blob: 741e4fa3e4749e0285c39f65dc306e644fb4032b [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
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: FIB frontend.
7 *
Linus Torvalds1da177e2005-04-16 15:20:36 -07008 * Authors: Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru>
9 *
10 * This program is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU General Public License
12 * as published by the Free Software Foundation; either version
13 * 2 of the License, or (at your option) any later version.
14 */
15
Linus Torvalds1da177e2005-04-16 15:20:36 -070016#include <linux/module.h>
17#include <asm/uaccess.h>
18#include <asm/system.h>
19#include <linux/bitops.h>
Randy Dunlap4fc268d2006-01-11 12:17:47 -080020#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070021#include <linux/types.h>
22#include <linux/kernel.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070023#include <linux/mm.h>
24#include <linux/string.h>
25#include <linux/socket.h>
26#include <linux/sockios.h>
27#include <linux/errno.h>
28#include <linux/in.h>
29#include <linux/inet.h>
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -020030#include <linux/inetdevice.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070031#include <linux/netdevice.h>
Thomas Graf18237302006-08-04 23:04:54 -070032#include <linux/if_addr.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070033#include <linux/if_arp.h>
34#include <linux/skbuff.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070035#include <linux/init.h>
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070036#include <linux/list.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070037
38#include <net/ip.h>
39#include <net/protocol.h>
40#include <net/route.h>
41#include <net/tcp.h>
42#include <net/sock.h>
43#include <net/icmp.h>
44#include <net/arp.h>
45#include <net/ip_fib.h>
Thomas Graf63f34442007-03-22 11:55:17 -070046#include <net/rtnetlink.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070047
Linus Torvalds1da177e2005-04-16 15:20:36 -070048#ifndef CONFIG_IP_MULTIPLE_TABLES
49
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -080050static int __net_init fib4_rules_init(struct net *net)
Pavel Emelyanovc3e9a352007-11-06 23:34:04 -080051{
Denis V. Lunev93456b62008-01-10 03:23:38 -080052 struct fib_table *local_table, *main_table;
53
Stephen Hemminger7f9b8052008-01-14 23:14:20 -080054 local_table = fib_hash_table(RT_TABLE_LOCAL);
Denis V. Lunev93456b62008-01-10 03:23:38 -080055 if (local_table == NULL)
Denis V. Lunevdbb50162008-01-10 03:21:49 -080056 return -ENOMEM;
57
Stephen Hemminger7f9b8052008-01-14 23:14:20 -080058 main_table = fib_hash_table(RT_TABLE_MAIN);
Denis V. Lunev93456b62008-01-10 03:23:38 -080059 if (main_table == NULL)
Denis V. Lunevdbb50162008-01-10 03:21:49 -080060 goto fail;
61
Denis V. Lunev93456b62008-01-10 03:23:38 -080062 hlist_add_head_rcu(&local_table->tb_hlist,
Denis V. Luneve4aef8a2008-01-10 03:28:24 -080063 &net->ipv4.fib_table_hash[TABLE_LOCAL_INDEX]);
Denis V. Lunev93456b62008-01-10 03:23:38 -080064 hlist_add_head_rcu(&main_table->tb_hlist,
Denis V. Luneve4aef8a2008-01-10 03:28:24 -080065 &net->ipv4.fib_table_hash[TABLE_MAIN_INDEX]);
Denis V. Lunevdbb50162008-01-10 03:21:49 -080066 return 0;
67
68fail:
Denis V. Lunev93456b62008-01-10 03:23:38 -080069 kfree(local_table);
Denis V. Lunevdbb50162008-01-10 03:21:49 -080070 return -ENOMEM;
Pavel Emelyanovc3e9a352007-11-06 23:34:04 -080071}
Linus Torvalds1da177e2005-04-16 15:20:36 -070072#else
73
Denis V. Lunev8ad49422008-01-10 03:24:11 -080074struct fib_table *fib_new_table(struct net *net, u32 id)
Linus Torvalds1da177e2005-04-16 15:20:36 -070075{
76 struct fib_table *tb;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070077 unsigned int h;
Linus Torvalds1da177e2005-04-16 15:20:36 -070078
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070079 if (id == 0)
80 id = RT_TABLE_MAIN;
Denis V. Lunev8ad49422008-01-10 03:24:11 -080081 tb = fib_get_table(net, id);
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070082 if (tb)
83 return tb;
Stephen Hemminger7f9b8052008-01-14 23:14:20 -080084
85 tb = fib_hash_table(id);
Linus Torvalds1da177e2005-04-16 15:20:36 -070086 if (!tb)
87 return NULL;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070088 h = id & (FIB_TABLE_HASHSZ - 1);
Denis V. Luneve4aef8a2008-01-10 03:28:24 -080089 hlist_add_head_rcu(&tb->tb_hlist, &net->ipv4.fib_table_hash[h]);
Linus Torvalds1da177e2005-04-16 15:20:36 -070090 return tb;
91}
92
Denis V. Lunev8ad49422008-01-10 03:24:11 -080093struct fib_table *fib_get_table(struct net *net, u32 id)
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070094{
95 struct fib_table *tb;
96 struct hlist_node *node;
Denis V. Luneve4aef8a2008-01-10 03:28:24 -080097 struct hlist_head *head;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -070098 unsigned int h;
Linus Torvalds1da177e2005-04-16 15:20:36 -070099
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700100 if (id == 0)
101 id = RT_TABLE_MAIN;
102 h = id & (FIB_TABLE_HASHSZ - 1);
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800103
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700104 rcu_read_lock();
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800105 head = &net->ipv4.fib_table_hash[h];
106 hlist_for_each_entry_rcu(tb, node, head, tb_hlist) {
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700107 if (tb->tb_id == id) {
108 rcu_read_unlock();
109 return tb;
110 }
111 }
112 rcu_read_unlock();
113 return NULL;
114}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700115#endif /* CONFIG_IP_MULTIPLE_TABLES */
116
Denis V. Lunev010278e2008-01-22 22:04:04 -0800117void fib_select_default(struct net *net,
118 const struct flowi *flp, struct fib_result *res)
Denis V. Lunev64c2d532008-01-22 22:03:33 -0800119{
120 struct fib_table *tb;
121 int table = RT_TABLE_MAIN;
122#ifdef CONFIG_IP_MULTIPLE_TABLES
123 if (res->r == NULL || res->r->action != FR_ACT_TO_TBL)
124 return;
125 table = res->r->table;
126#endif
Denis V. Lunev010278e2008-01-22 22:04:04 -0800127 tb = fib_get_table(net, table);
Denis V. Lunev64c2d532008-01-22 22:03:33 -0800128 if (FIB_RES_GW(*res) && FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK)
129 tb->tb_select_default(tb, flp, res);
130}
131
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800132static void fib_flush(struct net *net)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700133{
134 int flushed = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700135 struct fib_table *tb;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700136 struct hlist_node *node;
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800137 struct hlist_head *head;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700138 unsigned int h;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700139
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700140 for (h = 0; h < FIB_TABLE_HASHSZ; h++) {
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800141 head = &net->ipv4.fib_table_hash[h];
142 hlist_for_each_entry(tb, node, head, tb_hlist)
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700143 flushed += tb->tb_flush(tb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700144 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145
146 if (flushed)
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700147 rt_cache_flush(net, -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148}
149
150/*
151 * Find the first device with a given source address.
