1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1982, 1986, 1988, 1990, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
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 University 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.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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 *
31 * @(#)ip_output.c 8.3 (Berkeley) 1/21/94
32 */
33
34 #include <sys/cdefs.h>
35 __FBSDID("$FreeBSD$");
36
37 #include "opt_inet.h"
38 #include "opt_ratelimit.h"
39 #include "opt_ipsec.h"
40 #include "opt_mbuf_stress_test.h"
41 #include "opt_mpath.h"
42 #include "opt_route.h"
43 #include "opt_sctp.h"
44 #include "opt_rss.h"
45
46 #include <sys/param.h>
47 #include <sys/systm.h>
48 #include <sys/kernel.h>
49 #include <sys/lock.h>
50 #include <sys/malloc.h>
51 #include <sys/mbuf.h>
52 #include <sys/priv.h>
53 #include <sys/proc.h>
54 #include <sys/protosw.h>
55 #include <sys/rmlock.h>
56 #include <sys/sdt.h>
57 #include <sys/socket.h>
58 #include <sys/socketvar.h>
59 #include <sys/sysctl.h>
60 #include <sys/ucred.h>
61
62 #include <net/if.h>
63 #include <net/if_var.h>
64 #include <net/if_llatbl.h>
65 #include <net/netisr.h>
66 #include <net/pfil.h>
67 #include <net/route.h>
68 #ifdef RADIX_MPATH
69 #include <net/radix_mpath.h>
70 #endif
71 #include <net/rss_config.h>
72 #include <net/vnet.h>
73
74 #include <netinet/in.h>
75 #include <netinet/in_kdtrace.h>
76 #include <netinet/in_systm.h>
77 #include <netinet/ip.h>
78 #include <netinet/in_pcb.h>
79 #include <netinet/in_rss.h>
80 #include <netinet/in_var.h>
81 #include <netinet/ip_var.h>
82 #include <netinet/ip_options.h>
83
84 #include <netinet/udp.h>
85 #include <netinet/udp_var.h>
86
87 #ifdef SCTP
88 #include <netinet/sctp.h>
89 #include <netinet/sctp_crc32.h>
90 #endif
91
92 #include <netipsec/ipsec_support.h>
93
94 #include <machine/in_cksum.h>
95
96 #include <security/mac/mac_framework.h>
97
98 #ifdef MBUF_STRESS_TEST
99 static int mbuf_frag_size = 0;
100 SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW,
101 &mbuf_frag_size, 0, "Fragment outgoing mbufs to this size");
102 #endif
103
104 static void ip_mloopback(struct ifnet *, const struct mbuf *, int);
105
106
107 extern int in_mcast_loop;
108 extern struct protosw inetsw[];
109
110 static inline int
ip_output_pfil(struct mbuf ** mp,struct ifnet * ifp,struct inpcb * inp,struct sockaddr_in * dst,int * fibnum,int * error)111 ip_output_pfil(struct mbuf **mp, struct ifnet *ifp, struct inpcb *inp,
112 struct sockaddr_in *dst, int *fibnum, int *error)
113 {
114 struct m_tag *fwd_tag = NULL;
115 struct mbuf *m;
116 struct in_addr odst;
117 struct ip *ip;
118
119 m = *mp;
120 ip = mtod(m, struct ip *);
121
122 /* Run through list of hooks for output packets. */
123 odst.s_addr = ip->ip_dst.s_addr;
124 *error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp, PFIL_OUT, 0, inp);
125 m = *mp;
126 if ((*error) != 0 || m == NULL)
127 return 1; /* Finished */
128
129 ip = mtod(m, struct ip *);
130
131 /* See if destination IP address was changed by packet filter. */
132 if (odst.s_addr != ip->ip_dst.s_addr) {
133 m->m_flags |= M_SKIP_FIREWALL;
134 /* If destination is now ourself drop to ip_input(). */
135 if (in_localip(ip->ip_dst)) {
136 m->m_flags |= M_FASTFWD_OURS;
137 if (m->m_pkthdr.rcvif == NULL)
138 m->m_pkthdr.rcvif = V_loif;
139 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
140 m->m_pkthdr.csum_flags |=
141 CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
142 m->m_pkthdr.csum_data = 0xffff;
143 }
144 m->m_pkthdr.csum_flags |=
145 CSUM_IP_CHECKED | CSUM_IP_VALID;
146 #ifdef SCTP
147 if (m->m_pkthdr.csum_flags & CSUM_SCTP)
148 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
149 #endif
150 *error = netisr_queue(NETISR_IP, m);
151 return 1; /* Finished */
152 }
153
154 bzero(dst, sizeof(*dst));
155 dst->sin_family = AF_INET;
156 dst->sin_len = sizeof(*dst);
157 dst->sin_addr = ip->ip_dst;
158
159 return -1; /* Reloop */
160 }
161 /* See if fib was changed by packet filter. */
162 if ((*fibnum) != M_GETFIB(m)) {
163 m->m_flags |= M_SKIP_FIREWALL;
164 *fibnum = M_GETFIB(m);
165 return -1; /* Reloop for FIB change */
166 }
167
168 /* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */
169 if (m->m_flags & M_FASTFWD_OURS) {
170 if (m->m_pkthdr.rcvif == NULL)
171 m->m_pkthdr.rcvif = V_loif;
172 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
173 m->m_pkthdr.csum_flags |=
174 CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
175 m->m_pkthdr.csum_data = 0xffff;
176 }
177 #ifdef SCTP
178 if (m->m_pkthdr.csum_flags & CSUM_SCTP)
179 m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID;
180 #endif
181 m->m_pkthdr.csum_flags |=
182 CSUM_IP_CHECKED | CSUM_IP_VALID;
183
184 *error = netisr_queue(NETISR_IP, m);
185 return 1; /* Finished */
186 }
187 /* Or forward to some other address? */
188 if ((m->m_flags & M_IP_NEXTHOP) &&
189 ((fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL)) {
190 bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in));
191 m->m_flags |= M_SKIP_FIREWALL;
192 m->m_flags &= ~M_IP_NEXTHOP;
193 m_tag_delete(m, fwd_tag);
194
195 return -1; /* Reloop for CHANGE of dst */
196 }
197
198 return 0;
199 }
200
201 /*
202 * IP output. The packet in mbuf chain m contains a skeletal IP
203 * header (with len, off, ttl, proto, tos, src, dst).
204 * The mbuf chain containing the packet will be freed.
