Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * IP multicast routing support for mrouted 3.6/3.8 | |
3 | * | |
113aa838 | 4 | * (c) 1995 Alan Cox, <alan@lxorguk.ukuu.org.uk> |
1da177e4 LT |
5 | * 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 | * | |
1da177e4 LT |
12 | * 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 | |
f77f13e2 | 25 | * Relax this requirement to work with older peers. |
1da177e4 LT |
26 | * |
27 | */ | |
28 | ||
1da177e4 LT |
29 | #include <asm/uaccess.h> |
30 | #include <linux/types.h> | |
4fc268d2 | 31 | #include <linux/capability.h> |
1da177e4 LT |
32 | #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> | |
46f25dff | 48 | #include <linux/if_ether.h> |
5a0e3ad6 | 49 | #include <linux/slab.h> |
457c4cbc | 50 | #include <net/net_namespace.h> |
1da177e4 LT |
51 | #include <net/ip.h> |
52 | #include <net/protocol.h> | |
53 | #include <linux/skbuff.h> | |
14c85021 | 54 | #include <net/route.h> |
1da177e4 LT |
55 | #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> | |
709b46e8 | 62 | #include <linux/compat.h> |
bc3b2d7f | 63 | #include <linux/export.h> |
c5441932 | 64 | #include <net/ip_tunnels.h> |
1da177e4 | 65 | #include <net/checksum.h> |
dc5fc579 | 66 | #include <net/netlink.h> |
f0ad0860 | 67 | #include <net/fib_rules.h> |
d67b8c61 | 68 | #include <linux/netconf.h> |
1da177e4 | 69 | |
0c12295a | 70 | struct mr_table { |
f0ad0860 | 71 | struct list_head list; |
0c5c9fb5 | 72 | possible_net_t net; |
f0ad0860 | 73 | u32 id; |
4c968709 | 74 | struct sock __rcu *mroute_sk; |
0c12295a PM |
75 | struct timer_list ipmr_expire_timer; |
76 | struct list_head mfc_unres_queue; | |
77 | struct list_head mfc_cache_array[MFC_LINES]; | |
78 | struct vif_device vif_table[MAXVIFS]; | |
79 | int maxvif; | |
80 | atomic_t cache_resolve_queue_len; | |
53d6841d JP |
81 | bool mroute_do_assert; |
82 | bool mroute_do_pim; | |
0c12295a | 83 | int mroute_reg_vif_num; |
0c12295a PM |
84 | }; |
85 | ||
f0ad0860 PM |
86 | struct ipmr_rule { |
87 | struct fib_rule common; | |
88 | }; | |
89 | ||
90 | struct ipmr_result { | |
91 | struct mr_table *mrt; | |
92 | }; | |
93 | ||
c316c629 NA |
94 | static inline bool pimsm_enabled(void) |
95 | { | |
96 | return IS_BUILTIN(CONFIG_IP_PIMSM_V1) || IS_BUILTIN(CONFIG_IP_PIMSM_V2); | |
97 | } | |
98 | ||
1da177e4 | 99 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
a8cb16dd | 100 | * Note that the changes are semaphored via rtnl_lock. |
1da177e4 LT |
101 | */ |
102 | ||
103 | static DEFINE_RWLOCK(mrt_lock); | |
104 | ||
7ef8f65d | 105 | /* Multicast router control variables */ |
1da177e4 | 106 | |
0c12295a | 107 | #define VIF_EXISTS(_mrt, _idx) ((_mrt)->vif_table[_idx].dev != NULL) |
1da177e4 LT |
108 | |
109 | /* Special spinlock for queue of unresolved entries */ | |
110 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
111 | ||
112 | /* We return to original Alan's scheme. Hash table of resolved | |
a8cb16dd ED |
113 | * entries is changed only in process context and protected |
114 | * with weak lock mrt_lock. Queue of unresolved entries is protected | |
115 | * with strong spinlock mfc_unres_lock. | |
116 | * | |
117 | * In this case data path is free of exclusive locks at all. | |
1da177e4 LT |
