blob: 0394a8e6d978e55dcf267e804c7d01cd268591d7 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * IP multicast routing support for mrouted 3.6/3.8
3 *
Alan Cox113aa832008-10-13 19:01:08 -07004 * (c) 1995 Alan Cox, <alan@lxorguk.ukuu.org.uk>
Linus Torvalds1da177e2005-04-16 15:20:36 -07005 * Linux Consultancy and Custom Driver Development
6 *
7 * This program is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU General Public License
9 * as published by the Free Software Foundation; either version
10 * 2 of the License, or (at your option) any later version.
11 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070012 * Fixes:
13 * Michael Chastain : Incorrect size of copying.
14 * Alan Cox : Added the cache manager code
15 * Alan Cox : Fixed the clone/copy bug and device race.
16 * Mike McLagan : Routing by source
17 * Malcolm Beattie : Buffer handling fixes.
18 * Alexey Kuznetsov : Double buffer free and other fixes.
19 * SVR Anand : Fixed several multicast bugs and problems.
20 * Alexey Kuznetsov : Status, optimisations and more.
21 * Brad Parker : Better behaviour on mrouted upcall
22 * overflow.
23 * Carlos Picoto : PIMv1 Support
24 * Pavlin Ivanov Radoslavov: PIMv2 Registers must checksum only PIM header
Gilles Espinassef77f13e2010-03-29 15:41:47 +020025 * Relax this requirement to work with older peers.
Linus Torvalds1da177e2005-04-16 15:20:36 -070026 *
27 */
28
Linus Torvalds1da177e2005-04-16 15:20:36 -070029#include <asm/uaccess.h>
30#include <linux/types.h>
Randy Dunlap4fc268d2006-01-11 12:17:47 -080031#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070032#include <linux/errno.h>
33#include <linux/timer.h>
34#include <linux/mm.h>
35#include <linux/kernel.h>
36#include <linux/fcntl.h>
37#include <linux/stat.h>
38#include <linux/socket.h>
39#include <linux/in.h>
40#include <linux/inet.h>
41#include <linux/netdevice.h>
42#include <linux/inetdevice.h>
43#include <linux/igmp.h>
44#include <linux/proc_fs.h>
45#include <linux/seq_file.h>
46#include <linux/mroute.h>
47#include <linux/init.h>
Kris Katterjohn46f25df2006-01-05 16:35:42 -080048#include <linux/if_ether.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090049#include <linux/slab.h>
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020050#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070051#include <net/ip.h>
52#include <net/protocol.h>
53#include <linux/skbuff.h>
Arnaldo Carvalho de Melo14c85022005-12-27 02:43:12 -020054#include <net/route.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070055#include <net/sock.h>
56#include <net/icmp.h>
57#include <net/udp.h>
58#include <net/raw.h>
59#include <linux/notifier.h>
60#include <linux/if_arp.h>
61#include <linux/netfilter_ipv4.h>
Eric W. Biederman709b46e2011-01-29 16:15:56 +000062#include <linux/compat.h>
Paul Gortmakerbc3b2d72011-07-15 11:47:34 -040063#include <linux/export.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070064#include <net/ipip.h>
65#include <net/checksum.h>
Arnaldo Carvalho de Melodc5fc572007-03-25 23:06:12 -070066#include <net/netlink.h>
Patrick McHardyf0ad0862010-04-13 05:03:23 +000067#include <net/fib_rules.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070068
69#if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2)
70#define CONFIG_IP_PIMSM 1
71#endif
72
Patrick McHardy0c122952010-04-13 05:03:22 +000073struct mr_table {
Patrick McHardyf0ad0862010-04-13 05:03:23 +000074 struct list_head list;
Patrick McHardy8de53df2010-04-15 13:29:28 +020075#ifdef CONFIG_NET_NS
76 struct net *net;
77#endif
Patrick McHardyf0ad0862010-04-13 05:03:23 +000078 u32 id;
Eric Dumazet4c968702010-10-01 16:15:01 +000079 struct sock __rcu *mroute_sk;
Patrick McHardy0c122952010-04-13 05:03:22 +000080 struct timer_list ipmr_expire_timer;
81 struct list_head mfc_unres_queue;
82 struct list_head mfc_cache_array[MFC_LINES];
83 struct vif_device vif_table[MAXVIFS];
84 int maxvif;
85 atomic_t cache_resolve_queue_len;
86 int mroute_do_assert;
87 int mroute_do_pim;
88#if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2)
89 int mroute_reg_vif_num;
90#endif
91};
92
Patrick McHardyf0ad0862010-04-13 05:03:23 +000093struct ipmr_rule {
94 struct fib_rule common;
95};
96
97struct ipmr_result {
98 struct mr_table *mrt;
99};
100
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101/* Big lock, protecting vif table, mrt cache and mroute socket state.
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000102 * Note that the changes are semaphored via rtnl_lock.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103 */
104
105static DEFINE_RWLOCK(mrt_lock);
106
107/*
108 * Multicast router control variables
109 */
110
Patrick McHardy0c122952010-04-13 05:03:22 +0000111#define VIF_EXISTS(_mrt, _idx) ((_mrt)->vif_table[_idx].dev != NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700112
113/* Special spinlock for queue of unresolved entries */
114static DEFINE_SPINLOCK(mfc_unres_lock);
115
116/* We return to original Alan's scheme. Hash table of resolved
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000117 * entries is changed only in process context and protected
118 * with weak lock mrt_lock. Queue of unresolved entries is protected
119 * with strong spinlock mfc_unres_lock.
120 *
121 * In this case data path is free of exclusive locks at all.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122 */
123
Christoph Lametere18b8902006-12-06 20:33:20 -0800124static struct kmem_cache *mrt_cachep __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000126static struct mr_table *ipmr_new_table(struct net *net, u32 id);
Francesco Ruggeriacbb2192012-08-24 07:38:35 +0000127static void ipmr_free_table(struct mr_table *mrt);
128
Patrick McHardy0c122952010-04-13 05:03:22 +0000129static int ip_mr_forward(struct net *net, struct mr_table *mrt,
130 struct sk_buff *skb, struct mfc_cache *cache,
131 int local);
132static int ipmr_cache_report(struct mr_table *mrt,
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000133 struct sk_buff *pkt, vifi_t vifi, int assert);
Patrick McHardycb6a4e42010-04-26 16:02:08 +0200134static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
135 struct mfc_cache *c, struct rtmsg *rtm);
Francesco Ruggeriacbb2192012-08-24 07:38:35 +0000136static void mroute_clean_tables(struct mr_table *mrt);
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000137static void ipmr_expire_process(unsigned long arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700138
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000139#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
140#define ipmr_for_each_table(mrt, net) \
141 list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list)
142
143static struct mr_table *ipmr_get_table(struct net *net, u32 id)
144{
145 struct mr_table *mrt;
146
147 ipmr_for_each_table(mrt, net) {
148 if (mrt->id == id)
149 return mrt;
150 }
151 return NULL;
152}
153
David S. Millerda919812011-03-12 02:04:50 -0500154static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4,
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000155 struct mr_table **mrt)
156{
157 struct ipmr_result res;
158 struct fib_lookup_arg arg = { .result = &res, };
159 int err;
160
David S. Millerda919812011-03-12 02:04:50 -0500161 err = fib_rules_lookup(net->ipv4.mr_rules_ops,
162 flowi4_to_flowi(flp4), 0, &arg);
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000163 if (err < 0)
164 return err;
165 *mrt = res.mrt;
166 return 0;
167}
168
169static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp,
170 int flags, struct fib_lookup_arg *arg)
171{
172 struct ipmr_result *res = arg->result;
173 struct mr_table *mrt;
174
175 switch (rule->action) {
176 case FR_ACT_TO_TBL:
177 break;
178 case FR_ACT_UNREACHABLE:
179 return -ENETUNREACH;
180 case FR_ACT_PROHIBIT:
181 return -EACCES;
182 case FR_ACT_BLACKHOLE:
183 default:
184 return -EINVAL;
185 }
186
187 mrt = ipmr_get_table(rule->fr_net, rule->table);
188 if (mrt == NULL)
189 return -EAGAIN;
190 res->mrt = mrt;
191 return 0;
192}
193
194static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags)
195{
196 return 1;
197}
198
199static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = {
200 FRA_GENERIC_POLICY,
201};
202
203static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
204 struct fib_rule_hdr *frh, struct nlattr **tb)
205{
206 return 0;
207}
208
209static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
210 struct nlattr **tb)
211{
212 return 1;
213}
214
215static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
216 struct fib_rule_hdr *frh)
217{
218 frh->dst_len = 0;
219 frh->src_len = 0;
220 frh->tos = 0;
221 return 0;
222}
223
Andi Kleen04a6f822012-10-04 17:12:11 -0700224static const struct fib_rules_ops __net_initconst ipmr_rules_ops_template = {
Patrick McHardy25239ce2010-04-26 16:02:05 +0200225 .family = RTNL_FAMILY_IPMR,
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000226 .rule_size = sizeof(struct ipmr_rule),
227 .addr_size = sizeof(u32),
228 .action = ipmr_rule_action,
229 .match = ipmr_rule_match,
230 .configure = ipmr_rule_configure,
231 .compare = ipmr_rule_compare,
232 .default_pref = fib_default_rule_pref,
233 .fill = ipmr_rule_fill,
234 .nlgroup = RTNLGRP_IPV4_RULE,
235 .policy = ipmr_rule_policy,
236 .owner = THIS_MODULE,
237};
238
239static int __net_init ipmr_rules_init(struct net *net)
240{
241 struct fib_rules_ops *ops;
242 struct mr_table *mrt;
243 int err;
244
245 ops = fib_rules_register(&ipmr_rules_ops_template, net);
246 if (IS_ERR(ops))
247 return PTR_ERR(ops);
248
249 INIT_LIST_HEAD(&net->ipv4.mr_tables);
250
251 mrt = ipmr_new_table(net, RT_TABLE_DEFAULT);
252 if (mrt == NULL) {
253 err = -ENOMEM;
254 goto err1;
255 }
256
257 err = fib_default_rule_add(ops, 0x7fff, RT_TABLE_DEFAULT, 0);
258 if (err < 0)
259 goto err2;
260
261 net->ipv4.mr_rules_ops = ops;
262 return 0;
263
264err2:
265 kfree(mrt);
266err1:
267 fib_rules_unregister(ops);
268 return err;
269}
270
271static void __net_exit ipmr_rules_exit(struct net *net)
272{
273 struct mr_table *mrt, *next;
274
Eric Dumazet035320d2010-06-06 23:48:40 +0000275 list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) {
276 list_del(&mrt->list);
Francesco Ruggeriacbb2192012-08-24 07:38:35 +0000277 ipmr_free_table(mrt);
Eric Dumazet035320d2010-06-06 23:48:40 +0000278 }
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000279 fib_rules_unregister(net->ipv4.mr_rules_ops);
280}
281#else
282#define ipmr_for_each_table(mrt, net) \
283 for (mrt = net->ipv4.mrt; mrt; mrt = NULL)
284
285static struct mr_table *ipmr_get_table(struct net *net, u32 id)
286{
287 return net->ipv4.mrt;
288}
289
David S. Millerda919812011-03-12 02:04:50 -0500290static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4,
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000291 struct mr_table **mrt)
292{
293 *mrt = net->ipv4.mrt;
294 return 0;
295}
296
297static int __net_init ipmr_rules_init(struct net *net)
298{
299 net->ipv4.mrt = ipmr_new_table(net, RT_TABLE_DEFAULT);
300 return net->ipv4.mrt ? 0 : -ENOMEM;
301}
302
303static void __net_exit ipmr_rules_exit(struct net *net)
304{
Francesco Ruggeriacbb2192012-08-24 07:38:35 +0000305 ipmr_free_table(net->ipv4.mrt);
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000306}
307#endif
308
309static struct mr_table *ipmr_new_table(struct net *net, u32 id)
310{
311 struct mr_table *mrt;
312 unsigned int i;
313
314 mrt = ipmr_get_table(net, id);
315 if (mrt != NULL)
316 return mrt;
317
318 mrt = kzalloc(sizeof(*mrt), GFP_KERNEL);
319 if (mrt == NULL)
320 return NULL;
Patrick McHardy8de53df2010-04-15 13:29:28 +0200321 write_pnet(&mrt->net, net);
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000322 mrt->id = id;
323
324 /* Forwarding cache */
325 for (i = 0; i < MFC_LINES; i++)
326 INIT_LIST_HEAD(&mrt->mfc_cache_array[i]);
327
328 INIT_LIST_HEAD(&mrt->mfc_unres_queue);
329
330 setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process,
331 (unsigned long)mrt);
332
333#ifdef CONFIG_IP_PIMSM
334 mrt->mroute_reg_vif_num = -1;
335#endif
336#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
337 list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables);
338#endif
339 return mrt;
340}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341
Francesco Ruggeriacbb2192012-08-24 07:38:35 +0000342static void ipmr_free_table(struct mr_table *mrt)
343{
344 del_timer_sync(&mrt->ipmr_expire_timer);
345 mroute_clean_tables(mrt);
346 kfree(mrt);
347}
348
Linus Torvalds1da177e2005-04-16 15:20:36 -0700349/* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */
350
Wang Chend6070322008-07-14 20:55:26 -0700351static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v)
352{
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000353 struct net *net = dev_net(dev);
354
Wang Chend6070322008-07-14 20:55:26 -0700355 dev_close(dev);
356
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000357 dev = __dev_get_by_name(net, "tunl0");
Wang Chend6070322008-07-14 20:55:26 -0700358 if (dev) {
Stephen Hemminger5bc3eb72008-11-19 21:52:05 -0800359 const struct net_device_ops *ops = dev->netdev_ops;
Wang Chend6070322008-07-14 20:55:26 -0700360 struct ifreq ifr;
Wang Chend6070322008-07-14 20:55:26 -0700361 struct ip_tunnel_parm p;
362
363 memset(&p, 0, sizeof(p));
364 p.iph.daddr = v->vifc_rmt_addr.s_addr;
365 p.iph.saddr = v->vifc_lcl_addr.s_addr;
366 p.iph.version = 4;
367 p.iph.ihl = 5;
368 p.iph.protocol = IPPROTO_IPIP;
369 sprintf(p.name, "dvmrp%d", v->vifc_vifi);
370 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
371
Stephen Hemminger5bc3eb72008-11-19 21:52:05 -0800372 if (ops->ndo_do_ioctl) {
373 mm_segment_t oldfs = get_fs();
374
375 set_fs(KERNEL_DS);
376 ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL);
377 set_fs(oldfs);
378 }
Wang Chend6070322008-07-14 20:55:26 -0700379 }
380}
381
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382static
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000383struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700384{
