2 * Copyright (c) 1980, 1986, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)if.c 8.3 (Berkeley) 1/4/94
34 * $FreeBSD: src/sys/net/if.c,v 1.185 2004/03/13 02:35:03 brooks Exp $
35 * $DragonFly: src/sys/net/if.c,v 1.18 2004/06/04 07:45:45 hmp Exp $
38 #include "opt_compat.h"
39 #include "opt_inet6.h"
42 #include <sys/param.h>
43 #include <sys/malloc.h>
45 #include <sys/systm.h>
47 #include <sys/protosw.h>
48 #include <sys/socket.h>
49 #include <sys/socketvar.h>
50 #include <sys/socketops.h>
51 #include <sys/protosw.h>
52 #include <sys/kernel.h>
53 #include <sys/sockio.h>
54 #include <sys/syslog.h>
55 #include <sys/sysctl.h>
58 #include <net/if_arp.h>
59 #include <net/if_dl.h>
60 #include <net/if_types.h>
61 #include <net/if_var.h>
62 #include <net/radix.h>
63 #include <net/route.h>
64 #include <machine/stdarg.h>
66 #if defined(INET) || defined(INET6)
68 #include <netinet/in.h>
69 #include <netinet/in_var.h>
70 #include <netinet/if_ether.h>
72 #include <machine/clock.h> /* XXX: temporal workaround for fxp issue */
73 #include <netinet6/in6_var.h>
74 #include <netinet6/in6_ifattach.h>
78 #if defined(COMPAT_43)
79 #include <emulation/43bsd/43bsd_socket.h>
80 #endif /* COMPAT_43 */
83 * System initialization
86 static int ifconf (u_long, caddr_t, struct thread *);
87 static void ifinit (void *);
88 static void if_qflush (struct ifqueue *);
89 static void if_slowtimo (void *);
90 static void link_rtrequest (int, struct rtentry *, struct rt_addrinfo *);
91 static int if_rtdel (struct radix_node *, void *);
93 SYSINIT(interfaces, SI_SUB_PROTO_IF, SI_ORDER_FIRST, ifinit, NULL)
95 MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
96 MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");
97 MALLOC_DEFINE(M_CLONE, "clone", "interface cloning framework");
99 int ifqmaxlen = IFQ_MAXLEN;
100 struct ifnethead ifnet; /* depend on static init XXX */
104 * XXX: declare here to avoid to include many inet6 related files..
105 * should be more generalized?
107 extern void nd6_setmtu (struct ifnet *);
110 struct if_clone *if_clone_lookup (const char *, int *);
111 int if_clone_list (struct if_clonereq *);
113 LIST_HEAD(, if_clone) if_cloners = LIST_HEAD_INITIALIZER(if_cloners);
114 int if_cloners_count;
117 * Network interface utility routines.
119 * Routines with ifa_ifwith* names take sockaddr *'s as
131 TAILQ_FOREACH(ifp, &ifnet, if_link) {
132 if (ifp->if_snd.ifq_maxlen == 0) {
133 if_printf(ifp, "XXX: driver didn't set ifq_maxlen\n");
134 ifp->if_snd.ifq_maxlen = ifqmaxlen;
142 struct ifaddr **ifnet_addrs;
143 struct ifnet **ifindex2ifnet = NULL;
147 * Attach an interface to the
148 * list of "active" interfaces.
154 unsigned socksize, ifasize;
155 int namelen, masklen;
156 struct sockaddr_dl *sdl;
158 static int if_indexlim = 8;
166 TAILQ_INSERT_TAIL(&ifnet, ifp, if_link);
167 ifp->if_index = ++if_index;
170 * The old code would work if the interface passed a pre-existing
171 * chain of ifaddrs to this code. We don't trust our callers to
172 * properly initialize the tailq, however, so we no longer allow
173 * this unlikely case.
175 TAILQ_INIT(&ifp->if_addrhead);
176 TAILQ_INIT(&ifp->if_prefixhead);
177 LIST_INIT(&ifp->if_multiaddrs);
178 getmicrotime(&ifp->if_lastchange);
179 if (ifnet_addrs == 0 || if_index >= if_indexlim) {
180 unsigned n = (if_indexlim <<= 1) * sizeof(ifa);
181 caddr_t q = malloc(n, M_IFADDR, M_WAITOK);
184 bcopy((caddr_t)ifnet_addrs, (caddr_t)q, n/2);
185 free((caddr_t)ifnet_addrs, M_IFADDR);
187 ifnet_addrs = (struct ifaddr **)q;
189 /* grow ifindex2ifnet */
190 n = if_indexlim * sizeof(struct ifnet *);
191 q = malloc(n, M_IFADDR, M_WAITOK);
194 bcopy((caddr_t)ifindex2ifnet, q, n/2);
195 free((caddr_t)ifindex2ifnet, M_IFADDR);
197 ifindex2ifnet = (struct ifnet **)q;
200 ifindex2ifnet[if_index] = ifp;
203 * create a Link Level name for this device
205 namelen = strlen(ifp->if_xname);
206 #define _offsetof(t, m) ((int)((caddr_t)&((t *)0)->m))
207 masklen = _offsetof(struct sockaddr_dl, sdl_data[0]) + namelen;
208 socksize = masklen + ifp->if_addrlen;
209 #define ROUNDUP(a) (1 + (((a) - 1) | (sizeof(long) - 1)))
210 if (socksize < sizeof(*sdl))
211 socksize = sizeof(*sdl);
212 socksize = ROUNDUP(socksize);
213 ifasize = sizeof(*ifa) + 2 * socksize;
214 ifa = (struct ifaddr *)malloc(ifasize, M_IFADDR, M_WAITOK);
216 bzero((caddr_t)ifa, ifasize);
217 sdl = (struct sockaddr_dl *)(ifa + 1);
218 sdl->sdl_len = socksize;
219 sdl->sdl_family = AF_LINK;
220 bcopy(ifp->if_xname, sdl->sdl_data, namelen);
221 sdl->sdl_nlen = namelen;
222 sdl->sdl_index = ifp->if_index;
223 sdl->sdl_type = ifp->if_type;
224 ifnet_addrs[if_index - 1] = ifa;
226 ifa->ifa_rtrequest = link_rtrequest;
227 ifa->ifa_addr = (struct sockaddr *)sdl;
228 sdl = (struct sockaddr_dl *)(socksize + (caddr_t)sdl);
229 ifa->ifa_netmask = (struct sockaddr *)sdl;
230 sdl->sdl_len = masklen;
232 sdl->sdl_data[--namelen] = 0xff;
233 TAILQ_INSERT_HEAD(&ifp->if_addrhead, ifa, ifa_link);
236 /* Announce the interface. */
237 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
241 * Detach an interface, removing it from the
242 * list of "active" interfaces.
