xref: /titanic_51/usr/src/uts/common/inet/ip/ip_rts.c (revision f4ce81cfdad23bacfdb147be77d8d5fbe7673847)
1 /*
2  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
3  * Use is subject to license terms.
4  */
5 
6 /*
7  * Copyright (c) 1988, 1991, 1993
8  *	The Regents of the University of California.  All rights reserved.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. All advertising materials mentioning features or use of this software
19  *    must display the following acknowledgement:
20  *	This product includes software developed by the University of
21  *	California, Berkeley and its contributors.
22  * 4. Neither the name of the University nor the names of its contributors
23  *    may be used to endorse or promote products derived from this software
24  *    without specific prior written permission.
25  *
26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36  * SUCH DAMAGE.
37  *
38  *	@(#)rtsock.c	8.6 (Berkeley) 2/11/95
39  */
40 
41 /*
42  * This file contains routines that processes routing socket requests.
43  */
44 
45 #include <sys/types.h>
46 #include <sys/stream.h>
47 #include <sys/stropts.h>
48 #include <sys/ddi.h>
49 #include <sys/strsubr.h>
50 #include <sys/cmn_err.h>
51 #include <sys/debug.h>
52 #include <sys/policy.h>
53 #include <sys/zone.h>
54 
55 #include <sys/systm.h>
56 #include <sys/param.h>
57 #include <sys/socket.h>
58 #include <sys/strsun.h>
59 #include <net/if.h>
60 #include <net/route.h>
61 #include <netinet/in.h>
62 #include <net/if_dl.h>
63 #include <netinet/ip6.h>
64 
65 #include <inet/common.h>
66 #include <inet/ip.h>
67 #include <inet/ip6.h>
68 #include <inet/ip_if.h>
69 #include <inet/ip_ire.h>
70 #include <inet/ip_ftable.h>
71 #include <inet/ip_rts.h>
72 
73 #include <inet/ipclassifier.h>
74 
75 #include <sys/tsol/tndb.h>
76 #include <sys/tsol/tnet.h>
77 
78 #define	RTS_MSG_SIZE(type, rtm_addrs, af, sacnt) \
79 	(rts_data_msg_size(rtm_addrs, af, sacnt) + rts_header_msg_size(type))
80 
81 static size_t	rts_copyfromsockaddr(struct sockaddr *sa, in6_addr_t *addrp);
82 static void	rts_fill_msg(int type, int rtm_addrs, ipaddr_t dst,
83     ipaddr_t mask, ipaddr_t gateway, ipaddr_t src_addr, ipaddr_t brd_addr,
84     ipaddr_t author, ipaddr_t ifaddr, const ill_t *ill, mblk_t *mp,
85     const tsol_gc_t *);
86 static int	rts_getaddrs(rt_msghdr_t *rtm, in6_addr_t *dst_addrp,
87     in6_addr_t *gw_addrp, in6_addr_t *net_maskp, in6_addr_t *authorp,
88     in6_addr_t *if_addrp, in6_addr_t *src_addrp, ushort_t *indexp,
89     sa_family_t *afp, tsol_rtsecattr_t *rtsecattr, int *error);
90 static void	rts_getifdata(if_data_t *if_data, const ipif_t *ipif);
91 static int	rts_getmetrics(ire_t *ire, rt_metrics_t *metrics);
92 static mblk_t	*rts_rtmget(mblk_t *mp, ire_t *ire, ire_t *ifire,
93     const in6_addr_t *setsrc, tsol_ire_gw_secattr_t *attrp, sa_family_t af);
94 static void	rts_setmetrics(ire_t *ire, uint_t which, rt_metrics_t *metrics);
95 static ire_t	*ire_lookup_v4(ipaddr_t dst_addr, ipaddr_t net_mask,
96     ipaddr_t gw_addr, const ill_t *ill, zoneid_t zoneid,
97     const ts_label_t *tsl, int match_flags, ip_stack_t *ipst, ire_t **pifire,
98     ipaddr_t *v4setsrcp, tsol_ire_gw_secattr_t **gwattrp);
99 static ire_t	*ire_lookup_v6(const in6_addr_t *dst_addr_v6,
100     const in6_addr_t *net_mask_v6, const in6_addr_t *gw_addr_v6,
101     const ill_t *ill, zoneid_t zoneid, const ts_label_t *tsl, int match_flags,
102     ip_stack_t *ipst, ire_t **pifire,
103     in6_addr_t *v6setsrcp, tsol_ire_gw_secattr_t **gwattrp);
104 
105 /*
106  * Send `mp' to all eligible routing queues.  A queue is ineligible if:
107  *
108  *  1. SO_USELOOPBACK is off and it is not the originating queue.
109  *  2. RTA_UNDER_IPMP is on and RTSQ_UNDER_IPMP is not set in `flags'.
110  *  3. RTA_UNDER_IPMP is off and RTSQ_NORMAL is not set in `flags'.
111  *  4. It is not the same address family as `af', and `af' isn't AF_UNSPEC.
112  */
113 void
114 rts_queue_input(mblk_t *mp, conn_t *o_connp, sa_family_t af, uint_t flags,
115     ip_stack_t *ipst)
116 {
117 	mblk_t	*mp1;
118 	conn_t 	*connp, *next_connp;
119 
120 	/*
121 	 * Since we don't have an ill_t here, RTSQ_DEFAULT must already be
122 	 * resolved to one or more of RTSQ_NORMAL|RTSQ_UNDER_IPMP at this point.
123 	 */
124 	ASSERT(!(flags & RTSQ_DEFAULT));
125 
126 	mutex_enter(&ipst->ips_rts_clients->connf_lock);
127 	connp = ipst->ips_rts_clients->connf_head;
128 
129 	for (; connp != NULL; connp = next_connp) {
130 		next_connp = connp->conn_next;
131 		/*
132 		 * If there was a family specified when this routing socket was
133 		 * created and it doesn't match the family of the message to
134 		 * copy, then continue.
135 		 */
136 		if ((connp->conn_proto != AF_UNSPEC) &&
137 		    (connp->conn_proto != af))
138 			continue;
139 
140 		/*
141 		 * Queue the message only if the conn_t and flags match.
142 		 */
143 		if (connp->conn_rtaware & RTAW_UNDER_IPMP) {
144 			if (!(flags & RTSQ_UNDER_IPMP))
145 				continue;
146 		} else {
147 			if (!(flags & RTSQ_NORMAL))
148 				continue;
149 		}
150 		/*
151 		 * For the originating queue, we only copy the message upstream
152 		 * if loopback is set.  For others reading on the routing
153 		 * socket, we check if there is room upstream for a copy of the
154 		 * message.
155 		 */
156 		if ((o_connp == connp) && connp->conn_useloopback == 0) {
157 			connp = connp->conn_next;
158 			continue;
159 		}
160 		CONN_INC_REF(connp);
161 		mutex_exit(&ipst->ips_rts_clients->connf_lock);
162 		/* Pass to rts_input */
163 		if (IPCL_IS_NONSTR(connp) ? !connp->conn_flow_cntrld :
164 		    canputnext(connp->conn_rq)) {
165 			mp1 = dupmsg(mp);
166 			if (mp1 == NULL)
167 				mp1 = copymsg(mp);
168 			/* Note that we pass a NULL ira to rts_input */
169 			if (mp1 != NULL)
170 				(connp->conn_recv)(connp, mp1, NULL, NULL);
171 		}
172 
173 		mutex_enter(&ipst->ips_rts_clients->connf_lock);
174 		/* reload next_connp since conn_next may have changed */
175 		next_connp = connp->conn_next;
176 		CONN_DEC_REF(connp);
177 	}
178 	mutex_exit(&ipst->ips_rts_clients->connf_lock);
179 	freemsg(mp);
180 }
181 
182 /*
183  * Takes an ire and sends an ack to all the routing sockets. This
184  * routine is used
185  * - when a route is created/deleted through the ioctl interface.
186  * - when a stale redirect is deleted
187  */
188 void
189 ip_rts_rtmsg(int type, ire_t *ire, int error, ip_stack_t *ipst)
190 {
191 	mblk_t		*mp;
192 	rt_msghdr_t	*rtm;
193 	int		rtm_addrs = (RTA_DST | RTA_NETMASK | RTA_GATEWAY);
194 	sa_family_t	af;
195 	in6_addr_t	gw_addr_v6;
196 
197 	if (ire == NULL)
198 		return;
199 	ASSERT(ire->ire_ipversion == IPV4_VERSION ||
200 	    ire->ire_ipversion == IPV6_VERSION);
201 
202 	ASSERT(!(ire->ire_type & IRE_IF_CLONE));
203 
204 	if (ire->ire_flags & RTF_SETSRC)
205 		rtm_addrs |= RTA_SRC;
206 
207 	switch (ire->ire_ipversion) {
208 	case IPV4_VERSION:
209 		af = AF_INET;
210 		mp = rts_alloc_msg(type, rtm_addrs, af, 0);
211 		if (mp == NULL)
212 			return;
213 		rts_fill_msg(type, rtm_addrs, ire->ire_addr, ire->ire_mask,
214 		    ire->ire_gateway_addr, ire->ire_setsrc_addr, 0, 0, 0, NULL,
215 		    mp, NULL);
216 		break;
217 	case IPV6_VERSION:
218 		af = AF_INET6;
219 		mp = rts_alloc_msg(type, rtm_addrs, af, 0);
220 		if (mp == NULL)
221 			return;
222 		mutex_enter(&ire->ire_lock);
223 		gw_addr_v6 = ire->ire_gateway_addr_v6;
224 		mutex_exit(&ire->ire_lock);
225 		rts_fill_msg_v6(type, rtm_addrs, &ire->ire_addr_v6,
226 		    &ire->ire_mask_v6, &gw_addr_v6,
227 		    &ire->ire_setsrc_addr_v6, &ipv6_all_zeros, &ipv6_all_zeros,
228 		    &ipv6_all_zeros, NULL, mp, NULL);
229 		break;
230 	}
231 	rtm = (rt_msghdr_t *)mp->b_rptr;
232 	mp->b_wptr = (uchar_t *)&mp->b_rptr[rtm->rtm_msglen];
233 	rtm->rtm_addrs = rtm_addrs;
234 	rtm->rtm_flags = ire->ire_flags;
235 	if (error != 0)
236 		rtm->rtm_errno = error;
237 	else
238 		rtm->rtm_flags |= RTF_DONE;
239 	rts_queue_input(mp, NULL, af, RTSQ_ALL, ipst);
240 }
241 
242 /*
243  * This is a call from the RTS module
244  * indicating that this is a Routing Socket
245  * Stream. Insert this conn_t in routing
246  * socket client list.
247  */
248 void
249 ip_rts_register(conn_t *connp)
250 {
251 	ip_stack_t *ipst = connp->conn_netstack->netstack_ip;
252 
253 	connp->conn_useloopback = 1;
254 	ipcl_hash_insert_wildcard(ipst->ips_rts_clients, connp);
255 }
256 
257 /*
258  * This is a call from the RTS module indicating that it is closing.
259  */
260 void
261 ip_rts_unregister(conn_t *connp)
262 {
263 	ipcl_hash_remove(connp);
264 }
265 
266 /*
267  * Processes requests received on a routing socket. It extracts all the
268  * arguments and calls the appropriate function to process the request.
