net/ipv4/fib_rules.c

   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 __fib_lookup(struct net *net, struct flowi4 *flp,
  72                  struct fib_result *res, unsigned int flags)
  73 {
  74         struct fib_lookup_arg arg = {
  75                 .result = res,
  76                 .flags = flags,
  77         };
  78         int err;
  79
  80         /* update flow if oif or iif point to device enslaved to l3mdev */
  81         l3mdev_update_flow(net, flowi4_to_flowi(flp));
  82
  83         err = fib_rules_lookup(net->ipv4.rules_ops, flowi4_to_flowi(flp), 0, &arg);
  84 #ifdef CONFIG_IP_ROUTE_CLASSID
  85         if (arg.rule)
  86                 res->tclassid = ((struct fib4_rule *)arg.rule)->tclassid;
  87         else
  88                 res->tclassid = 0;
  89 #endif
  90
  91         if (err == -ESRCH)
  92                 err = -ENETUNREACH;
  93
  94         return err;
  95 }
  96 EXPORT_SYMBOL_GPL(__fib_lookup);
  97
  98 static int fib4_rule_action(struct fib_rule *rule, struct flowi *flp,
  99                             int flags, struct fib_lookup_arg *arg)
 100 {
 101         int err = -EAGAIN;
 102         struct fib_table *tbl;
 103         u32 tb_id;
 104
 105         switch (rule->action) {
 106         case FR_ACT_TO_TBL:
 107                 break;
 108
 109         case FR_ACT_UNREACHABLE:
 110                 return -ENETUNREACH;
 111
 112         case FR_ACT_PROHIBIT:
 113                 return -EACCES;
 114
 115         case FR_ACT_BLACKHOLE:
 116         default:
 117                 return -EINVAL;
 118         }
 119
 120         rcu_read_lock();
 121
 122         tb_id = fib_rule_get_table(rule, arg);
 123         tbl = fib_get_table(rule->fr_net, tb_id);
 124         if (tbl)
 125                 err = fib_table_lookup(tbl, &flp->u.ip4,
 126                                        (struct fib_result *)arg->result,
 127                                        arg->flags);
 128
 129         rcu_read_unlock();
 130         return err;
 131 }
 132
 133 static bool fib4_rule_suppress(struct fib_rule *rule, struct fib_lookup_arg *arg)
 134 {
 135         struct fib_result *result = (struct fib_result *) arg->result;
 136         struct net_device *dev = NULL;
 137
 138         if (result->fi)
 139                 dev = result->fi->fib_dev;
 140
 141         /* do not accept result if the route does
 142          * not meet the required prefix length
 143          */
 144         if (result->prefixlen <= rule->suppress_prefixlen)
 145                 goto suppress_route;
 146
 147         /* do not accept result if the route uses a device
 148          * belonging to a forbidden interface group
 149          */
 150         if (rule->suppress_ifgroup != -1 && dev && dev->group == rule->suppress_ifgroup)
 151                 goto suppress_route;
 152
 153         return false;
 154
 155 suppress_route:
 156         if (!(arg->flags & FIB_LOOKUP_NOREF))
 157                 fib_info_put(result->fi);
 158         return true;
 159 }
 160
 161 static int fib4_rule_match(struct fib_rule *rule, struct flowi *fl, int flags)
 162 {
 163         struct fib4_rule *r = (struct fib4_rule *) rule;
 164         struct flowi4 *fl4 = &fl->u.ip4;
 165         __be32 daddr = fl4->daddr;
 166         __be32 saddr = fl4->saddr;
 167
 168         if (((saddr ^ r->src) & r->srcmask) ||
 169             ((daddr ^ r->dst) & r->dstmask))
 170                 return 0;
 171
 172         if (r->tos && (r->tos != fl4->flowi4_tos))
 173                 return 0;
 174
 175         return 1;
 176 }
 177
 178 static struct fib_table *fib_empty_table(struct net *net)
 179 {
 180         u32 id;
 181
 182         for (id = 1; id <= RT_TABLE_MAX; id++)
 183                 if (!fib_get_table(net, id))
 184                         return fib_new_table(net, id);
 185         return NULL;
 186 }
 187
 188 static int call_fib_rule_notifier(struct notifier_block *nb, struct net *net,
 189                                   enum fib_event_type event_type,
 190                                   struct fib_rule *rule)