385 struct net_device *dev;
386
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000387 dev = __dev_get_by_name(net, "tunl0");
Linus Torvalds1da177e2005-04-16 15:20:36 -0700388
389 if (dev) {
Stephen Hemminger5bc3eb72008-11-19 21:52:05 -0800390 const struct net_device_ops *ops = dev->netdev_ops;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700391 int err;
392 struct ifreq ifr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700393 struct ip_tunnel_parm p;
394 struct in_device *in_dev;
395
396 memset(&p, 0, sizeof(p));
397 p.iph.daddr = v->vifc_rmt_addr.s_addr;
398 p.iph.saddr = v->vifc_lcl_addr.s_addr;
399 p.iph.version = 4;
400 p.iph.ihl = 5;
401 p.iph.protocol = IPPROTO_IPIP;
402 sprintf(p.name, "dvmrp%d", v->vifc_vifi);
Stephen Hemmingerba93ef72008-01-21 17:28:59 -0800403 ifr.ifr_ifru.ifru_data = (__force void __user *)&p;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700404
Stephen Hemminger5bc3eb72008-11-19 21:52:05 -0800405 if (ops->ndo_do_ioctl) {
406 mm_segment_t oldfs = get_fs();
407
408 set_fs(KERNEL_DS);
409 err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL);
410 set_fs(oldfs);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000411 } else {
Stephen Hemminger5bc3eb72008-11-19 21:52:05 -0800412 err = -EOPNOTSUPP;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000413 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 dev = NULL;
415
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000416 if (err == 0 &&
417 (dev = __dev_get_by_name(net, p.name)) != NULL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418 dev->flags |= IFF_MULTICAST;
419
Herbert Xue5ed6392005-10-03 14:35:55 -0700420 in_dev = __in_dev_get_rtnl(dev);
Herbert Xu71e27da2007-06-04 23:36:06 -0700421 if (in_dev == NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700422 goto failure;
Herbert Xu71e27da2007-06-04 23:36:06 -0700423
424 ipv4_devconf_setall(in_dev);
425 IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700426
427 if (dev_open(dev))
428 goto failure;
Wang Chen7dc00c82008-07-14 20:56:34 -0700429 dev_hold(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700430 }
431 }
432 return dev;
433
434failure:
435 /* allow the register to be completed before unregistering. */
436 rtnl_unlock();
437 rtnl_lock();
438
439 unregister_netdevice(dev);
440 return NULL;
441}
442
443#ifdef CONFIG_IP_PIMSM
444
Stephen Hemminger6fef4c02009-08-31 19:50:41 +0000445static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700446{
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000447 struct net *net = dev_net(dev);
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000448 struct mr_table *mrt;
David S. Millerda919812011-03-12 02:04:50 -0500449 struct flowi4 fl4 = {
450 .flowi4_oif = dev->ifindex,
451 .flowi4_iif = skb->skb_iif,
452 .flowi4_mark = skb->mark,
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000453 };
454 int err;
455
David S. Millerda919812011-03-12 02:04:50 -0500456 err = ipmr_fib_lookup(net, &fl4, &mrt);
Ben Greeare40dbc52010-07-15 13:22:33 +0000457 if (err < 0) {
458 kfree_skb(skb);
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000459 return err;
Ben Greeare40dbc52010-07-15 13:22:33 +0000460 }
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000461
Linus Torvalds1da177e2005-04-16 15:20:36 -0700462 read_lock(&mrt_lock);
Pavel Emelyanovcf3677a2008-05-21 14:17:33 -0700463 dev->stats.tx_bytes += skb->len;
464 dev->stats.tx_packets++;
Patrick McHardy0c122952010-04-13 05:03:22 +0000465 ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700466 read_unlock(&mrt_lock);
467 kfree_skb(skb);
Patrick McHardy6ed10652009-06-23 06:03:08 +0000468 return NETDEV_TX_OK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469}
470
Stephen Hemminger007c3832008-11-20 20:28:35 -0800471static const struct net_device_ops reg_vif_netdev_ops = {
472 .ndo_start_xmit = reg_vif_xmit,
473};
474
Linus Torvalds1da177e2005-04-16 15:20:36 -0700475static void reg_vif_setup(struct net_device *dev)
476{
477 dev->type = ARPHRD_PIMREG;
Kris Katterjohn46f25df2006-01-05 16:35:42 -0800478 dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700479 dev->flags = IFF_NOARP;
Stephen Hemminger007c3832008-11-20 20:28:35 -0800480 dev->netdev_ops = &reg_vif_netdev_ops,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700481 dev->destructor = free_netdev;
Tom Goff403dbb92009-06-14 03:16:13 -0700482 dev->features |= NETIF_F_NETNS_LOCAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700483}
484
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000485static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486{
487 struct net_device *dev;
488 struct in_device *in_dev;
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000489 char name[IFNAMSIZ];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000491 if (mrt->id == RT_TABLE_DEFAULT)
492 sprintf(name, "pimreg");
493 else
494 sprintf(name, "pimreg%u", mrt->id);
495
496 dev = alloc_netdev(0, name, reg_vif_setup);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700497
498 if (dev == NULL)
499 return NULL;
500
Tom Goff403dbb92009-06-14 03:16:13 -0700501 dev_net_set(dev, net);
502
Linus Torvalds1da177e2005-04-16 15:20:36 -0700503 if (register_netdevice(dev)) {
504 free_netdev(dev);
505 return NULL;
506 }
507 dev->iflink = 0;
508
Herbert Xu71e27da2007-06-04 23:36:06 -0700509 rcu_read_lock();
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000510 in_dev = __in_dev_get_rcu(dev);
511 if (!in_dev) {
Herbert Xu71e27da2007-06-04 23:36:06 -0700512 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700513 goto failure;
Herbert Xu71e27da2007-06-04 23:36:06 -0700514 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515
Herbert Xu71e27da2007-06-04 23:36:06 -0700516 ipv4_devconf_setall(in_dev);
517 IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0;
518 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519
520 if (dev_open(dev))
521 goto failure;
522
Wang Chen7dc00c82008-07-14 20:56:34 -0700523 dev_hold(dev);
524
Linus Torvalds1da177e2005-04-16 15:20:36 -0700525 return dev;
526
527failure:
528 /* allow the register to be completed before unregistering. */
529 rtnl_unlock();
530 rtnl_lock();
531
532 unregister_netdevice(dev);
533 return NULL;
534}
535#endif
536
Ben Hutchings2c530402012-07-10 10:55:09 +0000537/**
538 * vif_delete - Delete a VIF entry
Wang Chen7dc00c82008-07-14 20:56:34 -0700539 * @notify: Set to 1, if the caller is a notifier_call
Linus Torvalds1da177e2005-04-16 15:20:36 -0700540 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900541
Patrick McHardy0c122952010-04-13 05:03:22 +0000542static int vif_delete(struct mr_table *mrt, int vifi, int notify,
Eric Dumazetd17fa6f2009-10-28 05:21:38 +0000543 struct list_head *head)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700544{
545 struct vif_device *v;
546 struct net_device *dev;
547 struct in_device *in_dev;
548
Patrick McHardy0c122952010-04-13 05:03:22 +0000549 if (vifi < 0 || vifi >= mrt->maxvif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700550 return -EADDRNOTAVAIL;
551
Patrick McHardy0c122952010-04-13 05:03:22 +0000552 v = &mrt->vif_table[vifi];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700553
554 write_lock_bh(&mrt_lock);
555 dev = v->dev;
556 v->dev = NULL;
557
558 if (!dev) {
559 write_unlock_bh(&mrt_lock);
560 return -EADDRNOTAVAIL;
561 }
562
563#ifdef CONFIG_IP_PIMSM
Patrick McHardy0c122952010-04-13 05:03:22 +0000564 if (vifi == mrt->mroute_reg_vif_num)
565 mrt->mroute_reg_vif_num = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700566#endif
567
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000568 if (vifi + 1 == mrt->maxvif) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700569 int tmp;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000570
571 for (tmp = vifi - 1; tmp >= 0; tmp--) {
Patrick McHardy0c122952010-04-13 05:03:22 +0000572 if (VIF_EXISTS(mrt, tmp))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700573 break;
574 }
Patrick McHardy0c122952010-04-13 05:03:22 +0000575 mrt->maxvif = tmp+1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700576 }
577
578 write_unlock_bh(&mrt_lock);
579
580 dev_set_allmulti(dev, -1);
581
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000582 in_dev = __in_dev_get_rtnl(dev);
583 if (in_dev) {
Herbert Xu42f811b2007-06-04 23:34:44 -0700584 IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700585 ip_rt_multicast_event(in_dev);
586 }
587
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000588 if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify)
Eric Dumazetd17fa6f2009-10-28 05:21:38 +0000589 unregister_netdevice_queue(dev, head);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590
591 dev_put(dev);
592 return 0;
593}
594
Eric Dumazeta8c94862010-10-01 16:15:08 +0000595static void ipmr_cache_free_rcu(struct rcu_head *head)
596{
597 struct mfc_cache *c = container_of(head, struct mfc_cache, rcu);
598
599 kmem_cache_free(mrt_cachep, c);
600}
601
Benjamin Thery5c0a66f2009-01-22 04:56:17 +0000602static inline void ipmr_cache_free(struct mfc_cache *c)
603{
Eric Dumazeta8c94862010-10-01 16:15:08 +0000604 call_rcu(&c->rcu, ipmr_cache_free_rcu);
Benjamin Thery5c0a66f2009-01-22 04:56:17 +0000605}
606
Linus Torvalds1da177e2005-04-16 15:20:36 -0700607/* Destroy an unresolved cache entry, killing queued skbs
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000608 * and reporting error to netlink readers.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700609 */
610
Patrick McHardy0c122952010-04-13 05:03:22 +0000611static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700612{
Patrick McHardy8de53df2010-04-15 13:29:28 +0200613 struct net *net = read_pnet(&mrt->net);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700614 struct sk_buff *skb;
Patrick McHardy9ef1d4c2005-06-28 12:55:30 -0700615 struct nlmsgerr *e;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700616
Patrick McHardy0c122952010-04-13 05:03:22 +0000617 atomic_dec(&mrt->cache_resolve_queue_len);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700618
Jianjun Kongc354e122008-11-03 00:28:02 -0800619 while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) {
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700620 if (ip_hdr(skb)->version == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700621 struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr));
622 nlh->nlmsg_type = NLMSG_ERROR;
623 nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr));
624 skb_trim(skb, nlh->nlmsg_len);
Patrick McHardy9ef1d4c2005-06-28 12:55:30 -0700625 e = NLMSG_DATA(nlh);
626 e->error = -ETIMEDOUT;
627 memset(&e->msg, 0, sizeof(e->msg));
Thomas Graf2942e902006-08-15 00:30:25 -0700628
Eric W. Biederman15e47302012-09-07 20:12:54 +0000629 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000630 } else {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700631 kfree_skb(skb);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000632 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633 }
634
Benjamin Thery5c0a66f2009-01-22 04:56:17 +0000635 ipmr_cache_free(c);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636}
637
638
Patrick McHardye258beb2010-04-13 05:03:19 +0000639/* Timer process for the unresolved queue. */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700640
Patrick McHardye258beb2010-04-13 05:03:19 +0000641static void ipmr_expire_process(unsigned long arg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700642{
Patrick McHardy0c122952010-04-13 05:03:22 +0000643 struct mr_table *mrt = (struct mr_table *)arg;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700644 unsigned long now;
645 unsigned long expires;
Patrick McHardy862465f2010-04-13 05:03:21 +0000646 struct mfc_cache *c, *next;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647
648 if (!spin_trylock(&mfc_unres_lock)) {
Patrick McHardy0c122952010-04-13 05:03:22 +0000649 mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700650 return;
651 }
652
Patrick McHardy0c122952010-04-13 05:03:22 +0000653 if (list_empty(&mrt->mfc_unres_queue))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700654 goto out;
655
656 now = jiffies;
657 expires = 10*HZ;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700658
Patrick McHardy0c122952010-04-13 05:03:22 +0000659 list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660 if (time_after(c->mfc_un.unres.expires, now)) {
661 unsigned long interval = c->mfc_un.unres.expires - now;
662 if (interval < expires)
663 expires = interval;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700664 continue;
665 }
666
Patrick McHardy862465f2010-04-13 05:03:21 +0000667 list_del(&c->list);
Patrick McHardy0c122952010-04-13 05:03:22 +0000668 ipmr_destroy_unres(mrt, c);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700669 }
670
Patrick McHardy0c122952010-04-13 05:03:22 +0000671 if (!list_empty(&mrt->mfc_unres_queue))
672 mod_timer(&mrt->ipmr_expire_timer, jiffies + expires);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673
674out:
675 spin_unlock(&mfc_unres_lock);
676}
677
678/* Fill oifs list. It is called under write locked mrt_lock. */
679
Patrick McHardy0c122952010-04-13 05:03:22 +0000680static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache,
Patrick McHardyd658f8a2010-04-13 05:03:20 +0000681 unsigned char *ttls)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700682{
683 int vifi;
684
685 cache->mfc_un.res.minvif = MAXVIFS;
686 cache->mfc_un.res.maxvif = 0;
687 memset(cache->mfc_un.res.ttls, 255, MAXVIFS);
688
Patrick McHardy0c122952010-04-13 05:03:22 +0000689 for (vifi = 0; vifi < mrt->maxvif; vifi++) {
690 if (VIF_EXISTS(mrt, vifi) &&
Benjamin Therycf958ae32009-01-22 04:56:16 +0000691 ttls[vifi] && ttls[vifi] < 255) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700692 cache->mfc_un.res.ttls[vifi] = ttls[vifi];
693 if (cache->mfc_un.res.minvif > vifi)
694 cache->mfc_un.res.minvif = vifi;
695 if (cache->mfc_un.res.maxvif <= vifi)
696 cache->mfc_un.res.maxvif = vifi + 1;
697 }
698 }
