1/*	$NetBSD: rtsock.c,v 1.139 2011/12/31 20:41:58 christos Exp $	*/
2
3/*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * 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 project nor the names of its contributors
16 *    may be used to endorse or promote products derived from this software
17 *    without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32/*
33 * Copyright (c) 1988, 1991, 1993
34 *	The Regents of the University of California.  All rights reserved.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 *    notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 *    notice, this list of conditions and the following disclaimer in the
43 *    documentation and/or other materials provided with the distribution.
44 * 3. Neither the name of the University nor the names of its contributors
45 *    may be used to endorse or promote products derived from this software
46 *    without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 *
60 *	@(#)rtsock.c	8.7 (Berkeley) 10/12/95
61 */
62
63#include <sys/cdefs.h>
64__KERNEL_RCSID(0, "$NetBSD: rtsock.c,v 1.139 2011/12/31 20:41:58 christos Exp $");
65
66#ifdef _KERNEL_OPT
67#include "opt_inet.h"
68#include "opt_mpls.h"
69#include "opt_compat_netbsd.h"
70#endif
71
72#include <sys/param.h>
73#include <sys/systm.h>
74#include <sys/proc.h>
75#include <sys/mbuf.h>
76#include <sys/socket.h>
77#include <sys/socketvar.h>
78#include <sys/domain.h>
79#include <sys/protosw.h>
80#include <sys/sysctl.h>
81#include <sys/kauth.h>
82#include <sys/intr.h>
83#ifdef RTSOCK_DEBUG
84#include <netinet/in.h>
85#endif /* RTSOCK_DEBUG */
86
87#include <net/if.h>
88#include <net/route.h>
89#include <net/raw_cb.h>
90
91#include <netmpls/mpls.h>
92
93#if defined(COMPAT_14) || defined(COMPAT_50)
94#include <compat/net/if.h>
95#include <compat/net/route.h>
96#endif
97#ifdef COMPAT_RTSOCK
98#define	RTM_XVERSION	RTM_OVERSION
99#define	RT_XADVANCE(a,b) RT_OADVANCE(a,b)
100#define	RT_XROUNDUP(n)	RT_OROUNDUP(n)
101#define	PF_XROUTE	PF_OROUTE
102#define	rt_xmsghdr	rt_msghdr50
103#define	if_xmsghdr	if_msghdr	/* if_msghdr50 is for RTM_OIFINFO */
104#define	ifa_xmsghdr	ifa_msghdr50
105#define	if_xannouncemsghdr	if_announcemsghdr50
106#define	COMPATNAME(x)	compat_50_ ## x
107#define	DOMAINNAME	"oroute"
108CTASSERT(sizeof(struct ifa_xmsghdr) == 20);
109DOMAIN_DEFINE(compat_50_routedomain); /* forward declare and add to link set */
110#else
111#define	RTM_XVERSION	RTM_VERSION
112#define	RT_XADVANCE(a,b) RT_ADVANCE(a,b)
113#define	RT_XROUNDUP(n)	RT_ROUNDUP(n)
114#define	PF_XROUTE	PF_ROUTE
115#define	rt_xmsghdr	rt_msghdr
116#define	if_xmsghdr	if_msghdr
117#define	ifa_xmsghdr	ifa_msghdr
118#define	if_xannouncemsghdr	if_announcemsghdr
119#define	COMPATNAME(x)	x
120#define	DOMAINNAME	"route"
121CTASSERT(sizeof(struct ifa_xmsghdr) == 24);
122#ifdef COMPAT_50
123#define	COMPATCALL(name, args)	compat_50_ ## name args
124#endif
125DOMAIN_DEFINE(routedomain); /* forward declare and add to link set */
126#undef COMPAT_50
127#undef COMPAT_14
128#endif
129
130#ifndef COMPATCALL
131#define	COMPATCALL(name, args)	do { } while (/*CONSTCOND*/ 0)
132#endif
133
134struct route_info COMPATNAME(route_info) = {
135	.ri_dst = { .sa_len = 2, .sa_family = PF_XROUTE, },
136	.ri_src = { .sa_len = 2, .sa_family = PF_XROUTE, },
137	.ri_maxqlen = IFQ_MAXLEN,
138};
139
140#define	PRESERVED_RTF	(RTF_UP | RTF_GATEWAY | RTF_HOST | RTF_DONE | RTF_MASK)
141
142static void COMPATNAME(route_init)(void);
143static int COMPATNAME(route_output)(struct mbuf *, ...);
144static int COMPATNAME(route_usrreq)(struct socket *,
145	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *);
146
147static int rt_msg2(int, struct rt_addrinfo *, void *, struct rt_walkarg *, int *);
148static int rt_xaddrs(u_char, const char *, const char *, struct rt_addrinfo *);
149static struct mbuf *rt_makeifannouncemsg(struct ifnet *, int, int,
150    struct rt_addrinfo *);
151static void rt_setmetrics(int, const struct rt_xmsghdr *, struct rtentry *);
152static void rtm_setmetrics(const struct rtentry *, struct rt_xmsghdr *);
153static void sysctl_net_route_setup(struct sysctllog **);
154static int sysctl_dumpentry(struct rtentry *, void *);
155static int sysctl_iflist(int, struct rt_walkarg *, int);
156static int sysctl_rtable(SYSCTLFN_PROTO);
157static void rt_adjustcount(int, int);
158
159static void
160rt_adjustcount(int af, int cnt)
161{
162	struct route_cb * const cb = &COMPATNAME(route_info).ri_cb;
163
164	cb->any_count += cnt;
165
166	switch (af) {
167	case AF_INET:
168		cb->ip_count += cnt;
169		return;
170#ifdef INET6
171	case AF_INET6:
172		cb->ip6_count += cnt;
173		return;
174#endif
175	case AF_ISO:
176		cb->iso_count += cnt;
177		return;
178	case AF_MPLS:
179		cb->mpls_count += cnt;
180		return;
181	}
182}
183
184/*ARGSUSED*/
185int
186COMPATNAME(route_usrreq)(struct socket *so, int req, struct mbuf *m, struct mbuf *nam,
187	struct mbuf *control, struct lwp *l)
188{
189	int error = 0;
190	struct rawcb *rp = sotorawcb(so);
191	int s;
192
193	if (req == PRU_ATTACH) {
194		sosetlock(so);
195		rp = malloc(sizeof(*rp), M_PCB, M_WAITOK|M_ZERO);
196		so->so_pcb = rp;
197	}
198	if (req == PRU_DETACH && rp)
199		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
200	s = splsoftnet();
201
202	/*
203	 * Don't call raw_usrreq() in the attach case, because
204	 * we want to allow non-privileged processes to listen on
205	 * and send "safe" commands to the routing socket.