269  *
270  * RTA_SRC bit flag requests are sent by 'route -setsrc'.
271  *
272  * In general, this function does not consume the message supplied but rather
273  * sends the message upstream with an appropriate UNIX errno.
274  */
275 int
276 ip_rts_request_common(mblk_t *mp, conn_t *connp, cred_t *ioc_cr)
277 {
278 	rt_msghdr_t	*rtm = NULL;
279 	in6_addr_t	dst_addr_v6;
280 	in6_addr_t	src_addr_v6;
281 	in6_addr_t	gw_addr_v6;
282 	in6_addr_t	net_mask_v6;
283 	in6_addr_t	author_v6;
284 	in6_addr_t	if_addr_v6;
285 	mblk_t		*mp1;
286 	ire_t		*ire = NULL;
287 	ire_t		*ifire = NULL;
288 	ipaddr_t	v4setsrc;
289 	in6_addr_t	v6setsrc = ipv6_all_zeros;
290 	tsol_ire_gw_secattr_t *gwattr = NULL;
291 	int		error = 0;
292 	int		match_flags = MATCH_IRE_DSTONLY;
293 	int		match_flags_local = MATCH_IRE_TYPE | MATCH_IRE_GW;
294 	int		found_addrs;
295 	sa_family_t	af;
296 	ipaddr_t	dst_addr;
297 	ipaddr_t	gw_addr;
298 	ipaddr_t	src_addr;
299 	ipaddr_t	net_mask;
300 	ushort_t	index;
301 	boolean_t	gcgrp_xtraref = B_FALSE;
302 	tsol_gcgrp_addr_t ga;
303 	tsol_rtsecattr_t rtsecattr;
304 	struct rtsa_s	*rtsap = NULL;
305 	tsol_gcgrp_t	*gcgrp = NULL;
306 	tsol_gc_t	*gc = NULL;
307 	ts_label_t	*tsl = NULL;
308 	zoneid_t	zoneid;
309 	ip_stack_t	*ipst;
310 	ill_t   	*ill = NULL;
311 
312 	zoneid = connp->conn_zoneid;
313 	ipst = connp->conn_netstack->netstack_ip;
314 
315 	if (mp->b_cont != NULL && !pullupmsg(mp, -1)) {
316 		freemsg(mp);
317 		error =  EINVAL;
318 		goto done;
319 	}
320 	if ((mp->b_wptr - mp->b_rptr) < sizeof (rt_msghdr_t)) {
321 		freemsg(mp);
322 		error = EINVAL;
323 		goto done;
324 	}
325 
326 	/*
327 	 * Check the routing message for basic consistency including the
328 	 * version number and that the number of octets written is the same
329 	 * as specified by the rtm_msglen field.
330 	 *
331 	 * At this point, an error can be delivered back via rtm_errno.
332 	 */
333 	rtm = (rt_msghdr_t *)mp->b_rptr;
334 	if ((mp->b_wptr - mp->b_rptr) != rtm->rtm_msglen) {
335 		error = EINVAL;
336 		goto done;
337 	}
338 	if (rtm->rtm_version != RTM_VERSION) {
339 		error = EPROTONOSUPPORT;
340 		goto done;
341 	}
342 
343 	/* Only allow RTM_GET or RTM_RESOLVE for unprivileged process */
344 	if (rtm->rtm_type != RTM_GET &&
345 	    rtm->rtm_type != RTM_RESOLVE &&
346 	    (ioc_cr == NULL ||
347 	    secpolicy_ip_config(ioc_cr, B_FALSE) != 0)) {
348 		error = EPERM;
349 		goto done;
350 	}
351 
352 	found_addrs = rts_getaddrs(rtm, &dst_addr_v6, &gw_addr_v6, &net_mask_v6,
353 	    &author_v6, &if_addr_v6, &src_addr_v6, &index, &af, &rtsecattr,
354 	    &error);
355 
356 	if (error != 0)
357 		goto done;
358 
359 	if ((found_addrs & RTA_DST) == 0) {
360 		error = EINVAL;
361 		goto done;
362 	}
363 
364 	/*
365 	 * Based on the address family of the destination address, determine
366 	 * the destination, gateway and netmask and return the appropriate error
367 	 * if an unknown address family was specified (following the errno
368 	 * values that 4.4BSD-Lite2 returns.)
369 	 */
370 	switch (af) {
371 	case AF_INET:
372 		IN6_V4MAPPED_TO_IPADDR(&dst_addr_v6, dst_addr);
373 		IN6_V4MAPPED_TO_IPADDR(&src_addr_v6, src_addr);
374 		IN6_V4MAPPED_TO_IPADDR(&gw_addr_v6, gw_addr);
375 		if (((found_addrs & RTA_NETMASK) == 0) ||
376 		    (rtm->rtm_flags & RTF_HOST))
377 			net_mask = IP_HOST_MASK;
378 		else
379 			IN6_V4MAPPED_TO_IPADDR(&net_mask_v6, net_mask);
380 		break;
381 	case AF_INET6:
382 		if (((found_addrs & RTA_NETMASK) == 0) ||
383 		    (rtm->rtm_flags & RTF_HOST))
384 			net_mask_v6 = ipv6_all_ones;
385 		break;
386 	default:
387 		/*
388 		 * These errno values are meant to be compatible with
389 		 * 4.4BSD-Lite2 for the given message types.
390 		 */
391 		switch (rtm->rtm_type) {
392 		case RTM_ADD:
393 		case RTM_DELETE:
394 			error = ESRCH;
395 			goto done;
396 		case RTM_GET:
397 		case RTM_CHANGE:
398 			error = EAFNOSUPPORT;
399 			goto done;
400 		default:
401 			error = EOPNOTSUPP;
402 			goto done;
403 		}
404 	}
405 
406 	/*
407 	 * At this point, the address family must be something known.
408 	 */
409 	ASSERT(af == AF_INET || af == AF_INET6);
410 
411 	/* Handle RTA_IFP */
412 	if (index != 0) {
413 		ipif_t		*ipif;
414 lookup:
415 		ill = ill_lookup_on_ifindex(index, af == AF_INET6, ipst);
416 		if (ill == NULL) {
417 			error = EINVAL;
418 			goto done;
419 		}
420 
421 		/*
422 		 * Since all interfaces in an IPMP group must be equivalent,
423 		 * we prevent changes to a specific underlying interface's
424 		 * routing configuration.  However, for backward compatibility,
425 		 * we intepret a request to add a route on an underlying
426 		 * interface as a request to add a route on its IPMP interface.
427 		 */
428 		if (IS_UNDER_IPMP(ill)) {
429 			switch (rtm->rtm_type) {
430 			case RTM_CHANGE:
431 			case RTM_DELETE:
432 				error = EINVAL;
433 				goto done;
434 			case RTM_ADD:
435 				index = ipmp_ill_get_ipmp_ifindex(ill);
436 				ill_refrele(ill);
437 				if (index == 0) {
438 					ill = NULL; /* already refrele'd */
439 					error = EINVAL;
440 					goto done;
441 				}
442 				goto lookup;
443 			}
444 		}
445 
446 		match_flags |= MATCH_IRE_ILL;
447 		/*
448 		 * This provides the same zoneid as in Solaris 10
449 		 * that -ifp picks the zoneid from the first ipif on the ill.
450 		 * But it might not be useful since the first ipif will always
451 		 * have the same zoneid as the ill.
452 		 */
453 		ipif = ipif_get_next_ipif(NULL, ill);
454 		if (ipif != NULL) {
455 			zoneid = ipif->ipif_zoneid;
456 			ipif_refrele(ipif);
457 		}
458 	}
459 
460 	/*
461 	 * If a netmask was supplied in the message, then subsequent route
462 	 * lookups will attempt to match on the netmask as well.
463 	 */
464 	if ((found_addrs & RTA_NETMASK) != 0)
465 		match_flags |= MATCH_IRE_MASK;
466 
467 	/*
468 	 * We only process any passed-in route security attributes for
469 	 * either RTM_ADD or RTM_CHANGE message; We overload them
470 	 * to do an RTM_GET as a different label; ignore otherwise.
471 	 */
472 	if (rtm->rtm_type == RTM_ADD || rtm->rtm_type == RTM_CHANGE ||
473 	    rtm->rtm_type == RTM_GET) {
474 		ASSERT(rtsecattr.rtsa_cnt <= TSOL_RTSA_REQUEST_MAX);
475 		if (rtsecattr.rtsa_cnt > 0)
476 			rtsap = &rtsecattr.rtsa_attr[0];
477 	}
478 
479 	switch (rtm->rtm_type) {
480 	case RTM_ADD:
481 		/* if we are adding a route, gateway is a must */
482 		if ((found_addrs & RTA_GATEWAY) == 0) {
483 			error = EINVAL;
484 			goto done;
485 		}
486 
487 		/* Multirouting does not support net routes. */
488 		if ((rtm->rtm_flags & (RTF_MULTIRT | RTF_HOST)) ==
489 		    RTF_MULTIRT) {
490 			error = EADDRNOTAVAIL;
491 			goto done;
492 		}
493 
494 		/*
495 		 * Multirouting and user-specified source addresses
496 		 * do not support interface based routing.
497 		 * Assigning a source address to an interface based
498 		 * route is achievable by plumbing a new ipif and
499 		 * setting up the interface route via this ipif,
500 		 * though.
501 		 */
502 		if (rtm->rtm_flags & (RTF_MULTIRT | RTF_SETSRC)) {
503 			if ((rtm->rtm_flags & RTF_GATEWAY) == 0) {
504 				error = EADDRNOTAVAIL;
505 				goto done;
506 			}
507 		}
508 
509 		switch (af) {
510 		case AF_INET:
511 			if (src_addr != INADDR_ANY) {
512 				uint_t type;
513 
514 				/*
515 				 * The RTF_SETSRC flag is present, check that
516 				 * the supplied src address is not the loopback
517 				 * address. This would produce martian packets.
518 				 */
519 				if (src_addr == htonl(INADDR_LOOPBACK)) {
520 					error = EINVAL;
521 					goto done;
522 				}
523 				/*
524 				 * Also check that the supplied address is a
525 				 * valid, local one. Only allow IFF_UP ones
526 				 */
527 				type = ip_type_v4(src_addr, ipst);
528 				if (!(type & (IRE_LOCAL|IRE_LOOPBACK))) {
529 					error = EADDRNOTAVAIL;
530 					goto done;
531 				}
532 			} else {
533 				/*
534 				 * The RTF_SETSRC modifier must be associated
535 				 * to a non-null source address.
536 				 */
537 				if (rtm->rtm_flags & RTF_SETSRC) {
538 					error = EINVAL;
539 					goto done;
540 				}
541 			}
542 
543 			error = ip_rt_add(dst_addr, net_mask, gw_addr, src_addr,
544 			    rtm->rtm_flags, ill, &ire, B_FALSE,
545 			    rtsap, ipst, zoneid);
546 			if (ill != NULL)
547 				ASSERT(!MUTEX_HELD(&ill->ill_lock));
548 			break;
549 		case AF_INET6:
550 			if (!IN6_IS_ADDR_UNSPECIFIED(&src_addr_v6)) {
551 				uint_t type;
552 
553 				/*
554 				 * The RTF_SETSRC flag is present, check that
555 				 * the supplied src address is not the loopback
556 				 * address. This would produce martian packets.