118 | */ |
119 | ||
e18b890b | 120 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 | 121 | |
f0ad0860 | 122 | static struct mr_table *ipmr_new_table(struct net *net, u32 id); |
acbb219d FR |
123 | static void ipmr_free_table(struct mr_table *mrt); |
124 | ||
c4854ec8 RR |
125 | static void ip_mr_forward(struct net *net, struct mr_table *mrt, |
126 | struct sk_buff *skb, struct mfc_cache *cache, | |
127 | int local); | |
0c12295a | 128 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 129 | struct sk_buff *pkt, vifi_t vifi, int assert); |
cb6a4e46 PM |
130 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
131 | struct mfc_cache *c, struct rtmsg *rtm); | |
8cd3ac9f ND |
132 | static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc, |
133 | int cmd); | |
acbb219d | 134 | static void mroute_clean_tables(struct mr_table *mrt); |
f0ad0860 PM |
135 | static void ipmr_expire_process(unsigned long arg); |
136 | ||
137 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
138 | #define ipmr_for_each_table(mrt, net) \ | |
139 | list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list) | |
140 | ||
141 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
142 | { | |
143 | struct mr_table *mrt; | |
144 | ||
145 | ipmr_for_each_table(mrt, net) { | |
146 | if (mrt->id == id) | |
147 | return mrt; | |
148 | } | |
149 | return NULL; | |
150 | } | |
151 | ||
da91981b | 152 | static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4, |
f0ad0860 PM |
153 | struct mr_table **mrt) |
154 | { | |
f0ad0860 | 155 | int err; |
95f4a45d HFS |
156 | struct ipmr_result res; |
157 | struct fib_lookup_arg arg = { | |
158 | .result = &res, | |
159 | .flags = FIB_LOOKUP_NOREF, | |
160 | }; | |
f0ad0860 | 161 | |
da91981b DM |
162 | err = fib_rules_lookup(net->ipv4.mr_rules_ops, |
163 | flowi4_to_flowi(flp4), 0, &arg); | |
f0ad0860 PM |
164 | if (err < 0) |
165 | return err; | |
166 | *mrt = res.mrt; | |
167 | return 0; | |
168 | } | |
169 | ||
170 | static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp, | |
171 | int flags, struct fib_lookup_arg *arg) | |
172 | { | |
173 | struct ipmr_result *res = arg->result; | |
174 | struct mr_table *mrt; | |
1da177e4 | 175 | |
f0ad0860 PM |
176 | switch (rule->action) { |
177 | case FR_ACT_TO_TBL: | |
178 | break; | |
179 | case FR_ACT_UNREACHABLE: | |
180 | return -ENETUNREACH; | |
181 | case FR_ACT_PROHIBIT: | |
182 | return -EACCES; | |
183 | case FR_ACT_BLACKHOLE: | |
184 | default: | |
185 | return -EINVAL; | |
186 | } | |
187 | ||
188 | mrt = ipmr_get_table(rule->fr_net, rule->table); | |
51456b29 | 189 | if (!mrt) |
f0ad0860 PM |
190 | return -EAGAIN; |
191 | res->mrt = mrt; | |
192 | return 0; | |
193 | } | |
194 | ||
195 | static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags) | |
196 | { | |
197 | return 1; | |
198 | } | |
199 | ||
200 | static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = { | |
201 | FRA_GENERIC_POLICY, | |
202 | }; | |
203 | ||
204 | static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb, | |
205 | struct fib_rule_hdr *frh, struct nlattr **tb) | |
206 | { | |
207 | return 0; | |
208 | } | |
209 | ||
210 | static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh, | |