152 */
153
Denis V. Lunev1ab35272008-01-22 22:04:30 -0800154struct net_device * ip_dev_find(struct net *net, __be32 addr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155{
156 struct flowi fl = { .nl_u = { .ip4_u = { .daddr = addr } } };
157 struct fib_result res;
158 struct net_device *dev = NULL;
Pavel Emelyanov03cf7862007-10-23 21:17:27 -0700159 struct fib_table *local_table;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700160
161#ifdef CONFIG_IP_MULTIPLE_TABLES
162 res.r = NULL;
163#endif
164
Denis V. Lunev1ab35272008-01-22 22:04:30 -0800165 local_table = fib_get_table(net, RT_TABLE_LOCAL);
Pavel Emelyanov03cf7862007-10-23 21:17:27 -0700166 if (!local_table || local_table->tb_lookup(local_table, &fl, &res))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700167 return NULL;
168 if (res.type != RTN_LOCAL)
169 goto out;
170 dev = FIB_RES_DEV(res);
171
172 if (dev)
173 dev_hold(dev);
174out:
175 fib_res_put(&res);
176 return dev;
177}
178
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800179/*
180 * Find address type as if only "dev" was present in the system. If
181 * on_dev is NULL then all interfaces are taken into consideration.
182 */
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800183static inline unsigned __inet_dev_addr_type(struct net *net,
184 const struct net_device *dev,
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800185 __be32 addr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700186{
187 struct flowi fl = { .nl_u = { .ip4_u = { .daddr = addr } } };
188 struct fib_result res;
189 unsigned ret = RTN_BROADCAST;
Pavel Emelyanov03cf7862007-10-23 21:17:27 -0700190 struct fib_table *local_table;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700191
Jan Engelhardt1e637c72008-01-21 03:18:08 -0800192 if (ipv4_is_zeronet(addr) || ipv4_is_lbcast(addr))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700193 return RTN_BROADCAST;
Joe Perchesf97c1e02007-12-16 13:45:43 -0800194 if (ipv4_is_multicast(addr))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700195 return RTN_MULTICAST;
196
197#ifdef CONFIG_IP_MULTIPLE_TABLES
198 res.r = NULL;
199#endif
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900200
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800201 local_table = fib_get_table(net, RT_TABLE_LOCAL);
Pavel Emelyanov03cf7862007-10-23 21:17:27 -0700202 if (local_table) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203 ret = RTN_UNICAST;
Pavel Emelyanov03cf7862007-10-23 21:17:27 -0700204 if (!local_table->tb_lookup(local_table, &fl, &res)) {
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800205 if (!dev || dev == res.fi->fib_dev)
206 ret = res.type;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700207 fib_res_put(&res);
208 }
209 }
210 return ret;
211}
212
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800213unsigned int inet_addr_type(struct net *net, __be32 addr)
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800214{
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800215 return __inet_dev_addr_type(net, NULL, addr);
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800216}
217
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800218unsigned int inet_dev_addr_type(struct net *net, const struct net_device *dev,
219 __be32 addr)
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800220{
Eric W. Biederman6b175b22008-01-10 03:25:28 -0800221 return __inet_dev_addr_type(net, dev, addr);
Laszlo Attila Toth05538112007-12-04 23:28:46 -0800222}
223
Linus Torvalds1da177e2005-04-16 15:20:36 -0700224/* Given (packet source, input interface) and optional (dst, oif, tos):
225 - (main) check, that source is valid i.e. not broadcast or our local
226 address.
227 - figure out what "logical" interface this packet arrived
228 and calculate "specific destination" address.
229 - check, that packet arrived from expected physical interface.