557 				 */
558 				if (IN6_IS_ADDR_LOOPBACK(&src_addr_v6)) {
559 					error = EINVAL;
560 					goto done;
561 				}
562 				/*
563 				 * Also check that the supplied address is a
564 				 * valid, local one. Only allow UP ones.
565 				 */
566 				type = ip_type_v6(&src_addr_v6, ipst);
567 				if (!(type & (IRE_LOCAL|IRE_LOOPBACK))) {
568 					error = EADDRNOTAVAIL;
569 					goto done;
570 				}
571 
572 				error = ip_rt_add_v6(&dst_addr_v6, &net_mask_v6,
573 				    &gw_addr_v6, &src_addr_v6, rtm->rtm_flags,
574 				    ill, &ire, rtsap, ipst, zoneid);
575 				break;
576 			}
577 			/*
578 			 * The RTF_SETSRC modifier must be associated
579 			 * to a non-null source address.
580 			 */
581 			if (rtm->rtm_flags & RTF_SETSRC) {
582 				error = EINVAL;
583 				goto done;
584 			}
585 			error = ip_rt_add_v6(&dst_addr_v6, &net_mask_v6,
586 			    &gw_addr_v6, NULL, rtm->rtm_flags,
587 			    ill, &ire, rtsap, ipst, zoneid);
588 			if (ill != NULL)
589 				ASSERT(!MUTEX_HELD(&ill->ill_lock));
590 			break;
591 		}
592 		if (error != 0)
593 			goto done;
594 		ASSERT(ire != NULL);
595 		rts_setmetrics(ire, rtm->rtm_inits, &rtm->rtm_rmx);
596 		break;
597 	case RTM_DELETE:
598 		/* if we are deleting a route, gateway is a must */
599 		if ((found_addrs & RTA_GATEWAY) == 0) {
600 			error = EINVAL;
601 			goto done;
602 		}
603 		/*
604 		 * The RTF_SETSRC modifier does not make sense
605 		 * when deleting a route.
606 		 */
607 		if (rtm->rtm_flags & RTF_SETSRC) {
608 			error = EINVAL;
609 			goto done;
610 		}
611 
612 		switch (af) {
613 		case AF_INET:
614 			error = ip_rt_delete(dst_addr, net_mask, gw_addr,
615 			    found_addrs, rtm->rtm_flags, ill, B_FALSE,
616 			    ipst, zoneid);
617 			break;
618 		case AF_INET6:
619 			error = ip_rt_delete_v6(&dst_addr_v6, &net_mask_v6,
620 			    &gw_addr_v6, found_addrs, rtm->rtm_flags, ill,
621 			    ipst, zoneid);
622 			break;
623 		}
624 		break;
625 	case RTM_GET:
626 	case RTM_CHANGE:
627 		/*
628 		 * In the case of RTM_GET, the forwarding table should be
629 		 * searched recursively.  Also, if a gateway was
630 		 * specified then the gateway address must also be matched.
631 		 *
632 		 * In the case of RTM_CHANGE, the gateway address (if supplied)
633 		 * is the new gateway address so matching on the gateway address
634 		 * is not done.  This can lead to ambiguity when looking up the
635 		 * route to change as usually only the destination (and netmask,
636 		 * if supplied) is used for the lookup.  However if a RTA_IFP
637 		 * sockaddr is also supplied, it can disambiguate which route to
638 		 * change provided the ambigous routes are tied to distinct
639 		 * ill's (or interface indices).  If the routes are not tied to
640 		 * any particular interfaces (for example, with traditional
641 		 * gateway routes), then a RTA_IFP sockaddr will be of no use as
642 		 * it won't match any such routes.
643 		 * RTA_SRC is not supported for RTM_GET and RTM_CHANGE,
644 		 * except when RTM_CHANGE is combined to RTF_SETSRC.
645 		 */
646 		if (((found_addrs & RTA_SRC) != 0) &&
647 		    ((rtm->rtm_type == RTM_GET) ||
648 		    !(rtm->rtm_flags & RTF_SETSRC))) {
649 			error = EOPNOTSUPP;
650 			goto done;
651 		}
652 
653 		if (rtm->rtm_type == RTM_GET) {
654 			match_flags |= MATCH_IRE_SECATTR;
655 			match_flags_local |= MATCH_IRE_SECATTR;
656 			if ((found_addrs & RTA_GATEWAY) != 0)
657 				match_flags |= MATCH_IRE_GW;
658 			if (ioc_cr)
659 				tsl = crgetlabel(ioc_cr);
660 			if (rtsap != NULL) {
661 				if (rtsa_validate(rtsap) != 0) {
662 					error = EINVAL;
663 					goto done;
664 				}
665 				if (tsl != NULL &&
666 				    crgetzoneid(ioc_cr) != GLOBAL_ZONEID &&
667 				    (tsl->tsl_doi != rtsap->rtsa_doi ||
668 				    !bldominates(&tsl->tsl_label,
669 				    &rtsap->rtsa_slrange.lower_bound))) {
670 					error = EPERM;
671 					goto done;
672 				}
673 				tsl = labelalloc(
674 				    &rtsap->rtsa_slrange.lower_bound,
675 				    rtsap->rtsa_doi, KM_NOSLEEP);
676 			}
677 		}
678 		if (rtm->rtm_type == RTM_CHANGE) {
679 			if ((found_addrs & RTA_GATEWAY) &&
680 			    (rtm->rtm_flags & RTF_SETSRC)) {
681 				/*
682 				 * Do not want to change the gateway,
683 				 * but rather the source address.
684 				 */
685 				match_flags |= MATCH_IRE_GW;
686 			}
687 		}
688 
689 		/*
690 		 * If the netmask is all ones (either as supplied or as derived
691 		 * above), then first check for an IRE_LOOPBACK or
692 		 * IRE_LOCAL entry.
693 		 *
694 		 * If we didn't check for or find an IRE_LOOPBACK or IRE_LOCAL
695 		 * entry, then look for any other type of IRE.
696 		 */
697 		switch (af) {
698 		case AF_INET:
699 			if (net_mask == IP_HOST_MASK) {
700 				ire = ire_ftable_lookup_v4(dst_addr, 0, gw_addr,
701 				    IRE_LOCAL | IRE_LOOPBACK, NULL, zoneid,
702 				    tsl, match_flags_local, 0, ipst, NULL);
703 			}
704 			if (ire == NULL) {
705 				ire = ire_lookup_v4(dst_addr, net_mask,
706 				    gw_addr, ill, zoneid, tsl, match_flags,
707 				    ipst, &ifire, &v4setsrc, &gwattr);
708 				IN6_IPADDR_TO_V4MAPPED(v4setsrc, &v6setsrc);
709 			}
710 			break;
711 		case AF_INET6:
712 			if (IN6_ARE_ADDR_EQUAL(&net_mask_v6, &ipv6_all_ones)) {
713 				ire = ire_ftable_lookup_v6(&dst_addr_v6, NULL,
714 				    &gw_addr_v6, IRE_LOCAL | IRE_LOOPBACK, NULL,
715 				    zoneid, tsl, match_flags_local, 0, ipst,
716 				    NULL);
717 			}
718 			if (ire == NULL) {
719 				ire = ire_lookup_v6(&dst_addr_v6,
720 				    &net_mask_v6, &gw_addr_v6, ill, zoneid,
721 				    tsl, match_flags, ipst, &ifire, &v6setsrc,
722 				    &gwattr);
723 			}
724 			break;
725 		}
726 		if (tsl != NULL && tsl != crgetlabel(ioc_cr))
727 			label_rele(tsl);
728 
729 		if (ire == NULL) {
730 			error = ESRCH;
731 			goto done;
732 		}
733 		/*
734 		 * Want to return failure if we get an IRE_NOROUTE from
735 		 * ire_route_recursive
736 		 */
737 		if (ire->ire_type & IRE_NOROUTE) {
738 			ire_refrele(ire);
739 			ire = NULL;
740 			error = ESRCH;
741 			goto done;
742 		}
743 
744 		/* we know the IRE before we come here */
745 		switch (rtm->rtm_type) {
746 		case RTM_GET:
747 			mp1 = rts_rtmget(mp, ire, ifire, &v6setsrc, gwattr, af);
748 			if (mp1 == NULL) {
749 				error = ENOBUFS;
750 				goto done;
751 			}
752 			freemsg(mp);
753 			mp = mp1;
754 			rtm = (rt_msghdr_t *)mp->b_rptr;
755 			break;
756 		case RTM_CHANGE:
757 			/*
758 			 * Do not allow to the multirouting state of a route
759 			 * to be changed. This aims to prevent undesirable
760 			 * stages where both multirt and non-multirt routes
761 			 * for the same destination are declared.
762 			 */
763 			if ((ire->ire_flags & RTF_MULTIRT) !=
764 			    (rtm->rtm_flags & RTF_MULTIRT)) {
765 				error = EINVAL;
766 				goto done;
767 			}
768 			/*
769 			 * Note that we do not need to do
770 			 * ire_flush_cache_*(IRE_FLUSH_ADD) as a change
771 			 * in metrics or gateway will not affect existing
772 			 * routes since it does not create a more specific
773 			 * route.
774 			 */
775 			switch (af) {
776 			case AF_INET:
777 				if ((found_addrs & RTA_GATEWAY) != 0 &&
778 				    (ire->ire_gateway_addr != gw_addr)) {
779 					ire->ire_gateway_addr = gw_addr;
780 				}
781 
782 				if (rtsap != NULL) {
783 					ga.ga_af = AF_INET;
784 					IN6_IPADDR_TO_V4MAPPED(
785 					    ire->ire_gateway_addr, &ga.ga_addr);
786 
787 					gcgrp = gcgrp_lookup(&ga, B_TRUE);
788 					if (gcgrp == NULL) {
789 						error = ENOMEM;
790 						goto done;
791 					}
792 				}
793 
794 				if ((found_addrs & RTA_SRC) != 0 &&
795 				    (rtm->rtm_flags & RTF_SETSRC) != 0 &&
796 				    (ire->ire_setsrc_addr != src_addr)) {
797 					if (src_addr != INADDR_ANY) {
798 						uint_t type;
799 
800 						/*
801 						 * The RTF_SETSRC flag is
802 						 * present, check that the
803 						 * supplied src address is not
804 						 * the loopback address. This
805 						 * would produce martian
806 						 * packets.
807 						 */
808 						if (src_addr ==
809 						    htonl(INADDR_LOOPBACK)) {
810 							error = EINVAL;
811 							goto done;
812 						}
813 						/*
814 						 * Also check that the
815 						 * supplied addr is a valid
816 						 * local address.
