
1.354 ! ozaki-r 256: * Used to save the IP options in case a protocol wants to respond
1.1 cgd 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: */
1.354 ! ozaki-r 262: struct ip_srcrt {
! 263: int isr_nhops; /* number of hops */
! 264: struct in_addr isr_dst; /* final destination */
! 265: char isr_nop; /* one NOP to align */
! 266: char isr_hdr[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN & OFFSET */
! 267: struct in_addr isr_routes[MAX_IPOPTLEN/sizeof(struct in_addr)];
! 268: };
1.1 cgd 269:
1.295 dyoung 270: static int ip_drainwanted;
271:
1.354 ! ozaki-r 272: static void save_rte(struct mbuf *, u_char *, struct in_addr);
1.35 mycroft 273:
1.164 matt 274: #ifdef MBUFTRACE
1.234 dogcow 275: struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx");
276: struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx");
1.164 matt 277: #endif
278:
1.318 rmind 279: static void ipintr(void *);
1.312 rmind 280: static void ip_input(struct mbuf *);
1.329 ozaki-r 281: static void ip_forward(struct mbuf *, int, struct ifnet *);
1.311 rmind 282: static bool ip_dooptions(struct mbuf *);
1.346 ozaki-r 283: static struct in_ifaddr *ip_rtaddr(struct in_addr, struct psref *);
1.311 rmind 284: static void sysctl_net_inet_ip_setup(struct sysctllog **);
1.284 pooka 285:
1.333 ozaki-r 286: static struct in_ifaddr *ip_match_our_address(struct ifnet *, struct ip *,
287: int *);
288: static struct in_ifaddr *ip_match_our_address_broadcast(struct ifnet *,
289: struct ip *);
290:
1.344 ozaki-r 291: #ifdef NET_MPSAFE
292: #define SOFTNET_LOCK() mutex_enter(softnet_lock)
293: #define SOFTNET_UNLOCK() mutex_exit(softnet_lock)
294: #else
1.318 rmind 295: #define SOFTNET_LOCK() KASSERT(mutex_owned(softnet_lock))
296: #define SOFTNET_UNLOCK() KASSERT(mutex_owned(softnet_lock))
1.344 ozaki-r 297: #endif
1.318 rmind 298:
1.1 cgd 299: /*
300: * IP initialization: fill in IP protocol switch table.
301: * All protocols not implemented in kernel go to raw IP protocol handler.
302: */
1.8 mycroft 303: void
1.211 perry 304: ip_init(void)
1.1 cgd 305: {
1.199 matt 306: const struct protosw *pr;
1.1 cgd 307:
1.311 rmind 308: in_init();
1.284 pooka 309: sysctl_net_inet_ip_setup(NULL);
310:
1.311 rmind 311: pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW);
312: KASSERT(pr != NULL);
1.275 pooka 313:
1.319 ozaki-r 314: ip_pktq = pktq_create(IFQ_MAXLEN, ipintr, NULL);
1.318 rmind 315: KASSERT(ip_pktq != NULL);
316:
1.328 riastrad 317: for (u_int i = 0; i < IPPROTO_MAX; i++) {
318: ip_protox[i] = pr - inetsw;
1.311 rmind 319: }
1.328 riastrad 320: for (pr = inetdomain.dom_protosw;
321: pr < inetdomain.dom_protoswNPROTOSW; pr++)
1.1 cgd 322: if (pr->pr_domain->dom_family == PF_INET &&
323: pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
1.328 riastrad 324: ip_protox[pr->pr_protocol] = pr - inetsw;
1.192 jonathan 325:
1.288 rmind 326: ip_reass_init();
1.190 jonathan 327:
1.291 rmind 328: ip_ids = ip_id_init();
1.323 ozaki-r 329: ip_id = time_uptime & 0xfffff;
1.194 jonathan 330:
1.73 thorpej 331: #ifdef GATEWAY
1.311 rmind 332: ipflow_init();
1.73 thorpej 333: #endif
1.121 thorpej 334:
335: /* Register our Packet Filter hook. */
1.308 rmind 336: inet_pfil_hook = pfil_head_create(PFIL_TYPE_AF, (void *)AF_INET);
337: KASSERT(inet_pfil_hook != NULL);
1.135 thorpej 338:
1.164 matt 339: #ifdef MBUFTRACE
340: MOWNER_ATTACH(&ip_tx_mowner);
341: MOWNER_ATTACH(&ip_rx_mowner);
342: #endif /* MBUFTRACE */
1.266 thorpej 343:
344: ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS);
1.340 ozaki-r 345:
346: ipforward_rt_percpu = percpu_alloc(sizeof(struct route));
347: if (ipforward_rt_percpu == NULL)
348: panic("failed to allocate ipforward_rt_percpu");
1.352 ozaki-r 349:
350: ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout);
1.1 cgd 351: }
352:
1.333 ozaki-r 353: static struct in_ifaddr *
354: ip_match_our_address(struct ifnet *ifp, struct ip *ip, int *downmatch)
355: {
356: struct in_ifaddr *ia = NULL;
357: int checkif;
358:
359: /*
360: * Enable a consistency check between the destination address
361: * and the arrival interface for a unicast packet (the RFC 1122
362: * strong ES model) if IP forwarding is disabled and the packet
363: * is not locally generated.
364: *
365: * XXX - Checking also should be disabled if the destination
366: * address is ipnat'ed to a different interface.
367: *
368: * XXX - Checking is incompatible with IP aliases added
369: * to the loopback interface instead of the interface where
370: * the packets are received.
371: *
372: * XXX - We need to add a per ifaddr flag for this so that
373: * we get finer grain control.
374: */
375: checkif = ip_checkinterface && (ipforwarding == 0) &&
376: (ifp->if_flags & IFF_LOOPBACK) == 0;
377:
1.334 ozaki-r 378: IN_ADDRHASH_READER_FOREACH(ia, ip->ip_dst.s_addr) {
1.333 ozaki-r 379: if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) {
380: if (ia->ia4_flags & IN_IFF_NOTREADY)
381: continue;
382: if (checkif && ia->ia_ifp != ifp)
383: continue;
1.341 roy 384: if ((ia->ia_ifp->if_flags & IFF_UP) != 0 &&
385: (ia->ia4_flags & IN_IFF_DETACHED) == 0)
1.333 ozaki-r 386: break;
387: else
1.338 ozaki-r 388: (*downmatch)++;
1.333 ozaki-r 389: }
390: }
391:
392: return ia;
393: }
394:
395: static struct in_ifaddr *
396: ip_match_our_address_broadcast(struct ifnet *ifp, struct ip *ip)
397: {
398: struct in_ifaddr *ia = NULL;
399: struct ifaddr *ifa;
400:
1.336 ozaki-r 401: IFADDR_READER_FOREACH(ifa, ifp) {
1.333 ozaki-r 402: if (ifa->ifa_addr->sa_family != AF_INET)
403: continue;
404: ia = ifatoia(ifa);
1.341 roy 405: if (ia->ia4_flags & (IN_IFF_NOTREADY | IN_IFF_DETACHED))
1.333 ozaki-r 406: continue;
407: if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) ||
408: in_hosteq(ip->ip_dst, ia->ia_netbroadcast) ||
409: /*
410: * Look for all-0's host part (old broadcast addr),
411: * either for subnet or net.
412: */
413: ip->ip_dst.s_addr == ia->ia_subnet ||
414: ip->ip_dst.s_addr == ia->ia_net)
415: goto matched;
416: /*
417: * An interface with IP address zero accepts
418: * all packets that arrive on that interface.
419: */
420: if (in_nullhost(ia->ia_addr.sin_addr))
421: goto matched;
422: }
423: ia = NULL;
424:
425: matched:
426: return ia;
427: }
428:
1.1 cgd 429: /*
1.318 rmind 430: * IP software interrupt routine.
1.89 itojun 431: */
1.318 rmind 432: static void
433: ipintr(void *arg __unused)
1.89 itojun 434: {
435: struct mbuf *m;
1.286 tls 436:
1.318 rmind 437: KASSERT(cpu_softintr_p());
1.89 itojun 438:
1.344 ozaki-r 439: #ifndef NET_MPSAFE
1.268 ad 440: mutex_enter(softnet_lock);
1.344 ozaki-r 441: #endif
1.318 rmind 442: while ((m = pktq_dequeue(ip_pktq)) != NULL) {
1.89 itojun 443: ip_input(m);
444: }
1.344 ozaki-r 445: #ifndef NET_MPSAFE
1.268 ad 446: mutex_exit(softnet_lock);
1.344 ozaki-r 447: #endif
1.89 itojun 448: }
449:
450: /*
1.318 rmind 451: * IP input routine. Checksum and byte swap header. If fragmented
1.1 cgd 452: * try to reassemble. Process options. Pass to next level.
453: */
1.312 rmind 454: static void
1.89 itojun 455: ip_input(struct mbuf *m)
1.1 cgd 456: {
1.109 augustss 457: struct ip *ip = NULL;
1.339 ozaki-r 458: struct in_ifaddr *ia = NULL;
1.288 rmind 459: int hlen = 0, len;
1.100 itojun 460: int downmatch;
1.169 itojun 461: int srcrt = 0;
1.314 rmind 462: ifnet_t *ifp;
1.329 ozaki-r 463: struct psref psref;
1.339 ozaki-r 464: int s;
1.1 cgd 465:
1.318 rmind 466: KASSERTMSG(cpu_softintr_p(), "ip_input: not in the software "
467: "interrupt handler; synchronization assumptions violated");
468:
1.164 matt 469: MCLAIM(m, &ip_rx_mowner);
1.289 rmind 470: KASSERT((m->m_flags & M_PKTHDR) != 0);
1.329 ozaki-r 471:
472: ifp = m_get_rcvif_psref(m, &psref);
1.330 ozaki-r 473: if (__predict_false(ifp == NULL))
1.332 ozaki-r 474: goto out;
1.164 matt 475:
1.1 cgd 476: /*
477: * If no IP addresses have been set yet but the interfaces
478: * are receiving, can't do anything with incoming packets yet.