206	 */
207	if (req == PRU_ATTACH) {
208		if (l == NULL)
209			error = EACCES;
210		else
211			error = raw_attach(so, (int)(long)nam);
212	} else
213		error = raw_usrreq(so, req, m, nam, control, l);
214
215	rp = sotorawcb(so);
216	if (req == PRU_ATTACH && rp) {
217		if (error) {
218			free(rp, M_PCB);
219			splx(s);
220			return error;
221		}
222		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
223		rp->rcb_laddr = &COMPATNAME(route_info).ri_src;
224		rp->rcb_faddr = &COMPATNAME(route_info).ri_dst;
225		soisconnected(so);
226		so->so_options |= SO_USELOOPBACK;
227	}
228	splx(s);
229	return error;
230}
231
232/*ARGSUSED*/
233int
234COMPATNAME(route_output)(struct mbuf *m, ...)
235{
236	struct sockproto proto = { .sp_family = PF_XROUTE, };
237	struct rt_xmsghdr *rtm = NULL;
238	struct rt_xmsghdr *old_rtm = NULL;
239	struct rtentry *rt = NULL;
240	struct rtentry *saved_nrt = NULL;
241	struct rt_addrinfo info;
242	int len, error = 0;
243	struct ifnet *ifp = NULL;
244	struct ifaddr *ifa = NULL;
245	struct socket *so;
246	va_list ap;
247	sa_family_t family;
248
249	va_start(ap, m);
250	so = va_arg(ap, struct socket *);
251	va_end(ap);
252
253#define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0)
254	if (m == NULL || ((m->m_len < sizeof(int32_t)) &&
255	   (m = m_pullup(m, sizeof(int32_t))) == NULL))
256		return ENOBUFS;
257	if ((m->m_flags & M_PKTHDR) == 0)
258		panic("%s", __func__);
259	len = m->m_pkthdr.len;
260	if (len < sizeof(*rtm) ||
261	    len != mtod(m, struct rt_xmsghdr *)->rtm_msglen) {
262		info.rti_info[RTAX_DST] = NULL;
263		senderr(EINVAL);
264	}
265	R_Malloc(rtm, struct rt_xmsghdr *, len);
266	if (rtm == NULL) {
267		info.rti_info[RTAX_DST] = NULL;
268		senderr(ENOBUFS);
269	}
270	m_copydata(m, 0, len, rtm);
271	if (rtm->rtm_version != RTM_XVERSION) {
272		info.rti_info[RTAX_DST] = NULL;
273		senderr(EPROTONOSUPPORT);
274	}
275	rtm->rtm_pid = curproc->p_pid;
276	memset(&info, 0, sizeof(info));
277	info.rti_addrs = rtm->rtm_addrs;
278	if (rt_xaddrs(rtm->rtm_type, (const char *)(rtm + 1), len + (char *)rtm,
279	    &info)) {
280		senderr(EINVAL);
281	}
282	info.rti_flags = rtm->rtm_flags;
283#ifdef RTSOCK_DEBUG
284	if (info.rti_info[RTAX_DST]->sa_family == AF_INET) {
285		printf("%s: extracted info.rti_info[RTAX_DST] %s\n", __func__,
286		    inet_ntoa(((const struct sockaddr_in *)
287		    info.rti_info[RTAX_DST])->sin_addr));
288	}
289#endif /* RTSOCK_DEBUG */
290	if (info.rti_info[RTAX_DST] == NULL ||
291	    (info.rti_info[RTAX_DST]->sa_family >= AF_MAX)) {
292		senderr(EINVAL);
293	}
294	if (info.rti_info[RTAX_GATEWAY] != NULL &&
295	    (info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) {
296		senderr(EINVAL);
297	}
298
299	/*
300	 * Verify that the caller has the appropriate privilege; RTM_GET
301	 * is the only operation the non-superuser is allowed.
302	 */
303	if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE,
304	    0, rtm, NULL, NULL) != 0)
305		senderr(EACCES);
306
307	switch (rtm->rtm_type) {
308
309	case RTM_ADD:
310		if (info.rti_info[RTAX_GATEWAY] == NULL) {
311			senderr(EINVAL);
312		}
313		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
314		if (error == 0 && saved_nrt) {
315			rt_setmetrics(rtm->rtm_inits, rtm, saved_nrt);
316			saved_nrt->rt_refcnt--;
317		}
318		break;
319
320	case RTM_DELETE:
321		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
322		if (error == 0) {
323			(rt = saved_nrt)->rt_refcnt++;
324			goto report;
325		}
326		break;
327
328	case RTM_GET:
329	case RTM_CHANGE:
330	case RTM_LOCK:
331                /* XXX This will mask info.rti_info[RTAX_DST] with
332		 * info.rti_info[RTAX_NETMASK] before
333                 * searching.  It did not used to do that.  --dyoung
334		 */
335		error = rtrequest1(RTM_GET, &info, &rt);
336		if (error != 0)
337			senderr(error);
338		if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
339			if (memcmp(info.rti_info[RTAX_DST], rt_getkey(rt),
340			    info.rti_info[RTAX_DST]->sa_len) != 0)
341				senderr(ESRCH);
342			if (info.rti_info[RTAX_NETMASK] == NULL &&
343			    rt_mask(rt) != NULL)
344				senderr(ETOOMANYREFS);
345		}
346
347		switch (rtm->rtm_type) {
348		case RTM_GET:
349		report:
350			info.rti_info[RTAX_DST] = rt_getkey(rt);
351			info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
352			info.rti_info[RTAX_NETMASK] = rt_mask(rt);
353			info.rti_info[RTAX_TAG] = rt_gettag(rt);
354			if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) == 0)
355				;
356			else if ((ifp = rt->rt_ifp) != NULL) {
357				const struct ifaddr *rtifa;
358				info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
359                                /* rtifa used to be simply rt->rt_ifa.