817 						 */
818 						type = ip_type_v4(src_addr,
819 						    ipst);
820 						if (!(type &
821 						    (IRE_LOCAL|IRE_LOOPBACK))) {
822 							error = EADDRNOTAVAIL;
823 							goto done;
824 						}
825 						ire->ire_flags |= RTF_SETSRC;
826 						ire->ire_setsrc_addr =
827 						    src_addr;
828 					} else {
829 						ire->ire_flags &= ~RTF_SETSRC;
830 						ire->ire_setsrc_addr =
831 						    INADDR_ANY;
832 					}
833 					/*
834 					 * Let conn_ixa caching know that
835 					 * source address selection changed
836 					 */
837 					ip_update_source_selection(ipst);
838 				}
839 				ire_flush_cache_v4(ire, IRE_FLUSH_GWCHANGE);
840 				break;
841 			case AF_INET6:
842 				mutex_enter(&ire->ire_lock);
843 				if ((found_addrs & RTA_GATEWAY) != 0 &&
844 				    !IN6_ARE_ADDR_EQUAL(
845 				    &ire->ire_gateway_addr_v6, &gw_addr_v6)) {
846 					ire->ire_gateway_addr_v6 = gw_addr_v6;
847 				}
848 				mutex_exit(&ire->ire_lock);
849 
850 				if (rtsap != NULL) {
851 					ga.ga_af = AF_INET6;
852 					mutex_enter(&ire->ire_lock);
853 					ga.ga_addr = ire->ire_gateway_addr_v6;
854 					mutex_exit(&ire->ire_lock);
855 
856 					gcgrp = gcgrp_lookup(&ga, B_TRUE);
857 					if (gcgrp == NULL) {
858 						error = ENOMEM;
859 						goto done;
860 					}
861 				}
862 
863 				if ((found_addrs & RTA_SRC) != 0 &&
864 				    (rtm->rtm_flags & RTF_SETSRC) != 0 &&
865 				    !IN6_ARE_ADDR_EQUAL(
866 				    &ire->ire_setsrc_addr_v6, &src_addr_v6)) {
867 					if (!IN6_IS_ADDR_UNSPECIFIED(
868 					    &src_addr_v6)) {
869 						uint_t type;
870 
871 						/*
872 						 * The RTF_SETSRC flag is
873 						 * present, check that the
874 						 * supplied src address is not
875 						 * the loopback address. This
876 						 * would produce martian
877 						 * packets.
878 						 */
879 						if (IN6_IS_ADDR_LOOPBACK(
880 						    &src_addr_v6)) {
881 							error = EINVAL;
882 							goto done;
883 						}
884 						/*
885 						 * Also check that the
886 						 * supplied addr is a valid
887 						 * local address.
888 						 */
889 						type = ip_type_v6(&src_addr_v6,
890 						    ipst);
891 						if (!(type &
892 						    (IRE_LOCAL|IRE_LOOPBACK))) {
893 							error = EADDRNOTAVAIL;
894 							goto done;
895 						}
896 						mutex_enter(&ire->ire_lock);
897 						ire->ire_flags |= RTF_SETSRC;
898 						ire->ire_setsrc_addr_v6 =
899 						    src_addr_v6;
900 						mutex_exit(&ire->ire_lock);
901 					} else {
902 						mutex_enter(&ire->ire_lock);
903 						ire->ire_flags &= ~RTF_SETSRC;
904 						ire->ire_setsrc_addr_v6 =
905 						    ipv6_all_zeros;
906 						mutex_exit(&ire->ire_lock);
907 					}
908 					/*
909 					 * Let conn_ixa caching know that
910 					 * source address selection changed
911 					 */
912 					ip_update_source_selection(ipst);
913 				}
914 				ire_flush_cache_v6(ire, IRE_FLUSH_GWCHANGE);
915 				break;
916 			}
917 
918 			if (rtsap != NULL) {
919 				ASSERT(gcgrp != NULL);
920 
921 				/*
922 				 * Create and add the security attribute to
923 				 * prefix IRE; it will add a reference to the
924 				 * group upon allocating a new entry.  If it
925 				 * finds an already-existing entry for the
926 				 * security attribute, it simply returns it
927 				 * and no new group reference is made.
928 				 */
929 				gc = gc_create(rtsap, gcgrp, &gcgrp_xtraref);
930 				if (gc == NULL ||
931 				    (error = tsol_ire_init_gwattr(ire,
932 				    ire->ire_ipversion, gc)) != 0) {
933 					if (gc != NULL) {
934 						GC_REFRELE(gc);
935 					} else {
936 						/* gc_create failed */
937 						error = ENOMEM;
938 					}
939 					goto done;
940 				}
941 			}
942 			rts_setmetrics(ire, rtm->rtm_inits, &rtm->rtm_rmx);
943 			break;
944 		}
945 		break;
946 	default:
947 		error = EOPNOTSUPP;
948 		break;
949 	}
950 done:
951 	if (ire != NULL)
952 		ire_refrele(ire);
953 	if (ifire != NULL)
954 		ire_refrele(ifire);
955 	if (ill != NULL)
956 		ill_refrele(ill);
957 
958 	if (gcgrp_xtraref)
959 		GCGRP_REFRELE(gcgrp);
960 
961 	if (rtm != NULL) {
962 		ASSERT(mp->b_wptr <= mp->b_datap->db_lim);
963 		if (error != 0) {
964 			rtm->rtm_errno = error;
965 			/* Send error ACK */
966 			ip1dbg(("ip_rts_request: error %d\n", error));
967 		} else {
968 			rtm->rtm_flags |= RTF_DONE;
969 			/* OK ACK already set up by caller except this */
970 			ip2dbg(("ip_rts_request: OK ACK\n"));
971 		}
972 		rts_queue_input(mp, connp, af, RTSQ_ALL, ipst);
973 	}
974 	return (error);
975 }
976 
977 /*
978  * Helper function that can do recursive lookups including when
979  * MATCH_IRE_GW and/or MATCH_IRE_MASK is set.
980  */
981 static ire_t *
982 ire_lookup_v4(ipaddr_t dst_addr, ipaddr_t net_mask, ipaddr_t gw_addr,
983     const ill_t *ill, zoneid_t zoneid, const ts_label_t *tsl,
984     int match_flags, ip_stack_t *ipst, ire_t **pifire, ipaddr_t *v4setsrcp,
985     tsol_ire_gw_secattr_t **gwattrp)
986 {
987 	ire_t		*ire;
988 	ire_t		*ifire = NULL;
989 	uint_t		ire_type;
990 
991 	*pifire = NULL;
992 	*v4setsrcp = INADDR_ANY;
993 	*gwattrp = NULL;
994 
995 	/* Skip IRE_IF_CLONE */
996 	match_flags |= MATCH_IRE_TYPE;
997 	ire_type = (IRE_ONLINK|IRE_OFFLINK) & ~IRE_IF_CLONE;
998 
999 	/*
1000 	 * ire_route_recursive can't match gateway or mask thus if they are
1001 	 * set we have to do two steps of lookups
1002 	 */
1003 	if (match_flags & (MATCH_IRE_GW|MATCH_IRE_MASK)) {
1004 		ire = ire_ftable_lookup_v4(dst_addr, net_mask, gw_addr,
1005 		    ire_type, ill, zoneid, tsl, match_flags, 0, ipst, NULL);
1006 
1007 		if (ire == NULL ||(ire->ire_flags & (RTF_REJECT|RTF_BLACKHOLE)))
1008 			return (ire);
1009 
1010 		if (ire->ire_type & IRE_ONLINK)
1011 			return (ire);
1012 
1013 		if (ire->ire_flags & RTF_SETSRC) {
1014 			ASSERT(ire->ire_setsrc_addr != INADDR_ANY);
1015 			*v4setsrcp = ire->ire_setsrc_addr;
1016 			v4setsrcp = NULL;
1017 		}
1018 
1019 		/* The first ire_gw_secattr is passed back */
1020 		if (ire->ire_gw_secattr != NULL) {
1021 			*gwattrp = ire->ire_gw_secattr;
1022 			gwattrp = NULL;
1023 		}
1024 
1025 		/* Look for an interface ire recursively based on the gateway */
1026 		dst_addr = ire->ire_gateway_addr;
1027 		match_flags &= ~(MATCH_IRE_GW|MATCH_IRE_MASK);
1028 		ifire = ire_route_recursive_v4(dst_addr, ire_type, ill, zoneid,
1029 		    tsl, match_flags, B_FALSE, 0, ipst, v4setsrcp, gwattrp,
1030 		    NULL);
1031 	} else {
1032 		ire = ire_route_recursive_v4(dst_addr, ire_type, ill, zoneid,
1033 		    tsl, match_flags, B_FALSE, 0, ipst, v4setsrcp, gwattrp,
1034 		    NULL);
1035 	}
1036 	*pifire = ifire;
1037 	return (ire);
1038 }
1039 
1040 static ire_t *
1041 ire_lookup_v6(const in6_addr_t *dst_addr_v6,
1042     const in6_addr_t *net_mask_v6, const in6_addr_t *gw_addr_v6,
1043     const ill_t *ill, zoneid_t zoneid, const ts_label_t *tsl, int match_flags,
1044     ip_stack_t *ipst, ire_t **pifire,
1045     in6_addr_t *v6setsrcp, tsol_ire_gw_secattr_t **gwattrp)
1046 {
1047 	ire_t		*ire;
1048 	ire_t		*ifire = NULL;
1049 	uint_t		ire_type;
1050 
1051 	*pifire = NULL;
1052 	*v6setsrcp = ipv6_all_zeros;
1053 	*gwattrp = NULL;
1054 
1055 	/* Skip IRE_IF_CLONE */
1056 	match_flags |= MATCH_IRE_TYPE;
1057 	ire_type = (IRE_ONLINK|IRE_OFFLINK) & ~IRE_IF_CLONE;
1058 
1059 	/*
1060 	 * ire_route_recursive can't match gateway or mask thus if they are
1061 	 * set we have to do two steps of lookups
1062 	 */
1063 	if (match_flags & (MATCH_IRE_GW|MATCH_IRE_MASK)) {
1064 		in6_addr_t dst;
1065 
1066 		ire = ire_ftable_lookup_v6(dst_addr_v6, net_mask_v6,
1067 		    gw_addr_v6, ire_type, ill, zoneid, tsl, match_flags, 0,
1068 		    ipst, NULL);
1069 
1070 		if (ire == NULL ||(ire->ire_flags & (RTF_REJECT|RTF_BLACKHOLE)))
1071 			return (ire);
1072 
1073 		if (ire->ire_type & IRE_ONLINK)
1074 			return (ire);
1075 
1076 		if (ire->ire_flags & RTF_SETSRC) {
1077 			ASSERT(!IN6_IS_ADDR_UNSPECIFIED(
1078 			    &ire->ire_setsrc_addr_v6));
1079 			*v6setsrcp = ire->ire_setsrc_addr_v6;
1080 			v6setsrcp = NULL;
1081 		}
1082 
1083 		/* The first ire_gw_secattr is passed back */
1084 		if (ire->ire_gw_secattr != NULL) {
1085 			*gwattrp = ire->ire_gw_secattr;
1086 			gwattrp = NULL;
1087 		}
1088 
1089 		mutex_enter(&ire->ire_lock);
1090 		dst = ire->ire_gateway_addr_v6;
1091 		mutex_exit(&ire->ire_lock);
1092 		match_flags &= ~(MATCH_IRE_GW|MATCH_IRE_MASK);
1093 		ifire = ire_route_recursive_v6(&dst, ire_type, ill, zoneid, tsl,
1094 		    match_flags, B_FALSE, 0, ipst, v6setsrcp, gwattrp, NULL);
1095 	} else {
1096 		ire = ire_route_recursive_v6(dst_addr_v6, ire_type, ill, zoneid,
1097 		    tsl, match_flags, B_FALSE, 0, ipst, v6setsrcp, gwattrp,
1098 		    NULL);
1099 	}
1100 	*pifire = ifire;
1101 	return (ire);
1102 }
1103 
1104 
1105 /*
1106  * Handle IP_IOC_RTS_REQUEST ioctls
1107  */
1108 int
1109 ip_rts_request(queue_t *q, mblk_t *mp, cred_t *ioc_cr)
1110 {
1111 	conn_t	*connp = Q_TO_CONN(q);
1112 	IOCP	iocp = (IOCP)mp->b_rptr;
1113 	mblk_t	*mp1, *ioc_mp = mp;
1114 	int	error = 0;
1115 	ip_stack_t	*ipst;
1116 
1117 	ipst = connp->conn_netstack->netstack_ip;
1118 
1119 	ASSERT(mp->b_cont != NULL);
1120 	/* ioc_mp holds mp */
1121 	mp = mp->b_cont;
1122 
1123 	/*
1124 	 * The Routing Socket data starts on
1125 	 * next block. If there is no next block
1126 	 * this is an indication from routing module
1127 	 * that it is a routing socket stream queue.