1.318 rmind 479: * Note: we pre-check without locks held.
1.1 cgd 480: */
1.335 ozaki-r 481: if (IN_ADDRLIST_READER_EMPTY())
1.332 ozaki-r 482: goto out;
1.266 thorpej 483: IP_STATINC(IP_STAT_TOTAL);
1.318 rmind 484:
1.154 thorpej 485: /*
486: * If the IP header is not aligned, slurp it up into a new
487: * mbuf with space for link headers, in the event we forward
488: * it. Otherwise, if it is aligned, make sure the entire
489: * base IP header is in the first mbuf of the chain.
490: */
1.244 christos 491: if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) {
1.154 thorpej 492: if ((m = m_copyup(m, sizeof(struct ip),
493: (max_linkhdr + 3) & ~3)) == NULL) {
494: /* XXXJRT new stat, please */
1.266 thorpej 495: IP_STATINC(IP_STAT_TOOSMALL);
1.329 ozaki-r 496: goto out;
1.154 thorpej 497: }
498: } else if (__predict_false(m->m_len < sizeof (struct ip))) {
499: if ((m = m_pullup(m, sizeof (struct ip))) == NULL) {
1.266 thorpej 500: IP_STATINC(IP_STAT_TOOSMALL);
1.329 ozaki-r 501: goto out;
1.154 thorpej 502: }
1.1 cgd 503: }
504: ip = mtod(m, struct ip *);
1.13 mycroft 505: if (ip->ip_v != IPVERSION) {
1.266 thorpej 506: IP_STATINC(IP_STAT_BADVERS);
1.332 ozaki-r 507: goto out;
1.13 mycroft 508: }
1.1 cgd 509: hlen = ip->ip_hl << 2;
510: if (hlen < sizeof(struct ip)) { /* minimum header length */
1.266 thorpej 511: IP_STATINC(IP_STAT_BADHLEN);
1.332 ozaki-r 512: goto out;
1.1 cgd 513: }
514: if (hlen > m->m_len) {
1.298 liamjfoy 515: if ((m = m_pullup(m, hlen)) == NULL) {
1.266 thorpej 516: IP_STATINC(IP_STAT_BADHLEN);
1.329 ozaki-r 517: goto out;
1.1 cgd 518: }
519: ip = mtod(m, struct ip *);
520: }
1.98 thorpej 521:
1.85 hwr 522: /*
1.99 thorpej 523: * RFC1122: packets with a multicast source address are
1.98 thorpej 524: * not allowed.
1.85 hwr 525: */
526: if (IN_MULTICAST(ip->ip_src.s_addr)) {
1.266 thorpej 527: IP_STATINC(IP_STAT_BADADDR);
1.332 ozaki-r 528: goto out;
1.129 itojun 529: }
530:
531: /* 127/8 must not appear on wire - RFC1122 */
532: if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
533: (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
1.314 rmind 534: if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
1.266 thorpej 535: IP_STATINC(IP_STAT_BADADDR);
1.332 ozaki-r 536: goto out;
1.130 itojun 537: }
1.85 hwr 538: }
539:
1.135 thorpej 540: switch (m->m_pkthdr.csum_flags &
1.314 rmind 541: ((ifp->if_csum_flags_rx & M_CSUM_IPv4) |
1.135 thorpej 542: M_CSUM_IPv4_BAD)) {
543: case M_CSUM_IPv4|M_CSUM_IPv4_BAD:
544: INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad);
1.332 ozaki-r 545: IP_STATINC(IP_STAT_BADSUM);
546: goto out;
1.135 thorpej 547:
548: case M_CSUM_IPv4:
549: /* Checksum was okay. */
550: INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok);
551: break;
552:
553: default:
1.206 thorpej 554: /*
555: * Must compute it ourselves. Maybe skip checksum on
556: * loopback interfaces.
557: */
1.314 rmind 558: if (__predict_true(!(ifp->if_flags & IFF_LOOPBACK) ||
559: ip_do_loopback_cksum)) {
1.206 thorpej 560: INET_CSUM_COUNTER_INCR(&ip_swcsum);
1.332 ozaki-r 561: if (in_cksum(m, hlen) != 0) {
562: IP_STATINC(IP_STAT_BADSUM);
563: goto out;
564: }
1.206 thorpej 565: }
1.135 thorpej 566: break;
1.1 cgd 567: }
568:
1.121 thorpej 569: /* Retrieve the packet length. */
570: len = ntohs(ip->ip_len);
1.81 proff 571:
572: /*
573: * Check for additional length bogosity
574: */
1.84 proff 575: if (len < hlen) {
1.266 thorpej 576: IP_STATINC(IP_STAT_BADLEN);
1.332 ozaki-r 577: goto out;
1.81 proff 578: }
1.1 cgd 579:
580: /*
581: * Check that the amount of data in the buffers
582: * is as at least much as the IP header would have us expect.
583: * Trim mbufs if longer than we expect.
584: * Drop packet if shorter than we expect.
585: */
1.35 mycroft 586: if (m->m_pkthdr.len < len) {
1.266 thorpej 587: IP_STATINC(IP_STAT_TOOSHORT);
1.332 ozaki-r 588: goto out;
1.1 cgd 589: }
1.35 mycroft 590: if (m->m_pkthdr.len > len) {
1.1 cgd 591: if (m->m_len == m->m_pkthdr.len) {
1.35 mycroft 592: m->m_len = len;
593: m->m_pkthdr.len = len;
1.1 cgd 594: } else
1.35 mycroft 595: m_adj(m, len - m->m_pkthdr.len);
1.1 cgd 596: }
597:
1.64 thorpej 598: /*
599: * Assume that we can create a fast-forward IP flow entry
600: * based on this packet.
601: */
602: m->m_flags |= M_CANFASTFWD;
603:
1.33 mrg 604: /*
1.64 thorpej 605: * Run through list of hooks for input packets. If there are any
606: * filters which require that additional packets in the flow are
607: * not fast-forwarded, they must clear the M_CANFASTFWD flag.
608: * Note that filters must _never_ set this flag, as another filter
609: * in the list may have previously cleared it.
1.33 mrg 610: */
1.304 christos 611: #if defined(IPSEC)
1.317 christos 612: if (!ipsec_used || !ipsec_indone(m))
1.127 itojun 613: #else
614: if (1)
615: #endif
616: {
1.314 rmind 617: struct in_addr odst = ip->ip_dst;
1.318 rmind 618: bool freed;
1.169 itojun 619:
1.318 rmind 620: freed = pfil_run_hooks(inet_pfil_hook, &m, ifp, PFIL_IN) != 0;
621: if (freed || m == NULL) {
1.343 ozaki-r 622: m = NULL;
1.329 ozaki-r 623: goto out;
1.318 rmind 624: }
1.127 itojun 625: ip = mtod(m, struct ip *);
1.142 darrenr 626: hlen = ip->ip_hl << 2;
1.318 rmind 627:
1.205 darrenr 628: /*
629: * XXX The setting of "srcrt" here is to prevent ip_forward()
630: * from generating ICMP redirects for packets that have
631: * been redirected by a hook back out on to the same LAN that
632: * they came from and is not an indication that the packet
633: * is being inffluenced by source routing options. This
634: * allows things like
635: * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp"
636: * where tlp0 is both on the 1.1.1.0/24 network and is the
637: * default route for hosts on 1.1.1.0/24. Of course this
638: * also requires a "map tlp0 ..." to complete the story.
639: * One might argue whether or not this kind of network config.
1.212 perry 640: * should be supported in this manner...
1.205 darrenr 641: */
1.169 itojun 642: srcrt = (odst.s_addr != ip->ip_dst.s_addr);
1.127 itojun 643: }
1.123 thorpej 644:
645: #ifdef ALTQ
646: /* XXX Temporary until ALTQ is changed to use a pfil hook */
1.318 rmind 647: if (altq_input) {
648: SOFTNET_LOCK();
649: if ((*altq_input)(m, AF_INET) == 0) {
650: /* Packet dropped by traffic conditioner. */
651: SOFTNET_UNLOCK();
1.343 ozaki-r 652: m = NULL;
1.329 ozaki-r 653: goto out;
1.318 rmind 654: }
655: SOFTNET_UNLOCK();
1.123 thorpej 656: }
657: #endif
1.121 thorpej 658:
659: /*
1.1 cgd 660: * Process options and, if not destined for us,
661: * ship it on. ip_dooptions returns 1 when an
662: * error was detected (causing an icmp message
663: * to be sent and the original packet to be freed).
664: */
1.343 ozaki-r 665: if (hlen > sizeof (struct ip) && ip_dooptions(m)) {
666: m = NULL;
1.329 ozaki-r 667: goto out;
1.343 ozaki-r 668: }
1.1 cgd 669:
670: /*
671: * Check our list of addresses, to see if the packet is for us.