360                                 * If rt->rt_ifa != NULL, then
361                                 * rt_get_ifa() != NULL.  So this
362                                 * ought to still be safe. --dyoung
363				 */
364				rtifa = rt_get_ifa(rt);
365				info.rti_info[RTAX_IFA] = rtifa->ifa_addr;
366#ifdef RTSOCK_DEBUG
367				if (info.rti_info[RTAX_IFA]->sa_family ==
368				    AF_INET) {
369					printf("%s: copying out RTAX_IFA %s ",
370					    __func__, inet_ntoa(
371					    ((const struct sockaddr_in *)
372					    info.rti_info[RTAX_IFA])->sin_addr)
373					    );
374					printf("for info.rti_info[RTAX_DST] %s "
375					    "ifa_getifa %p ifa_seqno %p\n",
376					    inet_ntoa(
377					    ((const struct sockaddr_in *)
378					    info.rti_info[RTAX_DST])->sin_addr),
379					    (void *)rtifa->ifa_getifa,
380					    rtifa->ifa_seqno);
381				}
382#endif /* RTSOCK_DEBUG */
383				if (ifp->if_flags & IFF_POINTOPOINT) {
384					info.rti_info[RTAX_BRD] =
385					    rtifa->ifa_dstaddr;
386				} else
387					info.rti_info[RTAX_BRD] = NULL;
388				rtm->rtm_index = ifp->if_index;
389			} else {
390				info.rti_info[RTAX_IFP] = NULL;
391				info.rti_info[RTAX_IFA] = NULL;
392			}
393			(void)rt_msg2(rtm->rtm_type, &info, NULL, NULL, &len);
394			if (len > rtm->rtm_msglen) {
395				old_rtm = rtm;
396				R_Malloc(rtm, struct rt_xmsghdr *, len);
397				if (rtm == NULL)
398					senderr(ENOBUFS);
399				(void)memcpy(rtm, old_rtm, old_rtm->rtm_msglen);
400			}
401			(void)rt_msg2(rtm->rtm_type, &info, rtm, NULL, 0);
402			rtm->rtm_flags = rt->rt_flags;
403			rtm_setmetrics(rt, rtm);
404			rtm->rtm_addrs = info.rti_addrs;
405			break;
406
407		case RTM_CHANGE:
408			/*
409			 * new gateway could require new ifaddr, ifp;
410			 * flags may also be different; ifp may be specified
411			 * by ll sockaddr when protocol address is ambiguous
412			 */
413			if ((error = rt_getifa(&info)) != 0)
414				senderr(error);
415			if (info.rti_info[RTAX_GATEWAY] &&
416			    rt_setgate(rt, info.rti_info[RTAX_GATEWAY]))
417				senderr(EDQUOT);
418			if (info.rti_info[RTAX_TAG])
419				rt_settag(rt, info.rti_info[RTAX_TAG]);
420			/* new gateway could require new ifaddr, ifp;
421			   flags may also be different; ifp may be specified
422			   by ll sockaddr when protocol address is ambiguous */
423			if (info.rti_info[RTAX_IFP] &&
424			    (ifa = ifa_ifwithnet(info.rti_info[RTAX_IFP])) &&
425			    (ifp = ifa->ifa_ifp) && (info.rti_info[RTAX_IFA] ||
426			    info.rti_info[RTAX_GATEWAY])) {
427				if (info.rti_info[RTAX_IFA] == NULL ||
428				    (ifa = ifa_ifwithaddr(
429				    info.rti_info[RTAX_IFA])) == NULL)
430					ifa = ifaof_ifpforaddr(
431					    info.rti_info[RTAX_IFA] ?
432					    info.rti_info[RTAX_IFA] :
433					    info.rti_info[RTAX_GATEWAY], ifp);
434			} else if ((info.rti_info[RTAX_IFA] &&
435			    (ifa = ifa_ifwithaddr(info.rti_info[RTAX_IFA]))) ||
436			    (info.rti_info[RTAX_GATEWAY] &&
437			    (ifa = ifa_ifwithroute(rt->rt_flags,
438			    rt_getkey(rt), info.rti_info[RTAX_GATEWAY])))) {
439				ifp = ifa->ifa_ifp;
440			}
441			if (ifa) {
442				struct ifaddr *oifa = rt->rt_ifa;
443				if (oifa != ifa) {
444					if (oifa && oifa->ifa_rtrequest) {
445						oifa->ifa_rtrequest(RTM_DELETE,
446						    rt, &info);
447					}
448					rt_replace_ifa(rt, ifa);
449					rt->rt_ifp = ifp;
450				}
451			}
452			if (ifp && rt->rt_ifp != ifp)
453				rt->rt_ifp = ifp;
454			rt_setmetrics(rtm->rtm_inits, rtm, rt);
455			if (rt->rt_flags != info.rti_flags)
456				rt->rt_flags = (info.rti_flags & ~PRESERVED_RTF)
457				    | (rt->rt_flags & PRESERVED_RTF);
458			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
459				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
460			/*FALLTHROUGH*/
461		case RTM_LOCK:
462			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
463			rt->rt_rmx.rmx_locks |=
464			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
465			break;
466		}
467		break;
468
469	default:
470		senderr(EOPNOTSUPP);
471	}
472
473flush:
474	if (rtm) {
475		if (error)
476			rtm->rtm_errno = error;
477		else
478			rtm->rtm_flags |= RTF_DONE;
479	}
480	family = info.rti_info[RTAX_DST] ? info.rti_info[RTAX_DST]->sa_family :
481	    0;
482	/* We cannot free old_rtm until we have stopped using the
483	 * pointers in info, some of which may point to sockaddrs
484	 * in old_rtm.
485	 */
486	if (old_rtm != NULL)
487		Free(old_rtm);
488	if (rt)
489		rtfree(rt);
490    {
491	struct rawcb *rp = NULL;
492	/*
493	 * Check to see if we don't want our own messages.