230 */
231
Al Virod9c9df82006-09-26 21:28:14 -0700232int fib_validate_source(__be32 src, __be32 dst, u8 tos, int oif,
233 struct net_device *dev, __be32 *spec_dst, u32 *itag)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234{
235 struct in_device *in_dev;
236 struct flowi fl = { .nl_u = { .ip4_u =
237 { .daddr = src,
238 .saddr = dst,
239 .tos = tos } },
240 .iif = oif };
241 struct fib_result res;
242 int no_addr, rpf;
243 int ret;
Denis V. Lunev5b707aa2008-01-21 17:33:15 -0800244 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700245
246 no_addr = rpf = 0;
247 rcu_read_lock();
Herbert Xue5ed6392005-10-03 14:35:55 -0700248 in_dev = __in_dev_get_rcu(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700249 if (in_dev) {
250 no_addr = in_dev->ifa_list == NULL;
251 rpf = IN_DEV_RPFILTER(in_dev);
252 }
253 rcu_read_unlock();
254
255 if (in_dev == NULL)
256 goto e_inval;
257
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900258 net = dev_net(dev);
Denis V. Lunev5b707aa2008-01-21 17:33:15 -0800259 if (fib_lookup(net, &fl, &res))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700260 goto last_resort;
261 if (res.type != RTN_UNICAST)
262 goto e_inval_res;
263 *spec_dst = FIB_RES_PREFSRC(res);
264 fib_combine_itag(itag, &res);
265#ifdef CONFIG_IP_ROUTE_MULTIPATH
266 if (FIB_RES_DEV(res) == dev || res.fi->fib_nhs > 1)
267#else
268 if (FIB_RES_DEV(res) == dev)
269#endif
270 {
271 ret = FIB_RES_NH(res).nh_scope >= RT_SCOPE_HOST;
272 fib_res_put(&res);
273 return ret;
274 }
275 fib_res_put(&res);
276 if (no_addr)
277 goto last_resort;
278 if (rpf)
279 goto e_inval;
280 fl.oif = dev->ifindex;
281
282 ret = 0;
Denis V. Lunev5b707aa2008-01-21 17:33:15 -0800283 if (fib_lookup(net, &fl, &res) == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700284 if (res.type == RTN_UNICAST) {
285 *spec_dst = FIB_RES_PREFSRC(res);
286 ret = FIB_RES_NH(res).nh_scope >= RT_SCOPE_HOST;
287 }
288 fib_res_put(&res);
289 }
290 return ret;
291
292last_resort:
293 if (rpf)
294 goto e_inval;
295 *spec_dst = inet_select_addr(dev, 0, RT_SCOPE_UNIVERSE);
296 *itag = 0;
297 return 0;
298
299e_inval_res:
300 fib_res_put(&res);
301e_inval:
302 return -EINVAL;
303}
304
Al Viro81f7bf62006-09-27 18:40:00 -0700305static inline __be32 sk_extract_addr(struct sockaddr *addr)
Thomas Graf4e902c52006-08-17 18:14:52 -0700306{
307 return ((struct sockaddr_in *) addr)->sin_addr.s_addr;
308}
309
310static int put_rtax(struct nlattr *mx, int len, int type, u32 value)
311{
312 struct nlattr *nla;
313
314 nla = (struct nlattr *) ((char *) mx + len);
315 nla->nla_type = type;
316 nla->nla_len = nla_attr_size(4);
317 *(u32 *) nla_data(nla) = value;
318
319 return len + nla_total_size(4);
320}
321
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800322static int rtentry_to_fib_config(struct net *net, int cmd, struct rtentry *rt,
Thomas Graf4e902c52006-08-17 18:14:52 -0700323 struct fib_config *cfg)
324{
Al Viro6d85c102006-09-26 22:15:46 -0700325 __be32 addr;
Thomas Graf4e902c52006-08-17 18:14:52 -0700326 int plen;
327
328 memset(cfg, 0, sizeof(*cfg));
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800329 cfg->fc_nlinfo.nl_net = net;
Thomas Graf4e902c52006-08-17 18:14:52 -0700330
331 if (rt->rt_dst.sa_family != AF_INET)
332 return -EAFNOSUPPORT;
333
334 /*
335 * Check mask for validity:
336 * a) it must be contiguous.
337 * b) destination must have all host bits clear.
338 * c) if application forgot to set correct family (AF_INET),
339 * reject request unless it is absolutely clear i.e.
340 * both family and mask are zero.
341 */
342 plen = 32;
343 addr = sk_extract_addr(&rt->rt_dst);
344 if (!(rt->rt_flags & RTF_HOST)) {
Al Viro81f7bf62006-09-27 18:40:00 -0700345 __be32 mask = sk_extract_addr(&rt->rt_genmask);
Thomas Graf4e902c52006-08-17 18:14:52 -0700346
347 if (rt->rt_genmask.sa_family != AF_INET) {
348 if (mask || rt->rt_genmask.sa_family)
349 return -EAFNOSUPPORT;
350 }
351
352 if (bad_mask(mask, addr))
353 return -EINVAL;
354
355 plen = inet_mask_len(mask);
356 }
357
358 cfg->fc_dst_len = plen;
359 cfg->fc_dst = addr;
360
361 if (cmd != SIOCDELRT) {
362 cfg->fc_nlflags = NLM_F_CREATE;
363 cfg->fc_protocol = RTPROT_BOOT;
364 }
365
366 if (rt->rt_metric)
367 cfg->fc_priority = rt->rt_metric - 1;
368
369 if (rt->rt_flags & RTF_REJECT) {
370 cfg->fc_scope = RT_SCOPE_HOST;
371 cfg->fc_type = RTN_UNREACHABLE;
372 return 0;
373 }
374
375 cfg->fc_scope = RT_SCOPE_NOWHERE;
376 cfg->fc_type = RTN_UNICAST;
377
378 if (rt->rt_dev) {
379 char *colon;
380 struct net_device *dev;
381 char devname[IFNAMSIZ];
382
383 if (copy_from_user(devname, rt->rt_dev, IFNAMSIZ-1))
384 return -EFAULT;
385
386 devname[IFNAMSIZ-1] = 0;
387 colon = strchr(devname, ':');
388 if (colon)
389 *colon = 0;
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800390 dev = __dev_get_by_name(net, devname);
Thomas Graf4e902c52006-08-17 18:14:52 -0700391 if (!dev)
392 return -ENODEV;
393 cfg->fc_oif = dev->ifindex;
394 if (colon) {
395 struct in_ifaddr *ifa;
396 struct in_device *in_dev = __in_dev_get_rtnl(dev);
397 if (!in_dev)
398 return -ENODEV;
399 *colon = ':';
400 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next)
401 if (strcmp(ifa->ifa_label, devname) == 0)
402 break;
403 if (ifa == NULL)
404 return -ENODEV;
405 cfg->fc_prefsrc = ifa->ifa_local;
406 }
407 }
408
409 addr = sk_extract_addr(&rt->rt_gateway);
410 if (rt->rt_gateway.sa_family == AF_INET && addr) {
411 cfg->fc_gw = addr;
412 if (rt->rt_flags & RTF_GATEWAY &&