1.100 itojun 672: *
673: * Traditional 4.4BSD did not consult IFF_UP at all.
674: * The behavior here is to treat addresses on !IFF_UP interface
1.321 roy 675: * or IN_IFF_NOTREADY addresses as not mine.
1.1 cgd 676: */
1.100 itojun 677: downmatch = 0;
1.339 ozaki-r 678: s = pserialize_read_enter();
1.333 ozaki-r 679: ia = ip_match_our_address(ifp, ip, &downmatch);
1.339 ozaki-r 680: if (ia != NULL) {
681: pserialize_read_exit(s);
1.86 thorpej 682: goto ours;
1.339 ozaki-r 683: }
1.333 ozaki-r 684:
1.315 christos 685: if (ifp->if_flags & IFF_BROADCAST) {
1.333 ozaki-r 686: ia = ip_match_our_address_broadcast(ifp, ip);
1.339 ozaki-r 687: if (ia != NULL) {
688: pserialize_read_exit(s);
1.333 ozaki-r 689: goto ours;
1.339 ozaki-r 690: }
1.1 cgd 691: }
1.339 ozaki-r 692: pserialize_read_exit(s);
1.333 ozaki-r 693:
1.18 mycroft 694: if (IN_MULTICAST(ip->ip_dst.s_addr)) {
1.4 hpeyerl 695: #ifdef MROUTING
696: extern struct socket *ip_mrouter;
1.10 brezak 697:
1.4 hpeyerl 698: if (ip_mrouter) {
699: /*
700: * If we are acting as a multicast router, all
701: * incoming multicast packets are passed to the
702: * kernel-level multicast forwarding function.
703: * The packet is returned (relatively) intact; if
704: * ip_mforward() returns a non-zero value, the packet
705: * must be discarded, else it may be accepted below.
706: *
707: * (The IP ident field is put in the same byte order
708: * as expected when ip_mforward() is called from
709: * ip_output().)
710: */
1.318 rmind 711: SOFTNET_LOCK();
1.314 rmind 712: if (ip_mforward(m, ifp) != 0) {
1.318 rmind 713: SOFTNET_UNLOCK();
1.266 thorpej 714: IP_STATINC(IP_STAT_CANTFORWARD);
1.332 ozaki-r 715: goto out;
1.4 hpeyerl 716: }
1.318 rmind 717: SOFTNET_UNLOCK();
1.4 hpeyerl 718:
719: /*
720: * The process-level routing demon needs to receive
721: * all multicast IGMP packets, whether or not this
722: * host belongs to their destination groups.
723: */
1.318 rmind 724: if (ip->ip_p == IPPROTO_IGMP) {
1.4 hpeyerl 725: goto ours;
1.318 rmind 726: }
1.266 thorpej 727: IP_STATINC(IP_STAT_CANTFORWARD);
1.4 hpeyerl 728: }
729: #endif
730: /*
731: * See if we belong to the destination multicast group on the
732: * arrival interface.
733: */
1.316 rmind 734: if (!in_multi_group(ip->ip_dst, ifp, 0)) {
1.266 thorpej 735: IP_STATINC(IP_STAT_CANTFORWARD);
1.332 ozaki-r 736: goto out;
1.4 hpeyerl 737: }
738: goto ours;
739: }
1.19 mycroft 740: if (ip->ip_dst.s_addr == INADDR_BROADCAST ||
1.35 mycroft 741: in_nullhost(ip->ip_dst))
1.1 cgd 742: goto ours;
743:
744: /*
745: * Not for us; forward if possible and desirable.
746: */
747: if (ipforwarding == 0) {
1.329 ozaki-r 748: m_put_rcvif_psref(ifp, &psref);
1.266 thorpej 749: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 750: m_freem(m);
1.100 itojun 751: } else {
752: /*
753: * If ip_dst matched any of my address on !IFF_UP interface,
754: * and there's no IFF_UP interface that matches ip_dst,
755: * send icmp unreach. Forwarding it will result in in-kernel
756: * forwarding loop till TTL goes to 0.
757: */
758: if (downmatch) {
1.329 ozaki-r 759: m_put_rcvif_psref(ifp, &psref);
1.100 itojun 760: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0);
1.266 thorpej 761: IP_STATINC(IP_STAT_CANTFORWARD);
1.100 itojun 762: return;
763: }
1.304 christos 764: #ifdef IPSEC
1.305 rmind 765: /* Perform IPsec, if any. */
1.318 rmind 766: if (ipsec_used) {
767: SOFTNET_LOCK();
768: if (ipsec4_input(m, IP_FORWARDING |
769: (ip_directedbcast ? IP_ALLOWBROADCAST : 0)) != 0) {
770: SOFTNET_UNLOCK();
1.332 ozaki-r 771: goto out;
1.318 rmind 772: }
773: SOFTNET_UNLOCK();
1.173 jonathan 774: }
1.305 rmind 775: #endif
1.329 ozaki-r 776: ip_forward(m, srcrt, ifp);
777: m_put_rcvif_psref(ifp, &psref);
1.100 itojun 778: }
1.89 itojun 779: return;
1.1 cgd 780:
781: ours:
1.329 ozaki-r 782: m_put_rcvif_psref(ifp, &psref);
1.331 ozaki-r 783: ifp = NULL;
1.329 ozaki-r 784:
1.1 cgd 785: /*
786: * If offset or IP_MF are set, must reassemble.
787: */
1.155 itojun 788: if (ip->ip_off & ~htons(IP_DF|IP_RF)) {
1.289 rmind 789: /*
790: * Pass to IP reassembly mechanism.
791: */
1.290 rmind 792: if (ip_reass_packet(&m, ip) != 0) {
1.289 rmind 793: /* Failed; invalid fragment(s) or packet. */
1.332 ozaki-r 794: goto out;
1.16 cgd 795: }
1.290 rmind 796: if (m == NULL) {
1.289 rmind 797: /* More fragments should come; silently return. */
1.332 ozaki-r 798: goto out;
1.289 rmind 799: }
1.290 rmind 800: /*
801: * Reassembly is done, we have the final packet.
802: * Updated cached data in local variable(s).
803: */
1.289 rmind 804: ip = mtod(m, struct ip *);
805: hlen = ip->ip_hl << 2;
1.79 mycroft 806: }
1.128 itojun 807:
1.304 christos 808: #ifdef IPSEC
1.173 jonathan 809: /*
1.305 rmind 810: * Enforce IPsec policy checking if we are seeing last header.
811: * Note that we do not visit this with protocols with PCB layer
812: * code - like UDP/TCP/raw IP.
1.173 jonathan 813: */
1.317 christos 814: if (ipsec_used &&
1.328 riastrad 815: (inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) {
1.318 rmind 816: SOFTNET_LOCK();
1.305 rmind 817: if (ipsec4_input(m, 0) != 0) {
1.318 rmind 818: SOFTNET_UNLOCK();
1.332 ozaki-r 819: goto out;
1.173 jonathan 820: }
1.318 rmind 821: SOFTNET_UNLOCK();
1.173 jonathan 822: }
1.305 rmind 823: #endif
1.1 cgd 824:
825: /*
826: * Switch out to protocol's input routine.
827: */
1.82 aidan 828: #if IFA_STATS
1.339 ozaki-r 829: if (ia && ip) {
830: struct in_ifaddr *_ia;
831: /*
832: * Keep a reference from ip_match_our_address with psref
833: * is expensive, so explore ia here again.
834: */
835: s = pserialize_read_enter();
1.342 ozaki-r 836: _ia = in_get_ia(ip->ip_dst);
1.339 ozaki-r 837: _ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len);
838: pserialize_read_exit(s);
839: }
1.82 aidan 840: #endif
1.266 thorpej 841: IP_STATINC(IP_STAT_DELIVERED);
1.89 itojun 842:
1.318 rmind 843: const int off = hlen, nh = ip->ip_p;
844:
845: SOFTNET_LOCK();
1.328 riastrad 846: (*inetsw[ip_protox[nh]].pr_input)(m, off, nh);
1.318 rmind 847: SOFTNET_UNLOCK();
1.89 itojun 848: return;
1.135 thorpej 849:
1.329 ozaki-r 850: out:
851: m_put_rcvif_psref(ifp, &psref);
1.332 ozaki-r 852: if (m != NULL)
853: m_freem(m);
1.1 cgd 854: }
855:
856: /*
1.288 rmind 857: * IP timer processing.
1.1 cgd 858: */
1.8 mycroft 859: void
1.211 perry 860: ip_slowtimo(void)
1.1 cgd 861: {
1.268 ad 862:
1.348 ozaki-r 863: #ifndef NET_MPSAFE
1.268 ad 864: mutex_enter(softnet_lock);
865: KERNEL_LOCK(1, NULL);
1.348 ozaki-r 866: #endif
1.1 cgd 867:
1.288 rmind 868: ip_reass_slowtimo();
1.268 ad 869:
1.348 ozaki-r 870: #ifndef NET_MPSAFE
1.268 ad 871: KERNEL_UNLOCK_ONE(NULL);
872: mutex_exit(softnet_lock);
1.348 ozaki-r 873: #endif
1.1 cgd 874: }
875:
876: /*
1.288 rmind 877: * IP drain processing.