249 struct radix_node_head *rnh;
254 * Remove routes and flush queues.
260 * Remove address from ifnet_addrs[] and maybe decrement if_index.
261 * Clean up all addresses.
263 ifnet_addrs[ifp->if_index - 1] = 0;
264 while (if_index > 0 && ifnet_addrs[if_index - 1] == 0)
267 for (ifa = TAILQ_FIRST(&ifp->if_addrhead); ifa;
268 ifa = TAILQ_FIRST(&ifp->if_addrhead)) {
270 /* XXX: Ugly!! ad hoc just for INET */
271 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET) {
272 struct ifaliasreq ifr;
274 bzero(&ifr, sizeof(ifr));
275 ifr.ifra_addr = *ifa->ifa_addr;
276 if (ifa->ifa_dstaddr)
277 ifr.ifra_broadaddr = *ifa->ifa_dstaddr;
278 if (in_control(NULL, SIOCDIFADDR, (caddr_t)&ifr, ifp,
284 if (ifa->ifa_addr && ifa->ifa_addr->sa_family == AF_INET6) {
286 /* ifp_addrhead is already updated */
290 TAILQ_REMOVE(&ifp->if_addrhead, ifa, ifa_link);
296 * Remove all IPv6 kernel structs related to ifp. This should be done
297 * before removing routing entries below, since IPv6 interface direct
298 * routes are expected to be removed by the IPv6-specific kernel API.
299 * Otherwise, the kernel will detect some inconsistency and bark it.
305 * Delete all remaining routes using this interface
306 * Unfortuneatly the only way to do this is to slog through
307 * the entire routing table looking for routes which point
308 * to this interface...oh well...
310 for (i = 1; i <= AF_MAX; i++) {
311 if ((rnh = rt_tables[i]) == NULL)
313 (void) rnh->rnh_walktree(rnh, if_rtdel, ifp);
316 /* Announce that the interface is gone. */
317 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
319 TAILQ_REMOVE(&ifnet, ifp, if_link);
324 * Delete Routes for a Network Interface
326 * Called for each routing entry via the rnh->rnh_walktree() call above
327 * to delete all route entries referencing a detaching network interface.
330 * rn pointer to node in the routing table
331 * arg argument passed to rnh->rnh_walktree() - detaching interface
335 * errno failed - reason indicated
340 struct radix_node *rn;
343 struct rtentry *rt = (struct rtentry *)rn;
344 struct ifnet *ifp = arg;
347 if (rt->rt_ifp == ifp) {
350 * Protect (sorta) against walktree recursion problems
353 if ((rt->rt_flags & RTF_UP) == 0)
356 err = rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
357 rt_mask(rt), rt->rt_flags,
358 (struct rtentry **) NULL);
360 log(LOG_WARNING, "if_rtdel: error %d\n", err);
368 * Create a clone network interface.
371 if_clone_create(name, len)
375 struct if_clone *ifc;
377 int wildcard, bytoff, bitoff;
381 ifc = if_clone_lookup(name, &unit);
385 if (ifunit(name) != NULL)
389 wildcard = (unit < 0);
391 * Find a free unit if none was given.
394 while ((bytoff < ifc->ifc_bmlen)
395 && (ifc->ifc_units[bytoff] == 0xff))
397 if (bytoff >= ifc->ifc_bmlen)
399 while ((ifc->ifc_units[bytoff] & (1 << bitoff)) != 0)
401 unit = (bytoff << 3) + bitoff;
404 if (unit > ifc->ifc_maxunit)
407 err = (*ifc->ifc_create)(ifc, unit);
413 bitoff = unit - (bytoff << 3);
417 * Allocate the unit in the bitmap.