211 | struct nlattr **tb) | |
212 | { | |
213 | return 1; | |
214 | } | |
215 | ||
216 | static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb, | |
217 | struct fib_rule_hdr *frh) | |
218 | { | |
219 | frh->dst_len = 0; | |
220 | frh->src_len = 0; | |
221 | frh->tos = 0; | |
222 | return 0; | |
223 | } | |
224 | ||
04a6f82c | 225 | static const struct fib_rules_ops __net_initconst ipmr_rules_ops_template = { |
25239cee | 226 | .family = RTNL_FAMILY_IPMR, |
f0ad0860 PM |
227 | .rule_size = sizeof(struct ipmr_rule), |
228 | .addr_size = sizeof(u32), | |
229 | .action = ipmr_rule_action, | |
230 | .match = ipmr_rule_match, | |
231 | .configure = ipmr_rule_configure, | |
232 | .compare = ipmr_rule_compare, | |
f0ad0860 PM |
233 | .fill = ipmr_rule_fill, |
234 | .nlgroup = RTNLGRP_IPV4_RULE, | |
235 | .policy = ipmr_rule_policy, | |
236 | .owner = THIS_MODULE, | |
237 | }; | |
238 | ||
239 | static 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); | |
1113ebbc NA |
252 | if (IS_ERR(mrt)) { |
253 | err = PTR_ERR(mrt); | |
f0ad0860 PM |
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 | ||
264 | err2: | |
f243e5a7 | 265 | ipmr_free_table(mrt); |
f0ad0860 PM |
266 | err1: |
267 | fib_rules_unregister(ops); | |
268 | return err; | |
269 | } | |
270 | ||
271 | static void __net_exit ipmr_rules_exit(struct net *net) | |
272 | { | |
273 | struct mr_table *mrt, *next; | |
274 | ||
ed785309 | 275 | rtnl_lock(); |
035320d5 ED |
276 | list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) { |
277 | list_del(&mrt->list); | |
acbb219d | 278 | ipmr_free_table(mrt); |
035320d5 | 279 | } |
f0ad0860 | 280 | fib_rules_unregister(net->ipv4.mr_rules_ops); |
419df12f | 281 | rtnl_unlock(); |
f0ad0860 PM |
282 | } |
283 | #else | |
284 | #define ipmr_for_each_table(mrt, net) \ | |
285 | for (mrt = net->ipv4.mrt; mrt; mrt = NULL) | |
286 | ||
287 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
288 | { | |
289 | return net->ipv4.mrt; | |
290 | } | |
291 | ||
da91981b | 292 | static int ipmr_fib_lookup(struct net *net, struct flowi4 *flp4, |
f0ad0860 PM |
293 | struct mr_table **mrt) |
294 | { | |
295 | *mrt = net->ipv4.mrt; | |
296 | return 0; | |
297 | } | |
298 | ||
299 | static int __net_init ipmr_rules_init(struct net *net) | |
300 | { | |
1113ebbc NA |
301 | struct mr_table *mrt; |
302 | ||
303 | mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
304 | if (IS_ERR(mrt)) | |
305 | return PTR_ERR(mrt); | |
306 | net->ipv4.mrt = mrt; | |
307 | return 0; | |
f0ad0860 PM |
308 | } |
309 | ||
310 | static void __net_exit ipmr_rules_exit(struct net *net) | |
311 | { | |
ed785309 | 312 | rtnl_lock(); |
acbb219d | 313 | ipmr_free_table(net->ipv4.mrt); |
ed785309 WC |
314 | net->ipv4.mrt = NULL; |
315 | rtnl_unlock(); | |
f0ad0860 PM |
316 | } |
317 | #endif | |
318 | ||
319 | static struct mr_table *ipmr_new_table(struct net *net, u32 id) | |
320 | { | |
321 | struct mr_table *mrt; | |
322 | unsigned int i; | |
1da177e4 | 323 | |
1113ebbc NA |
324 | /* "pimreg%u" should not exceed 16 bytes (IFNAMSIZ) */ |
325 | if (id != RT_TABLE_DEFAULT && id >= 1000000000) | |
326 | return ERR_PTR(-EINVAL); | |
327 | ||
f0ad0860 | 328 | mrt = ipmr_get_table(net, id); |
00db4124 | 329 | if (mrt) |