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800413 inet_addr_type(net, addr) == RTN_UNICAST)
Thomas Graf4e902c52006-08-17 18:14:52 -0700414 cfg->fc_scope = RT_SCOPE_UNIVERSE;
415 }
416
417 if (cmd == SIOCDELRT)
418 return 0;
419
420 if (rt->rt_flags & RTF_GATEWAY && !cfg->fc_gw)
421 return -EINVAL;
422
423 if (cfg->fc_scope == RT_SCOPE_NOWHERE)
424 cfg->fc_scope = RT_SCOPE_LINK;
425
426 if (rt->rt_flags & (RTF_MTU | RTF_WINDOW | RTF_IRTT)) {
427 struct nlattr *mx;
428 int len = 0;
429
430 mx = kzalloc(3 * nla_total_size(4), GFP_KERNEL);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900431 if (mx == NULL)
Thomas Graf4e902c52006-08-17 18:14:52 -0700432 return -ENOMEM;
433
434 if (rt->rt_flags & RTF_MTU)
435 len = put_rtax(mx, len, RTAX_ADVMSS, rt->rt_mtu - 40);
436
437 if (rt->rt_flags & RTF_WINDOW)
438 len = put_rtax(mx, len, RTAX_WINDOW, rt->rt_window);
439
440 if (rt->rt_flags & RTF_IRTT)
441 len = put_rtax(mx, len, RTAX_RTT, rt->rt_irtt << 3);
442
443 cfg->fc_mx = mx;
444 cfg->fc_mx_len = len;
445 }
446
447 return 0;
448}
449
Linus Torvalds1da177e2005-04-16 15:20:36 -0700450/*
451 * Handle IP routing ioctl calls. These are used to manipulate the routing tables
452 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900453
Denis V. Lunev1bad1182008-01-10 03:29:53 -0800454int ip_rt_ioctl(struct net *net, unsigned int cmd, void __user *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700455{
Thomas Graf4e902c52006-08-17 18:14:52 -0700456 struct fib_config cfg;
457 struct rtentry rt;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700458 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700459
460 switch (cmd) {
461 case SIOCADDRT: /* Add a route */
462 case SIOCDELRT: /* Delete a route */
463 if (!capable(CAP_NET_ADMIN))
464 return -EPERM;
Thomas Graf4e902c52006-08-17 18:14:52 -0700465
466 if (copy_from_user(&rt, arg, sizeof(rt)))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700467 return -EFAULT;
Thomas Graf4e902c52006-08-17 18:14:52 -0700468
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469 rtnl_lock();
Denis V. Lunev1bad1182008-01-10 03:29:53 -0800470 err = rtentry_to_fib_config(net, cmd, &rt, &cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700471 if (err == 0) {
Thomas Graf4e902c52006-08-17 18:14:52 -0700472 struct fib_table *tb;
473
Linus Torvalds1da177e2005-04-16 15:20:36 -0700474 if (cmd == SIOCDELRT) {
Denis V. Lunev1bad1182008-01-10 03:29:53 -0800475 tb = fib_get_table(net, cfg.fc_table);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700476 if (tb)
Thomas Graf4e902c52006-08-17 18:14:52 -0700477 err = tb->tb_delete(tb, &cfg);
478 else
479 err = -ESRCH;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480 } else {
Denis V. Lunev1bad1182008-01-10 03:29:53 -0800481 tb = fib_new_table(net, cfg.fc_table);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700482 if (tb)
Thomas Graf4e902c52006-08-17 18:14:52 -0700483 err = tb->tb_insert(tb, &cfg);
484 else
485 err = -ENOBUFS;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486 }
Thomas Graf4e902c52006-08-17 18:14:52 -0700487
488 /* allocated by rtentry_to_fib_config() */
489 kfree(cfg.fc_mx);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490 }
491 rtnl_unlock();
492 return err;
493 }
494 return -EINVAL;
495}
496
Patrick McHardyef7c79e2007-06-05 12:38:30 -0700497const struct nla_policy rtm_ipv4_policy[RTA_MAX+1] = {
Thomas Graf4e902c52006-08-17 18:14:52 -0700498 [RTA_DST] = { .type = NLA_U32 },
499 [RTA_SRC] = { .type = NLA_U32 },
500 [RTA_IIF] = { .type = NLA_U32 },
501 [RTA_OIF] = { .type = NLA_U32 },
502 [RTA_GATEWAY] = { .type = NLA_U32 },
503 [RTA_PRIORITY] = { .type = NLA_U32 },
504 [RTA_PREFSRC] = { .type = NLA_U32 },
505 [RTA_METRICS] = { .type = NLA_NESTED },
Thomas Graf5176f912006-08-26 20:13:18 -0700506 [RTA_MULTIPATH] = { .len = sizeof(struct rtnexthop) },
Thomas Graf4e902c52006-08-17 18:14:52 -0700507 [RTA_FLOW] = { .type = NLA_U32 },
Thomas Graf4e902c52006-08-17 18:14:52 -0700508};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800510static int rtm_to_fib_config(struct net *net, struct sk_buff *skb,
511 struct nlmsghdr *nlh, struct fib_config *cfg)
Thomas Graf4e902c52006-08-17 18:14:52 -0700512{
513 struct nlattr *attr;
514 int err, remaining;
515 struct rtmsg *rtm;
516
517 err = nlmsg_validate(nlh, sizeof(*rtm), RTA_MAX, rtm_ipv4_policy);
518 if (err < 0)
519 goto errout;
520
521 memset(cfg, 0, sizeof(*cfg));
522
523 rtm = nlmsg_data(nlh);
Thomas Graf4e902c52006-08-17 18:14:52 -0700524 cfg->fc_dst_len = rtm->rtm_dst_len;
Thomas Graf4e902c52006-08-17 18:14:52 -0700525 cfg->fc_tos = rtm->rtm_tos;
526 cfg->fc_table = rtm->rtm_table;
527 cfg->fc_protocol = rtm->rtm_protocol;
528 cfg->fc_scope = rtm->rtm_scope;
529 cfg->fc_type = rtm->rtm_type;
530 cfg->fc_flags = rtm->rtm_flags;
531 cfg->fc_nlflags = nlh->nlmsg_flags;
532
533 cfg->fc_nlinfo.pid = NETLINK_CB(skb).pid;
534 cfg->fc_nlinfo.nlh = nlh;
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800535 cfg->fc_nlinfo.nl_net = net;
Thomas Graf4e902c52006-08-17 18:14:52 -0700536
Thomas Grafa0ee18b2007-03-24 20:32:54 -0700537 if (cfg->fc_type > RTN_MAX) {
538 err = -EINVAL;
539 goto errout;
540 }
541
Thomas Graf4e902c52006-08-17 18:14:52 -0700542 nlmsg_for_each_attr(attr, nlh, sizeof(struct rtmsg), remaining) {
Thomas Graf8f4c1f92007-09-12 14:44:36 +0200543 switch (nla_type(attr)) {
Thomas Graf4e902c52006-08-17 18:14:52 -0700544 case RTA_DST:
Al Viro17fb2c62006-09-26 22:15:25 -0700545 cfg->fc_dst = nla_get_be32(attr);
