2 * Copyright 1994, 1995 Massachusetts Institute of Technology
4 * Permission to use, copy, modify, and distribute this software and
5 * its documentation for any purpose and without fee is hereby
6 * granted, provided that both the above copyright notice and this
7 * permission notice appear in all copies, that both the above
8 * copyright notice and this permission notice appear in all
9 * supporting documentation, and that the name of M.I.T. not be used
10 * in advertising or publicity pertaining to distribution of the
11 * software without specific, written prior permission. M.I.T. makes
12 * no representations about the suitability of this software for any
13 * purpose. It is provided "as is" without express or implied
16 * THIS SOFTWARE IS PROVIDED BY M.I.T. ``AS IS''. M.I.T. DISCLAIMS
17 * ALL EXPRESS OR IMPLIED WARRANTIES WITH REGARD TO THIS SOFTWARE,
18 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
19 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT
20 * SHALL M.I.T. BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
21 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
22 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
23 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
24 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
25 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
26 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 #include <sys/cdefs.h>
31 __FBSDID("$FreeBSD$");
33 #include "opt_mpath.h"
35 #include <sys/param.h>
36 #include <sys/systm.h>
37 #include <sys/kernel.h>
38 #include <sys/sysctl.h>
39 #include <sys/socket.h>
43 #include <net/if_var.h>
44 #include <net/route.h>
45 #include <net/route_var.h>
46 #include <net/route/nhop.h>
47 #include <net/route/shared.h>
50 #include <netinet/in.h>
51 #include <netinet/in_var.h>
52 #include <netinet/ip.h>
53 #include <netinet/ip_icmp.h>
54 #include <netinet/ip_var.h>
56 extern int in_inithead(void **head, int off, u_int fibnum);
58 extern int in_detachhead(void **head, int off);
62 rib4_preadd(u_int fibnum, const struct sockaddr *addr, const struct sockaddr *mask,
63 struct nhop_object *nh)
65 const struct sockaddr_in *addr4 = (const struct sockaddr_in *)addr;
69 /* XXX: RTF_LOCAL && RTF_MULTICAST */
71 rt_flags = nhop_get_rtflags(nh);
73 if (rt_flags & RTF_HOST) {
76 * Backward compatibility:
77 * if the destination is broadcast,
78 * mark route as broadcast.
79 * This behavior was useful when route cloning
80 * was in place, so there was an explicit cloned
81 * route for every broadcasted address.
82 * Currently (2020-04) there is no kernel machinery
83 * to do route cloning, though someone might explicitly
84 * add these routes to support some cases with active-active
85 * load balancing. Given that, retain this support.
87 if (in_broadcast(addr4->sin_addr, nh->nh_ifp)) {
88 rt_flags |= RTF_BROADCAST;
89 nhop_set_rtflags(nh, rt_flags);
90 nh->nh_flags |= NHF_BROADCAST;
96 * inherit interface MTU if not set or
97 * check if MTU is too large.
99 if (nh->nh_mtu == 0) {
100 nh->nh_mtu = nh->nh_ifp->if_mtu;
101 } else if (nh->nh_mtu > nh->nh_ifp->if_mtu)
102 nh->nh_mtu = nh->nh_ifp->if_mtu;
104 /* Ensure that default route nhop has special flag */
105 const struct sockaddr_in *mask4 = (const struct sockaddr_in *)mask;
106 if ((rt_flags & RTF_HOST) == 0 && mask4->sin_addr.s_addr == 0)
107 nh->nh_flags |= NHF_DEFAULT;
109 /* Set nhop type to basic per-AF nhop */
110 if (nhop_get_type(nh) == 0) {
111 if (nh->nh_flags & NHF_GATEWAY)
112 nh_type = NH_TYPE_IPV4_ETHER_NHOP;
114 nh_type = NH_TYPE_IPV4_ETHER_RSLV;
116 nhop_set_type(nh, nh_type);
123 * Do what we need to do when inserting a route.
125 static struct radix_node *
126 in_addroute(void *v_arg, void *n_arg, struct radix_head *head,
127 struct radix_node *treenodes)
129 struct rtentry *rt = (struct rtentry *)treenodes;
130 struct sockaddr_in *sin = (struct sockaddr_in *)rt_key(rt);
133 * A little bit of help for both IP output and input:
134 * For host routes, we make sure that RTF_BROADCAST
135 * is set for anything that looks like a broadcast address.
136 * This way, we can avoid an expensive call to in_broadcast()
137 * in ip_output() most of the time (because the route passed
138 * to ip_output() is almost always a host route).
140 * We also do the same for local addresses, with the thought
141 * that this might one day be used to speed up ip_input().
143 * We also mark routes to multicast addresses as such, because
144 * it's easy to do and might be useful (but this is much more
145 * dubious since it's so easy to inspect the address).
147 if (rt->rt_flags & RTF_HOST) {
148 struct epoch_tracker et;
152 bcast = in_broadcast(sin->sin_addr, rt->rt_ifp);
155 rt->rt_flags |= RTF_BROADCAST;
156 else if (satosin(rt->rt_ifa->ifa_addr)->sin_addr.s_addr ==
157 sin->sin_addr.s_addr)
158 rt->rt_flags |= RTF_LOCAL;
160 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
161 rt->rt_flags |= RTF_MULTICAST;
163 if (rt->rt_ifp != NULL) {
167 * inherit interface MTU if not set or
168 * check if MTU is too large.
170 if (rt->rt_mtu == 0) {
171 rt->rt_mtu = rt->rt_ifp->if_mtu;
172 } else if (rt->rt_mtu > rt->rt_ifp->if_mtu)
173 rt->rt_mtu = rt->rt_ifp->if_mtu;
176 return (rn_addroute(v_arg, n_arg, head, treenodes));
179 static int _in_rt_was_here;
181 * Initialize our routing tree.
184 in_inithead(void **head, int off, u_int fibnum)
188 rh = rt_table_init(32, AF_INET, fibnum);
192 rh->rnh_preadd = rib4_preadd;
193 rh->rnh_addaddr = in_addroute;
195 rt_mpath_init_rnh(rh);
199 if (_in_rt_was_here == 0 ) {
207 in_detachhead(void **head, int off)
210 rt_table_destroy((struct rib_head *)(*head));
216 * This zaps old routes when the interface goes down or interface
217 * address is deleted. In the latter case, it deletes static routes
218 * that point to this address. If we don't do this, we may end up
219 * using the old address in the future. The ones we always want to
220 * get rid of are things like ARP entries, since the user might down
221 * the interface, walk over to a completely different network, and
224 struct in_ifadown_arg {
230 in_ifadownkill(const struct rtentry *rt, void *xap)
232 struct in_ifadown_arg *ap = xap;
234 if (rt->rt_ifa != ap->ifa)
237 if ((rt->rt_flags & RTF_STATIC) != 0 && ap->del == 0)
244 in_ifadown(struct ifaddr *ifa, int delete)
246 struct in_ifadown_arg arg;
248 KASSERT(ifa->ifa_addr->sa_family == AF_INET,
249 ("%s: wrong family", __func__));
254 rt_foreach_fib_walk_del(AF_INET, in_ifadownkill, &arg);
255 ifa->ifa_flags &= ~IFA_ROUTE; /* XXXlocking? */
259 * inet versions of rt functions. These have fib extensions and
260 * for now will just reference the _fib variants.
261 * eventually this order will be reversed,
264 in_rtalloc_ign(struct route *ro, u_long ignflags, u_int fibnum)
266 rtalloc_ign_fib(ro, ignflags, fibnum);