1128 	 * We need to support that for compatibility with SDP since
1129 	 * it has a contract private interface to use IP_IOC_RTS_REQUEST.
1130 	 * Note: SDP no longer uses IP_IOC_RTS_REQUEST - we can remove this.
1131 	 */
1132 	if (mp->b_cont == NULL) {
1133 		/*
1134 		 * This is a message from SDP
1135 		 * indicating that this is a Routing Socket
1136 		 * Stream. Insert this conn_t in routing
1137 		 * socket client list.
1138 		 */
1139 		connp->conn_useloopback = 1;
1140 		ipcl_hash_insert_wildcard(ipst->ips_rts_clients, connp);
1141 		goto done;
1142 	}
1143 	mp1 = dupmsg(mp->b_cont);
1144 	if (mp1 == NULL) {
1145 		error  = ENOBUFS;
1146 		goto done;
1147 	}
1148 	mp = mp1;
1149 
1150 	error = ip_rts_request_common(mp, connp, ioc_cr);
1151 done:
1152 	iocp->ioc_error = error;
1153 	ioc_mp->b_datap->db_type = M_IOCACK;
1154 	if (iocp->ioc_error != 0)
1155 		iocp->ioc_count = 0;
1156 	/* Note that we pass a NULL ira to rts_input */
1157 	(connp->conn_recv)(connp, ioc_mp, NULL, NULL);
1158 
1159 	/* conn was refheld in ip_wput_ioctl. */
1160 	CONN_OPER_PENDING_DONE(connp);
1161 
1162 	return (error);
1163 }
1164 
1165 /*
1166  * Build a reply to the RTM_GET request contained in the given message block
1167  * using the retrieved IRE of the destination address, the parent IRE (if it
1168  * exists) and the address family.
1169  *
1170  * Returns a pointer to a message block containing the reply if successful,
1171  * otherwise NULL is returned.
1172  */
1173 static mblk_t *
1174 rts_rtmget(mblk_t *mp, ire_t *ire, ire_t *ifire, const in6_addr_t *setsrc,
1175     tsol_ire_gw_secattr_t *attrp, sa_family_t af)
1176 {
1177 	rt_msghdr_t	*rtm;
1178 	rt_msghdr_t	*new_rtm;
1179 	mblk_t		*new_mp;
1180 	int		rtm_addrs;
1181 	int		rtm_flags;
1182 	tsol_gc_t	*gc = NULL;
1183 	tsol_gcgrp_t	*gcgrp = NULL;
1184 	ill_t		*ill;
1185 	ipif_t		*ipif = NULL;
1186 	ipaddr_t	brdaddr;	/* IFF_POINTOPOINT destination */
1187 	ipaddr_t	ifaddr;
1188 	in6_addr_t	brdaddr6;	/* IFF_POINTOPOINT destination */
1189 	in6_addr_t	ifaddr6;
1190 	ipaddr_t	v4setsrc;
1191 
1192 	rtm = (rt_msghdr_t *)mp->b_rptr;
1193 
1194 	/*
1195 	 * Find the ill used to send packets. This will be NULL in case
1196 	 * of a reject or blackhole.
1197 	 */
1198 	if (ifire != NULL)
1199 		ill = ire_nexthop_ill(ifire);
1200 	else
1201 		ill = ire_nexthop_ill(ire);
1202 
1203 	if (attrp != NULL) {
1204 		mutex_enter(&attrp->igsa_lock);
1205 		if ((gc = attrp->igsa_gc) != NULL) {
1206 			gcgrp = gc->gc_grp;
1207 			ASSERT(gcgrp != NULL);
1208 			rw_enter(&gcgrp->gcgrp_rwlock, RW_READER);
1209 		}
1210 		mutex_exit(&attrp->igsa_lock);
1211 	}
1212 
1213 	/*
1214 	 * Always return RTA_DST, RTA_GATEWAY and RTA_NETMASK.
1215 	 *
1216 	 * The 4.4BSD-Lite2 code (net/rtsock.c) returns both
1217 	 * RTA_IFP and RTA_IFA if either is defined, and also
1218 	 * returns RTA_BRD if the appropriate interface is
1219 	 * point-to-point.
1220 	 */
1221 	rtm_addrs = (RTA_DST | RTA_GATEWAY | RTA_NETMASK);
1222 	if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) && ill != NULL) {
1223 		rtm_addrs |= (RTA_IFP | RTA_IFA);
1224 		/*
1225 		 * We associate an IRE with an ILL, hence we don't exactly
1226 		 * know what might make sense for RTA_IFA and RTA_BRD. We
1227 		 * pick the first ipif on the ill.
1228 		 */
1229 		ipif = ipif_get_next_ipif(NULL, ill);
1230 		if (ipif != NULL) {
1231 			if (ipif->ipif_isv6)
1232 				ifaddr6 = ipif->ipif_v6lcl_addr;
1233 			else
1234 				ifaddr = ipif->ipif_lcl_addr;
1235 			if (ipif->ipif_flags & IPIF_POINTOPOINT) {
1236 				rtm_addrs |= RTA_BRD;
1237 				if (ipif->ipif_isv6)
1238 					brdaddr6 = ipif->ipif_v6pp_dst_addr;
1239 				else
1240 					brdaddr = ipif->ipif_pp_dst_addr;
1241 			}
1242 			ipif_refrele(ipif);
1243 		}
1244 	}
1245 
1246 	new_mp = rts_alloc_msg(RTM_GET, rtm_addrs, af, gc != NULL ? 1 : 0);
1247 	if (new_mp == NULL) {
1248 		if (gcgrp != NULL)
1249 			rw_exit(&gcgrp->gcgrp_rwlock);
1250 		if (ill != NULL)
1251 			ill_refrele(ill);
1252 		return (NULL);
1253 	}
1254 
1255 	/*
1256 	 * We set the destination address, gateway address,
1257 	 * netmask and flags in the RTM_GET response depending
1258 	 * on whether we found a parent IRE or not.
1259 	 * In particular, if we did find a parent IRE during the
1260 	 * recursive search, use that IRE's gateway address.
1261 	 * Otherwise, we use the IRE's source address for the
1262 	 * gateway address.
1263 	 */
1264 	ASSERT(af == AF_INET || af == AF_INET6);
1265 	switch (af) {
1266 	case AF_INET:
1267 		IN6_V4MAPPED_TO_IPADDR(setsrc, v4setsrc);
1268 		if (v4setsrc != INADDR_ANY)
1269 			rtm_addrs |= RTA_SRC;
1270 
1271 		rtm_flags = ire->ire_flags;
1272 		rts_fill_msg(RTM_GET, rtm_addrs, ire->ire_addr,
1273 		    ire->ire_mask, ire->ire_gateway_addr, v4setsrc,
1274 		    brdaddr, 0, ifaddr, ill, new_mp, gc);
1275 		break;
1276 	case AF_INET6:
1277 		if (!IN6_IS_ADDR_UNSPECIFIED(setsrc))
1278 			rtm_addrs |= RTA_SRC;
1279 
1280 		rtm_flags = ire->ire_flags;
1281 		rts_fill_msg_v6(RTM_GET, rtm_addrs, &ire->ire_addr_v6,
1282 		    &ire->ire_mask_v6, &ire->ire_gateway_addr_v6,
1283 		    setsrc, &brdaddr6, &ipv6_all_zeros,
1284 		    &ifaddr6, ill, new_mp, gc);
1285 		break;
1286 	}
1287 
1288 	if (gcgrp != NULL)
1289 		rw_exit(&gcgrp->gcgrp_rwlock);
1290 
1291 	new_rtm = (rt_msghdr_t *)new_mp->b_rptr;
1292 
1293 	/*
1294 	 * The rtm_msglen, rtm_version and rtm_type fields in
1295 	 * RTM_GET response are filled in by rts_fill_msg.
1296 	 *
1297 	 * rtm_addrs and rtm_flags are filled in based on what
1298 	 * was requested and the state of the IREs looked up
1299 	 * above.
1300 	 *
1301 	 * rtm_inits and rtm_rmx are filled in with metrics
1302 	 * based on whether a parent IRE was found or not.
1303 	 *
1304 	 * TODO: rtm_index and rtm_use should probably be
1305 	 * filled in with something resonable here and not just
1306 	 * copied from the request.
1307 	 */
1308 	new_rtm->rtm_index = rtm->rtm_index;
1309 	new_rtm->rtm_pid = rtm->rtm_pid;
1310 	new_rtm->rtm_seq = rtm->rtm_seq;
1311 	new_rtm->rtm_use = rtm->rtm_use;
1312 	new_rtm->rtm_addrs = rtm_addrs;
1313 	new_rtm->rtm_flags = rtm_flags;
1314 	new_rtm->rtm_inits = rts_getmetrics(ire, &new_rtm->rtm_rmx);
1315 	if (ill != NULL)
1316 		ill_refrele(ill);
1317 	return (new_mp);
1318 }
1319 
1320 /*
1321  * Fill the given if_data_t with interface statistics.
1322  */
1323 static void
1324 rts_getifdata(if_data_t *if_data, const ipif_t *ipif)
1325 {
1326 	if_data->ifi_type = ipif->ipif_ill->ill_type;
1327 						/* ethernet, tokenring, etc */
1328 	if_data->ifi_addrlen = 0;		/* media address length */
1329 	if_data->ifi_hdrlen = 0;		/* media header length */
1330 	if_data->ifi_mtu = ipif->ipif_ill->ill_mtu;	/* mtu */
1331 	if_data->ifi_metric = ipif->ipif_metric; /* metric (external only) */
1332 	if_data->ifi_baudrate = 0;		/* linespeed */
1333 
1334 	if_data->ifi_ipackets = 0;		/* packets received on if */
1335 	if_data->ifi_ierrors = 0;		/* input errors on interface */
1336 	if_data->ifi_opackets = 0;		/* packets sent on interface */
1337 	if_data->ifi_oerrors = 0;		/* output errors on if */
1338 	if_data->ifi_collisions = 0;		/* collisions on csma if */
1339 	if_data->ifi_ibytes = 0;		/* total number received */
1340 	if_data->ifi_obytes = 0;		/* total number sent */
1341 	if_data->ifi_imcasts = 0;		/* multicast packets received */
1342 	if_data->ifi_omcasts = 0;		/* multicast packets sent */
1343 	if_data->ifi_iqdrops = 0;		/* dropped on input */
1344 	if_data->ifi_noproto = 0;		/* destined for unsupported */
1345 						/* protocol. */
1346 }
1347 
1348 /*
1349  * Set the metrics on a forwarding table route.