1.1 cgd 878: */
1.8 mycroft 879: void
1.211 perry 880: ip_drain(void)
1.1 cgd 881: {
882:
1.268 ad 883: KERNEL_LOCK(1, NULL);
1.288 rmind 884: ip_reass_drain();
1.268 ad 885: KERNEL_UNLOCK_ONE(NULL);
1.1 cgd 886: }
887:
888: /*
1.311 rmind 889: * ip_dooptions: perform option processing on a datagram, possibly discarding
890: * it if bad options are encountered, or forwarding it if source-routed.
891: *
892: * => Returns true if packet has been forwarded/freed.
893: * => Returns false if the packet should be processed further.
1.1 cgd 894: */
1.311 rmind 895: static bool
1.211 perry 896: ip_dooptions(struct mbuf *m)
1.1 cgd 897: {
1.109 augustss 898: struct ip *ip = mtod(m, struct ip *);
899: u_char *cp, *cp0;
900: struct ip_timestamp *ipt;
901: struct in_ifaddr *ia;
1.1 cgd 902: int opt, optlen, cnt, off, code, type = ICMP_PARAMPROB, forward = 0;
1.104 thorpej 903: struct in_addr dst;
1.1 cgd 904: n_time ntime;
1.349 ozaki-r 905: struct ifaddr *ifa = NULL;
1.339 ozaki-r 906: int s;
1.1 cgd 907:
1.13 mycroft 908: dst = ip->ip_dst;
1.1 cgd 909: cp = (u_char *)(ip + 1);
910: cnt = (ip->ip_hl << 2) - sizeof (struct ip);
911: for (; cnt > 0; cnt -= optlen, cp += optlen) {
912: opt = cp[IPOPT_OPTVAL];
913: if (opt == IPOPT_EOL)
914: break;
915: if (opt == IPOPT_NOP)
916: optlen = 1;
917: else {
1.113 itojun 918: if (cnt < IPOPT_OLEN + sizeof(*cp)) {
919: code = &cp[IPOPT_OLEN] - (u_char *)ip;
920: goto bad;
921: }
1.1 cgd 922: optlen = cp[IPOPT_OLEN];
1.114 itojun 923: if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) {
1.1 cgd 924: code = &cp[IPOPT_OLEN] - (u_char *)ip;
925: goto bad;
926: }
927: }
928: switch (opt) {
929:
930: default:
931: break;
932:
933: /*
934: * Source routing with record.
935: * Find interface with current destination address.
936: * If none on this machine then drop if strictly routed,
937: * or do nothing if loosely routed.
938: * Record interface address and bring up next address
939: * component. If strictly routed make sure next
940: * address is on directly accessible net.
941: */
942: case IPOPT_LSRR:
1.346 ozaki-r 943: case IPOPT_SSRR: {
944: struct psref psref;
1.353 ozaki-r 945: struct sockaddr_in ipaddr = {
946: .sin_len = sizeof(ipaddr),
947: .sin_family = AF_INET,
948: };
949:
1.47 cjs 950: if (ip_allowsrcrt == 0) {
951: type = ICMP_UNREACH;
952: code = ICMP_UNREACH_NET_PROHIB;
953: goto bad;
954: }
1.114 itojun 955: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
956: code = &cp[IPOPT_OLEN] - (u_char *)ip;
957: goto bad;
958: }
1.1 cgd 959: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
960: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
961: goto bad;
962: }
963: ipaddr.sin_addr = ip->ip_dst;
1.339 ozaki-r 964:
965: s = pserialize_read_enter();
966: ifa = ifa_ifwithaddr(sintosa(&ipaddr));
967: if (ifa == NULL) {
968: pserialize_read_exit(s);
1.1 cgd 969: if (opt == IPOPT_SSRR) {
970: type = ICMP_UNREACH;
971: code = ICMP_UNREACH_SRCFAIL;
972: goto bad;
973: }
974: /*
975: * Loose routing, and not at next destination
976: * yet; nothing to do except forward.
977: */
978: break;
979: }
1.339 ozaki-r 980: pserialize_read_exit(s);
981:
1.1 cgd 982: off--; /* 0 origin */
1.112 sommerfe 983: if ((off + sizeof(struct in_addr)) > optlen) {
1.1 cgd 984: /*
985: * End of source route. Should be for us.
986: */
1.354 ! ozaki-r 987: save_rte(m, cp, ip->ip_src);
1.1 cgd 988: break;
989: }
990: /*
991: * locate outgoing interface
992: */
1.281 tsutsui 993: memcpy((void *)&ipaddr.sin_addr, (void *)(cp + off),
1.1 cgd 994: sizeof(ipaddr.sin_addr));
1.339 ozaki-r 995: if (opt == IPOPT_SSRR) {
1.346 ozaki-r 996: ifa = ifa_ifwithladdr_psref(sintosa(&ipaddr),
997: &psref);
1.339 ozaki-r 998: if (ifa != NULL)
999: ia = ifatoia(ifa);
1000: else
1001: ia = NULL;
1002: } else {
1.346 ozaki-r 1003: ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1.339 ozaki-r 1004: }
1005: if (ia == NULL) {
1.1 cgd 1006: type = ICMP_UNREACH;
1007: code = ICMP_UNREACH_SRCFAIL;
1008: goto bad;
1009: }
1010: ip->ip_dst = ipaddr.sin_addr;
1.244 christos 1011: bcopy((void *)&ia->ia_addr.sin_addr,
1012: (void *)(cp + off), sizeof(struct in_addr));
1.346 ozaki-r 1013: ia4_release(ia, &psref);
1.1 cgd 1014: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1.13 mycroft 1015: /*
1016: * Let ip_intr's mcast routing check handle mcast pkts
1017: */
1.18 mycroft 1018: forward = !IN_MULTICAST(ip->ip_dst.s_addr);
1.1 cgd 1019: break;
1.346 ozaki-r 1020: }
1.1 cgd 1021:
1.346 ozaki-r 1022: case IPOPT_RR: {
1023: struct psref psref;
1.353 ozaki-r 1024: struct sockaddr_in ipaddr = {
1025: .sin_len = sizeof(ipaddr),
1026: .sin_family = AF_INET,
1027: };
1028:
1.114 itojun 1029: if (optlen < IPOPT_OFFSET + sizeof(*cp)) {
1030: code = &cp[IPOPT_OLEN] - (u_char *)ip;
1031: goto bad;
1032: }
1.1 cgd 1033: if ((off = cp[IPOPT_OFFSET]) < IPOPT_MINOFF) {
1034: code = &cp[IPOPT_OFFSET] - (u_char *)ip;
1035: goto bad;
1036: }
1037: /*
1038: * If no space remains, ignore.
1039: */
1040: off--; /* 0 origin */
1.112 sommerfe 1041: if ((off + sizeof(struct in_addr)) > optlen)
1.1 cgd 1042: break;
1.281 tsutsui 1043: memcpy((void *)&ipaddr.sin_addr, (void *)(&ip->ip_dst),
1.1 cgd 1044: sizeof(ipaddr.sin_addr));
1045: /*
1046: * locate outgoing interface; if we're the destination,
1047: * use the incoming interface (should be same).