205 * The mbuf opt, if present, will not be freed.
206 * If route ro is present and has ro_rt initialized, route lookup would be
207 * skipped and ro->ro_rt would be used. If ro is present but ro->ro_rt is NULL,
208 * then result of route lookup is stored in ro->ro_rt.
209 *
210 * In the IP forwarding case, the packet will arrive with options already
211 * inserted, so must have a NULL opt pointer.
212 */
213 int
ip_output(struct mbuf * m,struct mbuf * opt,struct route * ro,int flags,struct ip_moptions * imo,struct inpcb * inp)214 ip_output(struct mbuf *m, struct mbuf *opt, struct route *ro, int flags,
215 struct ip_moptions *imo, struct inpcb *inp)
216 {
217 struct rm_priotracker in_ifa_tracker;
218 struct ip *ip;
219 struct ifnet *ifp = NULL; /* keep compiler happy */
220 struct mbuf *m0;
221 int hlen = sizeof (struct ip);
222 int mtu;
223 int error = 0;
224 struct sockaddr_in *dst;
225 const struct sockaddr_in *gw;
226 struct in_ifaddr *ia;
227 int isbroadcast;
228 uint16_t ip_len, ip_off;
229 struct route iproute;
230 struct rtentry *rte; /* cache for ro->ro_rt */
231 uint32_t fibnum;
232 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
233 int no_route_but_check_spd = 0;
234 #endif
235 M_ASSERTPKTHDR(m);
236
237 if (inp != NULL) {
238 INP_LOCK_ASSERT(inp);
239 M_SETFIB(m, inp->inp_inc.inc_fibnum);
240 if ((flags & IP_NODEFAULTFLOWID) == 0) {
241 m->m_pkthdr.flowid = inp->inp_flowid;
242 M_HASHTYPE_SET(m, inp->inp_flowtype);
243 }
244 }
245
246 if (ro == NULL) {
247 ro = &iproute;
248 bzero(ro, sizeof (*ro));
249 }
250
251 if (opt) {
252 int len = 0;
253 m = ip_insertoptions(m, opt, &len);
254 if (len != 0)
255 hlen = len; /* ip->ip_hl is updated above */
256 }
257 ip = mtod(m, struct ip *);
258 ip_len = ntohs(ip->ip_len);
259 ip_off = ntohs(ip->ip_off);
260
261 if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) {
262 ip->ip_v = IPVERSION;
263 ip->ip_hl = hlen >> 2;
264 ip_fillid(ip);
265 } else {
266 /* Header already set, fetch hlen from there */
267 hlen = ip->ip_hl << 2;
268 }
269 if ((flags & IP_FORWARDING) == 0)
270 IPSTAT_INC(ips_localout);
271
272 /*
273 * dst/gw handling:
274 *
275 * dst can be rewritten but always points to &ro->ro_dst.
276 * gw is readonly but can point either to dst OR rt_gateway,
277 * therefore we need restore gw if we're redoing lookup.
278 */
279 gw = dst = (struct sockaddr_in *)&ro->ro_dst;
280 fibnum = (inp != NULL) ? inp->inp_inc.inc_fibnum : M_GETFIB(m);
281 rte = ro->ro_rt;
282 if (rte == NULL) {
283 bzero(dst, sizeof(*dst));
284 dst->sin_family = AF_INET;
285 dst->sin_len = sizeof(*dst);
286 dst->sin_addr = ip->ip_dst;
287 }
288 NET_EPOCH_ENTER();
289 again:
290 /*
291 * Validate route against routing table additions;
292 * a better/more specific route might have been added.
293 */
294 if (inp)
295 RT_VALIDATE(ro, &inp->inp_rt_cookie, fibnum);
296 /*
297 * If there is a cached route,
298 * check that it is to the same destination
299 * and is still up. If not, free it and try again.
300 * The address family should also be checked in case of sharing the
301 * cache with IPv6.
302 * Also check whether routing cache needs invalidation.
303 */
304 rte = ro->ro_rt;
305 if (rte && ((rte->rt_flags & RTF_UP) == 0 ||
306 rte->rt_ifp == NULL ||
307 !RT_LINK_IS_UP(rte->rt_ifp) ||
308 dst->sin_family != AF_INET ||
309 dst->sin_addr.s_addr != ip->ip_dst.s_addr)) {
310 RO_INVALIDATE_CACHE(ro);
311 rte = NULL;
312 }
313 ia = NULL;
314 /*
315 * If routing to interface only, short circuit routing lookup.
316 * The use of an all-ones broadcast address implies this; an
317 * interface is specified by the broadcast address of an interface,
318 * or the destination address of a ptp interface.
319 */
320 if (flags & IP_SENDONES) {
321 if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst),
322 M_GETFIB(m)))) == NULL &&
323 (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst),
324 M_GETFIB(m)))) == NULL) {
325 IPSTAT_INC(ips_noroute);
326 error = ENETUNREACH;
327 goto bad;
328 }
329 ip->ip_dst.s_addr = INADDR_BROADCAST;
330 dst->sin_addr = ip->ip_dst;
331 ifp = ia->ia_ifp;
332 ip->ip_ttl = 1;
333 isbroadcast = 1;
334 } else if (flags & IP_ROUTETOIF) {
335 if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst),
336 M_GETFIB(m)))) == NULL &&
337 (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0,
338 M_GETFIB(m)))) == NULL) {
339 IPSTAT_INC(ips_noroute);
340 error = ENETUNREACH;
341 goto bad;
342 }
343 ifp = ia->ia_ifp;
344 ip->ip_ttl = 1;
345 isbroadcast = ifp->if_flags & IFF_BROADCAST ?
346 in_ifaddr_broadcast(dst->sin_addr, ia) : 0;
347 } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) &&
348 imo != NULL && imo->imo_multicast_ifp != NULL) {
349 /*
350 * Bypass the normal routing lookup for multicast
351 * packets if the interface is specified.
352 */
353 ifp = imo->imo_multicast_ifp;
354 IFP_TO_IA(ifp, ia, &in_ifa_tracker);
355 isbroadcast = 0; /* fool gcc */
356 } else {
357 /*
358 * We want to do any cloning requested by the link layer,
359 * as this is probably required in all cases for correct
360 * operation (as it is for ARP).