419 KASSERT((ifc->ifc_units[bytoff] & (1 << bitoff)) == 0,
420 ("%s: bit is already set", __func__));
421 ifc->ifc_units[bytoff] |= (1 << bitoff);
423 /* In the wildcard case, we need to update the name. */
425 for (dp = name; *dp != '\0'; dp++);
426 if (snprintf(dp, len - (dp-name), "%d", unit) >
427 len - (dp-name) - 1) {
429 * This can only be a programmer error and
430 * there's no straightforward way to recover if
433 panic("if_clone_create(): interface name too long");
442 * Destroy a clone network interface.
445 if_clone_destroy(name)
448 struct if_clone *ifc;
453 ifc = if_clone_lookup(name, &unit);
457 if (unit < ifc->ifc_minifs)
464 if (ifc->ifc_destroy == NULL)
467 (*ifc->ifc_destroy)(ifp);
470 * Compute offset in the bitmap and deallocate the unit.
473 bitoff = unit - (bytoff << 3);
474 KASSERT((ifc->ifc_units[bytoff] & (1 << bitoff)) != 0,
475 ("%s: bit is already cleared", __func__));
476 ifc->ifc_units[bytoff] &= ~(1 << bitoff);
481 * Look up a network interface cloner.
484 if_clone_lookup(name, unitp)
488 struct if_clone *ifc;
492 for (ifc = LIST_FIRST(&if_cloners); ifc != NULL;) {
493 for (cp = name, i = 0; i < ifc->ifc_namelen; i++, cp++) {
494 if (ifc->ifc_name[i] != *cp)
499 ifc = LIST_NEXT(ifc, ifc_list);
503 return ((struct if_clone *)NULL);
509 for (i = 0; *cp != '\0'; cp++) {
510 if (*cp < '0' || *cp > '9') {
511 /* Bogus unit number. */
514 i = (i * 10) + (*cp - '0');
524 * Register a network interface cloner.
528 struct if_clone *ifc;
535 KASSERT(ifc->ifc_minifs - 1 <= ifc->ifc_maxunit,
536 ("%s: %s requested more units then allowed (%d > %d)",
537 __func__, ifc->ifc_name, ifc->ifc_minifs,
538 ifc->ifc_maxunit + 1));
540 * Compute bitmap size and allocate it.
542 maxclone = ifc->ifc_maxunit + 1;
544 if ((len << 3) < maxclone)
546 ifc->ifc_units = malloc(len, M_CLONE, M_WAITOK | M_ZERO);
547 ifc->ifc_bmlen = len;
549 LIST_INSERT_HEAD(&if_cloners, ifc, ifc_list);
552 for (unit = 0; unit < ifc->ifc_minifs; unit++) {
553 err = (*ifc->ifc_create)(ifc, unit);
555 ("%s: failed to create required interface %s%d",
556 __func__, ifc->ifc_name, unit));
558 /* Allocate the unit in the bitmap. */
560 bitoff = unit - (bytoff << 3);
561 ifc->ifc_units[bytoff] |= (1 << bitoff);
566 * Unregister a network interface cloner.
570 struct if_clone *ifc;
573 LIST_REMOVE(ifc, ifc_list);
574 free(ifc->ifc_units, M_CLONE);
579 * Provide list of interface cloners to userspace.
583 struct if_clonereq *ifcr;
585 char outbuf[IFNAMSIZ], *dst;
586 struct if_clone *ifc;
587 int count, error = 0;
589 ifcr->ifcr_total = if_cloners_count;
590 if ((dst = ifcr->ifcr_buffer) == NULL) {
591 /* Just asking how many there are. */
595 if (ifcr->ifcr_count < 0)
598 count = (if_cloners_count < ifcr->ifcr_count) ?
599 if_cloners_count : ifcr->ifcr_count;
601 for (ifc = LIST_FIRST(&if_cloners); ifc != NULL && count != 0;
602 ifc = LIST_NEXT(ifc, ifc_list), count--, dst += IFNAMSIZ) {
603 strlcpy(outbuf, ifc->ifc_name, IFNAMSIZ);
604 error = copyout(outbuf, dst, IFNAMSIZ);
613 * Locate an interface based on a complete address.
618 struct sockaddr *addr;
623 #define equal(a1, a2) \
624 (bcmp((caddr_t)(a1), (caddr_t)(a2), ((struct sockaddr *)(a1))->sa_len) == 0)
625 TAILQ_FOREACH(ifp, &ifnet, if_link)
626 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
627 if (ifa->ifa_addr->sa_family != addr->sa_family)
629 if (equal(addr, ifa->ifa_addr))
631 if ((ifp->if_flags & IFF_BROADCAST) && ifa->ifa_broadaddr &&
632 /* IP6 doesn't have broadcast */
633 ifa->ifa_broadaddr->sa_len != 0 &&
634 equal(ifa->ifa_broadaddr, addr))
637 return ((struct ifaddr *)0);
640 * Locate the point to point interface with a given destination address.
644 ifa_ifwithdstaddr(addr)
645 struct sockaddr *addr;
650 TAILQ_FOREACH(ifp, &ifnet, if_link)
651 if (ifp->if_flags & IFF_POINTOPOINT)
652 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
653 if (ifa->ifa_addr->sa_family != addr->sa_family)
655 if (ifa->ifa_dstaddr && equal(addr, ifa->ifa_dstaddr))
658 return ((struct ifaddr *)0);
662 * Find an interface on a specific network. If many, choice
663 * is most specific found.
667 struct sockaddr *addr;
671 struct ifaddr *ifa_maybe = (struct ifaddr *) 0;
672 u_int af = addr->sa_family;
673 char *addr_data = addr->sa_data, *cplim;
676 * AF_LINK addresses can be looked up directly by their index number,
677 * so do that if we can.