1048: */
1.346 ozaki-r 1049: ifa = ifa_ifwithaddr_psref(sintosa(&ipaddr), &psref);
1.339 ozaki-r 1050: if (ifa == NULL) {
1.346 ozaki-r 1051: ia = ip_rtaddr(ipaddr.sin_addr, &psref);
1.339 ozaki-r 1052: if (ia == NULL) {
1053: type = ICMP_UNREACH;
1054: code = ICMP_UNREACH_HOST;
1055: goto bad;
1056: }
1057: } else {
1058: ia = ifatoia(ifa);
1.1 cgd 1059: }
1.244 christos 1060: bcopy((void *)&ia->ia_addr.sin_addr,
1061: (void *)(cp + off), sizeof(struct in_addr));
1.346 ozaki-r 1062: ia4_release(ia, &psref);
1.1 cgd 1063: cp[IPOPT_OFFSET] += sizeof(struct in_addr);
1064: break;
1.346 ozaki-r 1065: }
1.1 cgd 1066:
1067: case IPOPT_TS:
1068: code = cp - (u_char *)ip;
1069: ipt = (struct ip_timestamp *)cp;
1.114 itojun 1070: if (ipt->ipt_len < 4 || ipt->ipt_len > 40) {
1071: code = (u_char *)&ipt->ipt_len - (u_char *)ip;
1.1 cgd 1072: goto bad;
1.114 itojun 1073: }
1074: if (ipt->ipt_ptr < 5) {
1075: code = (u_char *)&ipt->ipt_ptr - (u_char *)ip;
1076: goto bad;
1077: }
1.15 cgd 1078: if (ipt->ipt_ptr > ipt->ipt_len - sizeof (int32_t)) {
1.114 itojun 1079: if (++ipt->ipt_oflw == 0) {
1080: code = (u_char *)&ipt->ipt_ptr -
1081: (u_char *)ip;
1.1 cgd 1082: goto bad;
1.114 itojun 1083: }
1.1 cgd 1084: break;
1085: }
1.104 thorpej 1086: cp0 = (cp + ipt->ipt_ptr - 1);
1.1 cgd 1087: switch (ipt->ipt_flg) {
1088:
1089: case IPOPT_TS_TSONLY:
1090: break;
1091:
1.329 ozaki-r 1092: case IPOPT_TS_TSANDADDR: {
1093: struct ifnet *rcvif;
1.339 ozaki-r 1094: int _s, _ss;
1.353 ozaki-r 1095: struct sockaddr_in ipaddr = {
1096: .sin_len = sizeof(ipaddr),
1097: .sin_family = AF_INET,
1098: };
1.329 ozaki-r 1099:
1.66 thorpej 1100: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1101: sizeof(struct in_addr) > ipt->ipt_len) {
1102: code = (u_char *)&ipt->ipt_ptr -
1103: (u_char *)ip;
1.1 cgd 1104: goto bad;
1.114 itojun 1105: }
1.13 mycroft 1106: ipaddr.sin_addr = dst;
1.339 ozaki-r 1107: _ss = pserialize_read_enter();
1108: rcvif = m_get_rcvif(m, &_s);
1.349 ozaki-r 1109: if (__predict_true(rcvif != NULL)) {
1110: ifa = ifaof_ifpforaddr(sintosa(&ipaddr),
1111: rcvif);
1112: }
1.339 ozaki-r 1113: m_put_rcvif(rcvif, &_s);
1114: if (ifa == NULL) {
1115: pserialize_read_exit(_ss);
1116: break;
1117: }
1118: ia = ifatoia(ifa);
1.104 thorpej 1119: bcopy(&ia->ia_addr.sin_addr,
1120: cp0, sizeof(struct in_addr));
1.339 ozaki-r 1121: pserialize_read_exit(_ss);
1.1 cgd 1122: ipt->ipt_ptr += sizeof(struct in_addr);
1123: break;
1.329 ozaki-r 1124: }
1.1 cgd 1125:
1.353 ozaki-r 1126: case IPOPT_TS_PRESPEC: {
1127: struct sockaddr_in ipaddr = {
1128: .sin_len = sizeof(ipaddr),
1129: .sin_family = AF_INET,
1130: };
1131:
1.66 thorpej 1132: if (ipt->ipt_ptr - 1 + sizeof(n_time) +
1.114 itojun 1133: sizeof(struct in_addr) > ipt->ipt_len) {
1134: code = (u_char *)&ipt->ipt_ptr -
1135: (u_char *)ip;
1.1 cgd 1136: goto bad;
1.114 itojun 1137: }
1.281 tsutsui 1138: memcpy(&ipaddr.sin_addr, cp0,
1.1 cgd 1139: sizeof(struct in_addr));
1.339 ozaki-r 1140: s = pserialize_read_enter();
1141: ifa = ifa_ifwithaddr(sintosa(&ipaddr));
1142: if (ifa == NULL) {
1143: pserialize_read_exit(s);
1.1 cgd 1144: continue;
1.339 ozaki-r 1145: }
1146: pserialize_read_exit(s);
1.1 cgd 1147: ipt->ipt_ptr += sizeof(struct in_addr);
1148: break;
1.353 ozaki-r 1149: }
1.1 cgd 1150:
1151: default:
1.114 itojun 1152: /* XXX can't take &ipt->ipt_flg */
1153: code = (u_char *)&ipt->ipt_ptr -
1154: (u_char *)ip + 1;
1.1 cgd 1155: goto bad;
1156: }
1157: ntime = iptime();
1.107 thorpej 1158: cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */
1.244 christos 1159: memmove((char *)cp + ipt->ipt_ptr - 1, cp0,
1.1 cgd 1160: sizeof(n_time));
1161: ipt->ipt_ptr += sizeof(n_time);
1162: }
1163: }
1164: if (forward) {
1.329 ozaki-r 1165: struct ifnet *rcvif;
1.339 ozaki-r 1166: struct psref _psref;
1.329 ozaki-r 1167:
1.26 thorpej 1168: if (ip_forwsrcrt == 0) {
1169: type = ICMP_UNREACH;
1170: code = ICMP_UNREACH_SRCFAIL;
1171: goto bad;
1172: }
1.329 ozaki-r 1173:
1.339 ozaki-r 1174: rcvif = m_get_rcvif_psref(m, &_psref);
1.330 ozaki-r 1175: if (__predict_false(rcvif == NULL)) {
1176: type = ICMP_UNREACH;
1177: code = ICMP_UNREACH_HOST;
1178: goto bad;
1179: }
1.329 ozaki-r 1180: ip_forward(m, 1, rcvif);
1.339 ozaki-r 1181: m_put_rcvif_psref(rcvif, &_psref);
1.311 rmind 1182: return true;
1.13 mycroft 1183: }
1.311 rmind 1184: return false;
1.1 cgd 1185: bad:
1.13 mycroft 1186: icmp_error(m, type, code, 0, 0);
1.266 thorpej 1187: IP_STATINC(IP_STAT_BADOPTIONS);
1.311 rmind 1188: return true;
1.1 cgd 1189: }
1190:
1191: /*
1.311 rmind 1192: * ip_rtaddr: given address of next destination (final or next hop),
1.1 cgd 1193: * return internet address info of interface to be used to get there.
1194: */
1.311 rmind 1195: static struct in_ifaddr *
1.346 ozaki-r 1196: ip_rtaddr(struct in_addr dst, struct psref *psref)
1.1 cgd 1197: {
1.249 dyoung 1198: struct rtentry *rt;
1199: union {
1200: struct sockaddr dst;
1201: struct sockaddr_in dst4;
1202: } u;
1.340 ozaki-r 1203: struct route *ro;
1.249 dyoung 1204:
1205: sockaddr_in_init(&u.dst4, &dst, 0);
1206:
1.340 ozaki-r 1207: ro = percpu_getref(ipforward_rt_percpu);
1208: rt = rtcache_lookup(ro, &u.dst);
1.345 ozaki-r 1209: if (rt == NULL) {
1210: percpu_putref(ipforward_rt_percpu);
1211: return NULL;
1212: }
1213:
1.346 ozaki-r 1214: ia4_acquire(ifatoia(rt->rt_ifa), psref);
1.345 ozaki-r 1215: rtcache_unref(rt, ro);
1.340 ozaki-r 1216: percpu_putref(ipforward_rt_percpu);
1.249 dyoung 1217:
1218: return ifatoia(rt->rt_ifa);
1.1 cgd 1219: }
1220:
1221: /*
1.311 rmind 1222: * save_rte: save incoming source route for use in replies, to be picked
1223: * up later by ip_srcroute if the receiver is interested.
1.1 cgd 1224: */
1.311 rmind 1225: static void
1.354 ! ozaki-r 1226: save_rte(struct mbuf *m, u_char *option, struct in_addr dst)
1.1 cgd 1227: {
1.354 ! ozaki-r 1228: struct ip_srcrt *isr;
! 1229: struct m_tag *mtag;
1.1 cgd 1230: unsigned olen;
1231:
1232: olen = option[IPOPT_OLEN];
1.354 ! ozaki-r 1233: if (olen > sizeof(isr->isr_hdr) + sizeof(isr->isr_routes))
! 1234: return;
! 1235:
! 1236: mtag = m_tag_get(PACKET_TAG_SRCROUTE, sizeof(*isr), M_NOWAIT);
! 1237: if (mtag == NULL)
1.1 cgd 1238: return;
1.354 ! ozaki-r 1239: isr = (struct ip_srcrt *)(mtag + 1);
! 1240:
! 1241: memcpy(isr->isr_hdr, option, olen);
! 1242: isr->isr_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr);
! 1243: isr->isr_dst = dst;
! 1244: m_tag_prepend(m, mtag);
1.1 cgd 1245: }
1246:
1247: /*
1248: * Retrieve incoming source route for use in replies,
1249: * in the same form used by setsockopt.
1250: * The first hop is placed before the options, will be removed later.
1251: */
1252: struct mbuf *
1.354 ! ozaki-r 1253: ip_srcroute(struct mbuf *m0)
1.1 cgd 1254: {
1.109 augustss 1255: struct in_addr *p, *q;
1256: struct mbuf *m;
1.354 ! ozaki-r 1257: struct ip_srcrt *isr;
! 1258: struct m_tag *mtag;
! 1259:
! 1260: mtag = m_tag_find(m0, PACKET_TAG_SRCROUTE, NULL);
! 1261: if (mtag == NULL)
! 1262: return NULL;
! 1263: isr = (struct ip_srcrt *)(mtag + 1);
1.1 cgd 1264:
1.354 ! ozaki-r 1265: if (isr->isr_nhops == 0)
1.237 dyoung 1266: return NULL;
1.354 ! ozaki-r 1267:
1.1 cgd 1268: m = m_get(M_DONTWAIT, MT_SOOPTS);
1.354 ! ozaki-r 1269: if (m == NULL)
1.237 dyoung 1270: return NULL;
1.1 cgd 1271:
1.164 matt 1272: MCLAIM(m, &inetdomain.dom_mowner);
1.354 ! ozaki-r 1273: #define OPTSIZ (sizeof(isr->isr_nop) + sizeof(isr->isr_hdr))
1.1 cgd 1274:
1.354 ! ozaki-r 1275: /* length is (nhops+1)*sizeof(addr) + sizeof(nop + header) */
! 1276: m->m_len = (isr->isr_nhops + 1) * sizeof(struct in_addr) + OPTSIZ;
1.1 cgd 1277:
1278: /*
1279: * First save first hop for return route
1280: */
1.354 ! ozaki-r 1281: p = &(isr->isr_routes[isr->isr_nhops - 1]);
1.1 cgd 1282: *(mtod(m, struct in_addr *)) = *p--;
1283:
1284: /*
1285: * Copy option fields and padding (nop) to mbuf.