361 */
362 if (rte == NULL) {
363 #ifdef RADIX_MPATH
364 rtalloc_mpath_fib(ro,
365 ntohl(ip->ip_src.s_addr ^ ip->ip_dst.s_addr),
366 fibnum);
367 #else
368 in_rtalloc_ign(ro, 0, fibnum);
369 #endif
370 rte = ro->ro_rt;
371 }
372 if (rte == NULL ||
373 (rte->rt_flags & RTF_UP) == 0 ||
374 rte->rt_ifp == NULL ||
375 !RT_LINK_IS_UP(rte->rt_ifp)) {
376 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
377 /*
378 * There is no route for this packet, but it is
379 * possible that a matching SPD entry exists.
380 */
381 no_route_but_check_spd = 1;
382 mtu = 0; /* Silence GCC warning. */
383 goto sendit;
384 #endif
385 IPSTAT_INC(ips_noroute);
386 error = EHOSTUNREACH;
387 goto bad;
388 }
389 ia = ifatoia(rte->rt_ifa);
390 ifp = rte->rt_ifp;
391 counter_u64_add(rte->rt_pksent, 1);
392 rt_update_ro_flags(ro);
393 if (rte->rt_flags & RTF_GATEWAY)
394 gw = (struct sockaddr_in *)rte->rt_gateway;
395 if (rte->rt_flags & RTF_HOST)
396 isbroadcast = (rte->rt_flags & RTF_BROADCAST);
397 else if (ifp->if_flags & IFF_BROADCAST)
398 isbroadcast = in_ifaddr_broadcast(gw->sin_addr, ia);
399 else
400 isbroadcast = 0;
401 }
402
403 /*
404 * Calculate MTU. If we have a route that is up, use that,
405 * otherwise use the interface's MTU.
406 */
407 if (rte != NULL && (rte->rt_flags & (RTF_UP|RTF_HOST)))
408 mtu = rte->rt_mtu;
409 else
410 mtu = ifp->if_mtu;
411 /* Catch a possible divide by zero later. */
412 KASSERT(mtu > 0, ("%s: mtu %d <= 0, rte=%p (rt_flags=0x%08x) ifp=%p",
413 __func__, mtu, rte, (rte != NULL) ? rte->rt_flags : 0, ifp));
414
415 if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
416 m->m_flags |= M_MCAST;
417 /*
418 * IP destination address is multicast. Make sure "gw"
419 * still points to the address in "ro". (It may have been
420 * changed to point to a gateway address, above.)
421 */
422 gw = dst;
423 /*
424 * See if the caller provided any multicast options
425 */
426 if (imo != NULL) {
427 ip->ip_ttl = imo->imo_multicast_ttl;
428 if (imo->imo_multicast_vif != -1)
429 ip->ip_src.s_addr =
430 ip_mcast_src ?
431 ip_mcast_src(imo->imo_multicast_vif) :
432 INADDR_ANY;
433 } else
434 ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL;
435 /*
436 * Confirm that the outgoing interface supports multicast.
437 */
438 if ((imo == NULL) || (imo->imo_multicast_vif == -1)) {
439 if ((ifp->if_flags & IFF_MULTICAST) == 0) {
440 IPSTAT_INC(ips_noroute);
441 error = ENETUNREACH;
442 goto bad;
443 }
444 }
445 /*
446 * If source address not specified yet, use address
447 * of outgoing interface.
448 */
449 if (ip->ip_src.s_addr == INADDR_ANY) {
450 /* Interface may have no addresses. */
451 if (ia != NULL)
452 ip->ip_src = IA_SIN(ia)->sin_addr;
453 }
454
455 if ((imo == NULL && in_mcast_loop) ||
456 (imo && imo->imo_multicast_loop)) {
457 /*
458 * Loop back multicast datagram if not expressly
459 * forbidden to do so, even if we are not a member
460 * of the group; ip_input() will filter it later,
461 * thus deferring a hash lookup and mutex acquisition
462 * at the expense of a cheap copy using m_copym().
463 */
464 ip_mloopback(ifp, m, hlen);
465 } else {
466 /*
467 * If we are acting as a multicast router, perform
468 * multicast forwarding as if the packet had just
469 * arrived on the interface to which we are about
470 * to send. The multicast forwarding function
471 * recursively calls this function, using the
472 * IP_FORWARDING flag to prevent infinite recursion.
473 *
474 * Multicasts that are looped back by ip_mloopback(),
475 * above, will be forwarded by the ip_input() routine,
476 * if necessary.
477 */
478 if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) {
479 /*
480 * If rsvp daemon is not running, do not
481 * set ip_moptions. This ensures that the packet
482 * is multicast and not just sent down one link
483 * as prescribed by rsvpd.
484 */
485 if (!V_rsvp_on)
486 imo = NULL;
487 if (ip_mforward &&
488 ip_mforward(ip, ifp, m, imo) != 0) {
489 m_freem(m);
490 goto done;
491 }
492 }
493 }
494
495 /*
496 * Multicasts with a time-to-live of zero may be looped-
497 * back, above, but must not be transmitted on a network.
498 * Also, multicasts addressed to the loopback interface
499 * are not sent -- the above call to ip_mloopback() will
500 * loop back a copy. ip_input() will drop the copy if
501 * this host does not belong to the destination group on
502 * the loopback interface.
503 */
504 if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) {
505 m_freem(m);
506 goto done;
507 }
508
509 goto sendit;
510 }
511
512 /*
513 * If the source address is not specified yet, use the address
514 * of the outoing interface.
515 */
516 if (ip->ip_src.s_addr == INADDR_ANY) {
517 /* Interface may have no addresses. */
518 if (ia != NULL) {
519 ip->ip_src = IA_SIN(ia)->sin_addr;
520 }
521 }
522
523 /*
524 * Look for broadcast address and
525 * verify user is allowed to send
526 * such a packet.
527 */
528 if (isbroadcast) {
529 if ((ifp->if_flags & IFF_BROADCAST) == 0) {
530 error = EADDRNOTAVAIL;
531 goto bad;
532 }
533 if ((flags & IP_ALLOWBROADCAST) == 0) {
534 error = EACCES;
535 goto bad;
536 }
537 /* don't allow broadcast messages to be fragmented */
538 if (ip_len > mtu) {
539 error = EMSGSIZE;
540 goto bad;
541 }
542 m->m_flags |= M_BCAST;
543 } else {
544 m->m_flags &= ~M_BCAST;
545 }
546
547 sendit:
548 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
549 if (IPSEC_ENABLED(ipv4)) {
550 if ((error = IPSEC_OUTPUT(ipv4, m, inp)) != 0) {
551 if (error == EINPROGRESS)
552 error = 0;
553 goto done;
554 }
555 }
556 /*
557 * Check if there was a route for this packet; return error if not.