f0ad0860 PM |
330 | return mrt; |
331 | ||
332 | mrt = kzalloc(sizeof(*mrt), GFP_KERNEL); | |
51456b29 | 333 | if (!mrt) |
1113ebbc | 334 | return ERR_PTR(-ENOMEM); |
8de53dfb | 335 | write_pnet(&mrt->net, net); |
f0ad0860 PM |
336 | mrt->id = id; |
337 | ||
338 | /* Forwarding cache */ | |
339 | for (i = 0; i < MFC_LINES; i++) | |
340 | INIT_LIST_HEAD(&mrt->mfc_cache_array[i]); | |
341 | ||
342 | INIT_LIST_HEAD(&mrt->mfc_unres_queue); | |
343 | ||
344 | setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process, | |
345 | (unsigned long)mrt); | |
346 | ||
f0ad0860 | 347 | mrt->mroute_reg_vif_num = -1; |
f0ad0860 PM |
348 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES |
349 | list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables); | |
350 | #endif | |
351 | return mrt; | |
352 | } | |
1da177e4 | 353 | |
acbb219d FR |
354 | static void ipmr_free_table(struct mr_table *mrt) |
355 | { | |
356 | del_timer_sync(&mrt->ipmr_expire_timer); | |
357 | mroute_clean_tables(mrt); | |
358 | kfree(mrt); | |
359 | } | |
360 | ||
1da177e4 LT |
361 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ |
362 | ||
d607032d WC |
363 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
364 | { | |
4feb88e5 BT |
365 | struct net *net = dev_net(dev); |
366 | ||
d607032d WC |
367 | dev_close(dev); |
368 | ||
4feb88e5 | 369 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 370 | if (dev) { |
5bc3eb7e | 371 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 372 | struct ifreq ifr; |
d607032d WC |
373 | struct ip_tunnel_parm p; |
374 | ||
375 | memset(&p, 0, sizeof(p)); | |
376 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
377 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
378 | p.iph.version = 4; | |
379 | p.iph.ihl = 5; | |
380 | p.iph.protocol = IPPROTO_IPIP; | |
381 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
382 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
383 | ||
5bc3eb7e SH |
384 | if (ops->ndo_do_ioctl) { |
385 | mm_segment_t oldfs = get_fs(); | |
386 | ||
387 | set_fs(KERNEL_DS); | |
388 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
389 | set_fs(oldfs); | |
390 | } | |
d607032d WC |
391 | } |
392 | } | |
393 | ||
7ef8f65d | 394 | static struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
395 | { |
396 | struct net_device *dev; | |
397 | ||
4feb88e5 | 398 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
399 | |
400 | if (dev) { | |
5bc3eb7e | 401 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
402 | int err; |
403 | struct ifreq ifr; | |
1da177e4 LT |
404 | struct ip_tunnel_parm p; |
405 | struct in_device *in_dev; | |
406 | ||
407 | memset(&p, 0, sizeof(p)); | |
408 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
409 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
410 | p.iph.version = 4; | |
411 | p.iph.ihl = 5; | |
412 | p.iph.protocol = IPPROTO_IPIP; | |
413 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 414 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 415 | |
5bc3eb7e SH |
416 | if (ops->ndo_do_ioctl) { |
417 | mm_segment_t oldfs = get_fs(); | |
418 | ||
419 | set_fs(KERNEL_DS); | |
420 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
421 | set_fs(oldfs); | |