1286: */
1.354 ! ozaki-r 1287: isr->isr_nop = IPOPT_NOP;
! 1288: isr->isr_hdr[IPOPT_OFFSET] = IPOPT_MINOFF;
! 1289: memmove(mtod(m, char *) + sizeof(struct in_addr), &isr->isr_nop,
1.244 christos 1290: OPTSIZ);
1291: q = (struct in_addr *)(mtod(m, char *) +
1.1 cgd 1292: sizeof(struct in_addr) + OPTSIZ);
1293: #undef OPTSIZ
1294: /*
1295: * Record return path as an IP source route,
1296: * reversing the path (pointers are now aligned).
1297: */
1.354 ! ozaki-r 1298: while (p >= isr->isr_routes) {
1.1 cgd 1299: *q++ = *p--;
1300: }
1301: /*
1302: * Last hop goes to final destination.
1303: */
1.354 ! ozaki-r 1304: *q = isr->isr_dst;
! 1305: m_tag_delete(m0, mtag);
! 1306: return m;
1.1 cgd 1307: }
1308:
1.139 matt 1309: const int inetctlerrmap[PRC_NCMDS] = {
1.256 yamt 1310: [PRC_MSGSIZE] = EMSGSIZE,
1311: [PRC_HOSTDEAD] = EHOSTDOWN,
1312: [PRC_HOSTUNREACH] = EHOSTUNREACH,
1313: [PRC_UNREACH_NET] = EHOSTUNREACH,
1314: [PRC_UNREACH_HOST] = EHOSTUNREACH,
1315: [PRC_UNREACH_PROTOCOL] = ECONNREFUSED,
1316: [PRC_UNREACH_PORT] = ECONNREFUSED,
1317: [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH,
1318: [PRC_PARAMPROB] = ENOPROTOOPT,
1.1 cgd 1319: };
1320:
1.295 dyoung 1321: void
1322: ip_fasttimo(void)
1323: {
1324: if (ip_drainwanted) {
1325: ip_drain();
1326: ip_drainwanted = 0;
1327: }
1328: }
1329:
1330: void
1331: ip_drainstub(void)
1332: {
1333: ip_drainwanted = 1;
1334: }
1335:
1.1 cgd 1336: /*
1337: * Forward a packet. If some error occurs return the sender
1338: * an icmp packet. Note we can't always generate a meaningful
1339: * icmp message because icmp doesn't have a large enough repertoire
1340: * of codes and types.
1341: *
1342: * If not forwarding, just drop the packet. This could be confusing
1343: * if ipforwarding was zero but some routing protocol was advancing
1344: * us as a gateway to somewhere. However, we must let the routing
1345: * protocol deal with that.
1346: *
1347: * The srcrt parameter indicates whether the packet is being forwarded
1348: * via a source route.
1349: */
1.313 rmind 1350: static void
1.329 ozaki-r 1351: ip_forward(struct mbuf *m, int srcrt, struct ifnet *rcvif)
1.1 cgd 1352: {
1.109 augustss 1353: struct ip *ip = mtod(m, struct ip *);
1354: struct rtentry *rt;
1.220 christos 1355: int error, type = 0, code = 0, destmtu = 0;
1.1 cgd 1356: struct mbuf *mcopy;
1.13 mycroft 1357: n_long dest;
1.249 dyoung 1358: union {
1359: struct sockaddr dst;
1360: struct sockaddr_in dst4;
1361: } u;
1.320 ozaki-r 1362: uint64_t *ips;
1.340 ozaki-r 1363: struct route *ro;
1.164 matt 1364:
1.318 rmind 1365: KASSERTMSG(cpu_softintr_p(), "ip_forward: not in the software "
1366: "interrupt handler; synchronization assumptions violated");
1367:
1.164 matt 1368: /*
1369: * We are now in the output path.
1370: */
1371: MCLAIM(m, &ip_tx_mowner);
1.135 thorpej 1372:
1373: /*
1374: * Clear any in-bound checksum flags for this packet.
1375: */
1376: m->m_pkthdr.csum_flags = 0;
1.1 cgd 1377:
1.13 mycroft 1378: dest = 0;
1.93 sommerfe 1379: if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) {
1.266 thorpej 1380: IP_STATINC(IP_STAT_CANTFORWARD);
1.1 cgd 1381: m_freem(m);
1382: return;
1383: }
1.318 rmind 1384:
1.1 cgd 1385: if (ip->ip_ttl <= IPTTLDEC) {
1.13 mycroft 1386: icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0);
1.1 cgd 1387: return;
1388: }
1389:
1.249 dyoung 1390: sockaddr_in_init(&u.dst4, &ip->ip_dst, 0);
1.318 rmind 1391:
1.340 ozaki-r 1392: ro = percpu_getref(ipforward_rt_percpu);
1.345 ozaki-r 1393: rt = rtcache_lookup(ro, &u.dst);
1394: if (rt == NULL) {
1.340 ozaki-r 1395: percpu_putref(ipforward_rt_percpu);
1.249 dyoung 1396: icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0);
1397: return;
1.1 cgd 1398: }
1399:
1400: /*
1.34 mycroft 1401: * Save at most 68 bytes of the packet in case
1.1 cgd 1402: * we need to generate an ICMP message to the src.
1.119 itojun 1403: * Pullup to avoid sharing mbuf cluster between m and mcopy.
1.1 cgd 1404: */
1.155 itojun 1405: mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT);
1.119 itojun 1406: if (mcopy)
1407: mcopy = m_pullup(mcopy, ip->ip_hl << 2);
1.1 cgd 1408:
1.221 christos 1409: ip->ip_ttl -= IPTTLDEC;
1410:
1.1 cgd 1411: /*
1412: * If forwarding packet using same interface that it came in on,
1413: * perhaps should send a redirect to sender to shortcut a hop.
1414: * Only send redirect if source is sending directly to us,
1415: * and if packet was not source routed (or has any options).
1416: * Also, don't send redirect if forwarding using a default route
1417: * or a route modified by a redirect.
1418: */
1.329 ozaki-r 1419: if (rt->rt_ifp == rcvif &&
1.1 cgd 1420: (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 &&
1.250 dyoung 1421: !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) &&
1.1 cgd 1422: ipsendredirects && !srcrt) {
1.19 mycroft 1423: if (rt->rt_ifa &&
1424: (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) ==
1425: ifatoia(rt->rt_ifa)->ia_subnet) {
1.77 thorpej 1426: if (rt->rt_flags & RTF_GATEWAY)
1427: dest = satosin(rt->rt_gateway)->sin_addr.s_addr;
1428: else
1429: dest = ip->ip_dst.s_addr;
1430: /*
1431: * Router requirements says to only send host
1432: * redirects.
1433: */
1434: type = ICMP_REDIRECT;
1435: code = ICMP_REDIRECT_HOST;
1.1 cgd 1436: }
1437: }
1.345 ozaki-r 1438: rtcache_unref(rt, ro);
1.1 cgd 1439:
1.340 ozaki-r 1440: error = ip_output(m, NULL, ro,
1.173 jonathan 1441: (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)),
1.296 plunky 1442: NULL, NULL);
1.173 jonathan 1443:
1.320 ozaki-r 1444: if (error) {
1.266 thorpej 1445: IP_STATINC(IP_STAT_CANTFORWARD);
1.320 ozaki-r 1446: goto error;
1447: }
1448:
1449: ips = IP_STAT_GETREF();
1450: ips[IP_STAT_FORWARD]++;
1451:
1452: if (type) {
1453: ips[IP_STAT_REDIRECTSENT]++;
1454: IP_STAT_PUTREF();
1455: goto redirect;
1456: }
1457:
1458: IP_STAT_PUTREF();
1459: if (mcopy) {
1.63 matt 1460: #ifdef GATEWAY
1.320 ozaki-r 1461: if (mcopy->m_flags & M_CANFASTFWD)
1.340 ozaki-r 1462: ipflow_create(ro, mcopy);
1.63 matt 1463: #endif
1.320 ozaki-r 1464: m_freem(mcopy);
1.1 cgd 1465: }
1.320 ozaki-r 1466:
1.340 ozaki-r 1467: percpu_putref(ipforward_rt_percpu);
1.320 ozaki-r 1468: return;
1469:
1470: redirect:
1471: error:
1.318 rmind 1472: if (mcopy == NULL) {
1.340 ozaki-r 1473: percpu_putref(ipforward_rt_percpu);
1.1 cgd 1474: return;
1.318 rmind 1475: }
1.13 mycroft 1476:
1.1 cgd 1477: switch (error) {
1478:
1479: case 0: /* forwarded, but need redirect */
1480: /* type, code set above */
1481: break;
1482:
1483: case ENETUNREACH: /* shouldn't happen, checked above */
1484: case EHOSTUNREACH:
1485: case ENETDOWN:
1486: case EHOSTDOWN:
1487: default:
1488: type = ICMP_UNREACH;
1489: code = ICMP_UNREACH_HOST;
1490: break;
1491:
1492: case EMSGSIZE:
1493: type = ICMP_UNREACH;
1494: code = ICMP_UNREACH_NEEDFRAG;
1.263 cube 1495:
1.345 ozaki-r 1496: if ((rt = rtcache_validate(ro)) != NULL) {
1.274 seanb 1497: destmtu = rt->rt_ifp->if_mtu;
1.345 ozaki-r 1498: rtcache_unref(rt, ro);
1499: }
1.305 rmind 1500: #ifdef IPSEC
1.317 christos 1501: if (ipsec_used)
1502: (void)ipsec4_forward(mcopy, &destmtu);
1.305 rmind 1503: #endif
1.266 thorpej 1504: IP_STATINC(IP_STAT_CANTFRAG);
1.1 cgd 1505: break;
1506:
1507: case ENOBUFS:
1.143 itojun 1508: /*
1.311 rmind 1509: * Do not generate ICMP_SOURCEQUENCH as required in RFC 1812,
1510: * Requirements for IP Version 4 Routers. Source quench can
1511: * big problem under DoS attacks or if the underlying
1512: * interface is rate-limited.