699}
700
Patrick McHardy0c122952010-04-13 05:03:22 +0000701static int vif_add(struct net *net, struct mr_table *mrt,
702 struct vifctl *vifc, int mrtsock)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700703{
704 int vifi = vifc->vifc_vifi;
Patrick McHardy0c122952010-04-13 05:03:22 +0000705 struct vif_device *v = &mrt->vif_table[vifi];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706 struct net_device *dev;
707 struct in_device *in_dev;
Wang Chend6070322008-07-14 20:55:26 -0700708 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700709
710 /* Is vif busy ? */
Patrick McHardy0c122952010-04-13 05:03:22 +0000711 if (VIF_EXISTS(mrt, vifi))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700712 return -EADDRINUSE;
713
714 switch (vifc->vifc_flags) {
715#ifdef CONFIG_IP_PIMSM
716 case VIFF_REGISTER:
717 /*
718 * Special Purpose VIF in PIM
719 * All the packets will be sent to the daemon
720 */
Patrick McHardy0c122952010-04-13 05:03:22 +0000721 if (mrt->mroute_reg_vif_num >= 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700722 return -EADDRINUSE;
Patrick McHardyf0ad0862010-04-13 05:03:23 +0000723 dev = ipmr_reg_vif(net, mrt);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700724 if (!dev)
725 return -ENOBUFS;
Wang Chend6070322008-07-14 20:55:26 -0700726 err = dev_set_allmulti(dev, 1);
727 if (err) {
728 unregister_netdevice(dev);
Wang Chen7dc00c82008-07-14 20:56:34 -0700729 dev_put(dev);
Wang Chend6070322008-07-14 20:55:26 -0700730 return err;
731 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732 break;
733#endif
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900734 case VIFF_TUNNEL:
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000735 dev = ipmr_new_tunnel(net, vifc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700736 if (!dev)
737 return -ENOBUFS;
Wang Chend6070322008-07-14 20:55:26 -0700738 err = dev_set_allmulti(dev, 1);
739 if (err) {
740 ipmr_del_tunnel(dev, vifc);
Wang Chen7dc00c82008-07-14 20:56:34 -0700741 dev_put(dev);
Wang Chend6070322008-07-14 20:55:26 -0700742 return err;
743 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700744 break;
Ilia Kee5e81f2009-09-16 05:53:07 +0000745
746 case VIFF_USE_IFINDEX:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747 case 0:
Ilia Kee5e81f2009-09-16 05:53:07 +0000748 if (vifc->vifc_flags == VIFF_USE_IFINDEX) {
749 dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex);
Eric Dumazet95ae6b22010-09-15 04:04:31 +0000750 if (dev && __in_dev_get_rtnl(dev) == NULL) {
Ilia Kee5e81f2009-09-16 05:53:07 +0000751 dev_put(dev);
752 return -EADDRNOTAVAIL;
753 }
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000754 } else {
Ilia Kee5e81f2009-09-16 05:53:07 +0000755 dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000756 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700757 if (!dev)
758 return -EADDRNOTAVAIL;
Wang Chend6070322008-07-14 20:55:26 -0700759 err = dev_set_allmulti(dev, 1);
Wang Chen7dc00c82008-07-14 20:56:34 -0700760 if (err) {
761 dev_put(dev);
Wang Chend6070322008-07-14 20:55:26 -0700762 return err;
Wang Chen7dc00c82008-07-14 20:56:34 -0700763 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700764 break;
765 default:
766 return -EINVAL;
767 }
768
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000769 in_dev = __in_dev_get_rtnl(dev);
770 if (!in_dev) {
Dan Carpenterd0490cf2009-11-11 02:03:54 +0000771 dev_put(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700772 return -EADDRNOTAVAIL;
Dan Carpenterd0490cf2009-11-11 02:03:54 +0000773 }
Herbert Xu42f811b2007-06-04 23:34:44 -0700774 IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700775 ip_rt_multicast_event(in_dev);
776
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000777 /* Fill in the VIF structures */
778
Jianjun Kongc354e122008-11-03 00:28:02 -0800779 v->rate_limit = vifc->vifc_rate_limit;
780 v->local = vifc->vifc_lcl_addr.s_addr;
781 v->remote = vifc->vifc_rmt_addr.s_addr;
782 v->flags = vifc->vifc_flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700783 if (!mrtsock)
784 v->flags |= VIFF_STATIC;
Jianjun Kongc354e122008-11-03 00:28:02 -0800785 v->threshold = vifc->vifc_threshold;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786 v->bytes_in = 0;
787 v->bytes_out = 0;
788 v->pkt_in = 0;
789 v->pkt_out = 0;
790 v->link = dev->ifindex;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000791 if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700792 v->link = dev->iflink;
793
794 /* And finish update writing critical data */
795 write_lock_bh(&mrt_lock);
Jianjun Kongc354e122008-11-03 00:28:02 -0800796 v->dev = dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700797#ifdef CONFIG_IP_PIMSM
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000798 if (v->flags & VIFF_REGISTER)
Patrick McHardy0c122952010-04-13 05:03:22 +0000799 mrt->mroute_reg_vif_num = vifi;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700800#endif
Patrick McHardy0c122952010-04-13 05:03:22 +0000801 if (vifi+1 > mrt->maxvif)
802 mrt->maxvif = vifi+1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700803 write_unlock_bh(&mrt_lock);
804 return 0;
805}
806
Eric Dumazeta8c94862010-10-01 16:15:08 +0000807/* called with rcu_read_lock() */
Patrick McHardy0c122952010-04-13 05:03:22 +0000808static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt,
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000809 __be32 origin,
810 __be32 mcastgrp)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700811{
Jianjun Kongc354e122008-11-03 00:28:02 -0800812 int line = MFC_HASH(mcastgrp, origin);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700813 struct mfc_cache *c;
814
Eric Dumazeta8c94862010-10-01 16:15:08 +0000815 list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) {
Patrick McHardy862465f2010-04-13 05:03:21 +0000816 if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp)
817 return c;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700818 }
Patrick McHardy862465f2010-04-13 05:03:21 +0000819 return NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700820}
821
822/*
823 * Allocate a multicast cache entry
824 */
Patrick McHardyd658f8a2010-04-13 05:03:20 +0000825static struct mfc_cache *ipmr_cache_alloc(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700826{
Jianjun Kongc354e122008-11-03 00:28:02 -0800827 struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL);
Eric Dumazeta8c94862010-10-01 16:15:08 +0000828
829 if (c)
830 c->mfc_un.res.minvif = MAXVIFS;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700831 return c;
832}
833
Patrick McHardyd658f8a2010-04-13 05:03:20 +0000834static struct mfc_cache *ipmr_cache_alloc_unres(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700835{
Jianjun Kongc354e122008-11-03 00:28:02 -0800836 struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC);
Eric Dumazeta8c94862010-10-01 16:15:08 +0000837
838 if (c) {
839 skb_queue_head_init(&c->mfc_un.unres.unresolved);
840 c->mfc_un.unres.expires = jiffies + 10*HZ;
841 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700842 return c;
843}
844
845/*
846 * A cache entry has gone into a resolved state from queued
847 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900848
Patrick McHardy0c122952010-04-13 05:03:22 +0000849static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt,
850 struct mfc_cache *uc, struct mfc_cache *c)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700851{
852 struct sk_buff *skb;
Patrick McHardy9ef1d4c2005-06-28 12:55:30 -0700853 struct nlmsgerr *e;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700854
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000855 /* Play the pending entries through our router */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700856
Jianjun Kongc354e122008-11-03 00:28:02 -0800857 while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) {
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700858 if (ip_hdr(skb)->version == 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700859 struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr));
860
Patrick McHardycb6a4e42010-04-26 16:02:08 +0200861 if (__ipmr_fill_mroute(mrt, skb, c, NLMSG_DATA(nlh)) > 0) {
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000862 nlh->nlmsg_len = skb_tail_pointer(skb) -
863 (u8 *)nlh;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700864 } else {
865 nlh->nlmsg_type = NLMSG_ERROR;
866 nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr));
867 skb_trim(skb, nlh->nlmsg_len);
Patrick McHardy9ef1d4c2005-06-28 12:55:30 -0700868 e = NLMSG_DATA(nlh);
869 e->error = -EMSGSIZE;
870 memset(&e->msg, 0, sizeof(e->msg));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700871 }
Thomas Graf2942e902006-08-15 00:30:25 -0700872
Eric W. Biederman15e47302012-09-07 20:12:54 +0000873 rtnl_unicast(skb, net, NETLINK_CB(skb).portid);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000874 } else {
Patrick McHardy0c122952010-04-13 05:03:22 +0000875 ip_mr_forward(net, mrt, skb, c, 0);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000876 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877 }
878}
879
880/*
881 * Bounce a cache query up to mrouted. We could use netlink for this but mrouted
882 * expects the following bizarre scheme.
883 *
884 * Called under mrt_lock.
885 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900886
Patrick McHardy0c122952010-04-13 05:03:22 +0000887static int ipmr_cache_report(struct mr_table *mrt,
Benjamin Thery4feb88e2009-01-22 04:56:23 +0000888 struct sk_buff *pkt, vifi_t vifi, int assert)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700889{
890 struct sk_buff *skb;
Arnaldo Carvalho de Meloc9bdd4b2007-03-12 20:09:15 -0300891 const int ihl = ip_hdrlen(pkt);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700892 struct igmphdr *igmp;
893 struct igmpmsg *msg;
Eric Dumazet4c968702010-10-01 16:15:01 +0000894 struct sock *mroute_sk;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700895 int ret;
896
897#ifdef CONFIG_IP_PIMSM
898 if (assert == IGMPMSG_WHOLEPKT)
899 skb = skb_realloc_headroom(pkt, sizeof(struct iphdr));
900 else
901#endif
902 skb = alloc_skb(128, GFP_ATOMIC);
903
Stephen Hemminger132adf52007-03-08 20:44:43 -0800904 if (!skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700905 return -ENOBUFS;
906
907#ifdef CONFIG_IP_PIMSM
908 if (assert == IGMPMSG_WHOLEPKT) {
909 /* Ugly, but we have no choice with this interface.
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000910 * Duplicate old header, fix ihl, length etc.
911 * And all this only to mangle msg->im_msgtype and
912 * to set msg->im_mbz to "mbz" :-)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913 */
Arnaldo Carvalho de Melo878c8142007-03-11 22:38:29 -0300914 skb_push(skb, sizeof(struct iphdr));
915 skb_reset_network_header(skb);
Arnaldo Carvalho de Melobadff6d2007-03-13 13:06:52 -0300916 skb_reset_transport_header(skb);
Arnaldo Carvalho de Melo0272ffc2007-03-12 20:05:39 -0300917 msg = (struct igmpmsg *)skb_network_header(skb);
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700918 memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700919 msg->im_msgtype = IGMPMSG_WHOLEPKT;
920 msg->im_mbz = 0;
Patrick McHardy0c122952010-04-13 05:03:22 +0000921 msg->im_vif = mrt->mroute_reg_vif_num;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700922 ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2;
923 ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) +
924 sizeof(struct iphdr));
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900925 } else
Linus Torvalds1da177e2005-04-16 15:20:36 -0700926#endif
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900927 {
928
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000929 /* Copy the IP header */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700930
Arnaldo Carvalho de Melo27a884d2007-04-19 20:29:13 -0700931 skb->network_header = skb->tail;
Arnaldo Carvalho de Meloddc7b8e2007-03-15 21:42:27 -0300932 skb_put(skb, ihl);
Arnaldo Carvalho de Melo27d7ff42007-03-31 11:55:19 -0300933 skb_copy_to_linear_data(skb, pkt->data, ihl);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000934 ip_hdr(skb)->protocol = 0; /* Flag to the kernel this is a route add */
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700935 msg = (struct igmpmsg *)skb_network_header(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936 msg->im_vif = vifi;
Eric Dumazetadf30902009-06-02 05:19:30 +0000937 skb_dst_set(skb, dst_clone(skb_dst(pkt)));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700938
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000939 /* Add our header */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700940
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000941 igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700942 igmp->type =
943 msg->im_msgtype = assert;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000944 igmp->code = 0;
945 ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700946 skb->transport_header = skb->network_header;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900947 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700948
Eric Dumazet4c968702010-10-01 16:15:01 +0000949 rcu_read_lock();
950 mroute_sk = rcu_dereference(mrt->mroute_sk);
951 if (mroute_sk == NULL) {
952 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -0700953 kfree_skb(skb);
954 return -EINVAL;
955 }
956
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000957 /* Deliver to mrouted */
958
Eric Dumazet4c968702010-10-01 16:15:01 +0000959 ret = sock_queue_rcv_skb(mroute_sk, skb);
960 rcu_read_unlock();
Benjamin Thery70a269e2009-01-22 04:56:15 +0000961 if (ret < 0) {
Joe Perchese87cc472012-05-13 21:56:26 +0000962 net_warn_ratelimited("mroute: pending queue full, dropping entries\n");
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963 kfree_skb(skb);
964 }
965
966 return ret;
967}
968
969/*
970 * Queue a packet for resolution. It gets locked cache entry!