680 struct sockaddr_dl *sdl = (struct sockaddr_dl *)addr;
681 if (sdl->sdl_index && sdl->sdl_index <= if_index)
682 return (ifnet_addrs[sdl->sdl_index - 1]);
686 * Scan though each interface, looking for ones that have
687 * addresses in this address family.
689 TAILQ_FOREACH(ifp, &ifnet, if_link) {
690 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
691 char *cp, *cp2, *cp3;
693 if (ifa->ifa_addr->sa_family != af)
695 if (af == AF_INET && ifp->if_flags & IFF_POINTOPOINT) {
697 * This is a bit broken as it doesn't
698 * take into account that the remote end may
699 * be a single node in the network we are
701 * The trouble is that we don't know the
702 * netmask for the remote end.
704 if (ifa->ifa_dstaddr != 0
705 && equal(addr, ifa->ifa_dstaddr))
709 * if we have a special address handler,
710 * then use it instead of the generic one.
712 if (ifa->ifa_claim_addr) {
713 if ((*ifa->ifa_claim_addr)(ifa, addr)) {
721 * Scan all the bits in the ifa's address.
722 * If a bit dissagrees with what we are
723 * looking for, mask it with the netmask
724 * to see if it really matters.
727 if (ifa->ifa_netmask == 0)
730 cp2 = ifa->ifa_addr->sa_data;
731 cp3 = ifa->ifa_netmask->sa_data;
732 cplim = ifa->ifa_netmask->sa_len
733 + (char *)ifa->ifa_netmask;
735 if ((*cp++ ^ *cp2++) & *cp3++)
736 goto next; /* next address! */
738 * If the netmask of what we just found
739 * is more specific than what we had before
740 * (if we had one) then remember the new one
741 * before continuing to search
742 * for an even better one.
744 if (ifa_maybe == 0 ||
745 rn_refines((caddr_t)ifa->ifa_netmask,
746 (caddr_t)ifa_maybe->ifa_netmask))
755 * Find an interface address specific to an interface best matching
759 ifaof_ifpforaddr(addr, ifp)
760 struct sockaddr *addr;
764 char *cp, *cp2, *cp3;
766 struct ifaddr *ifa_maybe = 0;
767 u_int af = addr->sa_family;
771 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
772 if (ifa->ifa_addr->sa_family != af)
776 if (ifa->ifa_netmask == 0) {
777 if (equal(addr, ifa->ifa_addr) ||
778 (ifa->ifa_dstaddr && equal(addr, ifa->ifa_dstaddr)))
782 if (ifp->if_flags & IFF_POINTOPOINT) {
783 if (equal(addr, ifa->ifa_dstaddr))
787 cp2 = ifa->ifa_addr->sa_data;
788 cp3 = ifa->ifa_netmask->sa_data;
789 cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask;
790 for (; cp3 < cplim; cp3++)
791 if ((*cp++ ^ *cp2++) & *cp3)
800 #include <net/route.h>
803 * Default action when installing a route with a Link Level gateway.
804 * Lookup an appropriate real ifa to point to.
805 * This should be moved to /sys/net/link.c eventually.
808 link_rtrequest(cmd, rt, info)
811 struct rt_addrinfo *info;
814 struct sockaddr *dst;
817 if (cmd != RTM_ADD || ((ifa = rt->rt_ifa) == 0) ||
818 ((ifp = ifa->ifa_ifp) == 0) || ((dst = rt_key(rt)) == 0))
820 ifa = ifaof_ifpforaddr(dst, ifp);
825 if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest)
826 ifa->ifa_rtrequest(cmd, rt, info);
831 * Mark an interface down and notify protocols of
833 * NOTE: must be called at splnet or eqivalent.
836 if_unroute(ifp, flag, fam)
842 ifp->if_flags &= ~flag;
843 getmicrotime(&ifp->if_lastchange);
844 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
845 if (fam == PF_UNSPEC || (fam == ifa->ifa_addr->sa_family))
846 pfctlinput(PRC_IFDOWN, ifa->ifa_addr);
847 if_qflush(&ifp->if_snd);
852 * Mark an interface up and notify protocols of
854 * NOTE: must be called at splnet or eqivalent.
857 if_route(ifp, flag, fam)
863 ifp->if_flags |= flag;
864 getmicrotime(&ifp->if_lastchange);
865 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link)
866 if (fam == PF_UNSPEC || (fam == ifa->ifa_addr->sa_family))
867 pfctlinput(PRC_IFUP, ifa->ifa_addr);
875 * Mark an interface down and notify protocols of
877 * NOTE: must be called at splnet or eqivalent.
884 if_unroute(ifp, IFF_UP, AF_UNSPEC);
888 * Mark an interface up and notify protocols of
890 * NOTE: must be called at splnet or eqivalent.
897 if_route(ifp, IFF_UP, AF_UNSPEC);
901 * Flush an interface queue.
910 while ((m = n) != 0) {
920 * Handle interface watchdog timer routines. Called
921 * from softclock, we decrement timers (if set) and
922 * call the appropriate interface routine on expiration.
931 TAILQ_FOREACH(ifp, &ifnet, if_link) {
932 if (ifp->if_timer == 0 || --ifp->if_timer)
934 if (ifp->if_watchdog)
935 (*ifp->if_watchdog)(ifp);
938 timeout(if_slowtimo, (void *)0, hz / IFNET_SLOWHZ);
942 * Map interface name to
943 * interface structure pointer.