558 */
559 if (no_route_but_check_spd) {
560 IPSTAT_INC(ips_noroute);
561 error = EHOSTUNREACH;
562 goto bad;
563 }
564 /* Update variables that are affected by ipsec4_output(). */
565 ip = mtod(m, struct ip *);
566 hlen = ip->ip_hl << 2;
567 #endif /* IPSEC */
568
569 /* Jump over all PFIL processing if hooks are not active. */
570 if (PFIL_HOOKED(&V_inet_pfil_hook)) {
571 switch (ip_output_pfil(&m, ifp, inp, dst, &fibnum, &error)) {
572 case 1: /* Finished */
573 goto done;
574
575 case 0: /* Continue normally */
576 ip = mtod(m, struct ip *);
577 break;
578
579 case -1: /* Need to try again */
580 /* Reset everything for a new round */
581 RO_RTFREE(ro);
582 ro->ro_prepend = NULL;
583 rte = NULL;
584 gw = dst;
585 ip = mtod(m, struct ip *);
586 goto again;
587
588 }
589 }
590
591 /* IN_LOOPBACK must not appear on the wire - RFC1122. */
592 if (IN_LOOPBACK(ntohl(ip->ip_dst.s_addr)) ||
593 IN_LOOPBACK(ntohl(ip->ip_src.s_addr))) {
594 if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
595 IPSTAT_INC(ips_badaddr);
596 error = EADDRNOTAVAIL;
597 goto bad;
598 }
599 }
600
601 m->m_pkthdr.csum_flags |= CSUM_IP;
602 if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) {
603 in_delayed_cksum(m);
604 m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
605 }
606 #ifdef SCTP
607 if (m->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) {
608 sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2));
609 m->m_pkthdr.csum_flags &= ~CSUM_SCTP;
610 }
611 #endif
612
613 /*
614 * If small enough for interface, or the interface will take
615 * care of the fragmentation for us, we can just send directly.
616 */
617 if (ip_len <= mtu ||
618 (m->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0) {
619 ip->ip_sum = 0;
620 if (m->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) {
621 ip->ip_sum = in_cksum(m, hlen);
622 m->m_pkthdr.csum_flags &= ~CSUM_IP;
623 }
624
625 /*
626 * Record statistics for this interface address.
627 * With CSUM_TSO the byte/packet count will be slightly
628 * incorrect because we count the IP+TCP headers only
629 * once instead of for every generated packet.
630 */
631 if (!(flags & IP_FORWARDING) && ia) {
632 if (m->m_pkthdr.csum_flags & CSUM_TSO)
633 counter_u64_add(ia->ia_ifa.ifa_opackets,
634 m->m_pkthdr.len / m->m_pkthdr.tso_segsz);
635 else
636 counter_u64_add(ia->ia_ifa.ifa_opackets, 1);
637
638 counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len);
639 }
640 #ifdef MBUF_STRESS_TEST
641 if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size)
642 m = m_fragment(m, M_NOWAIT, mbuf_frag_size);
643 #endif
644 /*
645 * Reset layer specific mbuf flags
646 * to avoid confusing lower layers.
647 */
648 m_clrprotoflags(m);
649 IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL);
650 #ifdef RATELIMIT
651 if (inp != NULL) {
652 if (inp->inp_flags2 & INP_RATE_LIMIT_CHANGED)
653 in_pcboutput_txrtlmt(inp, ifp, m);
654 /* stamp send tag on mbuf */
655 m->m_pkthdr.snd_tag = inp->inp_snd_tag;
656 m->m_pkthdr.csum_flags |= CSUM_SND_TAG;
657 } else {
658 m->m_pkthdr.snd_tag = NULL;
659 }
660 #endif
661 error = (*ifp->if_output)(ifp, m,
662 (const struct sockaddr *)gw, ro);
663 #ifdef RATELIMIT
664 /* check for route change */
665 if (error == EAGAIN)
666 in_pcboutput_eagain(inp);
667 #endif
668 goto done;
669 }
670
671 /* Balk when DF bit is set or the interface didn't support TSO. */
672 if ((ip_off & IP_DF) || (m->m_pkthdr.csum_flags & CSUM_TSO)) {
673 error = EMSGSIZE;
674 IPSTAT_INC(ips_cantfrag);
675 goto bad;
676 }
677
678 /*
679 * Too large for interface; fragment if possible. If successful,
680 * on return, m will point to a list of packets to be sent.
681 */
682 error = ip_fragment(ip, &m, mtu, ifp->if_hwassist);
683 if (error)
684 goto bad;
685 for (; m; m = m0) {
686 m0 = m->m_nextpkt;
687 m->m_nextpkt = 0;
688 if (error == 0) {
689 /* Record statistics for this interface address. */
690 if (ia != NULL) {
691 counter_u64_add(ia->ia_ifa.ifa_opackets, 1);
692 counter_u64_add(ia->ia_ifa.ifa_obytes,
693 m->m_pkthdr.len);
694 }
695 /*
696 * Reset layer specific mbuf flags
697 * to avoid confusing upper layers.
698 */
699 m_clrprotoflags(m);
700
701 IP_PROBE(send, NULL, NULL, mtod(m, struct ip *), ifp,
702 mtod(m, struct ip *), NULL);
703 #ifdef RATELIMIT
704 if (inp != NULL) {
705 if (inp->inp_flags2 & INP_RATE_LIMIT_CHANGED)
706 in_pcboutput_txrtlmt(inp, ifp, m);
707 /* stamp send tag on mbuf */
708 m->m_pkthdr.snd_tag = inp->inp_snd_tag;
709 m->m_pkthdr.csum_flags |= CSUM_SND_TAG;
710 } else {
711 m->m_pkthdr.snd_tag = NULL;
712 }
713 #endif
714 error = (*ifp->if_output)(ifp, m,
715 (const struct sockaddr *)gw, ro);
716 #ifdef RATELIMIT
717 /* check for route change */
718 if (error == EAGAIN)
719 in_pcboutput_eagain(inp);
720 #endif
721 } else
722 m_freem(m);
723 }
724
725 if (error == 0)
726 IPSTAT_INC(ips_fragmented);
727
728 done:
729 if (ro == &iproute)
730 RO_RTFREE(ro);
731 else if (rte == NULL)
732 /*
733 * If the caller supplied a route but somehow the reference
734 * to it has been released need to prevent the caller
735 * calling RTFREE on it again.