Thomas Graf4e902c52006-08-17 18:14:52 -0700546 break;
Thomas Graf4e902c52006-08-17 18:14:52 -0700547 case RTA_OIF:
548 cfg->fc_oif = nla_get_u32(attr);
549 break;
550 case RTA_GATEWAY:
Al Viro17fb2c62006-09-26 22:15:25 -0700551 cfg->fc_gw = nla_get_be32(attr);
Thomas Graf4e902c52006-08-17 18:14:52 -0700552 break;
553 case RTA_PRIORITY:
554 cfg->fc_priority = nla_get_u32(attr);
555 break;
556 case RTA_PREFSRC:
Al Viro17fb2c62006-09-26 22:15:25 -0700557 cfg->fc_prefsrc = nla_get_be32(attr);
Thomas Graf4e902c52006-08-17 18:14:52 -0700558 break;
559 case RTA_METRICS:
560 cfg->fc_mx = nla_data(attr);
561 cfg->fc_mx_len = nla_len(attr);
562 break;
563 case RTA_MULTIPATH:
564 cfg->fc_mp = nla_data(attr);
565 cfg->fc_mp_len = nla_len(attr);
566 break;
567 case RTA_FLOW:
568 cfg->fc_flow = nla_get_u32(attr);
569 break;
Thomas Graf4e902c52006-08-17 18:14:52 -0700570 case RTA_TABLE:
571 cfg->fc_table = nla_get_u32(attr);
572 break;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700573 }
574 }
Thomas Graf4e902c52006-08-17 18:14:52 -0700575
Linus Torvalds1da177e2005-04-16 15:20:36 -0700576 return 0;
Thomas Graf4e902c52006-08-17 18:14:52 -0700577errout:
578 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700579}
580
Jianjun Kong6ed25332008-11-03 00:25:16 -0800581static int inet_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700582{
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900583 struct net *net = sock_net(skb->sk);
Thomas Graf4e902c52006-08-17 18:14:52 -0700584 struct fib_config cfg;
585 struct fib_table *tb;
586 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700587
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800588 err = rtm_to_fib_config(net, skb, nlh, &cfg);
Thomas Graf4e902c52006-08-17 18:14:52 -0700589 if (err < 0)
590 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700591
Denis V. Lunev8ad49422008-01-10 03:24:11 -0800592 tb = fib_get_table(net, cfg.fc_table);
Thomas Graf4e902c52006-08-17 18:14:52 -0700593 if (tb == NULL) {
594 err = -ESRCH;
595 goto errout;
596 }
597
598 err = tb->tb_delete(tb, &cfg);
599errout:
600 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700601}
602
Jianjun Kong6ed25332008-11-03 00:25:16 -0800603static int inet_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh, void *arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700604{
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900605 struct net *net = sock_net(skb->sk);
Thomas Graf4e902c52006-08-17 18:14:52 -0700606 struct fib_config cfg;
607 struct fib_table *tb;
608 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700609
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800610 err = rtm_to_fib_config(net, skb, nlh, &cfg);
Thomas Graf4e902c52006-08-17 18:14:52 -0700611 if (err < 0)
612 goto errout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700613
Denis V. Lunev226b0b4a52008-01-10 03:30:24 -0800614 tb = fib_new_table(net, cfg.fc_table);
Thomas Graf4e902c52006-08-17 18:14:52 -0700615 if (tb == NULL) {
616 err = -ENOBUFS;
617 goto errout;
618 }
619
620 err = tb->tb_insert(tb, &cfg);
621errout:
622 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700623}
624
Thomas Graf63f34442007-03-22 11:55:17 -0700625static int inet_dump_fib(struct sk_buff *skb, struct netlink_callback *cb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700626{
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900627 struct net *net = sock_net(skb->sk);
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700628 unsigned int h, s_h;
629 unsigned int e = 0, s_e;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700630 struct fib_table *tb;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700631 struct hlist_node *node;
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800632 struct hlist_head *head;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700633 int dumped = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634
Thomas Grafbe403ea2006-08-17 18:15:17 -0700635 if (nlmsg_len(cb->nlh) >= sizeof(struct rtmsg) &&
636 ((struct rtmsg *) nlmsg_data(cb->nlh))->rtm_flags & RTM_F_CLONED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637 return ip_rt_dump(skb, cb);
638
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700639 s_h = cb->args[0];
640 s_e = cb->args[1];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700641
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700642 for (h = s_h; h < FIB_TABLE_HASHSZ; h++, s_e = 0) {
643 e = 0;
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800644 head = &net->ipv4.fib_table_hash[h];
645 hlist_for_each_entry(tb, node, head, tb_hlist) {
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700646 if (e < s_e)
647 goto next;
648 if (dumped)
649 memset(&cb->args[2], 0, sizeof(cb->args) -
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900650 2 * sizeof(cb->args[0]));
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700651 if (tb->tb_dump(tb, skb, cb) < 0)
652 goto out;
653 dumped = 1;
654next:
655 e++;
656 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700657 }
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700658out:
659 cb->args[1] = e;
660 cb->args[0] = h;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700661
662 return skb->len;
663}
664
665/* Prepare and feed intra-kernel routing request.
666 Really, it should be netlink message, but :-( netlink
667 can be not configured, so that we feed it directly
668 to fib engine. It is legal, because all events occur
669 only when netlink is already locked.