946 ifunit(const char *name)
951 * Search all the interfaces for this name/number
954 TAILQ_FOREACH(ifp, &ifnet, if_link) {
955 if (strncmp(ifp->if_xname, name, IFNAMSIZ) == 0)
963 * Map interface name in a sockaddr_dl to
964 * interface structure pointer.
970 char ifname[IFNAMSIZ+1];
971 struct sockaddr_dl *sdl = (struct sockaddr_dl *)sa;
973 if ( (sa->sa_family != AF_LINK) || (sdl->sdl_nlen == 0) ||
974 (sdl->sdl_nlen > IFNAMSIZ) )
978 * ifunit wants a null-terminated name. It may not be null-terminated
979 * in the sockaddr. We don't want to change the caller's sockaddr,
980 * and there might not be room to put the trailing null anyway, so we
981 * make a local copy that we know we can null terminate safely.
984 bcopy(sdl->sdl_data, ifname, sdl->sdl_nlen);
985 ifname[sdl->sdl_nlen] = '\0';
986 return ifunit(ifname);
994 ifioctl(struct socket *so, u_long cmd, caddr_t data, struct thread *td)
1002 size_t namelen, onamelen;
1003 char new_name[IFNAMSIZ];
1005 struct sockaddr_dl *sdl;
1011 return (ifconf(cmd, data, td));
1013 ifr = (struct ifreq *)data;
1018 if ((error = suser(td)) != 0)
1020 return ((cmd == SIOCIFCREATE) ?
1021 if_clone_create(ifr->ifr_name, sizeof(ifr->ifr_name)) :
1022 if_clone_destroy(ifr->ifr_name));
1024 case SIOCIFGCLONERS:
1025 return (if_clone_list((struct if_clonereq *)data));
1028 ifp = ifunit(ifr->ifr_name);
1034 ifr->ifr_flags = ifp->if_flags;
1035 ifr->ifr_flagshigh = ifp->if_flags >> 16;
1039 ifr->ifr_reqcap = ifp->if_capabilities;
1040 ifr->ifr_curcap = ifp->if_capenable;
1044 ifr->ifr_metric = ifp->if_metric;
1048 ifr->ifr_mtu = ifp->if_mtu;
1052 ifr->ifr_phys = ifp->if_physical;
1059 new_flags = (ifr->ifr_flags & 0xffff) |
1060 (ifr->ifr_flagshigh << 16);
1061 if (ifp->if_flags & IFF_SMART) {
1062 /* Smart drivers twiddle their own routes */
1063 } else if (ifp->if_flags & IFF_UP &&
1064 (new_flags & IFF_UP) == 0) {
1068 } else if (new_flags & IFF_UP &&
1069 (ifp->if_flags & IFF_UP) == 0) {
1074 ifp->if_flags = (ifp->if_flags & IFF_CANTCHANGE) |
1075 (new_flags &~ IFF_CANTCHANGE);
1076 if (new_flags & IFF_PPROMISC) {
1077 /* Permanently promiscuous mode requested */
1078 ifp->if_flags |= IFF_PROMISC;
1079 } else if (ifp->if_pcount == 0) {
1080 ifp->if_flags &= ~IFF_PROMISC;
1083 (void) (*ifp->if_ioctl)(ifp, cmd, data,
1084 td->td_proc->p_ucred);
1085 getmicrotime(&ifp->if_lastchange);
1092 if (ifr->ifr_reqcap & ~ifp->if_capabilities)
1094 (void) (*ifp->if_ioctl)(ifp, cmd, data, td->td_proc->p_ucred);
1101 error = copyinstr(ifr->ifr_data, new_name, IFNAMSIZ, NULL);
1104 if (new_name[0] == '\0')
1106 if (ifunit(new_name) != NULL)
1109 /* Announce the departure of the interface. */
1110 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
1112 strlcpy(ifp->if_xname, new_name, sizeof(ifp->if_xname));
1113 ifa = TAILQ_FIRST(&ifp->if_addrhead);
1114 /* XXX IFA_LOCK(ifa); */
1115 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
1116 namelen = strlen(new_name);
1117 onamelen = sdl->sdl_nlen;
1119 * Move the address if needed. This is safe because we
1120 * allocate space for a name of length IFNAMSIZ when we
1121 * create this in if_attach().
1123 if (namelen != onamelen) {
1124 bcopy(sdl->sdl_data + onamelen,
1125 sdl->sdl_data + namelen, sdl->sdl_alen);
1127 bcopy(new_name, sdl->sdl_data, namelen);
1128 sdl->sdl_nlen = namelen;
1129 sdl = (struct sockaddr_dl *)ifa->ifa_netmask;
1130 bzero(sdl->sdl_data, onamelen);
1131 while (namelen != 0)
1132 sdl->sdl_data[--namelen] = 0xff;
1133 /* XXX IFA_UNLOCK(ifa) */
1135 /* Announce the return of the interface. */
1136 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
1143 ifp->if_metric = ifr->ifr_metric;
1144 getmicrotime(&ifp->if_lastchange);
1153 error = (*ifp->if_ioctl)(ifp, cmd, data, td->td_proc->p_ucred);
1155 getmicrotime(&ifp->if_lastchange);
1160 u_long oldmtu = ifp->if_mtu;
1165 if (ifp->if_ioctl == NULL)
1166 return (EOPNOTSUPP);
1167 if (ifr->ifr_mtu < IF_MINMTU || ifr->ifr_mtu > IF_MAXMTU)
1169 error = (*ifp->if_ioctl)(ifp, cmd, data, td->td_proc->p_ucred);
1171 getmicrotime(&ifp->if_lastchange);
1175 * If the link MTU changed, do network layer specific procedure.