736 */
737 ro->ro_rt = NULL;
738 NET_EPOCH_EXIT();
739 return (error);
740 bad:
741 m_freem(m);
742 goto done;
743 }
744
745 /*
746 * Create a chain of fragments which fit the given mtu. m_frag points to the
747 * mbuf to be fragmented; on return it points to the chain with the fragments.
748 * Return 0 if no error. If error, m_frag may contain a partially built
749 * chain of fragments that should be freed by the caller.
750 *
751 * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist)
752 */
753 int
ip_fragment(struct ip * ip,struct mbuf ** m_frag,int mtu,u_long if_hwassist_flags)754 ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu,
755 u_long if_hwassist_flags)
756 {
757 int error = 0;
758 int hlen = ip->ip_hl << 2;
759 int len = (mtu - hlen) & ~7; /* size of payload in each fragment */
760 int off;
761 struct mbuf *m0 = *m_frag; /* the original packet */
762 int firstlen;
763 struct mbuf **mnext;
764 int nfrags;
765 uint16_t ip_len, ip_off;
766
767 ip_len = ntohs(ip->ip_len);
768 ip_off = ntohs(ip->ip_off);
769
770 if (ip_off & IP_DF) { /* Fragmentation not allowed */
771 IPSTAT_INC(ips_cantfrag);
772 return EMSGSIZE;
773 }
774
775 /*
776 * Must be able to put at least 8 bytes per fragment.
777 */
778 if (len < 8)
779 return EMSGSIZE;
780
781 /*
782 * If the interface will not calculate checksums on
783 * fragmented packets, then do it here.
784 */
785 if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
786 in_delayed_cksum(m0);
787 m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
788 }
789 #ifdef SCTP
790 if (m0->m_pkthdr.csum_flags & CSUM_SCTP) {
791 sctp_delayed_cksum(m0, hlen);
792 m0->m_pkthdr.csum_flags &= ~CSUM_SCTP;
793 }
794 #endif
795 if (len > PAGE_SIZE) {
796 /*
797 * Fragment large datagrams such that each segment
798 * contains a multiple of PAGE_SIZE amount of data,
799 * plus headers. This enables a receiver to perform
800 * page-flipping zero-copy optimizations.
801 *
802 * XXX When does this help given that sender and receiver
803 * could have different page sizes, and also mtu could
804 * be less than the receiver's page size ?
805 */
806 int newlen;
807
808 off = MIN(mtu, m0->m_pkthdr.len);
809
810 /*
811 * firstlen (off - hlen) must be aligned on an
812 * 8-byte boundary
813 */
814 if (off < hlen)
815 goto smart_frag_failure;
816 off = ((off - hlen) & ~7) + hlen;
817 newlen = (~PAGE_MASK) & mtu;
818 if ((newlen + sizeof (struct ip)) > mtu) {
819 /* we failed, go back the default */
820 smart_frag_failure:
821 newlen = len;
822 off = hlen + len;
823 }
824 len = newlen;
825
826 } else {
827 off = hlen + len;
828 }
829
830 firstlen = off - hlen;
831 mnext = &m0->m_nextpkt; /* pointer to next packet */
832
833 /*
834 * Loop through length of segment after first fragment,
835 * make new header and copy data of each part and link onto chain.
836 * Here, m0 is the original packet, m is the fragment being created.
837 * The fragments are linked off the m_nextpkt of the original
838 * packet, which after processing serves as the first fragment.
839 */
840 for (nfrags = 1; off < ip_len; off += len, nfrags++) {
841 struct ip *mhip; /* ip header on the fragment */
842 struct mbuf *m;
843 int mhlen = sizeof (struct ip);
844
845 m = m_gethdr(M_NOWAIT, MT_DATA);
846 if (m == NULL) {
847 error = ENOBUFS;
848 IPSTAT_INC(ips_odropped);
849 goto done;
850 }
851 /*
852 * Make sure the complete packet header gets copied
853 * from the originating mbuf to the newly created
854 * mbuf. This also ensures that existing firewall
855 * classification(s), VLAN tags and so on get copied
856 * to the resulting fragmented packet(s):
857 */
858 if (m_dup_pkthdr(m, m0, M_NOWAIT) == 0) {
859 m_free(m);
860 error = ENOBUFS;
861 IPSTAT_INC(ips_odropped);
862 goto done;
863 }
864 /*
865 * In the first mbuf, leave room for the link header, then
866 * copy the original IP header including options. The payload
867 * goes into an additional mbuf chain returned by m_copym().
868 */
869 m->m_data += max_linkhdr;
870 mhip = mtod(m, struct ip *);
871 *mhip = *ip;
872 if (hlen > sizeof (struct ip)) {
873 mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
874 mhip->ip_v = IPVERSION;
875 mhip->ip_hl = mhlen >> 2;
876 }
877 m->m_len = mhlen;
878 /* XXX do we need to add ip_off below ? */
879 mhip->ip_off = ((off - hlen) >> 3) + ip_off;
880 if (off + len >= ip_len)
881 len = ip_len - off;
882 else
883 mhip->ip_off |= IP_MF;
884 mhip->ip_len = htons((u_short)(len + mhlen));
885 m->m_next = m_copym(m0, off, len, M_NOWAIT);
886 if (m->m_next == NULL) { /* copy failed */
887 m_free(m);
888 error = ENOBUFS; /* ??? */
889 IPSTAT_INC(ips_odropped);
890 goto done;
891 }
892 m->m_pkthdr.len = mhlen + len;
893 #ifdef MAC
894 mac_netinet_fragment(m0, m);
895 #endif
896 mhip->ip_off = htons(mhip->ip_off);
897 mhip->ip_sum = 0;
898 if (m->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) {
899 mhip->ip_sum = in_cksum(m, mhlen);
900 m->m_pkthdr.csum_flags &= ~CSUM_IP;
901 }
902 *mnext = m;
903 mnext = &m->m_nextpkt;
904 }
905 IPSTAT_ADD(ips_ofragments, nfrags);
906
907 /*
908 * Update first fragment by trimming what's been copied out
909 * and updating header.