1350  */
1351 static void
1352 rts_setmetrics(ire_t *ire, uint_t which, rt_metrics_t *metrics)
1353 {
1354 	clock_t		rtt;
1355 	clock_t		rtt_sd;
1356 	ill_t		*ill;
1357 	ifrt_t		*ifrt;
1358 	mblk_t		*mp;
1359 	in6_addr_t	gw_addr_v6;
1360 
1361 	/* Need to add back some metrics to the IRE? */
1362 	/*
1363 	 * Bypass obtaining the lock and searching ill_saved_ire_mp in the
1364 	 * common case of no metrics.
1365 	 */
1366 	if (which == 0)
1367 		return;
1368 	ire->ire_metrics.iulp_set = B_TRUE;
1369 
1370 	/*
1371 	 * iulp_rtt and iulp_rtt_sd are in milliseconds, but 4.4BSD-Lite2's
1372 	 * <net/route.h> says: rmx_rtt and rmx_rttvar are stored as
1373 	 * microseconds.
1374 	 */
1375 	if (which & RTV_RTT)
1376 		rtt = metrics->rmx_rtt / 1000;
1377 	if (which & RTV_RTTVAR)
1378 		rtt_sd = metrics->rmx_rttvar / 1000;
1379 
1380 	/*
1381 	 * Update the metrics in the IRE itself.
1382 	 */
1383 	mutex_enter(&ire->ire_lock);
1384 	if (which & RTV_MTU)
1385 		ire->ire_metrics.iulp_mtu = metrics->rmx_mtu;
1386 	if (which & RTV_RTT)
1387 		ire->ire_metrics.iulp_rtt = rtt;
1388 	if (which & RTV_SSTHRESH)
1389 		ire->ire_metrics.iulp_ssthresh = metrics->rmx_ssthresh;
1390 	if (which & RTV_RTTVAR)
1391 		ire->ire_metrics.iulp_rtt_sd = rtt_sd;
1392 	if (which & RTV_SPIPE)
1393 		ire->ire_metrics.iulp_spipe = metrics->rmx_sendpipe;
1394 	if (which & RTV_RPIPE)
1395 		ire->ire_metrics.iulp_rpipe = metrics->rmx_recvpipe;
1396 	mutex_exit(&ire->ire_lock);
1397 
1398 	/*
1399 	 * Search through the ifrt_t chain hanging off the ILL in order to
1400 	 * reflect the metric change there.
1401 	 */
1402 	ill = ire->ire_ill;
1403 	if (ill == NULL)
1404 		return;
1405 	ASSERT((ill->ill_isv6 && ire->ire_ipversion == IPV6_VERSION) ||
1406 	    ((!ill->ill_isv6 && ire->ire_ipversion == IPV4_VERSION)));
1407 	if (ill->ill_isv6) {
1408 		mutex_enter(&ire->ire_lock);
1409 		gw_addr_v6 = ire->ire_gateway_addr_v6;
1410 		mutex_exit(&ire->ire_lock);
1411 	}
1412 	mutex_enter(&ill->ill_saved_ire_lock);
1413 	for (mp = ill->ill_saved_ire_mp; mp != NULL; mp = mp->b_cont) {
1414 		/*
1415 		 * On a given ill, the tuple of address, gateway, mask,
1416 		 * ire_type and zoneid unique for each saved IRE.
1417 		 */
1418 		ifrt = (ifrt_t *)mp->b_rptr;
1419 		if (ill->ill_isv6) {
1420 			if (!IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6addr,
1421 			    &ire->ire_addr_v6) ||
1422 			    !IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6gateway_addr,
1423 			    &gw_addr_v6) ||
1424 			    !IN6_ARE_ADDR_EQUAL(&ifrt->ifrt_v6mask,
1425 			    &ire->ire_mask_v6))
1426 				continue;
1427 		} else {
1428 			if (ifrt->ifrt_addr != ire->ire_addr ||
1429 			    ifrt->ifrt_gateway_addr != ire->ire_gateway_addr ||
1430 			    ifrt->ifrt_mask != ire->ire_mask)
1431 				continue;
1432 		}
1433 		if (ifrt->ifrt_zoneid != ire->ire_zoneid ||
1434 		    ifrt->ifrt_type != ire->ire_type)
1435 			continue;
1436 
1437 		if (which & RTV_MTU)
1438 			ifrt->ifrt_metrics.iulp_mtu = metrics->rmx_mtu;
1439 		if (which & RTV_RTT)
1440 			ifrt->ifrt_metrics.iulp_rtt = rtt;
1441 		if (which & RTV_SSTHRESH) {
1442 			ifrt->ifrt_metrics.iulp_ssthresh =
1443 			    metrics->rmx_ssthresh;
1444 		}
1445 		if (which & RTV_RTTVAR)
1446 			ifrt->ifrt_metrics.iulp_rtt_sd = metrics->rmx_rttvar;
1447 		if (which & RTV_SPIPE)
1448 			ifrt->ifrt_metrics.iulp_spipe = metrics->rmx_sendpipe;
1449 		if (which & RTV_RPIPE)
1450 			ifrt->ifrt_metrics.iulp_rpipe = metrics->rmx_recvpipe;
1451 		break;
1452 	}
1453 	mutex_exit(&ill->ill_saved_ire_lock);
1454 
1455 	/*
1456 	 * Update any IRE_IF_CLONE hanging created from this IRE_IF so they
1457 	 * get any new iulp_mtu.
1458 	 * We do that by deleting them; ire_create_if_clone will pick
1459 	 * up the new metrics.
1460 	 */
1461 	if ((ire->ire_type & IRE_INTERFACE) && ire->ire_dep_children != 0)
1462 		ire_dep_delete_if_clone(ire);
1463 }
1464 
1465 /*
1466  * Get the metrics from a forwarding table route.
1467  */
1468 static int
1469 rts_getmetrics(ire_t *ire, rt_metrics_t *metrics)
1470 {
1471 	int	metrics_set = 0;
1472 
1473 	bzero(metrics, sizeof (rt_metrics_t));
1474 
1475 	/*
1476 	 * iulp_rtt and iulp_rtt_sd are in milliseconds, but 4.4BSD-Lite2's
1477 	 * <net/route.h> says: rmx_rtt and rmx_rttvar are stored as
1478 	 * microseconds.
1479 	 */
1480 	metrics->rmx_rtt = ire->ire_metrics.iulp_rtt * 1000;
1481 	metrics_set |= RTV_RTT;
1482 	metrics->rmx_mtu = ire->ire_metrics.iulp_mtu;
1483 	metrics_set |= RTV_MTU;
1484 	metrics->rmx_ssthresh = ire->ire_metrics.iulp_ssthresh;
1485 	metrics_set |= RTV_SSTHRESH;
1486 	metrics->rmx_rttvar = ire->ire_metrics.iulp_rtt_sd * 1000;
1487 	metrics_set |= RTV_RTTVAR;
1488 	metrics->rmx_sendpipe = ire->ire_metrics.iulp_spipe;
1489 	metrics_set |= RTV_SPIPE;
1490 	metrics->rmx_recvpipe = ire->ire_metrics.iulp_rpipe;
1491 	metrics_set |= RTV_RPIPE;
1492 	return (metrics_set);
1493 }
1494 
1495 /*
1496  * Given two sets of metrics (src and dst), use the dst values if they are
1497  * set. If a dst value is not set but the src value is set, then we use
1498  * the src value.
1499  * dst is updated with the new values.
1500  * This is used to merge information from a dce_t and ire_metrics, where the
1501  * dce values takes precedence.
1502  */
1503 void
1504 rts_merge_metrics(iulp_t *dst, const iulp_t *src)
1505 {
1506 	if (!src->iulp_set)
1507 		return;
1508 
1509 	if (dst->iulp_ssthresh == 0)
1510 		dst->iulp_ssthresh = src->iulp_ssthresh;
1511 	if (dst->iulp_rtt == 0)
1512 		dst->iulp_rtt = src->iulp_rtt;
1513 	if (dst->iulp_rtt_sd == 0)
1514 		dst->iulp_rtt_sd = src->iulp_rtt_sd;
1515 	if (dst->iulp_spipe == 0)
1516 		dst->iulp_spipe = src->iulp_spipe;
1517 	if (dst->iulp_rpipe == 0)
1518 		dst->iulp_rpipe = src->iulp_rpipe;
1519 	if (dst->iulp_rtomax == 0)
1520 		dst->iulp_rtomax = src->iulp_rtomax;
1521 	if (dst->iulp_sack == 0)
1522 		dst->iulp_sack = src->iulp_sack;
1523 	if (dst->iulp_tstamp_ok == 0)
1524 		dst->iulp_tstamp_ok = src->iulp_tstamp_ok;
1525 	if (dst->iulp_wscale_ok == 0)
1526 		dst->iulp_wscale_ok = src->iulp_wscale_ok;
1527 	if (dst->iulp_ecn_ok == 0)
1528 		dst->iulp_ecn_ok = src->iulp_ecn_ok;
1529 	if (dst->iulp_pmtud_ok == 0)
1530 		dst->iulp_pmtud_ok = src->iulp_pmtud_ok;
1531 	if (dst->iulp_mtu == 0)
1532 		dst->iulp_mtu = src->iulp_mtu;
1533 }
1534 
1535 
1536 /*
1537  * Takes a pointer to a routing message and extracts necessary info by looking
1538  * at the rtm->rtm_addrs bits and store the requested sockaddrs in the pointers
1539  * passed (all of which must be valid).
1540  *
1541  * The bitmask of sockaddrs actually found in the message is returned, or zero
1542  * is returned in the case of an error.
1543  */
1544 static int
1545 rts_getaddrs(rt_msghdr_t *rtm, in6_addr_t *dst_addrp, in6_addr_t *gw_addrp,
1546     in6_addr_t *net_maskp, in6_addr_t *authorp, in6_addr_t *if_addrp,
1547     in6_addr_t *in_src_addrp, ushort_t *indexp, sa_family_t *afp,
1548     tsol_rtsecattr_t *rtsecattr, int *error)
1549 {
1550 	struct sockaddr *sa;
1551 	int	i;
1552 	int	addr_bits;
1553 	int	length;
1554 	int	found_addrs = 0;
1555 	caddr_t	cp;
1556 	size_t	size;
1557 	struct sockaddr_dl *sdl;
1558 
1559 	*dst_addrp = ipv6_all_zeros;
1560 	*gw_addrp = ipv6_all_zeros;
1561 	*net_maskp = ipv6_all_zeros;
1562 	*authorp = ipv6_all_zeros;
1563 	*if_addrp = ipv6_all_zeros;
1564 	*in_src_addrp = ipv6_all_zeros;
1565 	*indexp = 0;
1566 	*afp = AF_UNSPEC;
1567 	rtsecattr->rtsa_cnt = 0;
1568 	*error = 0;
1569 
1570 	/*
1571 	 * At present we handle only RTA_DST, RTA_GATEWAY, RTA_NETMASK, RTA_IFP,
1572 	 * RTA_IFA and RTA_AUTHOR.  The rest will be added as we need them.