a8cb16dd | 422 | } else { |
5bc3eb7e | 423 | err = -EOPNOTSUPP; |
a8cb16dd | 424 | } |
1da177e4 LT |
425 | dev = NULL; |
426 | ||
4feb88e5 BT |
427 | if (err == 0 && |
428 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 LT |
429 | dev->flags |= IFF_MULTICAST; |
430 | ||
e5ed6399 | 431 | in_dev = __in_dev_get_rtnl(dev); |
51456b29 | 432 | if (!in_dev) |
1da177e4 | 433 | goto failure; |
71e27da9 HX |
434 | |
435 | ipv4_devconf_setall(in_dev); | |
1d4c8c29 | 436 | neigh_parms_data_state_setall(in_dev->arp_parms); |
71e27da9 | 437 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; |
1da177e4 LT |
438 | |
439 | if (dev_open(dev)) | |
440 | goto failure; | |
7dc00c82 | 441 | dev_hold(dev); |
1da177e4 LT |
442 | } |
443 | } | |
444 | return dev; | |
445 | ||
446 | failure: | |
447 | /* allow the register to be completed before unregistering. */ | |
448 | rtnl_unlock(); | |
449 | rtnl_lock(); | |
450 | ||
451 | unregister_netdevice(dev); | |
452 | return NULL; | |
453 | } | |
454 | ||
c316c629 | 455 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) |
6fef4c0c | 456 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 457 | { |
4feb88e5 | 458 | struct net *net = dev_net(dev); |
f0ad0860 | 459 | struct mr_table *mrt; |
da91981b DM |
460 | struct flowi4 fl4 = { |
461 | .flowi4_oif = dev->ifindex, | |
6a662719 | 462 | .flowi4_iif = skb->skb_iif ? : LOOPBACK_IFINDEX, |
da91981b | 463 | .flowi4_mark = skb->mark, |
f0ad0860 PM |
464 | }; |
465 | int err; | |
466 | ||
da91981b | 467 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
e40dbc51 BG |
468 | if (err < 0) { |
469 | kfree_skb(skb); | |
f0ad0860 | 470 | return err; |
e40dbc51 | 471 | } |
4feb88e5 | 472 | |
1da177e4 | 473 | read_lock(&mrt_lock); |
cf3677ae PE |
474 | dev->stats.tx_bytes += skb->len; |
475 | dev->stats.tx_packets++; | |
0c12295a | 476 | ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT); |
1da177e4 LT |
477 | read_unlock(&mrt_lock); |
478 | kfree_skb(skb); | |
6ed10654 | 479 | return NETDEV_TX_OK; |
1da177e4 LT |
480 | } |
481 | ||
ee9b9596 ND |
482 | static int reg_vif_get_iflink(const struct net_device *dev) |
483 | { | |
484 | return 0; | |
485 | } | |
486 | ||
007c3838 SH |
487 | static const struct net_device_ops reg_vif_netdev_ops = { |
488 | .ndo_start_xmit = reg_vif_xmit, | |
ee9b9596 | 489 | .ndo_get_iflink = reg_vif_get_iflink, |
007c3838 SH |
490 | }; |
491 | ||
1da177e4 LT |
492 | static void reg_vif_setup(struct net_device *dev) |
493 | { | |
494 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 495 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 496 | dev->flags = IFF_NOARP; |
70cb4a45 | 497 | dev->netdev_ops = ®_vif_netdev_ops; |
1da177e4 | 498 | dev->destructor = free_netdev; |
403dbb97 | 499 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
500 | } |
501 | ||
f0ad0860 | 502 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) |
1da177e4 LT |
503 | { |
504 | struct net_device *dev; | |
505 | struct in_device *in_dev; | |
f0ad0860 | 506 | char name[IFNAMSIZ]; |
1da177e4 | 507 | |
f0ad0860 PM |
508 | if (mrt->id == RT_TABLE_DEFAULT) |
509 | sprintf(name, "pimreg"); | |
510 | else | |
511 | sprintf(name, "pimreg%u", mrt->id); | |
1da177e4 | 512 | |