670 */
671
Al Viro81f7bf62006-09-27 18:40:00 -0700672static void fib_magic(int cmd, int type, __be32 dst, int dst_len, struct in_ifaddr *ifa)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673{
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900674 struct net *net = dev_net(ifa->ifa_dev->dev);
Thomas Graf4e902c52006-08-17 18:14:52 -0700675 struct fib_table *tb;
676 struct fib_config cfg = {
677 .fc_protocol = RTPROT_KERNEL,
678 .fc_type = type,
679 .fc_dst = dst,
680 .fc_dst_len = dst_len,
681 .fc_prefsrc = ifa->ifa_local,
682 .fc_oif = ifa->ifa_dev->dev->ifindex,
683 .fc_nlflags = NLM_F_CREATE | NLM_F_APPEND,
Denis V. Lunev4d1169c2008-01-10 03:26:13 -0800684 .fc_nlinfo = {
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800685 .nl_net = net,
Denis V. Lunev4d1169c2008-01-10 03:26:13 -0800686 },
Thomas Graf4e902c52006-08-17 18:14:52 -0700687 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700688
689 if (type == RTN_UNICAST)
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800690 tb = fib_new_table(net, RT_TABLE_MAIN);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700691 else
Denis V. Lunev4b5d47d2008-01-10 03:29:23 -0800692 tb = fib_new_table(net, RT_TABLE_LOCAL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700693
694 if (tb == NULL)
695 return;
696
Thomas Graf4e902c52006-08-17 18:14:52 -0700697 cfg.fc_table = tb->tb_id;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698
Thomas Graf4e902c52006-08-17 18:14:52 -0700699 if (type != RTN_LOCAL)
700 cfg.fc_scope = RT_SCOPE_LINK;
701 else
702 cfg.fc_scope = RT_SCOPE_HOST;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700703
704 if (cmd == RTM_NEWROUTE)
Thomas Graf4e902c52006-08-17 18:14:52 -0700705 tb->tb_insert(tb, &cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706 else
Thomas Graf4e902c52006-08-17 18:14:52 -0700707 tb->tb_delete(tb, &cfg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700708}
709
Jamal Hadi Salim0ff60a42005-11-22 14:47:37 -0800710void fib_add_ifaddr(struct in_ifaddr *ifa)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700711{
712 struct in_device *in_dev = ifa->ifa_dev;
713 struct net_device *dev = in_dev->dev;
714 struct in_ifaddr *prim = ifa;
Al Viroa144ea42006-09-28 18:00:55 -0700715 __be32 mask = ifa->ifa_mask;
716 __be32 addr = ifa->ifa_local;
717 __be32 prefix = ifa->ifa_address&mask;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700718
719 if (ifa->ifa_flags&IFA_F_SECONDARY) {
720 prim = inet_ifa_byprefix(in_dev, prefix, mask);
721 if (prim == NULL) {
Stephen Hemmingera6db9012008-01-12 20:58:35 -0800722 printk(KERN_WARNING "fib_add_ifaddr: bug: prim == NULL\n");
Linus Torvalds1da177e2005-04-16 15:20:36 -0700723 return;
724 }
725 }
726
727 fib_magic(RTM_NEWROUTE, RTN_LOCAL, addr, 32, prim);
728
729 if (!(dev->flags&IFF_UP))
730 return;
731
732 /* Add broadcast address, if it is explicitly assigned. */
Al Viroa144ea42006-09-28 18:00:55 -0700733 if (ifa->ifa_broadcast && ifa->ifa_broadcast != htonl(0xFFFFFFFF))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734 fib_magic(RTM_NEWROUTE, RTN_BROADCAST, ifa->ifa_broadcast, 32, prim);
735
Joe Perchesf97c1e02007-12-16 13:45:43 -0800736 if (!ipv4_is_zeronet(prefix) && !(ifa->ifa_flags&IFA_F_SECONDARY) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -0700737 (prefix != addr || ifa->ifa_prefixlen < 32)) {
738 fib_magic(RTM_NEWROUTE, dev->flags&IFF_LOOPBACK ? RTN_LOCAL :
739 RTN_UNICAST, prefix, ifa->ifa_prefixlen, prim);
740
741 /* Add network specific broadcasts, when it takes a sense */
742 if (ifa->ifa_prefixlen < 31) {
743 fib_magic(RTM_NEWROUTE, RTN_BROADCAST, prefix, 32, prim);
744 fib_magic(RTM_NEWROUTE, RTN_BROADCAST, prefix|~mask, 32, prim);
745 }
746 }
747}
748
749static void fib_del_ifaddr(struct in_ifaddr *ifa)
750{
751 struct in_device *in_dev = ifa->ifa_dev;
752 struct net_device *dev = in_dev->dev;
753 struct in_ifaddr *ifa1;
754 struct in_ifaddr *prim = ifa;
Al Viroa144ea42006-09-28 18:00:55 -0700755 __be32 brd = ifa->ifa_address|~ifa->ifa_mask;
756 __be32 any = ifa->ifa_address&ifa->ifa_mask;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700757#define LOCAL_OK 1
758#define BRD_OK 2
759#define BRD0_OK 4
760#define BRD1_OK 8
761 unsigned ok = 0;
762
763 if (!(ifa->ifa_flags&IFA_F_SECONDARY))
764 fib_magic(RTM_DELROUTE, dev->flags&IFF_LOOPBACK ? RTN_LOCAL :
765 RTN_UNICAST, any, ifa->ifa_prefixlen, prim);
766 else {
767 prim = inet_ifa_byprefix(in_dev, any, ifa->ifa_mask);
768 if (prim == NULL) {
Stephen Hemmingera6db9012008-01-12 20:58:35 -0800769 printk(KERN_WARNING "fib_del_ifaddr: bug: prim == NULL\n");
Linus Torvalds1da177e2005-04-16 15:20:36 -0700770 return;
771 }
772 }
773
774 /* Deletion is more complicated than add.
775 We should take care of not to delete too much :-)
776
777 Scan address list to be sure that addresses are really gone.
778 */
779
780 for (ifa1 = in_dev->ifa_list; ifa1; ifa1 = ifa1->ifa_next) {
781 if (ifa->ifa_local == ifa1->ifa_local)
782 ok |= LOCAL_OK;
783 if (ifa->ifa_broadcast == ifa1->ifa_broadcast)
784 ok |= BRD_OK;
785 if (brd == ifa1->ifa_broadcast)
786 ok |= BRD1_OK;
787 if (any == ifa1->ifa_broadcast)
788 ok |= BRD0_OK;
789 }
790
791 if (!(ok&BRD_OK))
792 fib_magic(RTM_DELROUTE, RTN_BROADCAST, ifa->ifa_broadcast, 32, prim);
793 if (!(ok&BRD1_OK))
794 fib_magic(RTM_DELROUTE, RTN_BROADCAST, brd, 32, prim);
795 if (!(ok&BRD0_OK))
796 fib_magic(RTM_DELROUTE, RTN_BROADCAST, any, 32, prim);
797 if (!(ok&LOCAL_OK)) {
798 fib_magic(RTM_DELROUTE, RTN_LOCAL, ifa->ifa_local, 32, prim);
799
800 /* Check, that this local address finally disappeared. */
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900801 if (inet_addr_type(dev_net(dev), ifa->ifa_local) != RTN_LOCAL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802 /* And the last, but not the least thing.
803 We must flush stray FIB entries.
804
805 First of all, we scan fib_info list searching
806 for stray nexthop entries, then ignite fib_flush.
