1 /* 2 * INET An implementation of the TCP/IP protocol suite for the LINUX 3 * operating system. INET is implemented using the BSD Socket 4 * interface as the means of communication with the user level. 5 * 6 * IPv4 Forwarding Information Base: policy rules. 7 * 8 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru> 9 * Thomas Graf <tgraf@suug.ch> 10 * 11 * This program is free software; you can redistribute it and/or 12 * modify it under the terms of the GNU General Public License 13 * as published by the Free Software Foundation; either version 14 * 2 of the License, or (at your option) any later version. 15 * 16 * Fixes: 17 * Rani Assaf : local_rule cannot be deleted 18 * Marc Boucher : routing by fwmark 19 */ 20 21 #include <linux/types.h> 22 #include <linux/kernel.h> 23 #include <linux/netdevice.h> 24 #include <linux/netlink.h> 25 #include <linux/inetdevice.h> 26 #include <linux/init.h> 27 #include <linux/list.h> 28 #include <linux/rcupdate.h> 29 #include <linux/export.h> 30 #include <net/ip.h> 31 #include <net/route.h> 32 #include <net/tcp.h> 33 #include <net/ip_fib.h> 34 #include <net/fib_rules.h> 35 36 struct fib4_rule { 37 struct fib_rule common; 38 u8 dst_len; 39 u8 src_len; 40 u8 tos; 41 __be32 src; 42 __be32 srcmask; 43 __be32 dst; 44 __be32 dstmask; 45 #ifdef CONFIG_IP_ROUTE_CLASSID 46 u32 tclassid; 47 #endif 48 }; 49 50 static bool fib4_rule_matchall(const struct fib_rule *rule) 51 { 52 struct fib4_rule *r = container_of(rule, struct fib4_rule, common); 53 54 if (r->dst_len || r->src_len || r->tos) 55 return false; 56 return fib_rule_matchall(rule); 57 } 58 59 bool fib4_rule_default(const struct fib_rule *rule) 60 { 61 if (!fib4_rule_matchall(rule) || rule->action != FR_ACT_TO_TBL || 62 rule->l3mdev) 63 return false; 64 if (rule->table != RT_TABLE_LOCAL && rule->table != RT_TABLE_MAIN && 65 rule->table != RT_TABLE_DEFAULT) 66 return false; 67 return true; 68 } 69 EXPORT_SYMBOL_GPL(fib4_rule_default); 70 71 int fib4_rules_dump(struct net *net, struct notifier_block *nb) 72 { 73 return fib_rules_dump(net, nb, AF_INET); 74 } 75 76 unsigned int fib4_rules_seq_read(struct net *net) 77 { 78 return fib_rules_seq_read(net, AF_INET); 79 } 80 81 int __fib_lookup(struct net *net, struct flowi4 *flp, 82 struct fib_result *res, unsigned int flags) 83 { 84 struct fib_lookup_arg arg = { 85 .result = res, 86 .flags = flags, 87 }; 88 int err; 89 90 /* update flow if oif or iif point to device enslaved to l3mdev */ 91 l3mdev_update_flow(net, flowi4_to_flowi(flp)); 92 93 err = fib_rules_lookup(net->ipv4.rules_ops, flowi4_to_flowi(flp), 0, &arg); 94 #ifdef CONFIG_IP_ROUTE_CLASSID 95 if (arg.rule) 96 res->tclassid = ((struct fib4_rule *)arg.rule)->tclassid; 97 else 98 res->tclassid = 0; 99 #endif 100 101 if (err == -ESRCH) 102 err = -ENETUNREACH; 103 104 return err; 105 } 106 EXPORT_SYMBOL_GPL(__fib_lookup); 107 108 static int fib4_rule_action(struct fib_rule *rule, struct flowi *flp, 109 int flags, struct fib_lookup_arg *arg) 110 { 111 int err = -EAGAIN; 112 struct fib_table *tbl; 113 u32 tb_id; 114 115 switch (rule->action) { 116 case FR_ACT_TO_TBL: 117 break; 118 119 case FR_ACT_UNREACHABLE: 120 return -ENETUNREACH; 121 122 case FR_ACT_PROHIBIT: 123 return -EACCES; 124 125 case FR_ACT_BLACKHOLE: 126 default: 127 return -EINVAL; 128 } 129 130 rcu_read_lock(); 131 132 tb_id = fib_rule_get_table(rule, arg); 133 tbl = fib_get_table(rule->fr_net, tb_id); 134 if (tbl) 