1177 if (ifp->if_mtu != oldmtu) {
1191 /* Don't allow group membership on non-multicast interfaces. */
1192 if ((ifp->if_flags & IFF_MULTICAST) == 0)
1195 /* Don't let users screw up protocols' entries. */
1196 if (ifr->ifr_addr.sa_family != AF_LINK)
1199 if (cmd == SIOCADDMULTI) {
1200 struct ifmultiaddr *ifma;
1201 error = if_addmulti(ifp, &ifr->ifr_addr, &ifma);
1203 error = if_delmulti(ifp, &ifr->ifr_addr);
1206 getmicrotime(&ifp->if_lastchange);
1209 case SIOCSIFPHYADDR:
1210 case SIOCDIFPHYADDR:
1212 case SIOCSIFPHYADDR_IN6:
1214 case SIOCSLIFPHYADDR:
1216 case SIOCSIFGENERIC:
1220 if (ifp->if_ioctl == 0)
1221 return (EOPNOTSUPP);
1222 error = (*ifp->if_ioctl)(ifp, cmd, data, td->td_proc->p_ucred);
1224 getmicrotime(&ifp->if_lastchange);
1228 ifs = (struct ifstat *)data;
1229 ifs->ascii[0] = '\0';
1231 case SIOCGIFPSRCADDR:
1232 case SIOCGIFPDSTADDR:
1233 case SIOCGLIFPHYADDR:
1235 case SIOCGIFGENERIC:
1236 if (ifp->if_ioctl == 0)
1237 return (EOPNOTSUPP);
1238 return ((*ifp->if_ioctl)(ifp, cmd, data, td->td_proc->p_ucred));
1244 return if_setlladdr(ifp,
1245 ifr->ifr_addr.sa_data, ifr->ifr_addr.sa_len);
1248 oif_flags = ifp->if_flags;
1249 if (so->so_proto == 0)
1250 return (EOPNOTSUPP);
1252 error = so_pru_control(so, cmd, data, ifp, td);
1259 case SIOCSIFDSTADDR:
1261 case SIOCSIFBRDADDR:
1262 case SIOCSIFNETMASK:
1263 #if BYTE_ORDER != BIG_ENDIAN
1264 if (ifr->ifr_addr.sa_family == 0 &&
1265 ifr->ifr_addr.sa_len < 16) {
1266 ifr->ifr_addr.sa_family = ifr->ifr_addr.sa_len;
1267 ifr->ifr_addr.sa_len = 16;
1270 if (ifr->ifr_addr.sa_len == 0)
1271 ifr->ifr_addr.sa_len = 16;
1279 case OSIOCGIFDSTADDR:
1280 cmd = SIOCGIFDSTADDR;
1283 case OSIOCGIFBRDADDR:
1284 cmd = SIOCGIFBRDADDR;
1287 case OSIOCGIFNETMASK:
1288 cmd = SIOCGIFNETMASK;
1290 error = so_pru_control(so, cmd, data, ifp, td);
1294 case OSIOCGIFDSTADDR:
1295 case OSIOCGIFBRDADDR:
1296 case OSIOCGIFNETMASK:
1297 *(u_short *)&ifr->ifr_addr = ifr->ifr_addr.sa_family;
1301 #endif /* COMPAT_43 */
1303 if ((oif_flags ^ ifp->if_flags) & IFF_UP) {
1305 DELAY(100);/* XXX: temporary workaround for fxp issue*/
1306 if (ifp->if_flags & IFF_UP) {
1320 * Set/clear promiscuous mode on interface ifp based on the truth value
1321 * of pswitch. The calls are reference counted so that only the first
1322 * "on" request actually has an effect, as does the final "off" request.
1323 * Results are undefined if the "off" and "on" requests are not matched.
1326 ifpromisc(ifp, pswitch)
1334 oldflags = ifp->if_flags;
1335 if (ifp->if_flags & IFF_PPROMISC) {
1336 /* Do nothing if device is in permanently promiscuous mode */
1337 ifp->if_pcount += pswitch ? 1 : -1;
1342 * If the device is not configured up, we cannot put it in
1345 if ((ifp->if_flags & IFF_UP) == 0)
1347 if (ifp->if_pcount++ != 0)
1349 ifp->if_flags |= IFF_PROMISC;
1350 log(LOG_INFO, "%s: promiscuous mode enabled\n",
1353 if (--ifp->if_pcount > 0)
1355 ifp->if_flags &= ~IFF_PROMISC;
1356 log(LOG_INFO, "%s: promiscuous mode disabled\n",
1359 ifr.ifr_flags = ifp->if_flags;
1360 ifr.ifr_flagshigh = ifp->if_flags >> 16;
1361 error = (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t)&ifr,
1362 (struct ucred *)NULL);
1366 ifp->if_flags = oldflags;
1371 * Return interface configuration
1372 * of system. List may be used
1373 * in later ioctl's (above) to get
1374 * other information.