 191 {
 192         struct fib_rule_notifier_info info = {
 193                 .rule = rule,
 194         };
 195
 196         return call_fib_notifier(nb, net, event_type, &info.info);
 197 }
 198
 199 static int call_fib_rule_notifiers(struct net *net,
 200                                    enum fib_event_type event_type,
 201                                    struct fib_rule *rule)
 202 {
 203         struct fib_rule_notifier_info info = {
 204                 .rule = rule,
 205         };
 206
 207         return call_fib_notifiers(net, event_type, &info.info);
 208 }
 209
 210 /* Called with rcu_read_lock() */
 211 void fib_rules_notify(struct net *net, struct notifier_block *nb)
 212 {
 213         struct fib_rules_ops *ops = net->ipv4.rules_ops;
 214         struct fib_rule *rule;
 215
 216         list_for_each_entry_rcu(rule, &ops->rules_list, list)
 217                 call_fib_rule_notifier(nb, net, FIB_EVENT_RULE_ADD, rule);
 218 }
 219
 220 static const struct nla_policy fib4_rule_policy[FRA_MAX+1] = {
 221         FRA_GENERIC_POLICY,
 222         [FRA_FLOW]      = { .type = NLA_U32 },
 223 };
 224
 225 static int fib4_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
 226                                struct fib_rule_hdr *frh,
 227                                struct nlattr **tb)
 228 {
 229         struct net *net = sock_net(skb->sk);
 230         int err = -EINVAL;
 231         struct fib4_rule *rule4 = (struct fib4_rule *) rule;
 232
 233         if (frh->tos & ~IPTOS_TOS_MASK)
 234                 goto errout;
 235
 236         /* split local/main if they are not already split */
 237         err = fib_unmerge(net);
 238         if (err)
 239                 goto errout;
 240
 241         if (rule->table == RT_TABLE_UNSPEC && !rule->l3mdev) {
 242                 if (rule->action == FR_ACT_TO_TBL) {
 243                         struct fib_table *table;
 244
 245                         table = fib_empty_table(net);
 246                         if (!table) {
 247                                 err = -ENOBUFS;
 248                                 goto errout;
 249                         }
 250
 251                         rule->table = table->tb_id;
 252                 }
 253         }
 254
 255         if (frh->src_len)
 256                 rule4->src = nla_get_in_addr(tb[FRA_SRC]);
 257
 258         if (frh->dst_len)
 259                 rule4->dst = nla_get_in_addr(tb[FRA_DST]);
 260
 261 #ifdef CONFIG_IP_ROUTE_CLASSID
 262         if (tb[FRA_FLOW]) {
 263                 rule4->tclassid = nla_get_u32(tb[FRA_FLOW]);
 264                 if (rule4->tclassid)
 265                         net->ipv4.fib_num_tclassid_users++;
 266         }
 267 #endif
 268
 269         rule4->src_len = frh->src_len;
 270         rule4->srcmask = inet_make_mask(rule4->src_len);
 271         rule4->dst_len = frh->dst_len;
 272         rule4->dstmask = inet_make_mask(rule4->dst_len);
 273         rule4->tos = frh->tos;
 274
 275         net->ipv4.fib_has_custom_rules = true;
 276         call_fib_rule_notifiers(net, FIB_EVENT_RULE_ADD, rule);
 277
 278         err = 0;
 279 errout:
 280         return err;
 281 }
 282
 283 static int fib4_rule_delete(struct fib_rule *rule)
 284 {
 285         struct net *net = rule->fr_net;
 286         int err;
 287
 288         /* split local/main if they are not already split */
 289         err = fib_unmerge(net);
 290         if (err)
 291                 goto errout;
 292
 293 #ifdef CONFIG_IP_ROUTE_CLASSID
 294         if (((struct fib4_rule *)rule)->tclassid)
 295                 net->ipv4.fib_num_tclassid_users--;
 296 #endif
 297         net->ipv4.fib_has_custom_rules = true;
 298         call_fib_rule_notifiers(net, FIB_EVENT_RULE_DEL, rule);
 299 errout:
 300         return err;
 301 }
 302
 303 static int fib4_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
 304                              struct nlattr **tb)
 305 {
 306         struct fib4_rule *rule4 = (struct fib4_rule *) rule;
 307