c835a677 | 513 | dev = alloc_netdev(0, name, NET_NAME_UNKNOWN, reg_vif_setup); |
1da177e4 | 514 | |
51456b29 | 515 | if (!dev) |
1da177e4 LT |
516 | return NULL; |
517 | ||
403dbb97 TG |
518 | dev_net_set(dev, net); |
519 | ||
1da177e4 LT |
520 | if (register_netdevice(dev)) { |
521 | free_netdev(dev); | |
522 | return NULL; | |
523 | } | |
1da177e4 | 524 | |
71e27da9 | 525 | rcu_read_lock(); |
a8cb16dd ED |
526 | in_dev = __in_dev_get_rcu(dev); |
527 | if (!in_dev) { | |
71e27da9 | 528 | rcu_read_unlock(); |
1da177e4 | 529 | goto failure; |
71e27da9 | 530 | } |
1da177e4 | 531 | |
71e27da9 | 532 | ipv4_devconf_setall(in_dev); |
1d4c8c29 | 533 | neigh_parms_data_state_setall(in_dev->arp_parms); |
71e27da9 HX |
534 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; |
535 | rcu_read_unlock(); | |
1da177e4 LT |
536 | |
537 | if (dev_open(dev)) | |
538 | goto failure; | |
539 | ||
7dc00c82 WC |
540 | dev_hold(dev); |
541 | ||
1da177e4 LT |
542 | return dev; |
543 | ||
544 | failure: | |
545 | /* allow the register to be completed before unregistering. */ | |
546 | rtnl_unlock(); | |
547 | rtnl_lock(); | |
548 | ||
549 | unregister_netdevice(dev); | |
550 | return NULL; | |
551 | } | |
c316c629 NA |
552 | |
553 | /* called with rcu_read_lock() */ | |
554 | static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb, | |
555 | unsigned int pimlen) | |
556 | { | |
557 | struct net_device *reg_dev = NULL; | |
558 | struct iphdr *encap; | |
559 | ||
560 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); | |
7ef8f65d | 561 | /* Check that: |
c316c629 NA |
562 | * a. packet is really sent to a multicast group |
563 | * b. packet is not a NULL-REGISTER | |
564 | * c. packet is not truncated | |
565 | */ | |
566 | if (!ipv4_is_multicast(encap->daddr) || | |
567 | encap->tot_len == 0 || | |
568 | ntohs(encap->tot_len) + pimlen > skb->len) | |
569 | return 1; | |
570 | ||
571 | read_lock(&mrt_lock); | |
572 | if (mrt->mroute_reg_vif_num >= 0) | |
573 | reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev; | |
574 | read_unlock(&mrt_lock); | |
575 | ||
576 | if (!reg_dev) | |
577 | return 1; | |
578 | ||
579 | skb->mac_header = skb->network_header; | |
580 | skb_pull(skb, (u8 *)encap - skb->data); | |
581 | skb_reset_network_header(skb); | |
582 | skb->protocol = htons(ETH_P_IP); | |
583 | skb->ip_summed = CHECKSUM_NONE; | |
584 | ||
585 | skb_tunnel_rx(skb, reg_dev, dev_net(reg_dev)); | |
586 | ||
587 | netif_rx(skb); | |
588 | ||
589 | return NET_RX_SUCCESS; | |
590 | } | |
591 | #else | |
592 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) | |
593 | { | |
594 | return NULL; | |
595 | } | |
1da177e4 LT |
596 | #endif |
597 | ||
2c53040f BH |
598 | /** |
599 | * vif_delete - Delete a VIF entry | |
7dc00c82 | 600 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 601 | */ |
0c12295a | 602 | static int vif_delete(struct mr_table *mrt, int vifi, int notify, |
d17fa6fa | 603 | struct list_head *head) |
1da177e4 LT |
604 | { |
605 | struct vif_device *v; | |
606 | struct net_device *dev; | |
607 | struct in_device *in_dev; | |
608 | ||
0c12295a | 609 | if (vifi < 0 || vifi >= mrt->maxvif) |
1da177e4 LT |
610 | return -EADDRNOTAVAIL; |
611 | ||
0c12295a | 612 | v = &mrt->vif_table[vifi]; |