135 err = fib_table_lookup(tbl, &flp->u.ip4, 136 (struct fib_result *)arg->result, 137 arg->flags); 138 139 rcu_read_unlock(); 140 return err; 141 } 142 143 static bool fib4_rule_suppress(struct fib_rule *rule, struct fib_lookup_arg *arg) 144 { 145 struct fib_result *result = (struct fib_result *) arg->result; 146 struct net_device *dev = NULL; 147 148 if (result->fi) 149 dev = result->fi->fib_dev; 150 151 /* do not accept result if the route does 152 * not meet the required prefix length 153 */ 154 if (result->prefixlen <= rule->suppress_prefixlen) 155 goto suppress_route; 156 157 /* do not accept result if the route uses a device 158 * belonging to a forbidden interface group 159 */ 160 if (rule->suppress_ifgroup != -1 && dev && dev->group == rule->suppress_ifgroup) 161 goto suppress_route; 162 163 return false; 164 165 suppress_route: 166 if (!(arg->flags & FIB_LOOKUP_NOREF)) 167 fib_info_put(result->fi); 168 return true; 169 } 170 171 static int fib4_rule_match(struct fib_rule *rule, struct flowi *fl, int flags) 172 { 173 struct fib4_rule *r = (struct fib4_rule *) rule; 174 struct flowi4 *fl4 = &fl->u.ip4; 175 __be32 daddr = fl4->daddr; 176 __be32 saddr = fl4->saddr; 177 178 if (((saddr ^ r->src) & r->srcmask) || 179 ((daddr ^ r->dst) & r->dstmask)) 180 return 0; 181 182 if (r->tos && (r->tos != fl4->flowi4_tos)) 183 return 0; 184 185 return 1; 186 } 187 188 static struct fib_table *fib_empty_table(struct net *net) 189 { 190 u32 id; 191 192 for (id = 1; id <= RT_TABLE_MAX; id++) 193 if (!fib_get_table(net, id)) 194 return fib_new_table(net, id); 195 return NULL; 196 } 197 198 static const struct nla_policy fib4_rule_policy[FRA_MAX+1] = { 199 FRA_GENERIC_POLICY, 200 [FRA_FLOW] = { .type = NLA_U32 }, 201 }; 202 203 static int fib4_rule_configure(struct fib_rule *rule, struct sk_buff *skb, 204 struct fib_rule_hdr *frh, 205 struct nlattr **tb) 206 { 207 struct net *net = sock_net(skb->sk); 208 int err = -EINVAL; 209 struct fib4_rule *rule4 = (struct fib4_rule *) rule; 210 211 if (frh->tos & ~IPTOS_TOS_MASK) 212 goto errout; 213 214 /* split local/main if they are not already split */ 215 err = fib_unmerge(net); 216 if (err) 217 goto errout; 218 219 if (rule->table == RT_TABLE_UNSPEC && !rule->l3mdev) { 220 if (rule->action == FR_ACT_TO_TBL) { 221 struct fib_table *table; 222 223 table = fib_empty_table(net); 224 if (!table) { 225 err = -ENOBUFS; 226 goto errout; 227 } 228 229 rule->table = table->tb_id; 230 } 231 } 232 233 if (frh->src_len) 234 rule4->src = nla_get_in_addr(tb[FRA_SRC]); 235 236 if (frh->dst_len) 237 rule4->dst = nla_get_in_addr(tb[FRA_DST]); 238 239 #ifdef CONFIG_IP_ROUTE_CLASSID 240 if (tb[FRA_FLOW]) { 241 rule4->tclassid = nla_get_u32(tb[FRA_FLOW]); 242 if (rule4->tclassid) 243 net->ipv4.fib_num_tclassid_users++; 244 } 245 #endif 246 247 rule4->src_len = frh->src_len; 248 rule4->srcmask = inet_make_mask(rule4->src_len); 249 rule4->dst_len = frh->dst_len; 250 rule4->dstmask = inet_make_mask(rule4->dst_len); 251 rule4->tos = frh->tos; 252 253 net->ipv4.fib_has_custom_rules = true; 254 255 err = 0; 256 errout: 257 return err; 258 } 259 260 static int fib4_rule_delete(struct fib_rule *rule) 261 { 262 struct net *net = rule->fr_net; 263 int err; 264 265 /* split local/main if they are not already split */ 266 err = fib_unmerge(net); 267 if (err) 268 goto errout; 269 270 #ifdef CONFIG_IP_ROUTE_CLASSID 271 if (((struct fib4_rule *)rule)->tclassid) 272 