1378 ifconf(u_long cmd, caddr_t data, struct thread *td)
1380 struct ifconf *ifc = (struct ifconf *)data;
1383 struct sockaddr *sa;
1384 struct ifreq ifr, *ifrp;
1385 int space = ifc->ifc_len, error = 0;
1387 ifrp = ifc->ifc_req;
1388 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1391 if (space <= sizeof (ifr))
1393 if (strlcpy(ifr.ifr_name, ifp->if_xname, sizeof(ifr.ifr_name))
1394 >= sizeof(ifr.ifr_name)) {
1395 error = ENAMETOOLONG;
1400 ifa = ifp->if_addrhead.tqh_first;
1401 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1402 if (space <= sizeof(ifr))
1405 if (td->td_proc->p_ucred->cr_prison &&
1410 if (cmd == OSIOCGIFCONF) {
1411 struct osockaddr *osa =
1412 (struct osockaddr *)&ifr.ifr_addr;
1414 osa->sa_family = sa->sa_family;
1415 error = copyout((caddr_t)&ifr, (caddr_t)ifrp,
1420 if (sa->sa_len <= sizeof(*sa)) {
1422 error = copyout((caddr_t)&ifr, (caddr_t)ifrp,
1426 if (space < sizeof (ifr) + sa->sa_len -
1429 space -= sa->sa_len - sizeof(*sa);
1430 error = copyout((caddr_t)&ifr, (caddr_t)ifrp,
1431 sizeof (ifr.ifr_name));
1433 error = copyout((caddr_t)sa,
1434 (caddr_t)&ifrp->ifr_addr, sa->sa_len);
1435 ifrp = (struct ifreq *)
1436 (sa->sa_len + (caddr_t)&ifrp->ifr_addr);
1440 space -= sizeof (ifr);
1445 bzero((caddr_t)&ifr.ifr_addr, sizeof(ifr.ifr_addr));
1446 error = copyout((caddr_t)&ifr, (caddr_t)ifrp,
1450 space -= sizeof (ifr);
1454 ifc->ifc_len -= space;
1459 * Just like if_promisc(), but for all-multicast-reception mode.
1462 if_allmulti(ifp, onswitch)
1471 if (ifp->if_amcount++ == 0) {
1472 ifp->if_flags |= IFF_ALLMULTI;
1473 ifr.ifr_flags = ifp->if_flags;
1474 ifr.ifr_flagshigh = ifp->if_flags >> 16;
1475 error = ifp->if_ioctl(ifp, SIOCSIFFLAGS, (caddr_t)&ifr,
1476 (struct ucred *)NULL);
1479 if (ifp->if_amcount > 1) {
1482 ifp->if_amcount = 0;
1483 ifp->if_flags &= ~IFF_ALLMULTI;
1484 ifr.ifr_flags = ifp->if_flags;
1485 ifr.ifr_flagshigh = ifp->if_flags >> 16;
1486 error = ifp->if_ioctl(ifp, SIOCSIFFLAGS, (caddr_t)&ifr,
1487 (struct ucred *)NULL);
1498 * Add a multicast listenership to the interface in question.
1499 * The link layer provides a routine which converts
1502 if_addmulti(ifp, sa, retifma)
1503 struct ifnet *ifp; /* interface to manipulate */
1504 struct sockaddr *sa; /* address to add */
1505 struct ifmultiaddr **retifma;
1507 struct sockaddr *llsa, *dupsa;
1509 struct ifmultiaddr *ifma;
1512 * If the matching multicast address already exists
1513 * then don't add a new one, just add a reference
1515 LIST_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1516 if (equal(sa, ifma->ifma_addr)) {
1517 ifma->ifma_refcount++;
1525 * Give the link layer a chance to accept/reject it, and also
1526 * find out which AF_LINK address this maps to, if it isn't one
1529 if (ifp->if_resolvemulti) {
1530 error = ifp->if_resolvemulti(ifp, &llsa, sa);
1531 if (error) return error;
1536 MALLOC(ifma, struct ifmultiaddr *, sizeof *ifma, M_IFMADDR, M_WAITOK);
1537 MALLOC(dupsa, struct sockaddr *, sa->sa_len, M_IFMADDR, M_WAITOK);
1538 bcopy(sa, dupsa, sa->sa_len);
1540 ifma->ifma_addr = dupsa;
1541 ifma->ifma_lladdr = llsa;
1542 ifma->ifma_ifp = ifp;
1543 ifma->ifma_refcount = 1;
1544 ifma->ifma_protospec = 0;
1545 rt_newmaddrmsg(RTM_NEWMADDR, ifma);
1548 * Some network interfaces can scan the address list at
1549 * interrupt time; lock them out.
1552 LIST_INSERT_HEAD(&ifp->if_multiaddrs, ifma, ifma_link);
1557 LIST_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1558 if (equal(ifma->ifma_addr, llsa))
1562 ifma->ifma_refcount++;
1564 MALLOC(ifma, struct ifmultiaddr *, sizeof *ifma,
1565 M_IFMADDR, M_WAITOK);
1566 MALLOC(dupsa, struct sockaddr *, llsa->sa_len,
1567 M_IFMADDR, M_WAITOK);
1568 bcopy(llsa, dupsa, llsa->sa_len);
1569 ifma->ifma_addr = dupsa;
1570 ifma->ifma_ifp = ifp;
1571 ifma->ifma_refcount = 1;
1573 LIST_INSERT_HEAD(&ifp->if_multiaddrs, ifma, ifma_link);
1578 * We are certain we have added something, so call down to the
1579 * interface to let them know about it.