910 */
911 m_adj(m0, hlen + firstlen - ip_len);
912 m0->m_pkthdr.len = hlen + firstlen;
913 ip->ip_len = htons((u_short)m0->m_pkthdr.len);
914 ip->ip_off = htons(ip_off | IP_MF);
915 ip->ip_sum = 0;
916 if (m0->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) {
917 ip->ip_sum = in_cksum(m0, hlen);
918 m0->m_pkthdr.csum_flags &= ~CSUM_IP;
919 }
920
921 done:
922 *m_frag = m0;
923 return error;
924 }
925
926 void
in_delayed_cksum(struct mbuf * m)927 in_delayed_cksum(struct mbuf *m)
928 {
929 struct ip *ip;
930 struct udphdr *uh;
931 uint16_t cklen, csum, offset;
932
933 ip = mtod(m, struct ip *);
934 offset = ip->ip_hl << 2 ;
935
936 if (m->m_pkthdr.csum_flags & CSUM_UDP) {
937 /* if udp header is not in the first mbuf copy udplen */
938 if (offset + sizeof(struct udphdr) > m->m_len) {
939 m_copydata(m, offset + offsetof(struct udphdr,
940 uh_ulen), sizeof(cklen), (caddr_t)&cklen);
941 cklen = ntohs(cklen);
942 } else {
943 uh = (struct udphdr *)mtodo(m, offset);
944 cklen = ntohs(uh->uh_ulen);
945 }
946 csum = in_cksum_skip(m, cklen + offset, offset);
947 if (csum == 0)
948 csum = 0xffff;
949 } else {
950 cklen = ntohs(ip->ip_len);
951 csum = in_cksum_skip(m, cklen, offset);
952 }
953 offset += m->m_pkthdr.csum_data; /* checksum offset */
954
955 if (offset + sizeof(csum) > m->m_len)
956 m_copyback(m, offset, sizeof(csum), (caddr_t)&csum);
957 else
958 *(u_short *)mtodo(m, offset) = csum;
959 }
960
961 /*
962 * IP socket option processing.
963 */
964 int
ip_ctloutput(struct socket * so,struct sockopt * sopt)965 ip_ctloutput(struct socket *so, struct sockopt *sopt)
966 {
967 struct inpcb *inp = sotoinpcb(so);
968 int error, optval;
969 #ifdef RSS
970 uint32_t rss_bucket;
971 int retval;
972 #endif
973
974 error = optval = 0;
975 if (sopt->sopt_level != IPPROTO_IP) {
976 error = EINVAL;
977
978 if (sopt->sopt_level == SOL_SOCKET &&
979 sopt->sopt_dir == SOPT_SET) {
980 switch (sopt->sopt_name) {
981 case SO_REUSEADDR:
982 INP_WLOCK(inp);
983 if ((so->so_options & SO_REUSEADDR) != 0)
984 inp->inp_flags2 |= INP_REUSEADDR;
985 else
986 inp->inp_flags2 &= ~INP_REUSEADDR;
987 INP_WUNLOCK(inp);
988 error = 0;
989 break;
990 case SO_REUSEPORT:
991 INP_WLOCK(inp);
992 if ((so->so_options & SO_REUSEPORT) != 0)
993 inp->inp_flags2 |= INP_REUSEPORT;
994 else
995 inp->inp_flags2 &= ~INP_REUSEPORT;
996 INP_WUNLOCK(inp);
997 error = 0;
998 break;
999 case SO_REUSEPORT_LB:
1000 INP_WLOCK(inp);
1001 if ((so->so_options & SO_REUSEPORT_LB) != 0)
1002 inp->inp_flags2 |= INP_REUSEPORT_LB;
1003 else
1004 inp->inp_flags2 &= ~INP_REUSEPORT_LB;
1005 INP_WUNLOCK(inp);
1006 error = 0;
1007 break;
1008 case SO_SETFIB:
1009 INP_WLOCK(inp);
1010 inp->inp_inc.inc_fibnum = so->so_fibnum;
1011 INP_WUNLOCK(inp);
1012 error = 0;
1013 break;
1014 case SO_MAX_PACING_RATE:
1015 #ifdef RATELIMIT
1016 INP_WLOCK(inp);
1017 inp->inp_flags2 |= INP_RATE_LIMIT_CHANGED;
1018 INP_WUNLOCK(inp);
1019 error = 0;
1020 #else
1021 error = EOPNOTSUPP;
1022 #endif
1023 break;
1024 default:
1025 break;
1026 }
1027 }
1028 return (error);
1029 }
1030
1031 switch (sopt->sopt_dir) {
1032 case SOPT_SET:
1033 switch (sopt->sopt_name) {
1034 case IP_OPTIONS:
1035 #ifdef notyet
1036 case IP_RETOPTS:
1037 #endif
1038 {
1039 struct mbuf *m;
1040 if (sopt->sopt_valsize > MLEN) {
1041 error = EMSGSIZE;
1042 break;
1043 }
1044 m = m_get(sopt->sopt_td ? M_WAITOK : M_NOWAIT, MT_DATA);
1045 if (m == NULL) {
1046 error = ENOBUFS;
1047 break;
1048 }
1049 m->m_len = sopt->sopt_valsize;
1050 error = sooptcopyin(sopt, mtod(m, char *), m->m_len,
1051 m->m_len);
1052 if (error) {
1053 m_free(m);
1054 break;
1055 }
1056 INP_WLOCK(inp);
1057 error = ip_pcbopts(inp, sopt->sopt_name, m);
1058 INP_WUNLOCK(inp);
1059 return (error);
1060 }
1061
1062 case IP_BINDANY:
1063 if (sopt->sopt_td != NULL) {
1064 error = priv_check(sopt->sopt_td,
1065 PRIV_NETINET_BINDANY);
1066 if (error)
1067 break;
1068 }
1069 /* FALLTHROUGH */
1070 case IP_BINDMULTI:
1071 #ifdef RSS
1072 case IP_RSS_LISTEN_BUCKET:
1073 #endif
1074 case IP_TOS:
1075 case IP_TTL:
1076 case IP_MINTTL:
1077 case IP_RECVOPTS:
1078 case IP_RECVRETOPTS:
1079 case IP_ORIGDSTADDR:
1080 case IP_RECVDSTADDR:
1081 case IP_RECVTTL:
1082 case IP_RECVIF:
1083 case IP_ONESBCAST:
1084 case IP_DONTFRAG:
1085 case IP_RECVTOS:
1086 case IP_RECVFLOWID:
1087 #ifdef RSS
1088 case IP_RECVRSSBUCKETID:
1089 #endif
1090 error = sooptcopyin(sopt, &optval, sizeof optval,