494	 */
495	if ((so->so_options & SO_USELOOPBACK) == 0) {
496		if (COMPATNAME(route_info).ri_cb.any_count <= 1) {
497			if (rtm)
498				Free(rtm);
499			m_freem(m);
500			return error;
501		}
502		/* There is another listener, so construct message */
503		rp = sotorawcb(so);
504	}
505	if (rtm) {
506		m_copyback(m, 0, rtm->rtm_msglen, rtm);
507		if (m->m_pkthdr.len < rtm->rtm_msglen) {
508			m_freem(m);
509			m = NULL;
510		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
511			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
512		Free(rtm);
513	}
514	if (rp)
515		rp->rcb_proto.sp_family = 0; /* Avoid us */
516	if (family)
517		proto.sp_protocol = family;
518	if (m)
519		raw_input(m, &proto, &COMPATNAME(route_info).ri_src,
520		    &COMPATNAME(route_info).ri_dst);
521	if (rp)
522		rp->rcb_proto.sp_family = PF_XROUTE;
523    }
524	return error;
525}
526
527static void
528rt_setmetrics(int which, const struct rt_xmsghdr *in, struct rtentry *out)
529{
530#define metric(f, e) if (which & (f)) out->rt_rmx.e = in->rtm_rmx.e;
531	metric(RTV_RPIPE, rmx_recvpipe);
532	metric(RTV_SPIPE, rmx_sendpipe);
533	metric(RTV_SSTHRESH, rmx_ssthresh);
534	metric(RTV_RTT, rmx_rtt);
535	metric(RTV_RTTVAR, rmx_rttvar);
536	metric(RTV_HOPCOUNT, rmx_hopcount);
537	metric(RTV_MTU, rmx_mtu);
538	metric(RTV_EXPIRE, rmx_expire);
539#undef metric
540}
541
542static void
543rtm_setmetrics(const struct rtentry *in, struct rt_xmsghdr *out)
544{
545#define metric(e) out->rtm_rmx.e = in->rt_rmx.e;
546	metric(rmx_recvpipe);
547	metric(rmx_sendpipe);
548	metric(rmx_ssthresh);
549	metric(rmx_rtt);
550	metric(rmx_rttvar);
551	metric(rmx_hopcount);
552	metric(rmx_mtu);
553	metric(rmx_expire);
554#undef metric
555}
556
557static int
558rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim,
559    struct rt_addrinfo *rtinfo)
560{
561	const struct sockaddr *sa = NULL;	/* Quell compiler warning */
562	int i;
563
564	for (i = 0; i < RTAX_MAX && cp < cplim; i++) {
565		if ((rtinfo->rti_addrs & (1 << i)) == 0)
566			continue;
567		rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp;
568		RT_XADVANCE(cp, sa);
569	}
570
571	/*
572	 * Check for extra addresses specified, except RTM_GET asking
573	 * for interface info.
574	 */
575	if (rtmtype == RTM_GET) {
576		if (((rtinfo->rti_addrs &
577		    (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0 << i)) != 0)
578			return 1;
579	} else if ((rtinfo->rti_addrs & (~0 << i)) != 0)
580		return 1;
581	/* Check for bad data length.  */
582	if (cp != cplim) {
583		if (i == RTAX_NETMASK + 1 && sa != NULL &&
584		    cp - RT_XROUNDUP(sa->sa_len) + sa->sa_len == cplim)
585			/*
586			 * The last sockaddr was info.rti_info[RTAX_NETMASK].
587			 * We accept this for now for the sake of old
588			 * binaries or third party softwares.
589			 */
590			;
591		else
592			return 1;
593	}
594	return 0;
595}
596
597static int
598rt_getlen(int type)
599{
600#ifndef COMPAT_RTSOCK
601	CTASSERT(__alignof(struct ifa_msghdr) >= sizeof(uint64_t));
602	CTASSERT(__alignof(struct if_msghdr) >= sizeof(uint64_t));
603	CTASSERT(__alignof(struct if_announcemsghdr) >= sizeof(uint64_t));
604	CTASSERT(__alignof(struct rt_msghdr) >= sizeof(uint64_t));
605#endif
606
607	switch (type) {
608	case RTM_DELADDR:
609	case RTM_NEWADDR:
610	case RTM_CHGADDR:
611		return sizeof(struct ifa_xmsghdr);
612
613	case RTM_OOIFINFO:
614#ifdef COMPAT_14
615		return sizeof(struct if_msghdr14);
616#else
617#ifdef DIAGNOSTIC
618		printf("RTM_OOIFINFO\n");
619#endif
620		return -1;
621#endif
622	case RTM_OIFINFO:
623#ifdef COMPAT_50
624		return sizeof(struct if_msghdr50);
625#else
626#ifdef DIAGNOSTIC
627		printf("RTM_OIFINFO\n");
628#endif
629		return -1;
630#endif
631
632	case RTM_IFINFO:
633		return sizeof(struct if_xmsghdr);
634
635	case RTM_IFANNOUNCE:
636	case RTM_IEEE80211:
637		return sizeof(struct if_xannouncemsghdr);
638
639	default:
640		return sizeof(struct rt_xmsghdr);
641	}
642}
643
644
645struct mbuf *
646COMPATNAME(rt_msg1)(int type, struct rt_addrinfo *rtinfo, void *data, int datalen)
647{
648	struct rt_xmsghdr *rtm;
649	struct mbuf *m;
650	int i;
651	const struct sockaddr *sa;
652	int len, dlen;
653
654	m = m_gethdr(M_DONTWAIT, MT_DATA);
655	if (m == NULL)
656		return m;
657	MCLAIM(m, &COMPATNAME(routedomain).dom_mowner);
658
659	if ((len = rt_getlen(type)) == -1)
660		goto out;
661	if (len > MHLEN + MLEN)
662		panic("%s: message too long", __func__);
663	else if (len > MHLEN) {
664		m->m_next = m_get(M_DONTWAIT, MT_DATA);
665		if (m->m_next == NULL)
666			goto out;
667		MCLAIM(m->m_next, m->m_owner);
668		m->m_pkthdr.len = len;
669		m->m_len = MHLEN;
670		m->m_next->m_len = len - MHLEN;
671	} else {
672		m->m_pkthdr.len = m->m_len = len;
673	}
674	m->m_pkthdr.rcvif = NULL;
675	m_copyback(m, 0, datalen, data);
676	if (len > datalen)
677		(void)memset(mtod(m, char *) + datalen, 0, len - datalen);
678	rtm = mtod(m, struct rt_xmsghdr *);
679	for (i = 0; i < RTAX_MAX; i++) {
680		if ((sa = rtinfo->rti_info[i]) == NULL)
681			continue;
682		rtinfo->rti_addrs |= (1 << i);
683		dlen = RT_XROUNDUP(sa->sa_len);
684		m_copyback(m, len, sa->sa_len, sa);
685		if (dlen != sa->sa_len) {
686			/*
687			 * Up to 6 + 1 nul's since roundup is to
688			 * sizeof(uint64_t) (8 bytes)
689			 */
690			m_copyback(m, len + sa->sa_len,
691			    dlen - sa->sa_len, "\0\0\0\0\0\0");
692		}
693		len += dlen;
694	}
695	if (m->m_pkthdr.len != len)
696		goto out;
697	rtm->rtm_msglen = len;
698	rtm->rtm_version = RTM_XVERSION;
699	rtm->rtm_type = type;
700	return m;
701out:
702	m_freem(m);
703	return NULL;
704}
705
706/*
707 * rt_msg2
708 *
709 *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
710 *		returns the length of the message in 'lenp'.