1573 	 */
1574 	cp = (caddr_t)&rtm[1];
1575 	length = rtm->rtm_msglen;
1576 	for (i = 0; (i < RTA_NUMBITS) && ((cp - (caddr_t)rtm) < length); i++) {
1577 		/*
1578 		 * The address family we are working with starts out as
1579 		 * AF_UNSPEC, but is set to the one specified with the
1580 		 * destination address.
1581 		 *
1582 		 * If the "working" address family that has been set to
1583 		 * something other than AF_UNSPEC, then the address family of
1584 		 * subsequent sockaddrs must either be AF_UNSPEC (for
1585 		 * compatibility with older programs) or must be the same as our
1586 		 * "working" one.
1587 		 *
1588 		 * This code assumes that RTA_DST (1) comes first in the loop.
1589 		 */
1590 		sa = (struct sockaddr *)cp;
1591 		addr_bits = (rtm->rtm_addrs & (1 << i));
1592 		if (addr_bits == 0)
1593 			continue;
1594 		switch (addr_bits) {
1595 		case RTA_DST:
1596 			size = rts_copyfromsockaddr(sa, dst_addrp);
1597 			*afp = sa->sa_family;
1598 			break;
1599 		case RTA_GATEWAY:
1600 			if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC)
1601 				return (0);
1602 			size = rts_copyfromsockaddr(sa, gw_addrp);
1603 			break;
1604 		case RTA_NETMASK:
1605 			if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC)
1606 				return (0);
1607 			size = rts_copyfromsockaddr(sa, net_maskp);
1608 			break;
1609 		case RTA_IFP:
1610 			if (sa->sa_family != AF_LINK &&
1611 			    sa->sa_family != AF_UNSPEC)
1612 				return (0);
1613 			sdl = (struct sockaddr_dl *)cp;
1614 			*indexp = sdl->sdl_index;
1615 			size = sizeof (struct sockaddr_dl);
1616 			break;
1617 		case RTA_SRC:
1618 			/* Source address of the incoming packet */
1619 			size = rts_copyfromsockaddr(sa, in_src_addrp);
1620 			*afp = sa->sa_family;
1621 			break;
1622 		case RTA_IFA:
1623 			if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC)
1624 				return (0);
1625 			size = rts_copyfromsockaddr(sa, if_addrp);
1626 			break;
1627 		case RTA_AUTHOR:
1628 			if (sa->sa_family != *afp && sa->sa_family != AF_UNSPEC)
1629 				return (0);
1630 			size = rts_copyfromsockaddr(sa, authorp);
1631 			break;
1632 		default:
1633 			return (0);
1634 		}
1635 		if (size == 0)
1636 			return (0);
1637 		cp += size;
1638 		found_addrs |= addr_bits;
1639 	}
1640 
1641 	/*
1642 	 * Parse the routing message and look for any security-
1643 	 * related attributes for the route.  For each valid
1644 	 * attribute, allocate/obtain the corresponding kernel
1645 	 * route security attributes.
1646 	 */
1647 	if (((cp - (caddr_t)rtm) < length) && is_system_labeled()) {
1648 		*error = tsol_rtsa_init(rtm, rtsecattr, cp);
1649 		ASSERT(rtsecattr->rtsa_cnt <= TSOL_RTSA_REQUEST_MAX);
1650 	}
1651 
1652 	return (found_addrs);
1653 }
1654 
1655 /*
1656  * Fills the message with the given info.
1657  */
1658 static void
1659 rts_fill_msg(int type, int rtm_addrs, ipaddr_t dst, ipaddr_t mask,
1660     ipaddr_t gateway, ipaddr_t src_addr, ipaddr_t brd_addr, ipaddr_t author,
1661     ipaddr_t ifaddr, const ill_t *ill, mblk_t *mp,
1662     const tsol_gc_t *gc)
1663 {
1664 	rt_msghdr_t	*rtm;
1665 	sin_t		*sin;
1666 	size_t		data_size, header_size;
1667 	uchar_t		*cp;
1668 	int		i;
1669 
1670 	ASSERT(mp != NULL);
1671 	/*
1672 	 * First find the type of the message
1673 	 * and its length.
1674 	 */
1675 	header_size = rts_header_msg_size(type);
1676 	/*
1677 	 * Now find the size of the data
1678 	 * that follows the message header.
1679 	 */
1680 	data_size = rts_data_msg_size(rtm_addrs, AF_INET, gc != NULL ? 1 : 0);
1681 
1682 	rtm = (rt_msghdr_t *)mp->b_rptr;
1683 	mp->b_wptr = &mp->b_rptr[header_size];
1684 	cp = mp->b_wptr;
1685 	bzero(cp, data_size);
1686 	for (i = 0; i < RTA_NUMBITS; i++) {
1687 		sin = (sin_t *)cp;
1688 		switch (rtm_addrs & (1 << i)) {
1689 		case RTA_DST:
1690 			sin->sin_addr.s_addr = dst;
1691 			sin->sin_family = AF_INET;
1692 			cp += sizeof (sin_t);
1693 			break;
1694 		case RTA_GATEWAY:
1695 			sin->sin_addr.s_addr = gateway;
1696 			sin->sin_family = AF_INET;
1697 			cp += sizeof (sin_t);
1698 			break;
1699 		case RTA_NETMASK:
1700 			sin->sin_addr.s_addr = mask;
1701 			sin->sin_family = AF_INET;
1702 			cp += sizeof (sin_t);
1703 			break;
1704 		case RTA_IFP:
1705 			cp += ill_dls_info((struct sockaddr_dl *)cp, ill);
1706 			break;
1707 		case RTA_IFA:
1708 			sin->sin_addr.s_addr = ifaddr;
1709 			sin->sin_family = AF_INET;
1710 			cp += sizeof (sin_t);
1711 			break;
1712 		case RTA_SRC:
1713 			sin->sin_addr.s_addr = src_addr;
1714 			sin->sin_family = AF_INET;
1715 			cp += sizeof (sin_t);
1716 			break;
1717 		case RTA_AUTHOR:
1718 			sin->sin_addr.s_addr = author;
1719 			sin->sin_family = AF_INET;
1720 			cp += sizeof (sin_t);
1721 			break;
1722 		case RTA_BRD:
1723 			/*
1724 			 * RTA_BRD is used typically to specify a point-to-point
1725 			 * destination address.
1726 			 */
1727 			sin->sin_addr.s_addr = brd_addr;
1728 			sin->sin_family = AF_INET;
1729 			cp += sizeof (sin_t);
1730 			break;
1731 		}
1732 	}
1733 
1734 	if (gc != NULL) {
1735 		rtm_ext_t *rtm_ext;
1736 		struct rtsa_s *rp_dst;
1737 		tsol_rtsecattr_t *rsap;
1738 
1739 		ASSERT(gc->gc_grp != NULL);
1740 		ASSERT(RW_LOCK_HELD(&gc->gc_grp->gcgrp_rwlock));
1741 
1742 		rtm_ext = (rtm_ext_t *)cp;
1743 		rtm_ext->rtmex_type = RTMEX_GATEWAY_SECATTR;
1744 		rtm_ext->rtmex_len = TSOL_RTSECATTR_SIZE(1);
1745 
1746 		rsap = (tsol_rtsecattr_t *)(rtm_ext + 1);
1747 		rsap->rtsa_cnt = 1;
1748 		rp_dst = rsap->rtsa_attr;
1749 
1750 		ASSERT(gc->gc_db != NULL);
1751 		bcopy(&gc->gc_db->gcdb_attr, rp_dst, sizeof (*rp_dst));
1752 		cp = (uchar_t *)rp_dst;
1753 	}
1754 
1755 	mp->b_wptr = cp;
1756 	mp->b_cont = NULL;
1757 	/*
1758 	 * set the fields that are common to
1759 	 * to different messages.
1760 	 */
1761 	rtm->rtm_msglen = (short)(header_size + data_size);
1762 	rtm->rtm_version = RTM_VERSION;
1763 	rtm->rtm_type = (uchar_t)type;
1764 }
1765 
1766 /*
1767  * Allocates and initializes a routing socket message.
1768  * Note that sacnt is either zero or one.
1769  */
1770 mblk_t *
1771 rts_alloc_msg(int type, int rtm_addrs, sa_family_t af, uint_t sacnt)
1772 {
1773 	size_t	length;
1774 	mblk_t	*mp;
1775 
1776 	length = RTS_MSG_SIZE(type, rtm_addrs, af, sacnt);
1777 	mp = allocb(length, BPRI_MED);
1778 	if (mp == NULL)
1779 		return (mp);
1780 	bzero(mp->b_rptr, length);
1781 	return (mp);
1782 }
1783 
1784 /*
1785  * Returns the size of the routing
1786  * socket message header size.
1787  */
1788 size_t
1789 rts_header_msg_size(int type)
1790 {
1791 	switch (type) {
1792 	case RTM_DELADDR:
1793 	case RTM_NEWADDR:
1794 		return (sizeof (ifa_msghdr_t));
1795 	case RTM_IFINFO:
1796 		return (sizeof (if_msghdr_t));
1797 	default:
1798 		return (sizeof (rt_msghdr_t));
1799 	}
1800 }
1801 
1802 /*
1803  * Returns the size of the message needed with the given rtm_addrs and family.
1804  *
1805  * It is assumed that all of the sockaddrs (with the exception of RTA_IFP) are
1806  * of the same family (currently either AF_INET or AF_INET6).
1807  */
1808 size_t
1809 rts_data_msg_size(int rtm_addrs, sa_family_t af, uint_t sacnt)
1810 {
1811 	int	i;
1812 	size_t	length = 0;
1813 
1814 	for (i = 0; i < RTA_NUMBITS; i++) {
1815 		switch (rtm_addrs & (1 << i)) {
1816 		case RTA_IFP:
1817 			length += sizeof (struct sockaddr_dl);
1818 			break;
1819 		case RTA_DST:
1820 		case RTA_GATEWAY:
1821 		case RTA_NETMASK:
1822 		case RTA_SRC:
1823 		case RTA_IFA:
1824 		case RTA_AUTHOR:
1825 		case RTA_BRD:
1826 			ASSERT(af == AF_INET || af == AF_INET6);
1827 			switch (af) {
1828 			case AF_INET:
1829 				length += sizeof (sin_t);
1830 				break;
1831 			case AF_INET6:
1832 				length += sizeof (sin6_t);
1833 				break;
1834 			}
1835 			break;
1836 		}
1837 	}
1838 	if (sacnt > 0)
1839 		length += sizeof (rtm_ext_t) + TSOL_RTSECATTR_SIZE(sacnt);
1840 
1841 	return (length);
1842 }
1843 
1844 /*
1845  * This routine is called to generate a message to the routing
1846  * socket indicating that a redirect has occured, a routing lookup
1847  * has failed, or that a protocol has detected timeouts to a particular
1848  * destination. This routine is called for message types RTM_LOSING,
1849  * RTM_REDIRECT, and RTM_MISS.