807 */
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900808 if (fib_sync_down_addr(dev_net(dev), ifa->ifa_local))
809 fib_flush(dev_net(dev));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700810 }
811 }
812#undef LOCAL_OK
813#undef BRD_OK
814#undef BRD0_OK
815#undef BRD1_OK
816}
817
Robert Olsson246955f2005-06-20 13:36:39 -0700818static void nl_fib_lookup(struct fib_result_nl *frn, struct fib_table *tb )
819{
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900820
Robert Olsson246955f2005-06-20 13:36:39 -0700821 struct fib_result res;
Thomas Graf5f3008932006-11-09 15:21:41 -0800822 struct flowi fl = { .mark = frn->fl_mark,
Thomas Graf47dcf0c2006-11-09 15:20:38 -0800823 .nl_u = { .ip4_u = { .daddr = frn->fl_addr,
Robert Olsson246955f2005-06-20 13:36:39 -0700824 .tos = frn->fl_tos,
825 .scope = frn->fl_scope } } };
Alexey Kuznetsov1194ed02007-04-25 13:07:28 -0700826
Sergey Vlasov912a41a2007-04-27 02:17:19 -0700827#ifdef CONFIG_IP_MULTIPLE_TABLES
828 res.r = NULL;
829#endif
830
Alexey Kuznetsov1194ed02007-04-25 13:07:28 -0700831 frn->err = -ENOENT;
Robert Olsson246955f2005-06-20 13:36:39 -0700832 if (tb) {
833 local_bh_disable();
834
835 frn->tb_id = tb->tb_id;
836 frn->err = tb->tb_lookup(tb, &fl, &res);
837
838 if (!frn->err) {
839 frn->prefixlen = res.prefixlen;
840 frn->nh_sel = res.nh_sel;
841 frn->type = res.type;
842 frn->scope = res.scope;
Alexey Kuznetsov1194ed02007-04-25 13:07:28 -0700843 fib_res_put(&res);
Robert Olsson246955f2005-06-20 13:36:39 -0700844 }
845 local_bh_enable();
846 }
847}
848
David S. Miller28f7b0362007-10-10 21:32:39 -0700849static void nl_fib_input(struct sk_buff *skb)
Robert Olsson246955f2005-06-20 13:36:39 -0700850{
Denis V. Lunev6bd48fc2008-01-10 03:28:55 -0800851 struct net *net;
Robert Olsson246955f2005-06-20 13:36:39 -0700852 struct fib_result_nl *frn;
David S. Miller28f7b0362007-10-10 21:32:39 -0700853 struct nlmsghdr *nlh;
Robert Olsson246955f2005-06-20 13:36:39 -0700854 struct fib_table *tb;
David S. Miller28f7b0362007-10-10 21:32:39 -0700855 u32 pid;
Alexey Kuznetsov1194ed02007-04-25 13:07:28 -0700856
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900857 net = sock_net(skb->sk);
Arnaldo Carvalho de Melob529ccf2007-04-25 19:08:35 -0700858 nlh = nlmsg_hdr(skb);
Thomas Grafea865752005-12-01 14:30:00 -0800859 if (skb->len < NLMSG_SPACE(0) || skb->len < nlh->nlmsg_len ||
Denis V. Lunevd883a032007-12-21 02:01:53 -0800860 nlh->nlmsg_len < NLMSG_LENGTH(sizeof(*frn)))
Thomas Grafea865752005-12-01 14:30:00 -0800861 return;
Denis V. Lunevd883a032007-12-21 02:01:53 -0800862
863 skb = skb_clone(skb, GFP_KERNEL);
864 if (skb == NULL)
865 return;
866 nlh = nlmsg_hdr(skb);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900867
Robert Olsson246955f2005-06-20 13:36:39 -0700868 frn = (struct fib_result_nl *) NLMSG_DATA(nlh);
Denis V. Lunev6bd48fc2008-01-10 03:28:55 -0800869 tb = fib_get_table(net, frn->tb_id_in);
Robert Olsson246955f2005-06-20 13:36:39 -0700870
871 nl_fib_lookup(frn, tb);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900872
Alexey Kuznetsov1194ed02007-04-25 13:07:28 -0700873 pid = NETLINK_CB(skb).pid; /* pid of sending process */
Robert Olsson246955f2005-06-20 13:36:39 -0700874 NETLINK_CB(skb).pid = 0; /* from kernel */
Patrick McHardyac6d4392005-08-14 19:29:52 -0700875 NETLINK_CB(skb).dst_group = 0; /* unicast */
Denis V. Lunev6bd48fc2008-01-10 03:28:55 -0800876 netlink_unicast(net->ipv4.fibnl, skb, pid, MSG_DONTWAIT);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900877}
Robert Olsson246955f2005-06-20 13:36:39 -0700878
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -0800879static int nl_fib_lookup_init(struct net *net)
Robert Olsson246955f2005-06-20 13:36:39 -0700880{
Denis V. Lunev6bd48fc2008-01-10 03:28:55 -0800881 struct sock *sk;
882 sk = netlink_kernel_create(net, NETLINK_FIB_LOOKUP, 0,
883 nl_fib_input, NULL, THIS_MODULE);
884 if (sk == NULL)
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -0800885 return -EAFNOSUPPORT;
Denis V. Lunev6bd48fc2008-01-10 03:28:55 -0800886 net->ipv4.fibnl = sk;
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -0800887 return 0;
888}
889
890static void nl_fib_lookup_exit(struct net *net)
891{
Denis V. Lunevb7c6ba62008-01-28 14:41:19 -0800892 netlink_kernel_release(net->ipv4.fibnl);
Denis V. Lunev775516b2008-01-18 23:55:19 -0800893 net->ipv4.fibnl = NULL;
Robert Olsson246955f2005-06-20 13:36:39 -0700894}
895
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896static void fib_disable_ip(struct net_device *dev, int force)
897{
Denis V. Lunev85326fa2008-01-31 18:48:47 -0800898 if (fib_sync_down_dev(dev, force))
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900899 fib_flush(dev_net(dev));
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700900 rt_cache_flush(dev_net(dev), 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700901 arp_ifdown(dev);
902}
903
904static int fib_inetaddr_event(struct notifier_block *this, unsigned long event, void *ptr)
905{
Jianjun Kong6ed25332008-11-03 00:25:16 -0800906 struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700907 struct net_device *dev = ifa->ifa_dev->dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700908
909 switch (event) {
910 case NETDEV_UP:
911 fib_add_ifaddr(ifa);
912#ifdef CONFIG_IP_ROUTE_MULTIPATH
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700913 fib_sync_up(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700914#endif
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700915 rt_cache_flush(dev_net(dev), -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700916 break;
917 case NETDEV_DOWN:
918 fib_del_ifaddr(ifa);
Jayachandran C9fcc2e82005-10-27 15:10:01 -0700919 if (ifa->ifa_dev->ifa_list == NULL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700920 /* Last address was deleted from this interface.