net->ipv4.fib_num_tclassid_users--; 273 #endif 274 net->ipv4.fib_has_custom_rules = true; 275 errout: 276 return err; 277 } 278 279 static int fib4_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh, 280 struct nlattr **tb) 281 { 282 struct fib4_rule *rule4 = (struct fib4_rule *) rule; 283 284 if (frh->src_len && (rule4->src_len != frh->src_len)) 285 return 0; 286 287 if (frh->dst_len && (rule4->dst_len != frh->dst_len)) 288 return 0; 289 290 if (frh->tos && (rule4->tos != frh->tos)) 291 return 0; 292 293 #ifdef CONFIG_IP_ROUTE_CLASSID 294 if (tb[FRA_FLOW] && (rule4->tclassid != nla_get_u32(tb[FRA_FLOW]))) 295 return 0; 296 #endif 297 298 if (frh->src_len && (rule4->src != nla_get_in_addr(tb[FRA_SRC]))) 299 return 0; 300 301 if (frh->dst_len && (rule4->dst != nla_get_in_addr(tb[FRA_DST]))) 302 return 0; 303 304 return 1; 305 } 306 307 static int fib4_rule_fill(struct fib_rule *rule, struct sk_buff *skb, 308 struct fib_rule_hdr *frh) 309 { 310 struct fib4_rule *rule4 = (struct fib4_rule *) rule; 311 312 frh->dst_len = rule4->dst_len; 313 frh->src_len = rule4->src_len; 314 frh->tos = rule4->tos; 315 316 if ((rule4->dst_len && 317 nla_put_in_addr(skb, FRA_DST, rule4->dst)) || 318 (rule4->src_len && 319 nla_put_in_addr(skb, FRA_SRC, rule4->src))) 320 goto nla_put_failure; 321 #ifdef CONFIG_IP_ROUTE_CLASSID 322 if (rule4->tclassid && 323 nla_put_u32(skb, FRA_FLOW, rule4->tclassid)) 324 goto nla_put_failure; 325 #endif 326 return 0; 327 328 nla_put_failure: 329 return -ENOBUFS; 330 } 331 332 static size_t fib4_rule_nlmsg_payload(struct fib_rule *rule) 333 { 334 return nla_total_size(4) /* dst */ 335 + nla_total_size(4) /* src */ 336 + nla_total_size(4); /* flow */ 337 } 338 339 static void fib4_rule_flush_cache(struct fib_rules_ops *ops) 340 { 341 rt_cache_flush(ops->fro_net); 342 } 343 344 static const struct fib_rules_ops __net_initconst fib4_rules_ops_template = { 345 .family = AF_INET, 346 .rule_size = sizeof(struct fib4_rule), 347 .addr_size = sizeof(u32), 348 .action = fib4_rule_action, 349 .suppress = fib4_rule_suppress, 350 .match = fib4_rule_match, 351 .configure = fib4_rule_configure, 352 .delete = fib4_rule_delete, 353 .compare = fib4_rule_compare, 354 .fill = fib4_rule_fill, 355 .nlmsg_payload = fib4_rule_nlmsg_payload, 356 .flush_cache = fib4_rule_flush_cache, 357 .nlgroup = RTNLGRP_IPV4_RULE, 358 .policy = fib4_rule_policy, 359 .owner = THIS_MODULE, 360 }; 361 362 static int fib_default_rules_init(struct fib_rules_ops *ops) 363 { 364 int err; 365 366 err = fib_default_rule_add(ops, 0, RT_TABLE_LOCAL, 0); 367 if (err < 0) 368 return err; 369 err = fib_default_rule_add(ops, 0x7FFE, RT_TABLE_MAIN, 0); 370 if (err < 0) 371 return err; 372 err = fib_default_rule_add(ops, 0x7FFF, RT_TABLE_DEFAULT, 0); 373 if (err < 0) 374 return err; 375 return 0; 376 } 377 378 int __net_init fib4_rules_init(struct net *net) 379 { 380 int err; 381 struct fib_rules_ops *ops; 382 383 ops = fib_rules_register(&fib4_rules_ops_template, net); 384 if (IS_ERR(ops)) 385 return PTR_ERR(ops); 386 387 err = fib_default_rules_init(ops); 388 if (err < 0) 389 goto fail; 390 net->ipv4.rules_ops = ops; 391 net->ipv4.fib_has_custom_rules = false; 392 return 0; 393 394 fail: 395 /* also cleans all rules already added */ 396 fib_rules_unregister(ops); 397 return err; 398 } 399 400 void __net_exit fib4_rules_exit(struct net *net) 401 { 402 fib_rules_unregister(net->ipv4.rules_ops); 403 } 404