971 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900972
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973static int
Patrick McHardy0c122952010-04-13 05:03:22 +0000974ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700975{
Patrick McHardy862465f2010-04-13 05:03:21 +0000976 bool found = false;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700977 int err;
978 struct mfc_cache *c;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700979 const struct iphdr *iph = ip_hdr(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700980
981 spin_lock_bh(&mfc_unres_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +0000982 list_for_each_entry(c, &mrt->mfc_unres_queue, list) {
Patrick McHardye258beb2010-04-13 05:03:19 +0000983 if (c->mfc_mcastgrp == iph->daddr &&
Patrick McHardy862465f2010-04-13 05:03:21 +0000984 c->mfc_origin == iph->saddr) {
985 found = true;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700986 break;
Patrick McHardy862465f2010-04-13 05:03:21 +0000987 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988 }
989
Patrick McHardy862465f2010-04-13 05:03:21 +0000990 if (!found) {
Eric Dumazeta8cb16d2010-10-01 16:15:29 +0000991 /* Create a new entry if allowable */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700992
Patrick McHardy0c122952010-04-13 05:03:22 +0000993 if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 ||
Patrick McHardyd658f8a2010-04-13 05:03:20 +0000994 (c = ipmr_cache_alloc_unres()) == NULL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700995 spin_unlock_bh(&mfc_unres_lock);
996
997 kfree_skb(skb);
998 return -ENOBUFS;
999 }
1000
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001001 /* Fill in the new cache entry */
1002
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07001003 c->mfc_parent = -1;
1004 c->mfc_origin = iph->saddr;
1005 c->mfc_mcastgrp = iph->daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001007 /* Reflect first query at mrouted. */
1008
Patrick McHardy0c122952010-04-13 05:03:22 +00001009 err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE);
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001010 if (err < 0) {
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001011 /* If the report failed throw the cache entry
Linus Torvalds1da177e2005-04-16 15:20:36 -07001012 out - Brad Parker
1013 */
1014 spin_unlock_bh(&mfc_unres_lock);
1015
Benjamin Thery5c0a66f2009-01-22 04:56:17 +00001016 ipmr_cache_free(c);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001017 kfree_skb(skb);
1018 return err;
1019 }
1020
Patrick McHardy0c122952010-04-13 05:03:22 +00001021 atomic_inc(&mrt->cache_resolve_queue_len);
1022 list_add(&c->list, &mrt->mfc_unres_queue);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001023
David S. Miller278554b2010-05-12 00:05:35 -07001024 if (atomic_read(&mrt->cache_resolve_queue_len) == 1)
1025 mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001026 }
1027
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001028 /* See if we can append the packet */
1029
1030 if (c->mfc_un.unres.unresolved.qlen > 3) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001031 kfree_skb(skb);
1032 err = -ENOBUFS;
1033 } else {
Jianjun Kongc354e122008-11-03 00:28:02 -08001034 skb_queue_tail(&c->mfc_un.unres.unresolved, skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001035 err = 0;
1036 }
1037
1038 spin_unlock_bh(&mfc_unres_lock);
1039 return err;
1040}
1041
1042/*
1043 * MFC cache manipulation by user space mroute daemon
1044 */
1045
Patrick McHardy0c122952010-04-13 05:03:22 +00001046static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001047{
1048 int line;
Patrick McHardy862465f2010-04-13 05:03:21 +00001049 struct mfc_cache *c, *next;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001050
Jianjun Kongc354e122008-11-03 00:28:02 -08001051 line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001052
Patrick McHardy0c122952010-04-13 05:03:22 +00001053 list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001054 if (c->mfc_origin == mfc->mfcc_origin.s_addr &&
1055 c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) {
Eric Dumazeta8c94862010-10-01 16:15:08 +00001056 list_del_rcu(&c->list);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001057
Benjamin Thery5c0a66f2009-01-22 04:56:17 +00001058 ipmr_cache_free(c);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001059 return 0;
1060 }
1061 }
1062 return -ENOENT;
1063}
1064
Patrick McHardy0c122952010-04-13 05:03:22 +00001065static int ipmr_mfc_add(struct net *net, struct mr_table *mrt,
1066 struct mfcctl *mfc, int mrtsock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001067{
Patrick McHardy862465f2010-04-13 05:03:21 +00001068 bool found = false;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001069 int line;
Patrick McHardy862465f2010-04-13 05:03:21 +00001070 struct mfc_cache *uc, *c;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001071
Patrick McHardya50436f22010-03-17 06:04:14 +00001072 if (mfc->mfcc_parent >= MAXVIFS)
1073 return -ENFILE;
1074
Jianjun Kongc354e122008-11-03 00:28:02 -08001075 line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001076
Patrick McHardy0c122952010-04-13 05:03:22 +00001077 list_for_each_entry(c, &mrt->mfc_cache_array[line], list) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001078 if (c->mfc_origin == mfc->mfcc_origin.s_addr &&
Patrick McHardy862465f2010-04-13 05:03:21 +00001079 c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) {
1080 found = true;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001081 break;
Patrick McHardy862465f2010-04-13 05:03:21 +00001082 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001083 }
1084
Patrick McHardy862465f2010-04-13 05:03:21 +00001085 if (found) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001086 write_lock_bh(&mrt_lock);
1087 c->mfc_parent = mfc->mfcc_parent;
Patrick McHardy0c122952010-04-13 05:03:22 +00001088 ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001089 if (!mrtsock)
1090 c->mfc_flags |= MFC_STATIC;
1091 write_unlock_bh(&mrt_lock);
1092 return 0;
1093 }
1094
Joe Perchesf97c1e02007-12-16 13:45:43 -08001095 if (!ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001096 return -EINVAL;
1097
Patrick McHardyd658f8a2010-04-13 05:03:20 +00001098 c = ipmr_cache_alloc();
Jianjun Kongc354e122008-11-03 00:28:02 -08001099 if (c == NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001100 return -ENOMEM;
1101
Jianjun Kongc354e122008-11-03 00:28:02 -08001102 c->mfc_origin = mfc->mfcc_origin.s_addr;
1103 c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr;
1104 c->mfc_parent = mfc->mfcc_parent;
Patrick McHardy0c122952010-04-13 05:03:22 +00001105 ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001106 if (!mrtsock)
1107 c->mfc_flags |= MFC_STATIC;
1108
Eric Dumazeta8c94862010-10-01 16:15:08 +00001109 list_add_rcu(&c->list, &mrt->mfc_cache_array[line]);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001110
1111 /*
1112 * Check to see if we resolved a queued list. If so we
1113 * need to send on the frames and tidy up.
1114 */
Patrick McHardyb0ebb732010-04-15 13:29:28 +02001115 found = false;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001116 spin_lock_bh(&mfc_unres_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00001117 list_for_each_entry(uc, &mrt->mfc_unres_queue, list) {
Patrick McHardye258beb2010-04-13 05:03:19 +00001118 if (uc->mfc_origin == c->mfc_origin &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07001119 uc->mfc_mcastgrp == c->mfc_mcastgrp) {
Patrick McHardy862465f2010-04-13 05:03:21 +00001120 list_del(&uc->list);
Patrick McHardy0c122952010-04-13 05:03:22 +00001121 atomic_dec(&mrt->cache_resolve_queue_len);
Patrick McHardyb0ebb732010-04-15 13:29:28 +02001122 found = true;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001123 break;
1124 }
1125 }
Patrick McHardy0c122952010-04-13 05:03:22 +00001126 if (list_empty(&mrt->mfc_unres_queue))
1127 del_timer(&mrt->ipmr_expire_timer);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001128 spin_unlock_bh(&mfc_unres_lock);
1129
Patrick McHardyb0ebb732010-04-15 13:29:28 +02001130 if (found) {
Patrick McHardy0c122952010-04-13 05:03:22 +00001131 ipmr_cache_resolve(net, mrt, uc, c);
Benjamin Thery5c0a66f2009-01-22 04:56:17 +00001132 ipmr_cache_free(uc);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001133 }
1134 return 0;
1135}
1136
1137/*
1138 * Close the multicast socket, and clear the vif tables etc
1139 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001140
Patrick McHardy0c122952010-04-13 05:03:22 +00001141static void mroute_clean_tables(struct mr_table *mrt)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001142{
1143 int i;
Eric Dumazetd17fa6f2009-10-28 05:21:38 +00001144 LIST_HEAD(list);
Patrick McHardy862465f2010-04-13 05:03:21 +00001145 struct mfc_cache *c, *next;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001146
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001147 /* Shut down all active vif entries */
1148
Patrick McHardy0c122952010-04-13 05:03:22 +00001149 for (i = 0; i < mrt->maxvif; i++) {
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001150 if (!(mrt->vif_table[i].flags & VIFF_STATIC))
Patrick McHardy0c122952010-04-13 05:03:22 +00001151 vif_delete(mrt, i, 0, &list);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001152 }
Eric Dumazetd17fa6f2009-10-28 05:21:38 +00001153 unregister_netdevice_many(&list);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001154
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001155 /* Wipe the cache */
1156
Patrick McHardy862465f2010-04-13 05:03:21 +00001157 for (i = 0; i < MFC_LINES; i++) {
Patrick McHardy0c122952010-04-13 05:03:22 +00001158 list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) {
Eric Dumazeta8c94862010-10-01 16:15:08 +00001159 if (c->mfc_flags & MFC_STATIC)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001160 continue;
Eric Dumazeta8c94862010-10-01 16:15:08 +00001161 list_del_rcu(&c->list);
Benjamin Thery5c0a66f2009-01-22 04:56:17 +00001162 ipmr_cache_free(c);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001163 }
1164 }
1165
Patrick McHardy0c122952010-04-13 05:03:22 +00001166 if (atomic_read(&mrt->cache_resolve_queue_len) != 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001167 spin_lock_bh(&mfc_unres_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00001168 list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) {
Patrick McHardy862465f2010-04-13 05:03:21 +00001169 list_del(&c->list);
Patrick McHardy0c122952010-04-13 05:03:22 +00001170 ipmr_destroy_unres(mrt, c);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001171 }
1172 spin_unlock_bh(&mfc_unres_lock);
1173 }
1174}
1175
Eric Dumazet4c968702010-10-01 16:15:01 +00001176/* called from ip_ra_control(), before an RCU grace period,
1177 * we dont need to call synchronize_rcu() here
1178 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001179static void mrtsock_destruct(struct sock *sk)
1180{
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001181 struct net *net = sock_net(sk);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001182 struct mr_table *mrt;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001183
Linus Torvalds1da177e2005-04-16 15:20:36 -07001184 rtnl_lock();
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001185 ipmr_for_each_table(mrt, net) {
Eric Dumazet4c968702010-10-01 16:15:01 +00001186 if (sk == rtnl_dereference(mrt->mroute_sk)) {
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001187 IPV4_DEVCONF_ALL(net, MC_FORWARDING)--;
Stephen Hemmingera9b3cd72011-08-01 16:19:00 +00001188 RCU_INIT_POINTER(mrt->mroute_sk, NULL);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001189 mroute_clean_tables(mrt);
1190 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001191 }
1192 rtnl_unlock();
1193}
1194
1195/*
1196 * Socket options and virtual interface manipulation. The whole
1197 * virtual interface system is a complete heap, but unfortunately
1198 * that's how BSD mrouted happens to think. Maybe one day with a proper
1199 * MOSPF/PIM router set up we can clean this up.
1200 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001201
David S. Millerb7058842009-09-30 16:12:20 -07001202int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203{
1204 int ret;
1205 struct vifctl vif;
1206 struct mfcctl mfc;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001207 struct net *net = sock_net(sk);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001208 struct mr_table *mrt;
1209
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001210 if (sk->sk_type != SOCK_RAW ||
1211 inet_sk(sk)->inet_num != IPPROTO_IGMP)
1212 return -EOPNOTSUPP;
1213
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001214 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
1215 if (mrt == NULL)
1216 return -ENOENT;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001217
Stephen Hemminger132adf52007-03-08 20:44:43 -08001218 if (optname != MRT_INIT) {
Eric Dumazet33d480c2011-08-11 19:30:52 +00001219 if (sk != rcu_access_pointer(mrt->mroute_sk) &&
Eric W. Biederman52e804c2012-11-16 03:03:05 +00001220 !ns_capable(net->user_ns, CAP_NET_ADMIN))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001221 return -EACCES;
1222 }
1223
Stephen Hemminger132adf52007-03-08 20:44:43 -08001224 switch (optname) {
1225 case MRT_INIT:
Jianjun Kongc354e122008-11-03 00:28:02 -08001226 if (optlen != sizeof(int))
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001227 return -EINVAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001228
Stephen Hemminger132adf52007-03-08 20:44:43 -08001229 rtnl_lock();
Eric Dumazet4c968702010-10-01 16:15:01 +00001230 if (rtnl_dereference(mrt->mroute_sk)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001231 rtnl_unlock();
Stephen Hemminger132adf52007-03-08 20:44:43 -08001232 return -EADDRINUSE;
1233 }
1234
1235 ret = ip_ra_control(sk, 1, mrtsock_destruct);
1236 if (ret == 0) {
Eric Dumazetcf778b02012-01-12 04:41:32 +00001237 rcu_assign_pointer(mrt->mroute_sk, sk);
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001238 IPV4_DEVCONF_ALL(net, MC_FORWARDING)++;
Stephen Hemminger132adf52007-03-08 20:44:43 -08001239 }
1240 rtnl_unlock();
1241 return ret;
1242 case MRT_DONE:
Eric Dumazet33d480c2011-08-11 19:30:52 +00001243 if (sk != rcu_access_pointer(mrt->mroute_sk))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001244 return -EACCES;
1245 return ip_ra_control(sk, 0, NULL);
1246 case MRT_ADD_VIF:
1247 case MRT_DEL_VIF:
Jianjun Kongc354e122008-11-03 00:28:02 -08001248 if (optlen != sizeof(vif))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001249 return -EINVAL;
Jianjun Kongc354e122008-11-03 00:28:02 -08001250 if (copy_from_user(&vif, optval, sizeof(vif)))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001251 return -EFAULT;
1252 if (vif.vifc_vifi >= MAXVIFS)
1253 return -ENFILE;
1254 rtnl_lock();
Jianjun Kongc354e122008-11-03 00:28:02 -08001255 if (optname == MRT_ADD_VIF) {
Eric Dumazet4c968702010-10-01 16:15:01 +00001256 ret = vif_add(net, mrt, &vif,
1257 sk == rtnl_dereference(mrt->mroute_sk));
Stephen Hemminger132adf52007-03-08 20:44:43 -08001258 } else {
Patrick McHardy0c122952010-04-13 05:03:22 +00001259 ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL);
Stephen Hemminger132adf52007-03-08 20:44:43 -08001260 }
1261 rtnl_unlock();
1262 return ret;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001263
1264 /*
1265 * Manipulate the forwarding caches. These live
1266 * in a sort of kernel/user symbiosis.
1267 */
Stephen Hemminger132adf52007-03-08 20:44:43 -08001268 case MRT_ADD_MFC:
1269 case MRT_DEL_MFC:
Jianjun Kongc354e122008-11-03 00:28:02 -08001270 if (optlen != sizeof(mfc))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001271 return -EINVAL;
Jianjun Kongc354e122008-11-03 00:28:02 -08001272 if (copy_from_user(&mfc, optval, sizeof(mfc)))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001273 return -EFAULT;
1274 rtnl_lock();
Jianjun Kongc354e122008-11-03 00:28:02 -08001275 if (optname == MRT_DEL_MFC)
Patrick McHardy0c122952010-04-13 05:03:22 +00001276 ret = ipmr_mfc_delete(mrt, &mfc);
Stephen Hemminger132adf52007-03-08 20:44:43 -08001277 else
Eric Dumazet4c968702010-10-01 16:15:01 +00001278 ret = ipmr_mfc_add(net, mrt, &mfc,
1279 sk == rtnl_dereference(mrt->mroute_sk));
Stephen Hemminger132adf52007-03-08 20:44:43 -08001280 rtnl_unlock();
1281 return ret;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001282 /*
1283 * Control PIM assert.
1284 */
Stephen Hemminger132adf52007-03-08 20:44:43 -08001285 case MRT_ASSERT:
1286 {
1287 int v;
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001288 if (optlen != sizeof(v))
1289 return -EINVAL;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001290 if (get_user(v, (int __user *)optval))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001291 return -EFAULT;
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001292 mrt->mroute_do_assert = !!v;
Stephen Hemminger132adf52007-03-08 20:44:43 -08001293 return 0;
1294 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001295#ifdef CONFIG_IP_PIMSM
Stephen Hemminger132adf52007-03-08 20:44:43 -08001296 case MRT_PIM:
1297 {
Stephen Hemmingerba93ef72008-01-21 17:28:59 -08001298 int v;
1299
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001300 if (optlen != sizeof(v))
1301 return -EINVAL;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001302 if (get_user(v, (int __user *)optval))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001303 return -EFAULT;
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001304 v = !!v;
Stephen Hemmingerba93ef72008-01-21 17:28:59 -08001305
Stephen Hemminger132adf52007-03-08 20:44:43 -08001306 rtnl_lock();
1307 ret = 0;
Patrick McHardy0c122952010-04-13 05:03:22 +00001308 if (v != mrt->mroute_do_pim) {
1309 mrt->mroute_do_pim = v;
1310 mrt->mroute_do_assert = v;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001311 }
Stephen Hemminger132adf52007-03-08 20:44:43 -08001312 rtnl_unlock();
1313 return ret;
1314 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001315#endif
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001316#ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES
1317 case MRT_TABLE:
1318 {
1319 u32 v;
1320
1321 if (optlen != sizeof(u32))
1322 return -EINVAL;
1323 if (get_user(v, (u32 __user *)optval))
1324 return -EFAULT;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001325
1326 rtnl_lock();
1327 ret = 0;
Eric Dumazet4c968702010-10-01 16:15:01 +00001328 if (sk == rtnl_dereference(mrt->mroute_sk)) {
1329 ret = -EBUSY;
1330 } else {
1331 if (!ipmr_new_table(net, v))
1332 ret = -ENOMEM;
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001333 else
1334 raw_sk(sk)->ipmr_table = v;
Eric Dumazet4c968702010-10-01 16:15:01 +00001335 }
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001336 rtnl_unlock();
1337 return ret;
1338 }
1339#endif
Stephen Hemminger132adf52007-03-08 20:44:43 -08001340 /*
1341 * Spurious command, or MRT_VERSION which you cannot
1342 * set.