1da177e4 LT |
613 | |
614 | write_lock_bh(&mrt_lock); | |
615 | dev = v->dev; | |
616 | v->dev = NULL; | |
617 | ||
618 | if (!dev) { | |
619 | write_unlock_bh(&mrt_lock); | |
620 | return -EADDRNOTAVAIL; | |
621 | } | |
622 | ||
0c12295a PM |
623 | if (vifi == mrt->mroute_reg_vif_num) |
624 | mrt->mroute_reg_vif_num = -1; | |
1da177e4 | 625 | |
a8cb16dd | 626 | if (vifi + 1 == mrt->maxvif) { |
1da177e4 | 627 | int tmp; |
a8cb16dd ED |
628 | |
629 | for (tmp = vifi - 1; tmp >= 0; tmp--) { | |
0c12295a | 630 | if (VIF_EXISTS(mrt, tmp)) |
1da177e4 LT |
631 | break; |
632 | } | |
0c12295a | 633 | mrt->maxvif = tmp+1; |
1da177e4 LT |
634 | } |
635 | ||
636 | write_unlock_bh(&mrt_lock); | |
637 | ||
638 | dev_set_allmulti(dev, -1); | |
639 | ||
a8cb16dd ED |
640 | in_dev = __in_dev_get_rtnl(dev); |
641 | if (in_dev) { | |
42f811b8 | 642 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
d67b8c61 ND |
643 | inet_netconf_notify_devconf(dev_net(dev), |
644 | NETCONFA_MC_FORWARDING, | |
645 | dev->ifindex, &in_dev->cnf); | |
1da177e4 LT |
646 | ip_rt_multicast_event(in_dev); |
647 | } | |
648 | ||
a8cb16dd | 649 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify) |
d17fa6fa | 650 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
651 | |
652 | dev_put(dev); | |
653 | return 0; | |
654 | } | |
655 | ||
a8c9486b | 656 | static void ipmr_cache_free_rcu(struct rcu_head *head) |
5c0a66f5 | 657 | { |
a8c9486b ED |
658 | struct mfc_cache *c = container_of(head, struct mfc_cache, rcu); |
659 | ||
5c0a66f5 BT |
660 | kmem_cache_free(mrt_cachep, c); |
661 | } | |
662 | ||
a8c9486b ED |
663 | static inline void ipmr_cache_free(struct mfc_cache *c) |
664 | { | |
665 | call_rcu(&c->rcu, ipmr_cache_free_rcu); | |
666 | } | |
667 | ||
1da177e4 | 668 | /* Destroy an unresolved cache entry, killing queued skbs |
a8cb16dd | 669 | * and reporting error to netlink readers. |
1da177e4 | 670 | */ |
0c12295a | 671 | static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c) |
1da177e4 | 672 | { |
8de53dfb | 673 | struct net *net = read_pnet(&mrt->net); |
1da177e4 | 674 | struct sk_buff *skb; |
9ef1d4c7 | 675 | struct nlmsgerr *e; |
1da177e4 | 676 | |
0c12295a | 677 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 | 678 | |
c354e124 | 679 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 680 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
681 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
682 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 683 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
1da177e4 | 684 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 685 | e = nlmsg_data(nlh); |
9ef1d4c7 PM |
686 | e->error = -ETIMEDOUT; |
687 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 688 | |
15e47304 | 689 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 690 | } else { |
1da177e4 | 691 | kfree_skb(skb); |
a8cb16dd | 692 | } |
1da177e4 LT |
693 | } |
694 | ||
5c0a66f5 | 695 | ipmr_cache_free(c); |
1da177e4 LT |
696 | } |
697 | ||
e258beb2 | 698 | /* Timer process for the unresolved queue. */ |
e258beb2 | 699 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 700 | { |
0c12295a | 701 | struct mr_table *mrt = (struct mr_table *)arg; |
1da177e4 LT |