 308         if (frh->src_len && (rule4->src_len != frh->src_len))
 309                 return 0;
 310
 311         if (frh->dst_len && (rule4->dst_len != frh->dst_len))
 312                 return 0;
 313
 314         if (frh->tos && (rule4->tos != frh->tos))
 315                 return 0;
 316
 317 #ifdef CONFIG_IP_ROUTE_CLASSID
 318         if (tb[FRA_FLOW] && (rule4->tclassid != nla_get_u32(tb[FRA_FLOW])))
 319                 return 0;
 320 #endif
 321
 322         if (frh->src_len && (rule4->src != nla_get_in_addr(tb[FRA_SRC])))
 323                 return 0;
 324
 325         if (frh->dst_len && (rule4->dst != nla_get_in_addr(tb[FRA_DST])))
 326                 return 0;
 327
 328         return 1;
 329 }
 330
 331 static int fib4_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
 332                           struct fib_rule_hdr *frh)
 333 {
 334         struct fib4_rule *rule4 = (struct fib4_rule *) rule;
 335
 336         frh->dst_len = rule4->dst_len;
 337         frh->src_len = rule4->src_len;
 338         frh->tos = rule4->tos;
 339
 340         if ((rule4->dst_len &&
 341              nla_put_in_addr(skb, FRA_DST, rule4->dst)) ||
 342             (rule4->src_len &&
 343              nla_put_in_addr(skb, FRA_SRC, rule4->src)))
 344                 goto nla_put_failure;
 345 #ifdef CONFIG_IP_ROUTE_CLASSID
 346         if (rule4->tclassid &&
 347             nla_put_u32(skb, FRA_FLOW, rule4->tclassid))
 348                 goto nla_put_failure;
 349 #endif
 350         return 0;
 351
 352 nla_put_failure:
 353         return -ENOBUFS;
 354 }
 355
 356 static size_t fib4_rule_nlmsg_payload(struct fib_rule *rule)
 357 {
 358         return nla_total_size(4) /* dst */
 359                + nla_total_size(4) /* src */
 360                + nla_total_size(4); /* flow */
 361 }
 362
 363 static void fib4_rule_flush_cache(struct fib_rules_ops *ops)
 364 {
 365         rt_cache_flush(ops->fro_net);
 366 }
 367
 368 static const struct fib_rules_ops __net_initconst fib4_rules_ops_template = {
 369         .family         = AF_INET,
 370         .rule_size      = sizeof(struct fib4_rule),
 371         .addr_size      = sizeof(u32),
 372         .action         = fib4_rule_action,
 373         .suppress       = fib4_rule_suppress,
 374         .match          = fib4_rule_match,
 375         .configure      = fib4_rule_configure,
 376         .delete         = fib4_rule_delete,
 377         .compare        = fib4_rule_compare,
 378         .fill           = fib4_rule_fill,
 379         .nlmsg_payload  = fib4_rule_nlmsg_payload,
 380         .flush_cache    = fib4_rule_flush_cache,
 381         .nlgroup        = RTNLGRP_IPV4_RULE,
 382         .policy         = fib4_rule_policy,
 383         .owner          = THIS_MODULE,
 384 };
 385
 386 static int fib_default_rules_init(struct fib_rules_ops *ops)
 387 {
 388         int err;
 389
 390         err = fib_default_rule_add(ops, 0, RT_TABLE_LOCAL, 0);
 391         if (err < 0)
 392                 return err;
 393         err = fib_default_rule_add(ops, 0x7FFE, RT_TABLE_MAIN, 0);
 394         if (err < 0)
 395                 return err;
 396         err = fib_default_rule_add(ops, 0x7FFF, RT_TABLE_DEFAULT, 0);
 397         if (err < 0)
 398                 return err;
 399         return 0;
 400 }
 401
 402 int __net_init fib4_rules_init(struct net *net)
 403 {
 404         int err;
 405         struct fib_rules_ops *ops;
 406
 407         ops = fib_rules_register(&fib4_rules_ops_template, net);
 408         if (IS_ERR(ops))
 409                 return PTR_ERR(ops);
 410
 411         err = fib_default_rules_init(ops);
 412         if (err < 0)
 413                 goto fail;
 414         net->ipv4.rules_ops = ops;
 415         net->ipv4.fib_has_custom_rules = false;
 416         return 0;
 417
 418 fail:
 419         /* also cleans all rules already added */
 420         fib_rules_unregister(ops);
 421         return err;
 422 }
 423
 424 void __net_exit fib4_rules_exit(struct net *net)
 425 {
 426         fib_rules_unregister(net->ipv4.rules_ops);
 427 }