1343 */
1344 default:
1345 return -ENOPROTOOPT;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001346 }
1347}
1348
1349/*
1350 * Getsock opt support for the multicast routing system.
1351 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001352
Jianjun Kongc354e122008-11-03 00:28:02 -08001353int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001354{
1355 int olr;
1356 int val;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001357 struct net *net = sock_net(sk);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001358 struct mr_table *mrt;
1359
Eric Dumazet5e1859f2012-11-25 06:41:45 +00001360 if (sk->sk_type != SOCK_RAW ||
1361 inet_sk(sk)->inet_num != IPPROTO_IGMP)
1362 return -EOPNOTSUPP;
1363
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001364 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
1365 if (mrt == NULL)
1366 return -ENOENT;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001367
Jianjun Kongc354e122008-11-03 00:28:02 -08001368 if (optname != MRT_VERSION &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07001369#ifdef CONFIG_IP_PIMSM
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001370 optname != MRT_PIM &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07001371#endif
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001372 optname != MRT_ASSERT)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001373 return -ENOPROTOOPT;
1374
1375 if (get_user(olr, optlen))
1376 return -EFAULT;
1377
1378 olr = min_t(unsigned int, olr, sizeof(int));
1379 if (olr < 0)
1380 return -EINVAL;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001381
Jianjun Kongc354e122008-11-03 00:28:02 -08001382 if (put_user(olr, optlen))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001383 return -EFAULT;
Jianjun Kongc354e122008-11-03 00:28:02 -08001384 if (optname == MRT_VERSION)
1385 val = 0x0305;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001386#ifdef CONFIG_IP_PIMSM
Jianjun Kongc354e122008-11-03 00:28:02 -08001387 else if (optname == MRT_PIM)
Patrick McHardy0c122952010-04-13 05:03:22 +00001388 val = mrt->mroute_do_pim;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001389#endif
1390 else
Patrick McHardy0c122952010-04-13 05:03:22 +00001391 val = mrt->mroute_do_assert;
Jianjun Kongc354e122008-11-03 00:28:02 -08001392 if (copy_to_user(optval, &val, olr))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001393 return -EFAULT;
1394 return 0;
1395}
1396
1397/*
1398 * The IP multicast ioctl support routines.
1399 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001400
Linus Torvalds1da177e2005-04-16 15:20:36 -07001401int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg)
1402{
1403 struct sioc_sg_req sr;
1404 struct sioc_vif_req vr;
1405 struct vif_device *vif;
1406 struct mfc_cache *c;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001407 struct net *net = sock_net(sk);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001408 struct mr_table *mrt;
1409
1410 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
1411 if (mrt == NULL)
1412 return -ENOENT;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001413
Stephen Hemminger132adf52007-03-08 20:44:43 -08001414 switch (cmd) {
1415 case SIOCGETVIFCNT:
Jianjun Kongc354e122008-11-03 00:28:02 -08001416 if (copy_from_user(&vr, arg, sizeof(vr)))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001417 return -EFAULT;
Patrick McHardy0c122952010-04-13 05:03:22 +00001418 if (vr.vifi >= mrt->maxvif)
Stephen Hemminger132adf52007-03-08 20:44:43 -08001419 return -EINVAL;
1420 read_lock(&mrt_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00001421 vif = &mrt->vif_table[vr.vifi];
1422 if (VIF_EXISTS(mrt, vr.vifi)) {
Jianjun Kongc354e122008-11-03 00:28:02 -08001423 vr.icount = vif->pkt_in;
1424 vr.ocount = vif->pkt_out;
1425 vr.ibytes = vif->bytes_in;
1426 vr.obytes = vif->bytes_out;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001427 read_unlock(&mrt_lock);
Stephen Hemminger132adf52007-03-08 20:44:43 -08001428
Jianjun Kongc354e122008-11-03 00:28:02 -08001429 if (copy_to_user(arg, &vr, sizeof(vr)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001430 return -EFAULT;
Stephen Hemminger132adf52007-03-08 20:44:43 -08001431 return 0;
1432 }
1433 read_unlock(&mrt_lock);
1434 return -EADDRNOTAVAIL;
1435 case SIOCGETSGCNT:
Jianjun Kongc354e122008-11-03 00:28:02 -08001436 if (copy_from_user(&sr, arg, sizeof(sr)))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001437 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001438
Eric Dumazeta8c94862010-10-01 16:15:08 +00001439 rcu_read_lock();
Patrick McHardy0c122952010-04-13 05:03:22 +00001440 c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr);
Stephen Hemminger132adf52007-03-08 20:44:43 -08001441 if (c) {
1442 sr.pktcnt = c->mfc_un.res.pkt;
1443 sr.bytecnt = c->mfc_un.res.bytes;
1444 sr.wrong_if = c->mfc_un.res.wrong_if;
Eric Dumazeta8c94862010-10-01 16:15:08 +00001445 rcu_read_unlock();
Stephen Hemminger132adf52007-03-08 20:44:43 -08001446
Jianjun Kongc354e122008-11-03 00:28:02 -08001447 if (copy_to_user(arg, &sr, sizeof(sr)))
Stephen Hemminger132adf52007-03-08 20:44:43 -08001448 return -EFAULT;
1449 return 0;
1450 }
Eric Dumazeta8c94862010-10-01 16:15:08 +00001451 rcu_read_unlock();
Stephen Hemminger132adf52007-03-08 20:44:43 -08001452 return -EADDRNOTAVAIL;
1453 default:
1454 return -ENOIOCTLCMD;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001455 }
1456}
1457
Eric W. Biederman709b46e2011-01-29 16:15:56 +00001458#ifdef CONFIG_COMPAT
1459struct compat_sioc_sg_req {
1460 struct in_addr src;
1461 struct in_addr grp;
1462 compat_ulong_t pktcnt;
1463 compat_ulong_t bytecnt;
1464 compat_ulong_t wrong_if;
1465};
1466
David S. Millerca6b8bb02011-02-03 17:24:28 -08001467struct compat_sioc_vif_req {
1468 vifi_t vifi; /* Which iface */
1469 compat_ulong_t icount;
1470 compat_ulong_t ocount;
1471 compat_ulong_t ibytes;
1472 compat_ulong_t obytes;
1473};
1474
Eric W. Biederman709b46e2011-01-29 16:15:56 +00001475int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg)
1476{
David S. Miller0033d5a2011-02-03 17:21:31 -08001477 struct compat_sioc_sg_req sr;
David S. Millerca6b8bb02011-02-03 17:24:28 -08001478 struct compat_sioc_vif_req vr;
1479 struct vif_device *vif;
Eric W. Biederman709b46e2011-01-29 16:15:56 +00001480 struct mfc_cache *c;
1481 struct net *net = sock_net(sk);
1482 struct mr_table *mrt;
1483
1484 mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT);
1485 if (mrt == NULL)
1486 return -ENOENT;
1487
1488 switch (cmd) {
David S. Millerca6b8bb02011-02-03 17:24:28 -08001489 case SIOCGETVIFCNT:
1490 if (copy_from_user(&vr, arg, sizeof(vr)))
1491 return -EFAULT;
1492 if (vr.vifi >= mrt->maxvif)
1493 return -EINVAL;
1494 read_lock(&mrt_lock);
1495 vif = &mrt->vif_table[vr.vifi];
1496 if (VIF_EXISTS(mrt, vr.vifi)) {
1497 vr.icount = vif->pkt_in;
1498 vr.ocount = vif->pkt_out;
1499 vr.ibytes = vif->bytes_in;
1500 vr.obytes = vif->bytes_out;
1501 read_unlock(&mrt_lock);
1502
1503 if (copy_to_user(arg, &vr, sizeof(vr)))
1504 return -EFAULT;
1505 return 0;
1506 }
1507 read_unlock(&mrt_lock);
1508 return -EADDRNOTAVAIL;
Eric W. Biederman709b46e2011-01-29 16:15:56 +00001509 case SIOCGETSGCNT:
1510 if (copy_from_user(&sr, arg, sizeof(sr)))
1511 return -EFAULT;
1512
1513 rcu_read_lock();
1514 c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr);
1515 if (c) {
1516 sr.pktcnt = c->mfc_un.res.pkt;
1517 sr.bytecnt = c->mfc_un.res.bytes;
1518 sr.wrong_if = c->mfc_un.res.wrong_if;
1519 rcu_read_unlock();
1520
1521 if (copy_to_user(arg, &sr, sizeof(sr)))
1522 return -EFAULT;
1523 return 0;
1524 }
1525 rcu_read_unlock();
1526 return -EADDRNOTAVAIL;
1527 default:
1528 return -ENOIOCTLCMD;
1529 }
1530}
1531#endif
1532
Linus Torvalds1da177e2005-04-16 15:20:36 -07001533
1534static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr)
1535{
Eric W. Biedermane9dc8652007-09-12 13:02:17 +02001536 struct net_device *dev = ptr;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001537 struct net *net = dev_net(dev);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001538 struct mr_table *mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001539 struct vif_device *v;
1540 int ct;
Eric W. Biedermane9dc8652007-09-12 13:02:17 +02001541
Linus Torvalds1da177e2005-04-16 15:20:36 -07001542 if (event != NETDEV_UNREGISTER)
1543 return NOTIFY_DONE;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001544
1545 ipmr_for_each_table(mrt, net) {
1546 v = &mrt->vif_table[0];
1547 for (ct = 0; ct < mrt->maxvif; ct++, v++) {
1548 if (v->dev == dev)
RongQing.Lie92036a2011-11-23 23:10:52 +00001549 vif_delete(mrt, ct, 1, NULL);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001550 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001551 }
1552 return NOTIFY_DONE;
1553}
1554
1555
Jianjun Kongc354e122008-11-03 00:28:02 -08001556static struct notifier_block ip_mr_notifier = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001557 .notifier_call = ipmr_device_event,
1558};
1559
1560/*
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001561 * Encapsulate a packet by attaching a valid IPIP header to it.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001562 * This avoids tunnel drivers and other mess and gives us the speed so
1563 * important for multicast video.
1564 */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001565
Al Viro114c7842006-09-27 18:39:29 -07001566static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001567{
Arnaldo Carvalho de Melo8856dfa2007-03-10 19:40:39 -03001568 struct iphdr *iph;
Eric Dumazetb71d1d42011-04-22 04:53:02 +00001569 const struct iphdr *old_iph = ip_hdr(skb);
Arnaldo Carvalho de Melo8856dfa2007-03-10 19:40:39 -03001570
1571 skb_push(skb, sizeof(struct iphdr));
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -07001572 skb->transport_header = skb->network_header;
Arnaldo Carvalho de Melo8856dfa2007-03-10 19:40:39 -03001573 skb_reset_network_header(skb);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07001574 iph = ip_hdr(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001575
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001576 iph->version = 4;
Arnaldo Carvalho de Meloe023dd62007-03-12 20:09:36 -03001577 iph->tos = old_iph->tos;
1578 iph->ttl = old_iph->ttl;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001579 iph->frag_off = 0;
1580 iph->daddr = daddr;
1581 iph->saddr = saddr;
1582 iph->protocol = IPPROTO_IPIP;
1583 iph->ihl = 5;
1584 iph->tot_len = htons(skb->len);
Eric Dumazetadf30902009-06-02 05:19:30 +00001585 ip_select_ident(iph, skb_dst(skb), NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001586 ip_send_check(iph);
1587
Linus Torvalds1da177e2005-04-16 15:20:36 -07001588 memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
1589 nf_reset(skb);
1590}
1591
1592static inline int ipmr_forward_finish(struct sk_buff *skb)
1593{
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001594 struct ip_options *opt = &(IPCB(skb)->opt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001595
Eric Dumazetadf30902009-06-02 05:19:30 +00001596 IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS);
Vincent Bernat2d8dbb02012-06-05 03:41:42 +00001597 IP_ADD_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTOCTETS, skb->len);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001598
1599 if (unlikely(opt->optlen))
1600 ip_forward_options(skb);
1601
1602 return dst_output(skb);
1603}
1604
1605/*
1606 * Processing handlers for ipmr_forward
1607 */
1608
Patrick McHardy0c122952010-04-13 05:03:22 +00001609static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt,
1610 struct sk_buff *skb, struct mfc_cache *c, int vifi)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001611{
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07001612 const struct iphdr *iph = ip_hdr(skb);
Patrick McHardy0c122952010-04-13 05:03:22 +00001613 struct vif_device *vif = &mrt->vif_table[vifi];
Linus Torvalds1da177e2005-04-16 15:20:36 -07001614 struct net_device *dev;
1615 struct rtable *rt;
David S. Miller31e4543d2011-05-03 20:25:42 -07001616 struct flowi4 fl4;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001617 int encap = 0;
1618
1619 if (vif->dev == NULL)
1620 goto out_free;
1621
1622#ifdef CONFIG_IP_PIMSM
1623 if (vif->flags & VIFF_REGISTER) {
1624 vif->pkt_out++;
Jianjun Kongc354e122008-11-03 00:28:02 -08001625 vif->bytes_out += skb->len;
Pavel Emelyanovcf3677a2008-05-21 14:17:33 -07001626 vif->dev->stats.tx_bytes += skb->len;
1627 vif->dev->stats.tx_packets++;
Patrick McHardy0c122952010-04-13 05:03:22 +00001628 ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT);
Ilpo Järvinen69ebbf52009-02-06 23:46:51 -08001629 goto out_free;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001630 }
1631#endif
1632
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001633 if (vif->flags & VIFF_TUNNEL) {
David S. Miller31e4543d2011-05-03 20:25:42 -07001634 rt = ip_route_output_ports(net, &fl4, NULL,
David S. Miller78fbfd82011-03-12 00:00:52 -05001635 vif->remote, vif->local,
1636 0, 0,
1637 IPPROTO_IPIP,
1638 RT_TOS(iph->tos), vif->link);
David S. Millerb23dd4f2011-03-02 14:31:35 -08001639 if (IS_ERR(rt))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001640 goto out_free;
1641 encap = sizeof(struct iphdr);
1642 } else {
David S. Miller31e4543d2011-05-03 20:25:42 -07001643 rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0,
David S. Miller78fbfd82011-03-12 00:00:52 -05001644 0, 0,
1645 IPPROTO_IPIP,
1646 RT_TOS(iph->tos), vif->link);
David S. Millerb23dd4f2011-03-02 14:31:35 -08001647 if (IS_ERR(rt))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001648 goto out_free;
1649 }
1650
Changli Gaod8d1f302010-06-10 23:31:35 -07001651 dev = rt->dst.dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001652
Changli Gaod8d1f302010-06-10 23:31:35 -07001653 if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001654 /* Do not fragment multicasts. Alas, IPv4 does not
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001655 * allow to send ICMP, so that packets will disappear
1656 * to blackhole.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001657 */
1658
Pavel Emelyanov7c73a6f2008-07-16 20:20:11 -07001659 IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001660 ip_rt_put(rt);
1661 goto out_free;
1662 }
1663
Changli Gaod8d1f302010-06-10 23:31:35 -07001664 encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001665
1666 if (skb_cow(skb, encap)) {
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001667 ip_rt_put(rt);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001668 goto out_free;
1669 }
1670
1671 vif->pkt_out++;
Jianjun Kongc354e122008-11-03 00:28:02 -08001672 vif->bytes_out += skb->len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001673
Eric Dumazetadf30902009-06-02 05:19:30 +00001674 skb_dst_drop(skb);
Changli Gaod8d1f302010-06-10 23:31:35 -07001675 skb_dst_set(skb, &rt->dst);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07001676 ip_decrease_ttl(ip_hdr(skb));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001677
1678 /* FIXME: forward and output firewalls used to be called here.