711 *
712 * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
713 *	the message
714 * otherwise walkarg's w_needed is updated and if the user buffer is
715 *	specified and w_needed indicates space exists the information is copied
716 *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
717 *	if the allocation fails ENOBUFS is returned.
718 */
719static int
720rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct rt_walkarg *w,
721	int *lenp)
722{
723	int i;
724	int len, dlen, second_time = 0;
725	char *cp0, *cp = cpv;
726
727	rtinfo->rti_addrs = 0;
728again:
729	if ((len = rt_getlen(type)) == -1)
730		return EINVAL;
731
732	if ((cp0 = cp) != NULL)
733		cp += len;
734	for (i = 0; i < RTAX_MAX; i++) {
735		const struct sockaddr *sa;
736
737		if ((sa = rtinfo->rti_info[i]) == NULL)
738			continue;
739		rtinfo->rti_addrs |= (1 << i);
740		dlen = RT_XROUNDUP(sa->sa_len);
741		if (cp) {
742			int diff = dlen - sa->sa_len;
743			(void)memcpy(cp, sa, (size_t)sa->sa_len);
744			cp += sa->sa_len;
745			if (diff > 0) {
746				(void)memset(cp, 0, (size_t)diff);
747				cp += diff;
748			}
749		}
750		len += dlen;
751	}
752	if (cp == NULL && w != NULL && !second_time) {
753		struct rt_walkarg *rw = w;
754
755		rw->w_needed += len;
756		if (rw->w_needed <= 0 && rw->w_where) {
757			if (rw->w_tmemsize < len) {
758				if (rw->w_tmem)
759					free(rw->w_tmem, M_RTABLE);
760				rw->w_tmem = malloc(len, M_RTABLE, M_NOWAIT);
761				if (rw->w_tmem)
762					rw->w_tmemsize = len;
763				else
764					rw->w_tmemsize = 0;
765			}
766			if (rw->w_tmem) {
767				cp = rw->w_tmem;
768				second_time = 1;
769				goto again;
770			} else {
771				rw->w_tmemneeded = len;
772				return ENOBUFS;
773			}
774		}
775	}
776	if (cp) {
777		struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)cp0;
778
779		rtm->rtm_version = RTM_XVERSION;
780		rtm->rtm_type = type;
781		rtm->rtm_msglen = len;
782	}
783	if (lenp)
784		*lenp = len;
785	return 0;
786}
787
788/*
789 * This routine is called to generate a message from the routing
790 * socket indicating that a redirect has occurred, a routing lookup
791 * has failed, or that a protocol has detected timeouts to a particular
792 * destination.
793 */
794void
795COMPATNAME(rt_missmsg)(int type, const struct rt_addrinfo *rtinfo, int flags,
796    int error)
797{
798	struct rt_xmsghdr rtm;
799	struct mbuf *m;
800	const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
801	struct rt_addrinfo info = *rtinfo;
802
803	COMPATCALL(rt_missmsg, (type, rtinfo, flags, error));
804	if (COMPATNAME(route_info).ri_cb.any_count == 0)
805		return;
806	memset(&rtm, 0, sizeof(rtm));
807	rtm.rtm_flags = RTF_DONE | flags;
808	rtm.rtm_errno = error;
809	m = COMPATNAME(rt_msg1)(type, &info, &rtm, sizeof(rtm));
810	if (m == NULL)
811		return;
812	mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs;
813	COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0);
814}
815
816/*
817 * This routine is called to generate a message from the routing
818 * socket indicating that the status of a network interface has changed.
819 */
820void
821COMPATNAME(rt_ifmsg)(struct ifnet *ifp)
822{
823	struct if_xmsghdr ifm;
824	struct mbuf *m;
825	struct rt_addrinfo info;
826
827	COMPATCALL(rt_ifmsg, (ifp));
828	if (COMPATNAME(route_info).ri_cb.any_count == 0)
829		return;
830	(void)memset(&info, 0, sizeof(info));
831	(void)memset(&ifm, 0, sizeof(ifm));
832	ifm.ifm_index = ifp->if_index;
833	ifm.ifm_flags = ifp->if_flags;
834	ifm.ifm_data = ifp->if_data;
835	ifm.ifm_addrs = 0;
836	m = COMPATNAME(rt_msg1)(RTM_IFINFO, &info, &ifm, sizeof(ifm));
837	if (m == NULL)
838		return;
839	COMPATNAME(route_enqueue)(m, 0);
840#ifdef COMPAT_14
841	compat_14_rt_oifmsg(ifp);
842#endif
843#ifdef COMPAT_50
844	compat_50_rt_oifmsg(ifp);
845#endif
846}
847
848
849/*
850 * This is called to generate messages from the routing socket
851 * indicating a network interface has had addresses associated with it.
852 * if we ever reverse the logic and replace messages TO the routing
853 * socket indicate a request to configure interfaces, then it will
854 * be unnecessary as the routing socket will automatically generate
855 * copies of it.