1091 sizeof optval);
1092 if (error)
1093 break;
1094
1095 switch (sopt->sopt_name) {
1096 case IP_TOS:
1097 inp->inp_ip_tos = optval;
1098 break;
1099
1100 case IP_TTL:
1101 inp->inp_ip_ttl = optval;
1102 break;
1103
1104 case IP_MINTTL:
1105 if (optval >= 0 && optval <= MAXTTL)
1106 inp->inp_ip_minttl = optval;
1107 else
1108 error = EINVAL;
1109 break;
1110
1111 #define OPTSET(bit) do { \
1112 INP_WLOCK(inp); \
1113 if (optval) \
1114 inp->inp_flags |= bit; \
1115 else \
1116 inp->inp_flags &= ~bit; \
1117 INP_WUNLOCK(inp); \
1118 } while (0)
1119
1120 #define OPTSET2(bit, val) do { \
1121 INP_WLOCK(inp); \
1122 if (val) \
1123 inp->inp_flags2 |= bit; \
1124 else \
1125 inp->inp_flags2 &= ~bit; \
1126 INP_WUNLOCK(inp); \
1127 } while (0)
1128
1129 case IP_RECVOPTS:
1130 OPTSET(INP_RECVOPTS);
1131 break;
1132
1133 case IP_RECVRETOPTS:
1134 OPTSET(INP_RECVRETOPTS);
1135 break;
1136
1137 case IP_RECVDSTADDR:
1138 OPTSET(INP_RECVDSTADDR);
1139 break;
1140
1141 case IP_ORIGDSTADDR:
1142 OPTSET2(INP_ORIGDSTADDR, optval);
1143 break;
1144
1145 case IP_RECVTTL:
1146 OPTSET(INP_RECVTTL);
1147 break;
1148
1149 case IP_RECVIF:
1150 OPTSET(INP_RECVIF);
1151 break;
1152
1153 case IP_ONESBCAST:
1154 OPTSET(INP_ONESBCAST);
1155 break;
1156 case IP_DONTFRAG:
1157 OPTSET(INP_DONTFRAG);
1158 break;
1159 case IP_BINDANY:
1160 OPTSET(INP_BINDANY);
1161 break;
1162 case IP_RECVTOS:
1163 OPTSET(INP_RECVTOS);
1164 break;
1165 case IP_BINDMULTI:
1166 OPTSET2(INP_BINDMULTI, optval);
1167 break;
1168 case IP_RECVFLOWID:
1169 OPTSET2(INP_RECVFLOWID, optval);
1170 break;
1171 #ifdef RSS
1172 case IP_RSS_LISTEN_BUCKET:
1173 if ((optval >= 0) &&
1174 (optval < rss_getnumbuckets())) {
1175 inp->inp_rss_listen_bucket = optval;
1176 OPTSET2(INP_RSS_BUCKET_SET, 1);
1177 } else {
1178 error = EINVAL;
1179 }
1180 break;
1181 case IP_RECVRSSBUCKETID:
1182 OPTSET2(INP_RECVRSSBUCKETID, optval);
1183 break;
1184 #endif
1185 }
1186 break;
1187 #undef OPTSET
1188 #undef OPTSET2
1189
1190 /*
1191 * Multicast socket options are processed by the in_mcast
1192 * module.
1193 */
1194 case IP_MULTICAST_IF:
1195 case IP_MULTICAST_VIF:
1196 case IP_MULTICAST_TTL:
1197 case IP_MULTICAST_LOOP:
1198 case IP_ADD_MEMBERSHIP:
1199 case IP_DROP_MEMBERSHIP:
1200 case IP_ADD_SOURCE_MEMBERSHIP:
1201 case IP_DROP_SOURCE_MEMBERSHIP:
1202 case IP_BLOCK_SOURCE:
1203 case IP_UNBLOCK_SOURCE:
1204 case IP_MSFILTER:
1205 case MCAST_JOIN_GROUP:
1206 case MCAST_LEAVE_GROUP:
1207 case MCAST_JOIN_SOURCE_GROUP:
1208 case MCAST_LEAVE_SOURCE_GROUP:
1209 case MCAST_BLOCK_SOURCE:
1210 case MCAST_UNBLOCK_SOURCE:
1211 error = inp_setmoptions(inp, sopt);
1212 break;
1213
1214 case IP_PORTRANGE:
1215 error = sooptcopyin(sopt, &optval, sizeof optval,
1216 sizeof optval);
1217 if (error)
1218 break;
1219
1220 INP_WLOCK(inp);
1221 switch (optval) {
1222 case IP_PORTRANGE_DEFAULT:
1223 inp->inp_flags &= ~(INP_LOWPORT);
1224 inp->inp_flags &= ~(INP_HIGHPORT);
1225 break;
1226
1227 case IP_PORTRANGE_HIGH:
1228 inp->inp_flags &= ~(INP_LOWPORT);
1229 inp->inp_flags |= INP_HIGHPORT;
1230 break;
1231
1232 case IP_PORTRANGE_LOW:
1233 inp->inp_flags &= ~(INP_HIGHPORT);
1234 inp->inp_flags |= INP_LOWPORT;
1235 break;
1236
1237 default:
1238 error = EINVAL;
1239 break;
1240 }
1241 INP_WUNLOCK(inp);
1242 break;
1243
1244 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
1245 case IP_IPSEC_POLICY:
1246 if (IPSEC_ENABLED(ipv4)) {
1247 error = IPSEC_PCBCTL(ipv4, inp, sopt);
1248 break;
1249 }
1250 /* FALLTHROUGH */
1251 #endif /* IPSEC */
1252
1253 default:
1254 error = ENOPROTOOPT;
1255 break;
1256 }
1257 break;
1258
1259 case SOPT_GET:
1260 switch (sopt->sopt_name) {
1261 case IP_OPTIONS:
1262 case IP_RETOPTS:
1263 INP_RLOCK(inp);
1264 if (inp->inp_options) {
1265 struct mbuf *options;
1266
1267 options = m_copym(inp->inp_options, 0,
1268 M_COPYALL, M_NOWAIT);
1269 INP_RUNLOCK(inp);
1270 if (options != NULL) {
1271 error = sooptcopyout(sopt,
1272 mtod(options, char *),
1273 options->m_len);
1274 m_freem(options);
1275 } else
1276 error = ENOMEM;
1277 } else {
1278 INP_RUNLOCK(inp);
1279 sopt->sopt_valsize = 0;
1280 }
1281 break;
1282
1283 case IP_TOS:
1284 case IP_TTL:
1285 case IP_MINTTL:
1286 case IP_RECVOPTS:
1287 case IP_RECVRETOPTS:
1288 case IP_ORIGDSTADDR:
1289 case IP_RECVDSTADDR:
1290 case IP_RECVTTL:
1291 case IP_RECVIF:
1292 case IP_PORTRANGE:
1293 case IP_ONESBCAST:
1294 case IP_DONTFRAG:
