1 /*- 2 * Copyright 1994, 1995 Massachusetts Institute of Technology 3 * 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 14 * warranty. 15 * 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 27 * SUCH DAMAGE. 28 */ 29 30 #include <sys/cdefs.h> 31 __FBSDID("$FreeBSD$"); 32 33 #include "opt_mpath.h" 34 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> 40 #include <sys/mbuf.h> 41 42 #include <net/if.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> 48 #include <net/vnet.h> 49 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> 55 56 extern int in_inithead(void **head, int off, u_int fibnum); 57 #ifdef VIMAGE 58 extern int in_detachhead(void **head, int off); 59 #endif 60 61 static int 62 rib4_preadd(u_int fibnum, const struct sockaddr *addr, const struct sockaddr *mask, 63 struct nhop_object *nh) 64 { 65 const struct sockaddr_in *addr4 = (const struct sockaddr_in *)addr; 66 uint16_t nh_type; 67 int rt_flags; 68 69 /* XXX: RTF_LOCAL && RTF_MULTICAST */ 70 71 rt_flags = nhop_get_rtflags(nh); 72 73 if (rt_flags & RTF_HOST) { 74 75 /* 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. 86 */ 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; 91 } 92 } 93 94 /* 95 * Check route MTU: 96 * inherit interface MTU if not set or 97 * check if MTU is too large. 98 */ 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; 103 104 /* 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 != NULL && 107 mask4->sin_addr.s_addr == 0) 108 nh->nh_flags |= NHF_DEFAULT; 109 110 /* Set nhop type to basic per-AF nhop */ 111 if (nhop_get_type(nh) == 0) { 112 if (nh->nh_flags & NHF_GATEWAY) 113 nh_type = NH_TYPE_IPV4_ETHER_NHOP; 114 else 115 nh_type = NH_TYPE_IPV4_ETHER_RSLV; 116 117 nhop_set_type(nh, nh_type); 118 } 119 120 return (0); 121 } 122 123 /* 124 * Do what we need to do when inserting a route. 125 */ 126 static struct radix_node * 127 in_addroute(void *v_arg, void *n_arg, struct radix_head *head, 128 struct radix_node *treenodes) 129 { 130 struct rtentry *rt = (struct rtentry *)treenodes; 131 struct sockaddr_in *sin = (struct sockaddr_in *)rt_key(rt); 132 133 /* 134 * A little bit of help for both IP output and input: 135 * For host routes, we make sure that RTF_BROADCAST 136 * is set for anything that looks like a broadcast address. 137 * This way, we can avoid an expensive call to in_broadcast() 138 * in ip_output() most of the time (because the route passed 139 * to ip_output() is almost always a host route). 140 * 141 * We also do the same for local addresses, with the thought 142 * that this might one day be used to speed up ip_input(). 143 * 144 * We also mark routes to multicast addresses as such, because 145 * it's easy to do and might be useful (but this is much more 146 * dubious since it's so easy to inspect the address). 147 */ 148 if (rt->rt_flags & RTF_HOST) { 149 struct epoch_tracker et; 150 bool bcast; 151 152 NET_EPOCH_ENTER(et); 153 bcast = in_broadcast(sin->sin_addr, rt->rt_ifp); 154 NET_EPOCH_EXIT(et); 155 if (bcast) 156 rt->rt_flags |= RTF_BROADCAST; 157 else if (satosin(rt->rt_ifa->ifa_addr)->sin_addr.s_addr == 158 sin->sin_addr.s_addr) 159 rt->rt_flags |= RTF_LOCAL; 160 } 161 if (IN_MULTICAST(ntohl(sin->sin_addr.s_addr))) 162 rt->rt_flags |= RTF_MULTICAST; 163 164 if (rt->rt_ifp != NULL) { 165 166 /* 167 * Check route MTU: 168 * inherit interface MTU if not set or 169 * check if MTU is too large. 170 */ 171 if (rt->rt_mtu == 0) { 172 rt->rt_mtu = rt->rt_ifp->if_mtu; 173 } else if (rt->rt_mtu > rt->rt_ifp->if_mtu) 174 rt->rt_mtu = rt->rt_ifp->if_mtu; 175 } 176 177 return (rn_addroute(v_arg, n_arg, head, treenodes)); 178 } 179 180 static int _in_rt_was_here; 181 /* 182 * Initialize our routing tree. 183 */ 184 int 185 in_inithead(void **head, int off, u_int fibnum) 186 { 187 struct rib_head *rh; 188 189 rh = rt_table_init(32, AF_INET, fibnum); 190 if (rh == NULL) 191 return (0); 192 193 rh->rnh_preadd = rib4_preadd; 194 rh->rnh_addaddr = in_addroute; 195 #ifdef RADIX_MPATH 196 rt_mpath_init_rnh(rh); 197 #endif 198 *head = (void *)rh; 199 200 if (_in_rt_was_here == 0 ) { 201 _in_rt_was_here = 1; 202 } 203 return 1; 204 } 205 206 #ifdef VIMAGE 207 int 208 in_detachhead(void **head, int off) 209 { 210 211 rt_table_destroy((struct rib_head *)(*head)); 212 return (1); 213 } 214 #endif 215 216 /* 217 * This zaps old routes when the interface goes down or interface 218 * address is deleted. In the latter case, it deletes static routes 219 * that point to this address. If we don't do this, we may end up 220 * using the old address in the future. The ones we always want to 221 * get rid of are things like ARP entries, since the user might down 222 * the interface, walk over to a completely different network, and 223 * plug back in. 224 */ 225 struct in_ifadown_arg { 226 struct ifaddr *ifa; 227 int del; 228 }; 229 230 static int 231 in_ifadownkill(const struct rtentry *rt, const struct nhop_object *nh, 232 void *xap) 233 { 234 struct in_ifadown_arg *ap = xap; 235 236 if (nh->nh_ifa != ap->ifa) 237 return (0); 238 239 if ((nhop_get_rtflags(nh) & RTF_STATIC) != 0 && ap->del == 0) 240 return (0); 241 242 return (1); 243 } 244 245 void 246 in_ifadown(struct ifaddr *ifa, int delete) 247 { 248 struct in_ifadown_arg arg; 249 250 KASSERT(ifa->ifa_addr->sa_family == AF_INET, 251 ("%s: wrong family", __func__)); 252 253 arg.ifa = ifa; 254 arg.del = delete; 255 256 rt_foreach_fib_walk_del(AF_INET, in_ifadownkill, &arg); 257 ifa->ifa_flags &= ~IFA_ROUTE; /* XXXlocking? */ 258 } 259 260 /* 261 * inet versions of rt functions. These have fib extensions and 262 * for now will just reference the _fib variants. 263 * eventually this order will be reversed, 264 */ 265 void 266 in_rtalloc_ign(struct route *ro, u_long ignflags, u_int fibnum) 267 { 268 rtalloc_ign_fib(ro, ignflags, fibnum); 269 } 270 271