921 Disable IP.
922 */
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700923 fib_disable_ip(dev, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700924 } else {
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700925 rt_cache_flush(dev_net(dev), -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700926 }
927 break;
928 }
929 return NOTIFY_DONE;
930}
931
932static int fib_netdev_event(struct notifier_block *this, unsigned long event, void *ptr)
933{
934 struct net_device *dev = ptr;
Herbert Xue5ed6392005-10-03 14:35:55 -0700935 struct in_device *in_dev = __in_dev_get_rtnl(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936
937 if (event == NETDEV_UNREGISTER) {
938 fib_disable_ip(dev, 2);
939 return NOTIFY_DONE;
940 }
941
942 if (!in_dev)
943 return NOTIFY_DONE;
944
945 switch (event) {
946 case NETDEV_UP:
947 for_ifa(in_dev) {
948 fib_add_ifaddr(ifa);
949 } endfor_ifa(in_dev);
950#ifdef CONFIG_IP_ROUTE_MULTIPATH
951 fib_sync_up(dev);
952#endif
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700953 rt_cache_flush(dev_net(dev), -1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700954 break;
955 case NETDEV_DOWN:
956 fib_disable_ip(dev, 0);
957 break;
958 case NETDEV_CHANGEMTU:
959 case NETDEV_CHANGE:
Denis V. Lunev76e6ebf2008-07-05 19:00:44 -0700960 rt_cache_flush(dev_net(dev), 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700961 break;
962 }
963 return NOTIFY_DONE;
964}
965
966static struct notifier_block fib_inetaddr_notifier = {
Jianjun Kong6ed25332008-11-03 00:25:16 -0800967 .notifier_call = fib_inetaddr_event,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968};
969
970static struct notifier_block fib_netdev_notifier = {
Jianjun Kong6ed25332008-11-03 00:25:16 -0800971 .notifier_call = fib_netdev_event,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972};
973
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -0800974static int __net_init ip_fib_net_init(struct net *net)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700975{
Denis V. Lunevdce5cbe2008-01-31 18:44:53 -0800976 int err;
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700977 unsigned int i;
978
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800979 net->ipv4.fib_table_hash = kzalloc(
980 sizeof(struct hlist_head)*FIB_TABLE_HASHSZ, GFP_KERNEL);
981 if (net->ipv4.fib_table_hash == NULL)
982 return -ENOMEM;
983
Patrick McHardy1af5a8c2006-08-10 23:10:46 -0700984 for (i = 0; i < FIB_TABLE_HASHSZ; i++)
Denis V. Luneve4aef8a2008-01-10 03:28:24 -0800985 INIT_HLIST_HEAD(&net->ipv4.fib_table_hash[i]);
Pavel Emelyanovc3e9a352007-11-06 23:34:04 -0800986
Denis V. Lunevdce5cbe2008-01-31 18:44:53 -0800987 err = fib4_rules_init(net);
988 if (err < 0)
989 goto fail;
990 return 0;
991
992fail:
993 kfree(net->ipv4.fib_table_hash);
994 return err;
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -0800995}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700996
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -0800997static void __net_exit ip_fib_net_exit(struct net *net)
998{
999 unsigned int i;
Thomas Graf63f34442007-03-22 11:55:17 -07001000
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -08001001#ifdef CONFIG_IP_MULTIPLE_TABLES
1002 fib4_rules_exit(net);
1003#endif
1004
1005 for (i = 0; i < FIB_TABLE_HASHSZ; i++) {
1006 struct fib_table *tb;
1007 struct hlist_head *head;
1008 struct hlist_node *node, *tmp;
1009
Denis V. Luneve4aef8a2008-01-10 03:28:24 -08001010 head = &net->ipv4.fib_table_hash[i];
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -08001011 hlist_for_each_entry_safe(tb, node, tmp, head, tb_hlist) {
1012 hlist_del(node);
1013 tb->tb_flush(tb);
1014 kfree(tb);
1015 }
1016 }
Denis V. Luneve4aef8a2008-01-10 03:28:24 -08001017 kfree(net->ipv4.fib_table_hash);
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -08001018}
1019
1020static int __net_init fib_net_init(struct net *net)
1021{
1022 int error;
1023
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -08001024 error = ip_fib_net_init(net);
1025 if (error < 0)
1026 goto out;
1027 error = nl_fib_lookup_init(net);
1028 if (error < 0)
1029 goto out_nlfl;
1030 error = fib_proc_init(net);
1031 if (error < 0)
1032 goto out_proc;
1033out:
1034 return error;
1035
1036out_proc:
1037 nl_fib_lookup_exit(net);
1038out_nlfl:
1039 ip_fib_net_exit(net);
1040 goto out;
1041}
1042
1043static void __net_exit fib_net_exit(struct net *net)
1044{
1045 fib_proc_exit(net);
1046 nl_fib_lookup_exit(net);
1047 ip_fib_net_exit(net);
1048}
1049
1050static struct pernet_operations fib_net_ops = {
1051 .init = fib_net_init,
1052 .exit = fib_net_exit,
1053};
1054
1055void __init ip_fib_init(void)
1056{
Thomas Graf63f34442007-03-22 11:55:17 -07001057 rtnl_register(PF_INET, RTM_NEWROUTE, inet_rtm_newroute, NULL);
1058 rtnl_register(PF_INET, RTM_DELROUTE, inet_rtm_delroute, NULL);
1059 rtnl_register(PF_INET, RTM_GETROUTE, NULL, inet_dump_fib);
Denis V. Lunev7b1a74f2008-01-10 03:22:17 -08001060
1061 register_pernet_subsys(&fib_net_ops);
1062 register_netdevice_notifier(&fib_netdev_notifier);
1063 register_inetaddr_notifier(&fib_inetaddr_notifier);
Stephen Hemminger7f9b8052008-01-14 23:14:20 -08001064
1065 fib_hash_init();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001066}
1067
1068EXPORT_SYMBOL(inet_addr_type);
Laszlo Attila Toth05538112007-12-04 23:28:46 -08001069EXPORT_SYMBOL(inet_dev_addr_type);
Sean Heftya1e87332006-06-17 20:37:28 -07001070EXPORT_SYMBOL(ip_dev_find);