856 */
857void
858COMPATNAME(rt_newaddrmsg)(int cmd, struct ifaddr *ifa, int error,
859    struct rtentry *rt)
860{
861#define	cmdpass(__cmd, __pass)	(((__cmd) << 2) | (__pass))
862	struct rt_addrinfo info;
863	const struct sockaddr *sa;
864	int pass;
865	struct mbuf *m;
866	struct ifnet *ifp;
867	struct rt_xmsghdr rtm;
868	struct ifa_xmsghdr ifam;
869	int ncmd;
870
871	KASSERT(ifa != NULL);
872	ifp = ifa->ifa_ifp;
873	COMPATCALL(rt_newaddrmsg, (cmd, ifa, error, rt));
874	if (COMPATNAME(route_info).ri_cb.any_count == 0)
875		return;
876	for (pass = 1; pass < 3; pass++) {
877		memset(&info, 0, sizeof(info));
878		switch (cmdpass(cmd, pass)) {
879		case cmdpass(RTM_ADD, 1):
880		case cmdpass(RTM_CHANGE, 1):
881		case cmdpass(RTM_DELETE, 2):
882		case cmdpass(RTM_NEWADDR, 1):
883		case cmdpass(RTM_DELADDR, 1):
884		case cmdpass(RTM_CHGADDR, 1):
885			switch (cmd) {
886			case RTM_ADD:
887				ncmd = RTM_NEWADDR;
888				break;
889			case RTM_DELETE:
890				ncmd = RTM_DELADDR;
891				break;
892			case RTM_CHANGE:
893				ncmd = RTM_CHGADDR;
894				break;
895			default:
896				ncmd = cmd;
897			}
898			info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr;
899			KASSERT(ifp->if_dl != NULL);
900			info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
901			info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
902			info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
903			memset(&ifam, 0, sizeof(ifam));
904			ifam.ifam_index = ifp->if_index;
905			ifam.ifam_metric = ifa->ifa_metric;
906			ifam.ifam_flags = ifa->ifa_flags;
907			m = COMPATNAME(rt_msg1)(ncmd, &info, &ifam, sizeof(ifam));
908			if (m == NULL)
909				continue;
910			mtod(m, struct ifa_xmsghdr *)->ifam_addrs =
911			    info.rti_addrs;
912			break;
913		case cmdpass(RTM_ADD, 2):
914		case cmdpass(RTM_CHANGE, 2):
915		case cmdpass(RTM_DELETE, 1):
916			if (rt == NULL)
917				continue;
918			info.rti_info[RTAX_NETMASK] = rt_mask(rt);
919			info.rti_info[RTAX_DST] = sa = rt_getkey(rt);
920			info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
921			memset(&rtm, 0, sizeof(rtm));
922			rtm.rtm_index = ifp->if_index;
923			rtm.rtm_flags |= rt->rt_flags;
924			rtm.rtm_errno = error;
925			m = COMPATNAME(rt_msg1)(cmd, &info, &rtm, sizeof(rtm));
926			if (m == NULL)
927				continue;
928			mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs;
929			break;
930		default:
931			continue;
932		}
933#ifdef DIAGNOSTIC
934		if (m == NULL)
935			panic("%s: called with wrong command", __func__);
936#endif
937		COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0);
938	}
939#undef cmdpass
940}
941
942static struct mbuf *
943rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
944    struct rt_addrinfo *info)
945{
946	struct if_xannouncemsghdr ifan;
947
948	memset(info, 0, sizeof(*info));
949	memset(&ifan, 0, sizeof(ifan));
950	ifan.ifan_index = ifp->if_index;
951	strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name));
952	ifan.ifan_what = what;
953	return COMPATNAME(rt_msg1)(type, info, &ifan, sizeof(ifan));
954}
955
956/*
957 * This is called to generate routing socket messages indicating
958 * network interface arrival and departure.
959 */
960void
961COMPATNAME(rt_ifannouncemsg)(struct ifnet *ifp, int what)
962{
963	struct mbuf *m;
964	struct rt_addrinfo info;
965
966	COMPATCALL(rt_ifannouncemsg, (ifp, what));
967	if (COMPATNAME(route_info).ri_cb.any_count == 0)
968		return;
969	m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
970	if (m == NULL)
971		return;
972	COMPATNAME(route_enqueue)(m, 0);
973}
974
975/*
976 * This is called to generate routing socket messages indicating
977 * IEEE80211 wireless events.
978 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
979 */
980void
981COMPATNAME(rt_ieee80211msg)(struct ifnet *ifp, int what, void *data,
982	size_t data_len)
983{
984	struct mbuf *m;
985	struct rt_addrinfo info;
986
987	COMPATCALL(rt_ieee80211msg, (ifp, what, data, data_len));
988	if (COMPATNAME(route_info).ri_cb.any_count == 0)
989		return;
990	m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
991	if (m == NULL)
992		return;
993	/*
994	 * Append the ieee80211 data.  Try to stick it in the
995	 * mbuf containing the ifannounce msg; otherwise allocate
996	 * a new mbuf and append.
997	 *
998	 * NB: we assume m is a single mbuf.
999	 */
1000	if (data_len > M_TRAILINGSPACE(m)) {
1001		struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
1002		if (n == NULL) {
1003			m_freem(m);
1004			return;
1005		}
1006		(void)memcpy(mtod(n, void *), data, data_len);
1007		n->m_len = data_len;
1008		m->m_next = n;
1009	} else if (data_len > 0) {
1010		(void)memcpy(mtod(m, uint8_t *) + m->m_len, data, data_len);
1011		m->m_len += data_len;
1012	}
1013	if (m->m_flags & M_PKTHDR)
1014		m->m_pkthdr.len += data_len;
1015	mtod(m, struct if_xannouncemsghdr *)->ifan_msglen += data_len;
1016	COMPATNAME(route_enqueue)(m, 0);
1017}
1018
1019/*
1020 * This is used in dumping the kernel table via sysctl().
1021 */
1022static int
1023sysctl_dumpentry(struct rtentry *rt, void *v)
1024{
1025	struct rt_walkarg *w = v;
1026	int error = 0, size;
1027	struct rt_addrinfo info;
1028
1029	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1030		return 0;
1031	memset(&info, 0, sizeof(info));
1032	info.rti_info[RTAX_DST] = rt_getkey(rt);
1033	info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1034	info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1035	if (rt->rt_ifp) {
1036		const struct ifaddr *rtifa;
1037		info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr;
1038		/* rtifa used to be simply rt->rt_ifa.  If rt->rt_ifa != NULL,
1039		 * then rt_get_ifa() != NULL.  So this ought to still be safe.