1582 ifp->if_ioctl(ifp, SIOCADDMULTI, 0, (struct ucred *)NULL);
1589 * Remove a reference to a multicast address on this interface. Yell
1590 * if the request does not match an existing membership.
1593 if_delmulti(ifp, sa)
1595 struct sockaddr *sa;
1597 struct ifmultiaddr *ifma;
1600 LIST_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link)
1601 if (equal(sa, ifma->ifma_addr))
1606 if (ifma->ifma_refcount > 1) {
1607 ifma->ifma_refcount--;
1611 rt_newmaddrmsg(RTM_DELMADDR, ifma);
1612 sa = ifma->ifma_lladdr;
1614 LIST_REMOVE(ifma, ifma_link);
1616 * Make sure the interface driver is notified
1617 * in the case of a link layer mcast group being left.
1619 if (ifma->ifma_addr->sa_family == AF_LINK && sa == 0)
1620 ifp->if_ioctl(ifp, SIOCDELMULTI, 0, (struct ucred *)NULL);
1622 free(ifma->ifma_addr, M_IFMADDR);
1623 free(ifma, M_IFMADDR);
1628 * Now look for the link-layer address which corresponds to
1629 * this network address. It had been squirreled away in
1630 * ifma->ifma_lladdr for this purpose (so we don't have
1631 * to call ifp->if_resolvemulti() again), and we saved that
1632 * value in sa above. If some nasty deleted the
1633 * link-layer address out from underneath us, we can deal because
1634 * the address we stored was is not the same as the one which was
1635 * in the record for the link-layer address. (So we don't complain
1638 LIST_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link)
1639 if (equal(sa, ifma->ifma_addr))
1644 if (ifma->ifma_refcount > 1) {
1645 ifma->ifma_refcount--;
1650 LIST_REMOVE(ifma, ifma_link);
1651 ifp->if_ioctl(ifp, SIOCDELMULTI, 0, (struct ucred *)NULL);
1653 free(ifma->ifma_addr, M_IFMADDR);
1654 free(sa, M_IFMADDR);
1655 free(ifma, M_IFMADDR);
1661 * Set the link layer address on an interface.
1663 * At this time we only support certain types of interfaces,
1664 * and we don't allow the length of the address to change.
1667 if_setlladdr(struct ifnet *ifp, const u_char *lladdr, int len)
1669 struct sockaddr_dl *sdl;
1673 ifa = ifnet_addrs[ifp->if_index - 1];
1676 sdl = (struct sockaddr_dl *)ifa->ifa_addr;
1679 if (len != sdl->sdl_alen) /* don't allow length to change */
1681 switch (ifp->if_type) {
1682 case IFT_ETHER: /* these types use struct arpcom */
1687 bcopy(lladdr, ((struct arpcom *)ifp->if_softc)->ac_enaddr, len);
1690 bcopy(lladdr, LLADDR(sdl), len);
1696 * If the interface is already up, we need
1697 * to re-init it in order to reprogram its
1700 if ((ifp->if_flags & IFF_UP) != 0) {
1701 ifp->if_flags &= ~IFF_UP;
1702 ifr.ifr_flags = ifp->if_flags;
1703 ifr.ifr_flagshigh = ifp->if_flags >> 16;
1704 (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t)&ifr,
1705 (struct ucred *)NULL);
1706 ifp->if_flags |= IFF_UP;
1707 ifr.ifr_flags = ifp->if_flags;
1708 ifr.ifr_flagshigh = ifp->if_flags >> 16;
1709 (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t)&ifr,
1710 (struct ucred *)NULL);
1713 * Also send gratuitous ARPs to notify other nodes about
1714 * the address change.
1716 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
1717 if (ifa->ifa_addr != NULL &&
1718 ifa->ifa_addr->sa_family == AF_INET)
1719 arp_ifinit(ifp, ifa);
1726 struct ifmultiaddr *
1727 ifmaof_ifpforaddr(sa, ifp)
1728 struct sockaddr *sa;
1731 struct ifmultiaddr *ifma;
1733 LIST_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link)
1734 if (equal(ifma->ifma_addr, sa))
1741 * The name argument must be a pointer to storage which will last as
1742 * long as the interface does. For physical devices, the result of
1743 * device_get_name(dev) is a good choice and for pseudo-devices a
1744 * static string works well.
1747 if_initname(struct ifnet *ifp, const char *name, int unit)
1749 ifp->if_dname = name;
1750 ifp->if_dunit = unit;
1751 if (unit != IF_DUNIT_NONE)
1752 snprintf(ifp->if_xname, IFNAMSIZ, "%s%d", name, unit);
1754 strlcpy(ifp->if_xname, name, IFNAMSIZ);
1758 if_printf(struct ifnet *ifp, const char *fmt, ...)
1763 retval = printf("%s: ", ifp->if_xname);
1764 __va_start(ap, fmt);
1765 retval += vprintf(fmt, ap);
1770 SYSCTL_NODE(_net, PF_LINK, link, CTLFLAG_RW, 0, "Link layers");
1771 SYSCTL_NODE(_net_link, 0, generic, CTLFLAG_RW, 0, "Generic link-management");