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001679 * What do we do with netfilter? -- RR
1680 */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001681 if (vif->flags & VIFF_TUNNEL) {
1682 ip_encap(skb, vif->local, vif->remote);
1683 /* FIXME: extra output firewall step used to be here. --RR */
Pavel Emelyanov2f4c02d2008-05-21 14:16:14 -07001684 vif->dev->stats.tx_packets++;
1685 vif->dev->stats.tx_bytes += skb->len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001686 }
1687
1688 IPCB(skb)->flags |= IPSKB_FORWARDED;
1689
1690 /*
1691 * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally
1692 * not only before forwarding, but after forwarding on all output
1693 * interfaces. It is clear, if mrouter runs a multicasting
1694 * program, it should receive packets not depending to what interface
1695 * program is joined.
1696 * If we will not make it, the program will have to join on all
1697 * interfaces. On the other hand, multihoming host (or router, but
1698 * not mrouter) cannot join to more than one interface - it will
1699 * result in receiving multiple packets.
1700 */
Jan Engelhardt9bbc7682010-03-23 04:07:29 +01001701 NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, skb, skb->dev, dev,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001702 ipmr_forward_finish);
1703 return;
1704
1705out_free:
1706 kfree_skb(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001707}
1708
Patrick McHardy0c122952010-04-13 05:03:22 +00001709static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001710{
1711 int ct;
Patrick McHardy0c122952010-04-13 05:03:22 +00001712
1713 for (ct = mrt->maxvif-1; ct >= 0; ct--) {
1714 if (mrt->vif_table[ct].dev == dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001715 break;
1716 }
1717 return ct;
1718}
1719
1720/* "local" means that we should preserve one skb (for local delivery) */
1721
Patrick McHardy0c122952010-04-13 05:03:22 +00001722static int ip_mr_forward(struct net *net, struct mr_table *mrt,
1723 struct sk_buff *skb, struct mfc_cache *cache,
1724 int local)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001725{
1726 int psend = -1;
1727 int vif, ct;
1728
1729 vif = cache->mfc_parent;
1730 cache->mfc_un.res.pkt++;
1731 cache->mfc_un.res.bytes += skb->len;
1732
1733 /*
1734 * Wrong interface: drop packet and (maybe) send PIM assert.
1735 */
Patrick McHardy0c122952010-04-13 05:03:22 +00001736 if (mrt->vif_table[vif].dev != skb->dev) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001737 int true_vifi;
1738
David S. Millerc7537962010-11-11 17:07:48 -08001739 if (rt_is_output_route(skb_rtable(skb))) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001740 /* It is our own packet, looped back.
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001741 * Very complicated situation...
1742 *
1743 * The best workaround until routing daemons will be
1744 * fixed is not to redistribute packet, if it was
1745 * send through wrong interface. It means, that
1746 * multicast applications WILL NOT work for
1747 * (S,G), which have default multicast route pointing
1748 * to wrong oif. In any case, it is not a good
1749 * idea to use multicasting applications on router.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001750 */
1751 goto dont_forward;
1752 }
1753
1754 cache->mfc_un.res.wrong_if++;
Patrick McHardy0c122952010-04-13 05:03:22 +00001755 true_vifi = ipmr_find_vif(mrt, skb->dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001756
Patrick McHardy0c122952010-04-13 05:03:22 +00001757 if (true_vifi >= 0 && mrt->mroute_do_assert &&
Linus Torvalds1da177e2005-04-16 15:20:36 -07001758 /* pimsm uses asserts, when switching from RPT to SPT,
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001759 * so that we cannot check that packet arrived on an oif.
1760 * It is bad, but otherwise we would need to move pretty
1761 * large chunk of pimd to kernel. Ough... --ANK
Linus Torvalds1da177e2005-04-16 15:20:36 -07001762 */
Patrick McHardy0c122952010-04-13 05:03:22 +00001763 (mrt->mroute_do_pim ||
Benjamin Thery6f9374a2009-01-22 04:56:20 +00001764 cache->mfc_un.res.ttls[true_vifi] < 255) &&
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001765 time_after(jiffies,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001766 cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) {
1767 cache->mfc_un.res.last_assert = jiffies;
Patrick McHardy0c122952010-04-13 05:03:22 +00001768 ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001769 }
1770 goto dont_forward;
1771 }
1772
Patrick McHardy0c122952010-04-13 05:03:22 +00001773 mrt->vif_table[vif].pkt_in++;
1774 mrt->vif_table[vif].bytes_in += skb->len;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001775
1776 /*
1777 * Forward the frame
1778 */
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001779 for (ct = cache->mfc_un.res.maxvif - 1;
1780 ct >= cache->mfc_un.res.minvif; ct--) {
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07001781 if (ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001782 if (psend != -1) {
1783 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001784
Linus Torvalds1da177e2005-04-16 15:20:36 -07001785 if (skb2)
Patrick McHardy0c122952010-04-13 05:03:22 +00001786 ipmr_queue_xmit(net, mrt, skb2, cache,
1787 psend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001788 }
Jianjun Kongc354e122008-11-03 00:28:02 -08001789 psend = ct;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001790 }
1791 }
1792 if (psend != -1) {
1793 if (local) {
1794 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001795
Linus Torvalds1da177e2005-04-16 15:20:36 -07001796 if (skb2)
Patrick McHardy0c122952010-04-13 05:03:22 +00001797 ipmr_queue_xmit(net, mrt, skb2, cache, psend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001798 } else {
Patrick McHardy0c122952010-04-13 05:03:22 +00001799 ipmr_queue_xmit(net, mrt, skb, cache, psend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001800 return 0;
1801 }
1802 }
1803
1804dont_forward:
1805 if (!local)
1806 kfree_skb(skb);
1807 return 0;
1808}
1809
David S. Miller417da662011-05-03 19:42:43 -07001810static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb)
David S. Milleree3f1aa2011-03-09 14:06:20 -08001811{
David S. Miller417da662011-05-03 19:42:43 -07001812 struct rtable *rt = skb_rtable(skb);
1813 struct iphdr *iph = ip_hdr(skb);
David S. Millerda919812011-03-12 02:04:50 -05001814 struct flowi4 fl4 = {
David S. Miller417da662011-05-03 19:42:43 -07001815 .daddr = iph->daddr,
1816 .saddr = iph->saddr,
Julian Anastasovb0fe4a32011-07-23 02:00:41 +00001817 .flowi4_tos = RT_TOS(iph->tos),
David S. Miller4fd551d2012-07-17 14:39:44 -07001818 .flowi4_oif = (rt_is_output_route(rt) ?
1819 skb->dev->ifindex : 0),
1820 .flowi4_iif = (rt_is_output_route(rt) ?
Pavel Emelyanov1fb94892012-08-08 21:53:36 +00001821 LOOPBACK_IFINDEX :
David S. Miller4fd551d2012-07-17 14:39:44 -07001822 skb->dev->ifindex),
David Millerb4869882012-07-01 02:03:01 +00001823 .flowi4_mark = skb->mark,
David S. Milleree3f1aa2011-03-09 14:06:20 -08001824 };
1825 struct mr_table *mrt;
1826 int err;
1827
David S. Millerda919812011-03-12 02:04:50 -05001828 err = ipmr_fib_lookup(net, &fl4, &mrt);
David S. Milleree3f1aa2011-03-09 14:06:20 -08001829 if (err)
1830 return ERR_PTR(err);
1831 return mrt;
1832}
Linus Torvalds1da177e2005-04-16 15:20:36 -07001833
1834/*
1835 * Multicast packets for forwarding arrive here
Eric Dumazet4c968702010-10-01 16:15:01 +00001836 * Called with rcu_read_lock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07001837 */
1838
1839int ip_mr_input(struct sk_buff *skb)
1840{
1841 struct mfc_cache *cache;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001842 struct net *net = dev_net(skb->dev);
Eric Dumazet511c3f92009-06-02 05:14:27 +00001843 int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001844 struct mr_table *mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001845
1846 /* Packet is looped back after forward, it should not be
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001847 * forwarded second time, but still can be delivered locally.
Linus Torvalds1da177e2005-04-16 15:20:36 -07001848 */
Eric Dumazet4c968702010-10-01 16:15:01 +00001849 if (IPCB(skb)->flags & IPSKB_FORWARDED)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001850 goto dont_forward;
1851
David S. Miller417da662011-05-03 19:42:43 -07001852 mrt = ipmr_rt_fib_lookup(net, skb);
David S. Milleree3f1aa2011-03-09 14:06:20 -08001853 if (IS_ERR(mrt)) {
1854 kfree_skb(skb);
1855 return PTR_ERR(mrt);
Ben Greeare40dbc52010-07-15 13:22:33 +00001856 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001857 if (!local) {
Eric Dumazet4c968702010-10-01 16:15:01 +00001858 if (IPCB(skb)->opt.router_alert) {
1859 if (ip_call_ra_chain(skb))
1860 return 0;
1861 } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) {
1862 /* IGMPv1 (and broken IGMPv2 implementations sort of
1863 * Cisco IOS <= 11.2(8)) do not put router alert
1864 * option to IGMP packets destined to routable
1865 * groups. It is very bad, because it means
1866 * that we can forward NO IGMP messages.