1.143 itojun 1513: */
1514: if (mcopy)
1515: m_freem(mcopy);
1.340 ozaki-r 1516: percpu_putref(ipforward_rt_percpu);
1.143 itojun 1517: return;
1.1 cgd 1518: }
1.220 christos 1519: icmp_error(mcopy, type, code, dest, destmtu);
1.340 ozaki-r 1520: percpu_putref(ipforward_rt_percpu);
1.44 thorpej 1521: }
1522:
1523: void
1.211 perry 1524: ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip,
1525: struct mbuf *m)
1.44 thorpej 1526: {
1.311 rmind 1527: struct socket *so = inp->inp_socket;
1.329 ozaki-r 1528: ifnet_t *ifp;
1.311 rmind 1529: int inpflags = inp->inp_flags;
1.329 ozaki-r 1530: struct psref psref;
1.44 thorpej 1531:
1.329 ozaki-r 1532: ifp = m_get_rcvif_psref(m, &psref);
1.330 ozaki-r 1533: if (__predict_false(ifp == NULL))
1534: return; /* XXX should report error? */
1535:
1.311 rmind 1536: if (so->so_options & SO_TIMESTAMP
1.278 christos 1537: #ifdef SO_OTIMESTAMP
1.311 rmind 1538: || so->so_options & SO_OTIMESTAMP
1.278 christos 1539: #endif
1540: ) {
1.44 thorpej 1541: struct timeval tv;
1542:
1543: microtime(&tv);
1.278 christos 1544: #ifdef SO_OTIMESTAMP
1.311 rmind 1545: if (so->so_options & SO_OTIMESTAMP) {
1.278 christos 1546: struct timeval50 tv50;
1547: timeval_to_timeval50(&tv, &tv50);
1548: *mp = sbcreatecontrol((void *) &tv50, sizeof(tv50),
1549: SCM_OTIMESTAMP, SOL_SOCKET);
1550: } else
1551: #endif
1.244 christos 1552: *mp = sbcreatecontrol((void *) &tv, sizeof(tv),
1.44 thorpej 1553: SCM_TIMESTAMP, SOL_SOCKET);
1554: if (*mp)
1555: mp = &(*mp)->m_next;
1556: }
1.311 rmind 1557: if (inpflags & INP_RECVDSTADDR) {
1.244 christos 1558: *mp = sbcreatecontrol((void *) &ip->ip_dst,
1.44 thorpej 1559: sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP);
1560: if (*mp)
1561: mp = &(*mp)->m_next;
1562: }
1.311 rmind 1563: if (inpflags & INP_RECVPKTINFO) {
1.306 christos 1564: struct in_pktinfo ipi;
1.307 christos 1565: ipi.ipi_addr = ip->ip_src;
1.314 rmind 1566: ipi.ipi_ifindex = ifp->if_index;
1.306 christos 1567: *mp = sbcreatecontrol((void *) &ipi,
1568: sizeof(ipi), IP_RECVPKTINFO, IPPROTO_IP);
1569: if (*mp)
1570: mp = &(*mp)->m_next;
1571: }
1.311 rmind 1572: if (inpflags & INP_PKTINFO) {
1.306 christos 1573: struct in_pktinfo ipi;
1.307 christos 1574: ipi.ipi_addr = ip->ip_dst;
1.314 rmind 1575: ipi.ipi_ifindex = ifp->if_index;
1.306 christos 1576: *mp = sbcreatecontrol((void *) &ipi,
1577: sizeof(ipi), IP_PKTINFO, IPPROTO_IP);
1578: if (*mp)
1579: mp = &(*mp)->m_next;
1580: }
1.311 rmind 1581: if (inpflags & INP_RECVIF) {
1.44 thorpej 1582: struct sockaddr_dl sdl;
1583:
1.337 ozaki-r 1584: sockaddr_dl_init(&sdl, sizeof(sdl), ifp->if_index, 0, NULL, 0,
1585: NULL, 0);
1.251 dyoung 1586: *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP);
1.44 thorpej 1587: if (*mp)
1588: mp = &(*mp)->m_next;
1589: }
1.311 rmind 1590: if (inpflags & INP_RECVTTL) {
1.282 minskim 1591: *mp = sbcreatecontrol((void *) &ip->ip_ttl,
1592: sizeof(uint8_t), IP_RECVTTL, IPPROTO_IP);
1593: if (*mp)
1594: mp = &(*mp)->m_next;
1595: }
1.329 ozaki-r 1596: m_put_rcvif_psref(ifp, &psref);
1.13 mycroft 1597: }
1598:
1.189 atatat 1599: /*
1.228 elad 1600: * sysctl helper routine for net.inet.ip.forwsrcrt.
1601: */
1602: static int
1603: sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS)
1604: {
1605: int error, tmp;
1606: struct sysctlnode node;
1607:
1608: node = *rnode;
1609: tmp = ip_forwsrcrt;
1610: node.sysctl_data = &tmp;
1611: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1612: if (error || newp == NULL)
1613: return (error);
1614:
1.280 elad 1615: error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT,
1616: 0, NULL, NULL, NULL);
1617: if (error)
1618: return (error);
1.228 elad 1619:
1620: ip_forwsrcrt = tmp;
1621:
1622: return (0);
1623: }
1624:
1625: /*
1.189 atatat 1626: * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the
1627: * range of the new value and tweaks timers if it changes.
1628: */
1629: static int
1630: sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS)
1.13 mycroft 1631: {
1.189 atatat 1632: int error, tmp;
1633: struct sysctlnode node;
1634:
1.350 ozaki-r 1635: icmp_mtudisc_lock();
1636:
1.189 atatat 1637: node = *rnode;
1638: tmp = ip_mtudisc_timeout;
1639: node.sysctl_data = &tmp;
1640: error = sysctl_lookup(SYSCTLFN_CALL(&node));
1641: if (error || newp == NULL)
1.350 ozaki-r 1642: goto out;
1643: if (tmp < 0) {
1644: error = EINVAL;
1645: goto out;
1646: }
1.273 matt 1647:
1.189 atatat 1648: ip_mtudisc_timeout = tmp;
1649: rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout);
1.350 ozaki-r 1650: error = 0;
1651: out:
1652: icmp_mtudisc_unlock();
1.351 ozaki-r 1653: return error;
1.189 atatat 1654: }
1.54 lukem 1655:
1.266 thorpej 1656: static int
1657: sysctl_net_inet_ip_stats(SYSCTLFN_ARGS)
1658: {
1659:
1.271 thorpej 1660: return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS));
1.266 thorpej 1661: }
1.131 itojun 1662:
1.284 pooka 1663: static void
1664: sysctl_net_inet_ip_setup(struct sysctllog **clog)
1.189 atatat 1665: {
1.197 atatat 1666: sysctl_createv(clog, 0, NULL, NULL,
1667: CTLFLAG_PERMANENT,
1.203 atatat 1668: CTLTYPE_NODE, "inet",
1669: SYSCTL_DESCR("PF_INET related settings"),
1.189 atatat 1670: NULL, 0, NULL, 0,
1671: CTL_NET, PF_INET, CTL_EOL);
1.197 atatat 1672: sysctl_createv(clog, 0, NULL, NULL,
1673: CTLFLAG_PERMANENT,
1.203 atatat 1674: CTLTYPE_NODE, "ip",
1675: SYSCTL_DESCR("IPv4 related settings"),
1.189 atatat 1676: NULL, 0, NULL, 0,
1677: CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL);
1.212 perry 1678:
1.197 atatat 1679: sysctl_createv(clog, 0, NULL, NULL,
1680: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1681: CTLTYPE_INT, "forwarding",
1682: SYSCTL_DESCR("Enable forwarding of INET datagrams"),
1.189 atatat 1683: NULL, 0, &ipforwarding, 0,
1684: CTL_NET, PF_INET, IPPROTO_IP,
1685: IPCTL_FORWARDING, CTL_EOL);
1.197 atatat 1686: sysctl_createv(clog, 0, NULL, NULL,
1687: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1688: CTLTYPE_INT, "redirect",
1689: SYSCTL_DESCR("Enable sending of ICMP redirect messages"),
1.189 atatat 1690: NULL, 0, &ipsendredirects, 0,
1691: CTL_NET, PF_INET, IPPROTO_IP,
1692: IPCTL_SENDREDIRECTS, CTL_EOL);
1.197 atatat 1693: sysctl_createv(clog, 0, NULL, NULL,
1694: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1695: CTLTYPE_INT, "ttl",
1696: SYSCTL_DESCR("Default TTL for an INET datagram"),
1.189 atatat 1697: NULL, 0, &ip_defttl, 0,
1698: CTL_NET, PF_INET, IPPROTO_IP,
1699: IPCTL_DEFTTL, CTL_EOL);
1700: #ifdef IPCTL_DEFMTU
1.197 atatat 1701: sysctl_createv(clog, 0, NULL, NULL,
1702: CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */,
1.203 atatat 1703: CTLTYPE_INT, "mtu",
1704: SYSCTL_DESCR("Default MTA for an INET route"),
1.189 atatat 1705: NULL, 0, &ip_mtu, 0,
1706: CTL_NET, PF_INET, IPPROTO_IP,
1707: IPCTL_DEFMTU, CTL_EOL);