1040		 * --dyoung
1041		 */
1042		rtifa = rt_get_ifa(rt);
1043		info.rti_info[RTAX_IFA] = rtifa->ifa_addr;
1044		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1045			info.rti_info[RTAX_BRD] = rtifa->ifa_dstaddr;
1046	}
1047	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
1048		return error;
1049	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1050		struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)w->w_tmem;
1051
1052		rtm->rtm_flags = rt->rt_flags;
1053		rtm->rtm_use = rt->rt_use;
1054		rtm_setmetrics(rt, rtm);
1055		KASSERT(rt->rt_ifp != NULL);
1056		rtm->rtm_index = rt->rt_ifp->if_index;
1057		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1058		rtm->rtm_addrs = info.rti_addrs;
1059		if ((error = copyout(rtm, w->w_where, size)) != 0)
1060			w->w_where = NULL;
1061		else
1062			w->w_where = (char *)w->w_where + size;
1063	}
1064	return error;
1065}
1066
1067static int
1068sysctl_iflist(int af, struct rt_walkarg *w, int type)
1069{
1070	struct ifnet *ifp;
1071	struct ifaddr *ifa;
1072	struct	rt_addrinfo info;
1073	int	len, error = 0;
1074
1075	memset(&info, 0, sizeof(info));
1076	IFNET_FOREACH(ifp) {
1077		if (w->w_arg && w->w_arg != ifp->if_index)
1078			continue;
1079		if (IFADDR_EMPTY(ifp))
1080			continue;
1081		info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
1082		switch (type) {
1083		case NET_RT_IFLIST:
1084			error = rt_msg2(RTM_IFINFO, &info, NULL, w, &len);
1085			break;
1086#ifdef COMPAT_14
1087		case NET_RT_OOIFLIST:
1088			error = rt_msg2(RTM_OOIFINFO, &info, NULL, w, &len);
1089			break;
1090#endif
1091#ifdef COMPAT_50
1092		case NET_RT_OIFLIST:
1093			error = rt_msg2(RTM_OIFINFO, &info, NULL, w, &len);
1094			break;
1095#endif
1096		default:
1097			panic("sysctl_iflist(1)");
1098		}
1099		if (error)
1100			return error;
1101		info.rti_info[RTAX_IFP] = NULL;
1102		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1103			switch (type) {
1104			case NET_RT_IFLIST: {
1105				struct if_xmsghdr *ifm;
1106
1107				ifm = (struct if_xmsghdr *)w->w_tmem;
1108				ifm->ifm_index = ifp->if_index;
1109				ifm->ifm_flags = ifp->if_flags;
1110				ifm->ifm_data = ifp->if_data;
1111				ifm->ifm_addrs = info.rti_addrs;
1112				error = copyout(ifm, w->w_where, len);
1113				if (error)
1114					return error;
1115				w->w_where = (char *)w->w_where + len;
1116				break;
1117			}
1118
1119#ifdef COMPAT_14
1120			case NET_RT_OOIFLIST:
1121				error = compat_14_iflist(ifp, w, &info, len);
1122				if (error)
1123					return error;
1124				break;
1125#endif
1126#ifdef COMPAT_50
1127			case NET_RT_OIFLIST:
1128				error = compat_50_iflist(ifp, w, &info, len);
1129				if (error)
1130					return error;
1131				break;
1132#endif
1133			default:
1134				panic("sysctl_iflist(2)");
1135			}
1136		}
1137		IFADDR_FOREACH(ifa, ifp) {
1138			if (af && af != ifa->ifa_addr->sa_family)
1139				continue;
1140			info.rti_info[RTAX_IFA] = ifa->ifa_addr;
1141			info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1142			info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1143			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
1144				return error;
1145			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1146				struct ifa_xmsghdr *ifam;
1147
1148				ifam = (struct ifa_xmsghdr *)w->w_tmem;
1149				ifam->ifam_index = ifa->ifa_ifp->if_index;
1150				ifam->ifam_flags = ifa->ifa_flags;
1151				ifam->ifam_metric = ifa->ifa_metric;
1152				ifam->ifam_addrs = info.rti_addrs;
1153				error = copyout(w->w_tmem, w->w_where, len);
1154				if (error)
1155					return error;
1156				w->w_where = (char *)w->w_where + len;
1157			}
1158		}
1159		info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] =
1160		    info.rti_info[RTAX_BRD] = NULL;
1161	}
1162	return 0;
1163}
1164
1165static int
1166sysctl_rtable(SYSCTLFN_ARGS)
1167{
1168	void 	*where = oldp;
1169	size_t	*given = oldlenp;
1170	const void *new = newp;
1171	int	i, s, error = EINVAL;
1172	u_char  af;
1173	struct	rt_walkarg w;
1174
1175	if (namelen == 1 && name[0] == CTL_QUERY)
1176		return sysctl_query(SYSCTLFN_CALL(rnode));
1177
1178	if (new)
1179		return EPERM;
1180	if (namelen != 3)
1181		return EINVAL;
1182	af = name[0];
1183	w.w_tmemneeded = 0;
1184	w.w_tmemsize = 0;
1185	w.w_tmem = NULL;
1186again:
1187	/* we may return here if a later [re]alloc of the t_mem buffer fails */
1188	if (w.w_tmemneeded) {
1189		w.w_tmem = malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1190		w.w_tmemsize = w.w_tmemneeded;
1191		w.w_tmemneeded = 0;
1192	}
1193	w.w_op = name[1];
1194	w.w_arg = name[2];
1195	w.w_given = *given;
1196	w.w_needed = 0 - w.w_given;
1197	w.w_where = where;
1198