1867 */
1868 struct sock *mroute_sk;
1869
1870 mroute_sk = rcu_dereference(mrt->mroute_sk);
1871 if (mroute_sk) {
1872 nf_reset(skb);
1873 raw_rcv(mroute_sk, skb);
1874 return 0;
1875 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001876 }
1877 }
1878
Eric Dumazeta8c94862010-10-01 16:15:08 +00001879 /* already under rcu_read_lock() */
Patrick McHardy0c122952010-04-13 05:03:22 +00001880 cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001881
1882 /*
1883 * No usable cache entry
1884 */
Jianjun Kongc354e122008-11-03 00:28:02 -08001885 if (cache == NULL) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001886 int vif;
1887
1888 if (local) {
1889 struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC);
1890 ip_local_deliver(skb);
Eric Dumazeta8c94862010-10-01 16:15:08 +00001891 if (skb2 == NULL)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001892 return -ENOBUFS;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001893 skb = skb2;
1894 }
1895
Eric Dumazeta8c94862010-10-01 16:15:08 +00001896 read_lock(&mrt_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00001897 vif = ipmr_find_vif(mrt, skb->dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001898 if (vif >= 0) {
Eric Dumazet0eae88f2010-04-20 19:06:52 -07001899 int err2 = ipmr_cache_unresolved(mrt, vif, skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001900 read_unlock(&mrt_lock);
1901
Eric Dumazet0eae88f2010-04-20 19:06:52 -07001902 return err2;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001903 }
1904 read_unlock(&mrt_lock);
1905 kfree_skb(skb);
1906 return -ENODEV;
1907 }
1908
Eric Dumazeta8c94862010-10-01 16:15:08 +00001909 read_lock(&mrt_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00001910 ip_mr_forward(net, mrt, skb, cache, local);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001911 read_unlock(&mrt_lock);
1912
1913 if (local)
1914 return ip_local_deliver(skb);
1915
1916 return 0;
1917
1918dont_forward:
1919 if (local)
1920 return ip_local_deliver(skb);
1921 kfree_skb(skb);
1922 return 0;
1923}
1924
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001925#ifdef CONFIG_IP_PIMSM
Eric Dumazet55747a02010-10-01 16:14:55 +00001926/* called with rcu_read_lock() */
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001927static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb,
1928 unsigned int pimlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001929{
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001930 struct net_device *reg_dev = NULL;
1931 struct iphdr *encap;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001932
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001933 encap = (struct iphdr *)(skb_transport_header(skb) + pimlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001934 /*
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001935 * Check that:
1936 * a. packet is really sent to a multicast group
1937 * b. packet is not a NULL-REGISTER
1938 * c. packet is not truncated
Linus Torvalds1da177e2005-04-16 15:20:36 -07001939 */
Joe Perchesf97c1e02007-12-16 13:45:43 -08001940 if (!ipv4_is_multicast(encap->daddr) ||
Linus Torvalds1da177e2005-04-16 15:20:36 -07001941 encap->tot_len == 0 ||
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001942 ntohs(encap->tot_len) + pimlen > skb->len)
1943 return 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001944
1945 read_lock(&mrt_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00001946 if (mrt->mroute_reg_vif_num >= 0)
1947 reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001948 read_unlock(&mrt_lock);
1949
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001950 if (reg_dev == NULL)
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001951 return 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001952
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -07001953 skb->mac_header = skb->network_header;
Eric Dumazet55747a02010-10-01 16:14:55 +00001954 skb_pull(skb, (u8 *)encap - skb->data);
Arnaldo Carvalho de Melo31c77112007-03-10 19:04:55 -03001955 skb_reset_network_header(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001956 skb->protocol = htons(ETH_P_IP);
Eric Dumazet55747a02010-10-01 16:14:55 +00001957 skb->ip_summed = CHECKSUM_NONE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001958 skb->pkt_type = PACKET_HOST;
Eric Dumazetd19d56d2010-05-17 22:36:55 -07001959
1960 skb_tunnel_rx(skb, reg_dev);
1961
Linus Torvalds1da177e2005-04-16 15:20:36 -07001962 netif_rx(skb);
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001963
Eric Dumazet55747a02010-10-01 16:14:55 +00001964 return NET_RX_SUCCESS;
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001965}
1966#endif
1967
1968#ifdef CONFIG_IP_PIMSM_V1
1969/*
1970 * Handle IGMP messages of PIMv1
1971 */
1972
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00001973int pim_rcv_v1(struct sk_buff *skb)
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001974{
1975 struct igmphdr *pim;
Benjamin Thery4feb88e2009-01-22 04:56:23 +00001976 struct net *net = dev_net(skb->dev);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001977 struct mr_table *mrt;
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001978
1979 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr)))
1980 goto drop;
1981
1982 pim = igmp_hdr(skb);
1983
David S. Miller417da662011-05-03 19:42:43 -07001984 mrt = ipmr_rt_fib_lookup(net, skb);
David S. Milleree3f1aa2011-03-09 14:06:20 -08001985 if (IS_ERR(mrt))
1986 goto drop;
Patrick McHardy0c122952010-04-13 05:03:22 +00001987 if (!mrt->mroute_do_pim ||
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001988 pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER)
1989 goto drop;
1990
Patrick McHardyf0ad0862010-04-13 05:03:23 +00001991 if (__pim_rcv(mrt, skb, sizeof(*pim))) {
Ilpo Järvinenb1879202008-12-16 01:15:11 -08001992drop:
1993 kfree_skb(skb);
1994 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07001995 return 0;
1996}
1997#endif
1998
1999#ifdef CONFIG_IP_PIMSM_V2
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002000static int pim_rcv(struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002001{
2002 struct pimreghdr *pim;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002003 struct net *net = dev_net(skb->dev);
2004 struct mr_table *mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002005
Ilpo Järvinenb1879202008-12-16 01:15:11 -08002006 if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr)))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002007 goto drop;
2008
Arnaldo Carvalho de Melo9c702202007-04-25 18:04:18 -07002009 pim = (struct pimreghdr *)skb_transport_header(skb);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002010 if (pim->type != ((PIM_VERSION << 4) | (PIM_REGISTER)) ||
2011 (pim->flags & PIM_NULL_REGISTER) ||
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002012 (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 &&
Al Virod3bc23e2006-11-14 21:24:49 -08002013 csum_fold(skb_checksum(skb, 0, skb->len, 0))))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002014 goto drop;
2015
David S. Miller417da662011-05-03 19:42:43 -07002016 mrt = ipmr_rt_fib_lookup(net, skb);
David S. Milleree3f1aa2011-03-09 14:06:20 -08002017 if (IS_ERR(mrt))
2018 goto drop;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002019 if (__pim_rcv(mrt, skb, sizeof(*pim))) {
Ilpo Järvinenb1879202008-12-16 01:15:11 -08002020drop:
2021 kfree_skb(skb);
2022 }
Linus Torvalds1da177e2005-04-16 15:20:36 -07002023 return 0;
2024}
2025#endif
2026
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002027static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
2028 struct mfc_cache *c, struct rtmsg *rtm)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002029{
2030 int ct;
2031 struct rtnexthop *nhp;
Thomas Graf92a395e2012-06-26 23:36:13 +00002032 struct nlattr *mp_attr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002033
Nicolas Dichtel74381892010-03-25 23:45:35 +00002034 /* If cache is unresolved, don't try to parse IIF and OIF */
Dan Carpentered0f160a2010-05-26 00:38:56 -07002035 if (c->mfc_parent >= MAXVIFS)
Nicolas Dichtel74381892010-03-25 23:45:35 +00002036 return -ENOENT;
2037
Thomas Graf92a395e2012-06-26 23:36:13 +00002038 if (VIF_EXISTS(mrt, c->mfc_parent) &&
2039 nla_put_u32(skb, RTA_IIF, mrt->vif_table[c->mfc_parent].dev->ifindex) < 0)
2040 return -EMSGSIZE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002041
Thomas Graf92a395e2012-06-26 23:36:13 +00002042 if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH)))
2043 return -EMSGSIZE;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002044
2045 for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) {
Patrick McHardy0c122952010-04-13 05:03:22 +00002046 if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) {
Thomas Graf92a395e2012-06-26 23:36:13 +00002047 if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) {
2048 nla_nest_cancel(skb, mp_attr);
2049 return -EMSGSIZE;
2050 }
2051
Linus Torvalds1da177e2005-04-16 15:20:36 -07002052 nhp->rtnh_flags = 0;
2053 nhp->rtnh_hops = c->mfc_un.res.ttls[ct];
Patrick McHardy0c122952010-04-13 05:03:22 +00002054 nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002055 nhp->rtnh_len = sizeof(*nhp);
2056 }
2057 }
Thomas Graf92a395e2012-06-26 23:36:13 +00002058
2059 nla_nest_end(skb, mp_attr);
2060
Linus Torvalds1da177e2005-04-16 15:20:36 -07002061 rtm->rtm_type = RTN_MULTICAST;
2062 return 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002063}
2064
David S. Miller9a1b9492011-05-04 12:18:54 -07002065int ipmr_get_route(struct net *net, struct sk_buff *skb,
2066 __be32 saddr, __be32 daddr,
2067 struct rtmsg *rtm, int nowait)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002068{
Linus Torvalds1da177e2005-04-16 15:20:36 -07002069 struct mfc_cache *cache;
David S. Miller9a1b9492011-05-04 12:18:54 -07002070 struct mr_table *mrt;
2071 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002072
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002073 mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
2074 if (mrt == NULL)
2075 return -ENOENT;
2076
Eric Dumazeta8c94862010-10-01 16:15:08 +00002077 rcu_read_lock();
David S. Miller9a1b9492011-05-04 12:18:54 -07002078 cache = ipmr_cache_find(mrt, saddr, daddr);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002079
Jianjun Kongc354e122008-11-03 00:28:02 -08002080 if (cache == NULL) {
Alexey Kuznetsov72287492006-07-25 16:45:12 -07002081 struct sk_buff *skb2;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07002082 struct iphdr *iph;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002083 struct net_device *dev;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002084 int vif = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002085
2086 if (nowait) {
Eric Dumazeta8c94862010-10-01 16:15:08 +00002087 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002088 return -EAGAIN;
2089 }
2090
2091 dev = skb->dev;
Eric Dumazeta8c94862010-10-01 16:15:08 +00002092 read_lock(&mrt_lock);
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002093 if (dev)
2094 vif = ipmr_find_vif(mrt, dev);
2095 if (vif < 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002096 read_unlock(&mrt_lock);
Eric Dumazeta8c94862010-10-01 16:15:08 +00002097 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002098 return -ENODEV;
2099 }
Alexey Kuznetsov72287492006-07-25 16:45:12 -07002100 skb2 = skb_clone(skb, GFP_ATOMIC);
2101 if (!skb2) {
2102 read_unlock(&mrt_lock);
Eric Dumazeta8c94862010-10-01 16:15:08 +00002103 rcu_read_unlock();
Alexey Kuznetsov72287492006-07-25 16:45:12 -07002104 return -ENOMEM;
2105 }
2106
Arnaldo Carvalho de Meloe2d1bca2007-04-10 20:46:21 -07002107 skb_push(skb2, sizeof(struct iphdr));
2108 skb_reset_network_header(skb2);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07002109 iph = ip_hdr(skb2);
2110 iph->ihl = sizeof(struct iphdr) >> 2;
David S. Miller9a1b9492011-05-04 12:18:54 -07002111 iph->saddr = saddr;
2112 iph->daddr = daddr;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -07002113 iph->version = 0;
Patrick McHardy0c122952010-04-13 05:03:22 +00002114 err = ipmr_cache_unresolved(mrt, vif, skb2);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002115 read_unlock(&mrt_lock);
Eric Dumazeta8c94862010-10-01 16:15:08 +00002116 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002117 return err;
2118 }
2119
Eric Dumazeta8c94862010-10-01 16:15:08 +00002120 read_lock(&mrt_lock);
2121 if (!nowait && (rtm->rtm_flags & RTM_F_NOTIFY))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002122 cache->mfc_flags |= MFC_NOTIFY;
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002123 err = __ipmr_fill_mroute(mrt, skb, cache, rtm);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002124 read_unlock(&mrt_lock);
Eric Dumazeta8c94862010-10-01 16:15:08 +00002125 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002126 return err;
2127}
2128
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002129static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb,
Eric W. Biederman15e47302012-09-07 20:12:54 +00002130 u32 portid, u32 seq, struct mfc_cache *c)
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002131{
2132 struct nlmsghdr *nlh;
2133 struct rtmsg *rtm;
2134
Eric W. Biederman15e47302012-09-07 20:12:54 +00002135 nlh = nlmsg_put(skb, portid, seq, RTM_NEWROUTE, sizeof(*rtm), NLM_F_MULTI);
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002136 if (nlh == NULL)
2137 return -EMSGSIZE;
2138
2139 rtm = nlmsg_data(nlh);
2140 rtm->rtm_family = RTNL_FAMILY_IPMR;
2141 rtm->rtm_dst_len = 32;
2142 rtm->rtm_src_len = 32;
2143 rtm->rtm_tos = 0;
2144 rtm->rtm_table = mrt->id;
David S. Millerf3756b72012-04-01 20:39:02 -04002145 if (nla_put_u32(skb, RTA_TABLE, mrt->id))
2146 goto nla_put_failure;
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002147 rtm->rtm_type = RTN_MULTICAST;
2148 rtm->rtm_scope = RT_SCOPE_UNIVERSE;
2149 rtm->rtm_protocol = RTPROT_UNSPEC;
2150 rtm->rtm_flags = 0;
2151
David S. Millerf3756b72012-04-01 20:39:02 -04002152 if (nla_put_be32(skb, RTA_SRC, c->mfc_origin) ||
2153 nla_put_be32(skb, RTA_DST, c->mfc_mcastgrp))
2154 goto nla_put_failure;
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002155 if (__ipmr_fill_mroute(mrt, skb, c, rtm) < 0)
2156 goto nla_put_failure;
2157
2158 return nlmsg_end(skb, nlh);
2159
2160nla_put_failure:
2161 nlmsg_cancel(skb, nlh);
2162 return -EMSGSIZE;
2163}
2164
2165static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb)
2166{
2167 struct net *net = sock_net(skb->sk);
2168 struct mr_table *mrt;
2169 struct mfc_cache *mfc;
2170 unsigned int t = 0, s_t;
2171 unsigned int h = 0, s_h;
2172 unsigned int e = 0, s_e;
2173
2174 s_t = cb->args[0];
2175 s_h = cb->args[1];
2176 s_e = cb->args[2];
2177
Eric Dumazeta8c94862010-10-01 16:15:08 +00002178 rcu_read_lock();
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002179 ipmr_for_each_table(mrt, net) {
2180 if (t < s_t)
2181 goto next_table;
2182 if (t > s_t)
2183 s_h = 0;
2184 for (h = s_h; h < MFC_LINES; h++) {
Eric Dumazeta8c94862010-10-01 16:15:08 +00002185 list_for_each_entry_rcu(mfc, &mrt->mfc_cache_array[h], list) {
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002186 if (e < s_e)
2187 goto next_entry;
2188 if (ipmr_fill_mroute(mrt, skb,
Eric W. Biederman15e47302012-09-07 20:12:54 +00002189 NETLINK_CB(cb->skb).portid,
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002190 cb->nlh->nlmsg_seq,
2191 mfc) < 0)
2192 goto done;
2193next_entry:
2194 e++;
2195 }
2196 e = s_e = 0;
2197 }
2198 s_h = 0;
2199next_table:
2200 t++;
2201 }
2202done:
Eric Dumazeta8c94862010-10-01 16:15:08 +00002203 rcu_read_unlock();
Patrick McHardycb6a4e42010-04-26 16:02:08 +02002204
2205 cb->args[2] = e;
2206 cb->args[1] = h;
2207 cb->args[0] = t;
2208
2209 return skb->len;
2210}
2211
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002212#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -07002213/*
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002214 * The /proc interfaces to multicast routing :
2215 * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002216 */
2217struct ipmr_vif_iter {
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002218 struct seq_net_private p;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002219 struct mr_table *mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002220 int ct;
2221};
2222
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002223static struct vif_device *ipmr_vif_seq_idx(struct net *net,
2224 struct ipmr_vif_iter *iter,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002225 loff_t pos)
2226{
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002227 struct mr_table *mrt = iter->mrt;
Patrick McHardy0c122952010-04-13 05:03:22 +00002228
2229 for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) {
2230 if (!VIF_EXISTS(mrt, iter->ct))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002231 continue;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002232 if (pos-- == 0)