1295 case IP_BINDANY:
1296 case IP_RECVTOS:
1297 case IP_BINDMULTI:
1298 case IP_FLOWID:
1299 case IP_FLOWTYPE:
1300 case IP_RECVFLOWID:
1301 #ifdef RSS
1302 case IP_RSSBUCKETID:
1303 case IP_RECVRSSBUCKETID:
1304 #endif
1305 switch (sopt->sopt_name) {
1306
1307 case IP_TOS:
1308 optval = inp->inp_ip_tos;
1309 break;
1310
1311 case IP_TTL:
1312 optval = inp->inp_ip_ttl;
1313 break;
1314
1315 case IP_MINTTL:
1316 optval = inp->inp_ip_minttl;
1317 break;
1318
1319 #define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0)
1320 #define OPTBIT2(bit) (inp->inp_flags2 & bit ? 1 : 0)
1321
1322 case IP_RECVOPTS:
1323 optval = OPTBIT(INP_RECVOPTS);
1324 break;
1325
1326 case IP_RECVRETOPTS:
1327 optval = OPTBIT(INP_RECVRETOPTS);
1328 break;
1329
1330 case IP_RECVDSTADDR:
1331 optval = OPTBIT(INP_RECVDSTADDR);
1332 break;
1333
1334 case IP_ORIGDSTADDR:
1335 optval = OPTBIT2(INP_ORIGDSTADDR);
1336 break;
1337
1338 case IP_RECVTTL:
1339 optval = OPTBIT(INP_RECVTTL);
1340 break;
1341
1342 case IP_RECVIF:
1343 optval = OPTBIT(INP_RECVIF);
1344 break;
1345
1346 case IP_PORTRANGE:
1347 if (inp->inp_flags & INP_HIGHPORT)
1348 optval = IP_PORTRANGE_HIGH;
1349 else if (inp->inp_flags & INP_LOWPORT)
1350 optval = IP_PORTRANGE_LOW;
1351 else
1352 optval = 0;
1353 break;
1354
1355 case IP_ONESBCAST:
1356 optval = OPTBIT(INP_ONESBCAST);
1357 break;
1358 case IP_DONTFRAG:
1359 optval = OPTBIT(INP_DONTFRAG);
1360 break;
1361 case IP_BINDANY:
1362 optval = OPTBIT(INP_BINDANY);
1363 break;
1364 case IP_RECVTOS:
1365 optval = OPTBIT(INP_RECVTOS);
1366 break;
1367 case IP_FLOWID:
1368 optval = inp->inp_flowid;
1369 break;
1370 case IP_FLOWTYPE:
1371 optval = inp->inp_flowtype;
1372 break;
1373 case IP_RECVFLOWID:
1374 optval = OPTBIT2(INP_RECVFLOWID);
1375 break;
1376 #ifdef RSS
1377 case IP_RSSBUCKETID:
1378 retval = rss_hash2bucket(inp->inp_flowid,
1379 inp->inp_flowtype,
1380 &rss_bucket);
1381 if (retval == 0)
1382 optval = rss_bucket;
1383 else
1384 error = EINVAL;
1385 break;
1386 case IP_RECVRSSBUCKETID:
1387 optval = OPTBIT2(INP_RECVRSSBUCKETID);
1388 break;
1389 #endif
1390 case IP_BINDMULTI:
1391 optval = OPTBIT2(INP_BINDMULTI);
1392 break;
1393 }
1394 error = sooptcopyout(sopt, &optval, sizeof optval);
1395 break;
1396
1397 /*
1398 * Multicast socket options are processed by the in_mcast
1399 * module.
1400 */
1401 case IP_MULTICAST_IF:
1402 case IP_MULTICAST_VIF:
1403 case IP_MULTICAST_TTL:
1404 case IP_MULTICAST_LOOP:
1405 case IP_MSFILTER:
1406 error = inp_getmoptions(inp, sopt);
1407 break;
1408
1409 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
1410 case IP_IPSEC_POLICY:
1411 if (IPSEC_ENABLED(ipv4)) {
1412 error = IPSEC_PCBCTL(ipv4, inp, sopt);
1413 break;
1414 }
1415 /* FALLTHROUGH */
1416 #endif /* IPSEC */
1417
1418 default:
1419 error = ENOPROTOOPT;
1420 break;
1421 }
1422 break;
1423 }
1424 return (error);
1425 }
1426
1427 /*
1428 * Routine called from ip_output() to loop back a copy of an IP multicast
1429 * packet to the input queue of a specified interface. Note that this
1430 * calls the output routine of the loopback "driver", but with an interface
1431 * pointer that might NOT be a loopback interface -- evil, but easier than
1432 * replicating that code here.
1433 */
1434 static void
ip_mloopback(struct ifnet * ifp,const struct mbuf * m,int hlen)1435 ip_mloopback(struct ifnet *ifp, const struct mbuf *m, int hlen)
1436 {
1437 struct ip *ip;
1438 struct mbuf *copym;
1439
1440 /*
1441 * Make a deep copy of the packet because we're going to
1442 * modify the pack in order to generate checksums.
1443 */
1444 copym = m_dup(m, M_NOWAIT);
1445 if (copym != NULL && (!M_WRITABLE(copym) || copym->m_len < hlen))
1446 copym = m_pullup(copym, hlen);
1447 if (copym != NULL) {
1448 /* If needed, compute the checksum and mark it as valid. */
1449 if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
1450 in_delayed_cksum(copym);
1451 copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
1452 copym->m_pkthdr.csum_flags |=
1453 CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
1454 copym->m_pkthdr.csum_data = 0xffff;
1455 }
1456 /*
1457 * We don't bother to fragment if the IP length is greater
1458 * than the interface's MTU. Can this possibly matter?
1459 */
1460 ip = mtod(copym, struct ip *);
1461 ip->ip_sum = 0;
1462 ip->ip_sum = in_cksum(copym, hlen);
1463 if_simloop(ifp, copym, AF_INET, 0);
1464 }
1465 }
1466