1708: #endif /* IPCTL_DEFMTU */
1.197 atatat 1709: sysctl_createv(clog, 0, NULL, NULL,
1.228 elad 1710: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1711: CTLTYPE_INT, "forwsrcrt",
1712: SYSCTL_DESCR("Enable forwarding of source-routed "
1713: "datagrams"),
1.228 elad 1714: sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0,
1.189 atatat 1715: CTL_NET, PF_INET, IPPROTO_IP,
1716: IPCTL_FORWSRCRT, CTL_EOL);
1.197 atatat 1717: sysctl_createv(clog, 0, NULL, NULL,
1718: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1719: CTLTYPE_INT, "directed-broadcast",
1720: SYSCTL_DESCR("Enable forwarding of broadcast datagrams"),
1.189 atatat 1721: NULL, 0, &ip_directedbcast, 0,
1722: CTL_NET, PF_INET, IPPROTO_IP,
1723: IPCTL_DIRECTEDBCAST, CTL_EOL);
1.197 atatat 1724: sysctl_createv(clog, 0, NULL, NULL,
1725: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1726: CTLTYPE_INT, "allowsrcrt",
1727: SYSCTL_DESCR("Accept source-routed datagrams"),
1.189 atatat 1728: NULL, 0, &ip_allowsrcrt, 0,
1729: CTL_NET, PF_INET, IPPROTO_IP,
1730: IPCTL_ALLOWSRCRT, CTL_EOL);
1.311 rmind 1731:
1.197 atatat 1732: sysctl_createv(clog, 0, NULL, NULL,
1733: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1734: CTLTYPE_INT, "mtudisc",
1735: SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"),
1.189 atatat 1736: NULL, 0, &ip_mtudisc, 0,
1737: CTL_NET, PF_INET, IPPROTO_IP,
1738: IPCTL_MTUDISC, CTL_EOL);
1.197 atatat 1739: sysctl_createv(clog, 0, NULL, NULL,
1740: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1741: CTLTYPE_INT, "anonportmin",
1742: SYSCTL_DESCR("Lowest ephemeral port number to assign"),
1.189 atatat 1743: sysctl_net_inet_ip_ports, 0, &anonportmin, 0,
1744: CTL_NET, PF_INET, IPPROTO_IP,
1745: IPCTL_ANONPORTMIN, CTL_EOL);
1.197 atatat 1746: sysctl_createv(clog, 0, NULL, NULL,
1747: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1748: CTLTYPE_INT, "anonportmax",
1749: SYSCTL_DESCR("Highest ephemeral port number to assign"),
1.189 atatat 1750: sysctl_net_inet_ip_ports, 0, &anonportmax, 0,
1751: CTL_NET, PF_INET, IPPROTO_IP,
1752: IPCTL_ANONPORTMAX, CTL_EOL);
1.197 atatat 1753: sysctl_createv(clog, 0, NULL, NULL,
1754: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1755: CTLTYPE_INT, "mtudisctimeout",
1756: SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"),
1.300 dsl 1757: sysctl_net_inet_ip_pmtudto, 0, (void *)&ip_mtudisc_timeout, 0,
1.189 atatat 1758: CTL_NET, PF_INET, IPPROTO_IP,
1759: IPCTL_MTUDISCTIMEOUT, CTL_EOL);
1760: #ifndef IPNOPRIVPORTS
1.197 atatat 1761: sysctl_createv(clog, 0, NULL, NULL,
1762: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1763: CTLTYPE_INT, "lowportmin",
1764: SYSCTL_DESCR("Lowest privileged ephemeral port number "
1765: "to assign"),
1.189 atatat 1766: sysctl_net_inet_ip_ports, 0, &lowportmin, 0,
1767: CTL_NET, PF_INET, IPPROTO_IP,
1768: IPCTL_LOWPORTMIN, CTL_EOL);
1.197 atatat 1769: sysctl_createv(clog, 0, NULL, NULL,
1770: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1771: CTLTYPE_INT, "lowportmax",
1772: SYSCTL_DESCR("Highest privileged ephemeral port number "
1773: "to assign"),
1.189 atatat 1774: sysctl_net_inet_ip_ports, 0, &lowportmax, 0,
1775: CTL_NET, PF_INET, IPPROTO_IP,
1776: IPCTL_LOWPORTMAX, CTL_EOL);
1777: #endif /* IPNOPRIVPORTS */
1778: #if NGRE > 0
1.197 atatat 1779: sysctl_createv(clog, 0, NULL, NULL,
1780: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1781: CTLTYPE_INT, "grettl",
1782: SYSCTL_DESCR("Default TTL for a gre tunnel datagram"),
1.189 atatat 1783: NULL, 0, &ip_gre_ttl, 0,
1784: CTL_NET, PF_INET, IPPROTO_IP,
1785: IPCTL_GRE_TTL, CTL_EOL);
1786: #endif /* NGRE */
1.197 atatat 1787: sysctl_createv(clog, 0, NULL, NULL,
1788: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1789: CTLTYPE_INT, "checkinterface",
1790: SYSCTL_DESCR("Enable receive side of Strong ES model "
1791: "from RFC1122"),
1.189 atatat 1792: NULL, 0, &ip_checkinterface, 0,
1793: CTL_NET, PF_INET, IPPROTO_IP,
1794: IPCTL_CHECKINTERFACE, CTL_EOL);
1.197 atatat 1795: sysctl_createv(clog, 0, NULL, NULL,
1796: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1.203 atatat 1797: CTLTYPE_INT, "random_id",
1798: SYSCTL_DESCR("Assign random ip_id values"),
1.189 atatat 1799: NULL, 0, &ip_do_randomid, 0,
1800: CTL_NET, PF_INET, IPPROTO_IP,
1801: IPCTL_RANDOMID, CTL_EOL);
1.206 thorpej 1802: sysctl_createv(clog, 0, NULL, NULL,
1803: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1804: CTLTYPE_INT, "do_loopback_cksum",
1805: SYSCTL_DESCR("Perform IP checksum on loopback"),
1806: NULL, 0, &ip_do_loopback_cksum, 0,
1807: CTL_NET, PF_INET, IPPROTO_IP,
1808: IPCTL_LOOPBACKCKSUM, CTL_EOL);
1.219 elad 1809: sysctl_createv(clog, 0, NULL, NULL,
1810: CTLFLAG_PERMANENT,
1811: CTLTYPE_STRUCT, "stats",
1812: SYSCTL_DESCR("IP statistics"),
1.266 thorpej 1813: sysctl_net_inet_ip_stats, 0, NULL, 0,
1.219 elad 1814: CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS,
1815: CTL_EOL);
1.322 joerg 1816: #if NARP
1.321 roy 1817: sysctl_createv(clog, 0, NULL, NULL,
1818: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1819: CTLTYPE_INT, "dad_count",
1820: SYSCTL_DESCR("Number of Duplicate Address Detection "
1821: "probes to send"),
1822: NULL, 0, &ip_dad_count, 0,
1823: CTL_NET, PF_INET, IPPROTO_IP,
1824: IPCTL_DAD_COUNT, CTL_EOL);
1.322 joerg 1825: #endif
1.301 christos 1826:
1827: /* anonportalgo RFC6056 subtree */
1.302 christos 1828: const struct sysctlnode *portalgo_node;
1829: sysctl_createv(clog, 0, NULL, &portalgo_node,
1.301 christos 1830: CTLFLAG_PERMANENT,
1831: CTLTYPE_NODE, "anonportalgo",
1832: SYSCTL_DESCR("Anonymous Port Algorithm Selection (RFC 6056)"),
1833: NULL, 0, NULL, 0,
1834: CTL_NET, PF_INET, IPPROTO_IP, CTL_CREATE, CTL_EOL);
1.302 christos 1835: sysctl_createv(clog, 0, &portalgo_node, NULL,
1.301 christos 1836: CTLFLAG_PERMANENT,
1837: CTLTYPE_STRING, "available",
1838: SYSCTL_DESCR("available algorithms"),
1.302 christos 1839: sysctl_portalgo_available, 0, NULL, PORTALGO_MAXLEN,
1.301 christos 1840: CTL_CREATE, CTL_EOL);
1.302 christos 1841: sysctl_createv(clog, 0, &portalgo_node, NULL,
1.301 christos 1842: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1843: CTLTYPE_STRING, "selected",
1844: SYSCTL_DESCR("selected algorithm"),
1.303 christos 1845: sysctl_portalgo_selected4, 0, NULL, PORTALGO_MAXLEN,
1846: CTL_CREATE, CTL_EOL);
1847: sysctl_createv(clog, 0, &portalgo_node, NULL,
1848: CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1849: CTLTYPE_STRUCT, "reserve",
1850: SYSCTL_DESCR("bitmap of reserved ports"),
1851: sysctl_portalgo_reserve4, 0, NULL, 0,
1.301 christos 1852: CTL_CREATE, CTL_EOL);
1.1 cgd 1853: }
1.266 thorpej 1854:
1855: void
1856: ip_statinc(u_int stat)
1857: {
1858:
1859: KASSERT(stat < IP_NSTATS);
1860: IP_STATINC(stat);
1861: }
CVSweb <webmaster@jp.NetBSD.org>