Patrick McHardy0c122952010-04-13 05:03:22 +00002233 return &mrt->vif_table[iter->ct];
Linus Torvalds1da177e2005-04-16 15:20:36 -07002234 }
2235 return NULL;
2236}
2237
2238static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos)
Stephen Hemmingerba93ef72008-01-21 17:28:59 -08002239 __acquires(mrt_lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002240{
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002241 struct ipmr_vif_iter *iter = seq->private;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002242 struct net *net = seq_file_net(seq);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002243 struct mr_table *mrt;
2244
2245 mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
2246 if (mrt == NULL)
2247 return ERR_PTR(-ENOENT);
2248
2249 iter->mrt = mrt;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002250
Linus Torvalds1da177e2005-04-16 15:20:36 -07002251 read_lock(&mrt_lock);
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002252 return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002253 : SEQ_START_TOKEN;
2254}
2255
2256static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos)
2257{
2258 struct ipmr_vif_iter *iter = seq->private;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002259 struct net *net = seq_file_net(seq);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002260 struct mr_table *mrt = iter->mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002261
2262 ++*pos;
2263 if (v == SEQ_START_TOKEN)
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002264 return ipmr_vif_seq_idx(net, iter, 0);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002265
Patrick McHardy0c122952010-04-13 05:03:22 +00002266 while (++iter->ct < mrt->maxvif) {
2267 if (!VIF_EXISTS(mrt, iter->ct))
Linus Torvalds1da177e2005-04-16 15:20:36 -07002268 continue;
Patrick McHardy0c122952010-04-13 05:03:22 +00002269 return &mrt->vif_table[iter->ct];
Linus Torvalds1da177e2005-04-16 15:20:36 -07002270 }
2271 return NULL;
2272}
2273
2274static void ipmr_vif_seq_stop(struct seq_file *seq, void *v)
Stephen Hemmingerba93ef72008-01-21 17:28:59 -08002275 __releases(mrt_lock)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002276{
2277 read_unlock(&mrt_lock);
2278}
2279
2280static int ipmr_vif_seq_show(struct seq_file *seq, void *v)
2281{
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002282 struct ipmr_vif_iter *iter = seq->private;
2283 struct mr_table *mrt = iter->mrt;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002284
Linus Torvalds1da177e2005-04-16 15:20:36 -07002285 if (v == SEQ_START_TOKEN) {
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002286 seq_puts(seq,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002287 "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n");
2288 } else {
2289 const struct vif_device *vif = v;
2290 const char *name = vif->dev ? vif->dev->name : "none";
2291
2292 seq_printf(seq,
2293 "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n",
Patrick McHardy0c122952010-04-13 05:03:22 +00002294 vif - mrt->vif_table,
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002295 name, vif->bytes_in, vif->pkt_in,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002296 vif->bytes_out, vif->pkt_out,
2297 vif->flags, vif->local, vif->remote);
2298 }
2299 return 0;
2300}
2301
Stephen Hemmingerf6908082007-03-12 14:34:29 -07002302static const struct seq_operations ipmr_vif_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002303 .start = ipmr_vif_seq_start,
2304 .next = ipmr_vif_seq_next,
2305 .stop = ipmr_vif_seq_stop,
2306 .show = ipmr_vif_seq_show,
2307};
2308
2309static int ipmr_vif_open(struct inode *inode, struct file *file)
2310{
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002311 return seq_open_net(inode, file, &ipmr_vif_seq_ops,
2312 sizeof(struct ipmr_vif_iter));
Linus Torvalds1da177e2005-04-16 15:20:36 -07002313}
2314
Arjan van de Ven9a321442007-02-12 00:55:35 -08002315static const struct file_operations ipmr_vif_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002316 .owner = THIS_MODULE,
2317 .open = ipmr_vif_open,
2318 .read = seq_read,
2319 .llseek = seq_lseek,
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002320 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002321};
2322
2323struct ipmr_mfc_iter {
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002324 struct seq_net_private p;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002325 struct mr_table *mrt;
Patrick McHardy862465f2010-04-13 05:03:21 +00002326 struct list_head *cache;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002327 int ct;
2328};
2329
2330
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002331static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net,
2332 struct ipmr_mfc_iter *it, loff_t pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002333{
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002334 struct mr_table *mrt = it->mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002335 struct mfc_cache *mfc;
2336
Eric Dumazeta8c94862010-10-01 16:15:08 +00002337 rcu_read_lock();
Patrick McHardy862465f2010-04-13 05:03:21 +00002338 for (it->ct = 0; it->ct < MFC_LINES; it->ct++) {
Patrick McHardy0c122952010-04-13 05:03:22 +00002339 it->cache = &mrt->mfc_cache_array[it->ct];
Eric Dumazeta8c94862010-10-01 16:15:08 +00002340 list_for_each_entry_rcu(mfc, it->cache, list)
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002341 if (pos-- == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002342 return mfc;
Patrick McHardy862465f2010-04-13 05:03:21 +00002343 }
Eric Dumazeta8c94862010-10-01 16:15:08 +00002344 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002345
Linus Torvalds1da177e2005-04-16 15:20:36 -07002346 spin_lock_bh(&mfc_unres_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00002347 it->cache = &mrt->mfc_unres_queue;
Patrick McHardy862465f2010-04-13 05:03:21 +00002348 list_for_each_entry(mfc, it->cache, list)
Patrick McHardye258beb2010-04-13 05:03:19 +00002349 if (pos-- == 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002350 return mfc;
2351 spin_unlock_bh(&mfc_unres_lock);
2352
2353 it->cache = NULL;
2354 return NULL;
2355}
2356
2357
2358static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos)
2359{
2360 struct ipmr_mfc_iter *it = seq->private;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002361 struct net *net = seq_file_net(seq);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002362 struct mr_table *mrt;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002363
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002364 mrt = ipmr_get_table(net, RT_TABLE_DEFAULT);
2365 if (mrt == NULL)
2366 return ERR_PTR(-ENOENT);
2367
2368 it->mrt = mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002369 it->cache = NULL;
2370 it->ct = 0;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002371 return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002372 : SEQ_START_TOKEN;
2373}
2374
2375static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos)
2376{
2377 struct mfc_cache *mfc = v;
2378 struct ipmr_mfc_iter *it = seq->private;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002379 struct net *net = seq_file_net(seq);
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002380 struct mr_table *mrt = it->mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002381
2382 ++*pos;
2383
2384 if (v == SEQ_START_TOKEN)
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002385 return ipmr_mfc_seq_idx(net, seq->private, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002386
Patrick McHardy862465f2010-04-13 05:03:21 +00002387 if (mfc->list.next != it->cache)
2388 return list_entry(mfc->list.next, struct mfc_cache, list);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002389
Patrick McHardy0c122952010-04-13 05:03:22 +00002390 if (it->cache == &mrt->mfc_unres_queue)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002391 goto end_of_list;
2392
Patrick McHardy0c122952010-04-13 05:03:22 +00002393 BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002394
2395 while (++it->ct < MFC_LINES) {
Patrick McHardy0c122952010-04-13 05:03:22 +00002396 it->cache = &mrt->mfc_cache_array[it->ct];
Patrick McHardy862465f2010-04-13 05:03:21 +00002397 if (list_empty(it->cache))
2398 continue;
2399 return list_first_entry(it->cache, struct mfc_cache, list);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002400 }
2401
2402 /* exhausted cache_array, show unresolved */
Eric Dumazeta8c94862010-10-01 16:15:08 +00002403 rcu_read_unlock();
Patrick McHardy0c122952010-04-13 05:03:22 +00002404 it->cache = &mrt->mfc_unres_queue;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002405 it->ct = 0;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002406
Linus Torvalds1da177e2005-04-16 15:20:36 -07002407 spin_lock_bh(&mfc_unres_lock);
Patrick McHardy862465f2010-04-13 05:03:21 +00002408 if (!list_empty(it->cache))
2409 return list_first_entry(it->cache, struct mfc_cache, list);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002410
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002411end_of_list:
Linus Torvalds1da177e2005-04-16 15:20:36 -07002412 spin_unlock_bh(&mfc_unres_lock);
2413 it->cache = NULL;
2414
2415 return NULL;
2416}
2417
2418static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v)
2419{
2420 struct ipmr_mfc_iter *it = seq->private;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002421 struct mr_table *mrt = it->mrt;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002422
Patrick McHardy0c122952010-04-13 05:03:22 +00002423 if (it->cache == &mrt->mfc_unres_queue)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002424 spin_unlock_bh(&mfc_unres_lock);
Patrick McHardy0c122952010-04-13 05:03:22 +00002425 else if (it->cache == &mrt->mfc_cache_array[it->ct])
Eric Dumazeta8c94862010-10-01 16:15:08 +00002426 rcu_read_unlock();
Linus Torvalds1da177e2005-04-16 15:20:36 -07002427}
2428
2429static int ipmr_mfc_seq_show(struct seq_file *seq, void *v)
2430{
2431 int n;
2432
2433 if (v == SEQ_START_TOKEN) {
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002434 seq_puts(seq,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002435 "Group Origin Iif Pkts Bytes Wrong Oifs\n");
2436 } else {
2437 const struct mfc_cache *mfc = v;
2438 const struct ipmr_mfc_iter *it = seq->private;
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002439 const struct mr_table *mrt = it->mrt;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002440
Eric Dumazet0eae88f2010-04-20 19:06:52 -07002441 seq_printf(seq, "%08X %08X %-3hd",
2442 (__force u32) mfc->mfc_mcastgrp,
2443 (__force u32) mfc->mfc_origin,
Benjamin Thery1ea472e2008-12-03 22:21:47 -08002444 mfc->mfc_parent);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002445
Patrick McHardy0c122952010-04-13 05:03:22 +00002446 if (it->cache != &mrt->mfc_unres_queue) {
Benjamin Thery1ea472e2008-12-03 22:21:47 -08002447 seq_printf(seq, " %8lu %8lu %8lu",
2448 mfc->mfc_un.res.pkt,
2449 mfc->mfc_un.res.bytes,
2450 mfc->mfc_un.res.wrong_if);
Stephen Hemminger132adf52007-03-08 20:44:43 -08002451 for (n = mfc->mfc_un.res.minvif;
Eric Dumazeta8cb16d2010-10-01 16:15:29 +00002452 n < mfc->mfc_un.res.maxvif; n++) {
Patrick McHardy0c122952010-04-13 05:03:22 +00002453 if (VIF_EXISTS(mrt, n) &&
Benjamin Therycf958ae32009-01-22 04:56:16 +00002454 mfc->mfc_un.res.ttls[n] < 255)
2455 seq_printf(seq,
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002456 " %2d:%-3d",
Linus Torvalds1da177e2005-04-16 15:20:36 -07002457 n, mfc->mfc_un.res.ttls[n]);
2458 }
Benjamin Thery1ea472e2008-12-03 22:21:47 -08002459 } else {
2460 /* unresolved mfc_caches don't contain
2461 * pkt, bytes and wrong_if values
2462 */
2463 seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul);
Linus Torvalds1da177e2005-04-16 15:20:36 -07002464 }
2465 seq_putc(seq, '\n');
2466 }
2467 return 0;
2468}
2469
Stephen Hemmingerf6908082007-03-12 14:34:29 -07002470static const struct seq_operations ipmr_mfc_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002471 .start = ipmr_mfc_seq_start,
2472 .next = ipmr_mfc_seq_next,
2473 .stop = ipmr_mfc_seq_stop,
2474 .show = ipmr_mfc_seq_show,
2475};
2476
2477static int ipmr_mfc_open(struct inode *inode, struct file *file)
2478{
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002479 return seq_open_net(inode, file, &ipmr_mfc_seq_ops,
2480 sizeof(struct ipmr_mfc_iter));
Linus Torvalds1da177e2005-04-16 15:20:36 -07002481}
2482
Arjan van de Ven9a321442007-02-12 00:55:35 -08002483static const struct file_operations ipmr_mfc_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002484 .owner = THIS_MODULE,
2485 .open = ipmr_mfc_open,
2486 .read = seq_read,
2487 .llseek = seq_lseek,
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002488 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002489};
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002490#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -07002491
2492#ifdef CONFIG_IP_PIMSM_V2
Alexey Dobriyan32613092009-09-14 12:21:47 +00002493static const struct net_protocol pim_protocol = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07002494 .handler = pim_rcv,
Tom Goff403dbb92009-06-14 03:16:13 -07002495 .netns_ok = 1,
Linus Torvalds1da177e2005-04-16 15:20:36 -07002496};
2497#endif
2498
2499
2500/*
2501 * Setup for IP multicast routing
2502 */
Benjamin Therycf958ae32009-01-22 04:56:16 +00002503static int __net_init ipmr_net_init(struct net *net)
2504{
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002505 int err;
Benjamin Therycf958ae32009-01-22 04:56:16 +00002506
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002507 err = ipmr_rules_init(net);
2508 if (err < 0)
Benjamin Therycf958ae32009-01-22 04:56:16 +00002509 goto fail;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002510
2511#ifdef CONFIG_PROC_FS
2512 err = -ENOMEM;
2513 if (!proc_net_fops_create(net, "ip_mr_vif", 0, &ipmr_vif_fops))
2514 goto proc_vif_fail;
2515 if (!proc_net_fops_create(net, "ip_mr_cache", 0, &ipmr_mfc_fops))
2516 goto proc_cache_fail;
2517#endif
Benjamin Thery2bb8b262009-01-22 04:56:18 +00002518 return 0;
2519
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002520#ifdef CONFIG_PROC_FS
2521proc_cache_fail:
2522 proc_net_remove(net, "ip_mr_vif");
2523proc_vif_fail:
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002524 ipmr_rules_exit(net);
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002525#endif
Benjamin Therycf958ae32009-01-22 04:56:16 +00002526fail:
2527 return err;
2528}
2529
2530static void __net_exit ipmr_net_exit(struct net *net)
2531{
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002532#ifdef CONFIG_PROC_FS
2533 proc_net_remove(net, "ip_mr_cache");
2534 proc_net_remove(net, "ip_mr_vif");
2535#endif
Patrick McHardyf0ad0862010-04-13 05:03:23 +00002536 ipmr_rules_exit(net);
Benjamin Therycf958ae32009-01-22 04:56:16 +00002537}
2538
2539static struct pernet_operations ipmr_net_ops = {
2540 .init = ipmr_net_init,
2541 .exit = ipmr_net_exit,
2542};
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002543
Wang Chen03d2f892008-07-03 12:13:36 +08002544int __init ip_mr_init(void)
Linus Torvalds1da177e2005-04-16 15:20:36 -07002545{
Wang Chen03d2f892008-07-03 12:13:36 +08002546 int err;
2547
Linus Torvalds1da177e2005-04-16 15:20:36 -07002548 mrt_cachep = kmem_cache_create("ip_mrt_cache",
2549 sizeof(struct mfc_cache),
Eric Dumazeta8c94862010-10-01 16:15:08 +00002550 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC,
Paul Mundt20c2df82007-07-20 10:11:58 +09002551 NULL);
Wang Chen03d2f892008-07-03 12:13:36 +08002552 if (!mrt_cachep)
2553 return -ENOMEM;
2554
Benjamin Therycf958ae32009-01-22 04:56:16 +00002555 err = register_pernet_subsys(&ipmr_net_ops);
2556 if (err)
2557 goto reg_pernet_fail;
2558
Wang Chen03d2f892008-07-03 12:13:36 +08002559 err = register_netdevice_notifier(&ip_mr_notifier);
2560 if (err)
2561 goto reg_notif_fail;
Tom Goff403dbb92009-06-14 03:16:13 -07002562#ifdef CONFIG_IP_PIMSM_V2
2563 if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) {
Joe Perches058bd4d2012-03-11 18:36:11 +00002564 pr_err("%s: can't add PIM protocol\n", __func__);
Tom Goff403dbb92009-06-14 03:16:13 -07002565 err = -EAGAIN;
2566 goto add_proto_fail;
2567 }
2568#endif
Greg Rosec7ac8672011-06-10 01:27:09 +00002569 rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE,
2570 NULL, ipmr_rtm_dumproute, NULL);
Wang Chen03d2f892008-07-03 12:13:36 +08002571 return 0;
Benjamin Theryf6bb4512009-01-22 04:56:22 +00002572
Tom Goff403dbb92009-06-14 03:16:13 -07002573#ifdef CONFIG_IP_PIMSM_V2
2574add_proto_fail:
2575 unregister_netdevice_notifier(&ip_mr_notifier);
2576#endif
Benjamin Theryc3e38892008-11-19 14:07:41 -08002577reg_notif_fail:
Benjamin Therycf958ae32009-01-22 04:56:16 +00002578 unregister_pernet_subsys(&ipmr_net_ops);
2579reg_pernet_fail:
Benjamin Theryc3e38892008-11-19 14:07:41 -08002580 kmem_cache_destroy(mrt_cachep);
Wang Chen03d2f892008-07-03 12:13:36 +08002581 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -07002582}