1199	s = splsoftnet();
1200	switch (w.w_op) {
1201
1202	case NET_RT_DUMP:
1203	case NET_RT_FLAGS:
1204		for (i = 1; i <= AF_MAX; i++)
1205			if ((af == 0 || af == i) &&
1206			    (error = rt_walktree(i, sysctl_dumpentry, &w)))
1207				break;
1208		break;
1209
1210#ifdef COMPAT_14
1211	case NET_RT_OOIFLIST:
1212		error = sysctl_iflist(af, &w, w.w_op);
1213		break;
1214#endif
1215#ifdef COMPAT_50
1216	case NET_RT_OIFLIST:
1217		error = sysctl_iflist(af, &w, w.w_op);
1218		break;
1219#endif
1220	case NET_RT_IFLIST:
1221		error = sysctl_iflist(af, &w, w.w_op);
1222		break;
1223	}
1224	splx(s);
1225
1226	/* check to see if we couldn't allocate memory with NOWAIT */
1227	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1228		goto again;
1229
1230	if (w.w_tmem)
1231		free(w.w_tmem, M_RTABLE);
1232	w.w_needed += w.w_given;
1233	if (where) {
1234		*given = (char *)w.w_where - (char *)where;
1235		if (*given < w.w_needed)
1236			return ENOMEM;
1237	} else {
1238		*given = (11 * w.w_needed) / 10;
1239	}
1240	return error;
1241}
1242
1243/*
1244 * Routing message software interrupt routine
1245 */
1246static void
1247COMPATNAME(route_intr)(void *cookie)
1248{
1249	struct sockproto proto = { .sp_family = PF_XROUTE, };
1250	struct route_info * const ri = &COMPATNAME(route_info);
1251	struct mbuf *m;
1252	int s;
1253
1254	mutex_enter(softnet_lock);
1255	KERNEL_LOCK(1, NULL);
1256	while (!IF_IS_EMPTY(&ri->ri_intrq)) {
1257		s = splnet();
1258		IF_DEQUEUE(&ri->ri_intrq, m);
1259		splx(s);
1260		if (m == NULL)
1261			break;
1262		proto.sp_protocol = M_GETCTX(m, uintptr_t);
1263		raw_input(m, &proto, &ri->ri_src, &ri->ri_dst);
1264	}
1265	KERNEL_UNLOCK_ONE(NULL);
1266	mutex_exit(softnet_lock);
1267}
1268
1269/*
1270 * Enqueue a message to the software interrupt routine.
1271 */
1272void
1273COMPATNAME(route_enqueue)(struct mbuf *m, int family)
1274{
1275	struct route_info * const ri = &COMPATNAME(route_info);
1276	int s, wasempty;
1277
1278	s = splnet();
1279	if (IF_QFULL(&ri->ri_intrq)) {
1280		IF_DROP(&ri->ri_intrq);
1281		m_freem(m);
1282	} else {
1283		wasempty = IF_IS_EMPTY(&ri->ri_intrq);
1284		M_SETCTX(m, (uintptr_t)family);
1285		IF_ENQUEUE(&ri->ri_intrq, m);
1286		if (wasempty)
1287			softint_schedule(ri->ri_sih);
1288	}
1289	splx(s);
1290}
1291
1292static void
1293COMPATNAME(route_init)(void)
1294{
1295	struct route_info * const ri = &COMPATNAME(route_info);
1296
1297#ifndef COMPAT_RTSOCK
1298	rt_init();
1299#endif
1300
1301	sysctl_net_route_setup(NULL);
1302	ri->ri_intrq.ifq_maxlen = ri->ri_maxqlen;
1303	ri->ri_sih = softint_establish(SOFTINT_NET | SOFTINT_MPSAFE,
1304	    COMPATNAME(route_intr), NULL);
1305}
1306
1307/*
1308 * Definitions of protocols supported in the ROUTE domain.
1309 */
1310#ifndef COMPAT_RTSOCK
1311PR_WRAP_USRREQ(route_usrreq);
1312#else
1313PR_WRAP_USRREQ(compat_50_route_usrreq);
1314#endif
1315
1316static const struct protosw COMPATNAME(route_protosw)[] = {
1317	{
1318		.pr_type = SOCK_RAW,
1319		.pr_domain = &COMPATNAME(routedomain),
1320		.pr_flags = PR_ATOMIC|PR_ADDR,
1321		.pr_input = raw_input,
1322		.pr_output = COMPATNAME(route_output),
1323		.pr_ctlinput = raw_ctlinput,
1324		.pr_usrreq = COMPATNAME(route_usrreq_wrapper),
1325		.pr_init = raw_init,
1326	},
1327};
1328
1329struct domain COMPATNAME(routedomain) = {
1330	.dom_family = PF_XROUTE,
1331	.dom_name = DOMAINNAME,
1332	.dom_init = COMPATNAME(route_init),
1333	.dom_protosw = COMPATNAME(route_protosw),
1334	.dom_protoswNPROTOSW =
1335	    &COMPATNAME(route_protosw)[__arraycount(COMPATNAME(route_protosw))],
1336};
1337
1338static void
1339sysctl_net_route_setup(struct sysctllog **clog)
1340{
1341	const struct sysctlnode *rnode = NULL;
1342
1343	sysctl_createv(clog, 0, NULL, NULL,
1344		       CTLFLAG_PERMANENT,
1345		       CTLTYPE_NODE, "net", NULL,
1346		       NULL, 0, NULL, 0,
1347		       CTL_NET, CTL_EOL);
1348
1349	sysctl_createv(clog, 0, NULL, &rnode,
1350		       CTLFLAG_PERMANENT,
1351		       CTLTYPE_NODE, DOMAINNAME,
1352		       SYSCTL_DESCR("PF_ROUTE information"),
1353		       NULL, 0, NULL, 0,
1354		       CTL_NET, PF_XROUTE, CTL_EOL);
1355
1356	sysctl_createv(clog, 0, NULL, NULL,
1357		       CTLFLAG_PERMANENT,
1358		       CTLTYPE_NODE, "rtable",
1359		       SYSCTL_DESCR("Routing table information"),
1360		       sysctl_rtable, 0, NULL, 0,
1361		       CTL_NET, PF_XROUTE, 0 /* any protocol */, CTL_EOL);
1362
1363	sysctl_createv(clog, 0, &rnode, NULL,
1364		       CTLFLAG_PERMANENT,
1365		       CTLTYPE_STRUCT, "stats",
1366		       SYSCTL_DESCR("Routing statistics"),
1367		       NULL, 0, &rtstat, sizeof(rtstat),
1368		       CTL_CREATE, CTL_EOL);
1369}
1370