1850  */
1851 void
1852 ip_rts_change(int type, ipaddr_t dst_addr, ipaddr_t gw_addr, ipaddr_t net_mask,
1853     ipaddr_t source, ipaddr_t author, int flags, int error, int rtm_addrs,
1854     ip_stack_t *ipst)
1855 {
1856 	rt_msghdr_t	*rtm;
1857 	mblk_t		*mp;
1858 
1859 	if (rtm_addrs == 0)
1860 		return;
1861 	mp = rts_alloc_msg(type, rtm_addrs, AF_INET, 0);
1862 	if (mp == NULL)
1863 		return;
1864 	rts_fill_msg(type, rtm_addrs, dst_addr, net_mask, gw_addr, source, 0,
1865 	    author, 0, NULL, mp, NULL);
1866 	rtm = (rt_msghdr_t *)mp->b_rptr;
1867 	rtm->rtm_flags = flags;
1868 	rtm->rtm_errno = error;
1869 	rtm->rtm_flags |= RTF_DONE;
1870 	rtm->rtm_addrs = rtm_addrs;
1871 	rts_queue_input(mp, NULL, AF_INET, RTSQ_ALL, ipst);
1872 }
1873 
1874 /*
1875  * This routine is called to generate a message to the routing
1876  * socket indicating that the status of a network interface has changed.
1877  * Message type generated RTM_IFINFO.
1878  */
1879 void
1880 ip_rts_ifmsg(const ipif_t *ipif, uint_t flags)
1881 {
1882 	ip_rts_xifmsg(ipif, 0, 0, flags);
1883 }
1884 
1885 void
1886 ip_rts_xifmsg(const ipif_t *ipif, uint64_t set, uint64_t clear, uint_t flags)
1887 {
1888 	if_msghdr_t	*ifm;
1889 	mblk_t		*mp;
1890 	sa_family_t	af;
1891 	ip_stack_t	*ipst = ipif->ipif_ill->ill_ipst;
1892 
1893 	/*
1894 	 * This message should be generated only
1895 	 * when the physical device is changing
1896 	 * state.
1897 	 */
1898 	if (ipif->ipif_id != 0)
1899 		return;
1900 	if (ipif->ipif_isv6) {
1901 		af = AF_INET6;
1902 		mp = rts_alloc_msg(RTM_IFINFO, RTA_IFP, af, 0);
1903 		if (mp == NULL)
1904 			return;
1905 		rts_fill_msg_v6(RTM_IFINFO, RTA_IFP, &ipv6_all_zeros,
1906 		    &ipv6_all_zeros, &ipv6_all_zeros, &ipv6_all_zeros,
1907 		    &ipv6_all_zeros, &ipv6_all_zeros, &ipv6_all_zeros,
1908 		    ipif->ipif_ill, mp, NULL);
1909 	} else {
1910 		af = AF_INET;
1911 		mp = rts_alloc_msg(RTM_IFINFO, RTA_IFP, af, 0);
1912 		if (mp == NULL)
1913 			return;
1914 		rts_fill_msg(RTM_IFINFO, RTA_IFP, 0, 0, 0, 0, 0, 0, 0,
1915 		    ipif->ipif_ill, mp, NULL);
1916 	}
1917 	ifm = (if_msghdr_t *)mp->b_rptr;
1918 	ifm->ifm_index = ipif->ipif_ill->ill_phyint->phyint_ifindex;
1919 	ifm->ifm_flags = (ipif->ipif_flags | ipif->ipif_ill->ill_flags |
1920 	    ipif->ipif_ill->ill_phyint->phyint_flags | set) & ~clear;
1921 	rts_getifdata(&ifm->ifm_data, ipif);
1922 	ifm->ifm_addrs = RTA_IFP;
1923 
1924 	if (flags & RTSQ_DEFAULT) {
1925 		flags = RTSQ_ALL;
1926 		/*
1927 		 * If this message is for an underlying interface, prevent
1928 		 * "normal" (IPMP-unaware) routing sockets from seeing it.
1929 		 */
1930 		if (IS_UNDER_IPMP(ipif->ipif_ill))
1931 			flags &= ~RTSQ_NORMAL;
1932 	}
1933 
1934 	rts_queue_input(mp, NULL, af, flags, ipst);
1935 }
1936 
1937 /*
1938  * This is called to generate messages to the routing socket
1939  * indicating a network interface has had addresses associated with it.
1940  * The structure of the code is based on the 4.4BSD-Lite2 <net/rtsock.c>.
1941  */
1942 void
1943 ip_rts_newaddrmsg(int cmd, int error, const ipif_t *ipif, uint_t flags)
1944 {
1945 	int		pass;
1946 	int		ncmd;
1947 	int		rtm_addrs;
1948 	mblk_t		*mp;
1949 	ifa_msghdr_t	*ifam;
1950 	rt_msghdr_t	*rtm;
1951 	sa_family_t	af;
1952 	ip_stack_t	*ipst = ipif->ipif_ill->ill_ipst;
1953 
1954 	/*
1955 	 * Let conn_ixa caching know that source address selection
1956 	 * changed
1957 	 */
1958 	ip_update_source_selection(ipst);
1959 
1960 	if (ipif->ipif_isv6)
1961 		af = AF_INET6;
1962 	else
1963 		af = AF_INET;
1964 
1965 	if (flags & RTSQ_DEFAULT) {
1966 		flags = RTSQ_ALL;
1967 		/*
1968 		 * If this message is for an underlying interface, prevent
1969 		 * "normal" (IPMP-unaware) routing sockets from seeing it.
1970 		 */
1971 		if (IS_UNDER_IPMP(ipif->ipif_ill))
1972 			flags &= ~RTSQ_NORMAL;
1973 	}
1974 
1975 	/*
1976 	 * If the request is DELETE, send RTM_DELETE and RTM_DELADDR.
1977 	 * if the request is ADD, send RTM_NEWADDR and RTM_ADD.
1978 	 */
1979 	for (pass = 1; pass < 3; pass++) {
1980 		if ((cmd == RTM_ADD && pass == 1) ||
1981 		    (cmd == RTM_DELETE && pass == 2)) {
1982 			ncmd = ((cmd == RTM_ADD) ? RTM_NEWADDR : RTM_DELADDR);
1983 
1984 			rtm_addrs = (RTA_IFA | RTA_NETMASK | RTA_BRD | RTA_IFP);
1985 			mp = rts_alloc_msg(ncmd, rtm_addrs, af, 0);
1986 			if (mp == NULL)
1987 				continue;
1988 			switch (af) {
1989 			case AF_INET:
1990 				rts_fill_msg(ncmd, rtm_addrs, 0,
1991 				    ipif->ipif_net_mask, 0, ipif->ipif_lcl_addr,
1992 				    ipif->ipif_pp_dst_addr, 0,
1993 				    ipif->ipif_lcl_addr, ipif->ipif_ill,
1994 				    mp, NULL);
1995 				break;
1996 			case AF_INET6:
1997 				rts_fill_msg_v6(ncmd, rtm_addrs,
1998 				    &ipv6_all_zeros, &ipif->ipif_v6net_mask,
1999 				    &ipv6_all_zeros, &ipif->ipif_v6lcl_addr,
2000 				    &ipif->ipif_v6pp_dst_addr, &ipv6_all_zeros,
2001 				    &ipif->ipif_v6lcl_addr, ipif->ipif_ill,
2002 				    mp, NULL);
2003 				break;
2004 			}
2005 			ifam = (ifa_msghdr_t *)mp->b_rptr;
2006 			ifam->ifam_index =
2007 			    ipif->ipif_ill->ill_phyint->phyint_ifindex;
2008 			ifam->ifam_metric = ipif->ipif_metric;
2009 			ifam->ifam_flags = ((cmd == RTM_ADD) ? RTF_UP : 0);
2010 			ifam->ifam_addrs = rtm_addrs;
2011 			rts_queue_input(mp, NULL, af, flags, ipst);
2012 		}
2013 		if ((cmd == RTM_ADD && pass == 2) ||
2014 		    (cmd == RTM_DELETE && pass == 1)) {
2015 			rtm_addrs = (RTA_DST | RTA_NETMASK);
2016 			mp = rts_alloc_msg(cmd, rtm_addrs, af, 0);
2017 			if (mp == NULL)
2018 				continue;
2019 			switch (af) {
2020 			case AF_INET:
2021 				rts_fill_msg(cmd, rtm_addrs,
2022 				    ipif->ipif_lcl_addr, ipif->ipif_net_mask, 0,
2023 				    0, 0, 0, 0, NULL, mp, NULL);
2024 				break;
2025 			case AF_INET6:
2026 				rts_fill_msg_v6(cmd, rtm_addrs,
2027 				    &ipif->ipif_v6lcl_addr,
2028 				    &ipif->ipif_v6net_mask, &ipv6_all_zeros,
2029 				    &ipv6_all_zeros, &ipv6_all_zeros,
2030 				    &ipv6_all_zeros, &ipv6_all_zeros,
2031 				    NULL, mp, NULL);
2032 				break;
2033 			}
2034 			rtm = (rt_msghdr_t *)mp->b_rptr;
2035 			rtm->rtm_index =
2036 			    ipif->ipif_ill->ill_phyint->phyint_ifindex;
2037 			rtm->rtm_flags = ((cmd == RTM_ADD) ? RTF_UP : 0);
2038 			rtm->rtm_errno = error;
2039 			if (error == 0)
2040 				rtm->rtm_flags |= RTF_DONE;
2041 			rtm->rtm_addrs = rtm_addrs;
2042 			rts_queue_input(mp, NULL, af, flags, ipst);
2043 		}
2044 	}
2045 }
2046 
2047 /*
2048  * Based on the address family specified in a sockaddr, copy the address field
2049  * into an in6_addr_t.
2050  *
2051  * In the case of AF_UNSPEC, we assume the family is actually AF_INET for
2052  * compatibility with programs that leave the family cleared in the sockaddr.
2053  * Callers of rts_copyfromsockaddr should check the family themselves if they
2054  * wish to verify its value.
2055  *
2056  * In the case of AF_INET6, a check is made to ensure that address is not an
2057  * IPv4-mapped address.
2058  */
2059 size_t
2060 rts_copyfromsockaddr(struct sockaddr *sa, in6_addr_t *addrp)
2061 {
2062 	switch (sa->sa_family) {
2063 	case AF_INET:
2064 	case AF_UNSPEC:
2065 		IN6_IPADDR_TO_V4MAPPED(((sin_t *)sa)->sin_addr.s_addr, addrp);
2066 		return (sizeof (sin_t));
2067 	case AF_INET6:
2068 		*addrp = ((sin6_t *)sa)->sin6_addr;
2069 		if (IN6_IS_ADDR_V4MAPPED(addrp))
2070 			return (0);
2071 		return (sizeof (sin6_t));
2072 	default:
2073 		return (0);
2074 	}
2075 }
2076