702 | unsigned long now; |
703 | unsigned long expires; | |
862465f2 | 704 | struct mfc_cache *c, *next; |
1da177e4 LT |
705 | |
706 | if (!spin_trylock(&mfc_unres_lock)) { | |
0c12295a | 707 | mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
708 | return; |
709 | } | |
710 | ||
0c12295a | 711 | if (list_empty(&mrt->mfc_unres_queue)) |
1da177e4 LT |
712 | goto out; |
713 | ||
714 | now = jiffies; | |
715 | expires = 10*HZ; | |
1da177e4 | 716 | |
0c12295a | 717 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
1da177e4 LT |
718 | if (time_after(c->mfc_un.unres.expires, now)) { |
719 | unsigned long interval = c->mfc_un.unres.expires - now; | |
720 | if (interval < expires) | |
721 | expires = interval; | |
1da177e4 LT |
722 | continue; |
723 | } | |
724 | ||
862465f2 | 725 | list_del(&c->list); |
8cd3ac9f | 726 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
0c12295a | 727 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
728 | } |
729 | ||
0c12295a PM |
730 | if (!list_empty(&mrt->mfc_unres_queue)) |
731 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
732 | |
733 | out: | |
734 | spin_unlock(&mfc_unres_lock); | |
735 | } | |
736 | ||
737 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
0c12295a | 738 | static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache, |
d658f8a0 | 739 | unsigned char *ttls) |
1da177e4 LT |
740 | { |
741 | int vifi; | |
742 | ||
743 | cache->mfc_un.res.minvif = MAXVIFS; | |
744 | cache->mfc_un.res.maxvif = 0; | |
745 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
746 | ||
0c12295a PM |
747 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
748 | if (VIF_EXISTS(mrt, vifi) && | |
cf958ae3 | 749 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
750 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
751 | if (cache->mfc_un.res.minvif > vifi) | |
752 | cache->mfc_un.res.minvif = vifi; | |
753 | if (cache->mfc_un.res.maxvif <= vifi) | |
754 | cache->mfc_un.res.maxvif = vifi + 1; | |
755 | } | |
756 | } | |
757 | } | |
758 | ||
0c12295a PM |
759 | static int vif_add(struct net *net, struct mr_table *mrt, |
760 | struct vifctl *vifc, int mrtsock) | |
1da177e4 LT |
761 | { |
762 | int vifi = vifc->vifc_vifi; | |
0c12295a | 763 | struct vif_device *v = &mrt->vif_table[vifi]; |
1da177e4 LT |
764 | struct net_device *dev; |
765 | struct in_device *in_dev; | |
d607032d | 766 | int err; |
1da177e4 LT |
767 | |
768 | /* Is vif busy ? */ | |
0c12295a | 769 | if (VIF_EXISTS(mrt, vifi)) |
1da177e4 LT |
770 | return -EADDRINUSE; |
771 | ||
772 | switch (vifc->vifc_flags) { | |
1da177e4 | 773 | case VIFF_REGISTER: |
c316c629 NA |
774 | if (!pimsm_enabled()) |
775 | return -EINVAL; | |
776 | /* Special Purpose VIF in PIM | |
1da177e4 LT |
777 | * All the packets will be sent to the daemon |
778 | */ | |
0c12295a | 779 | if (mrt->mroute_reg_vif_num >= 0) |
1da177e4 | 780 | return -EADDRINUSE; |
f0ad0860 | 781 | dev = ipmr_reg_vif(net, mrt); |
1da177e4 LT |
782 | if (!dev) |
783 | return -ENOBUFS; | |
d607032d WC |
784 | err = dev_set_allmulti(dev, 1); |
785 | if (err) { | |
786 | unregister_netdevice(dev); | |
7dc00c82 | 787 | dev_put(dev); |
d607032d WC |
788 | return err; |
789 | } | |
1da177e4 | 790 | break; |
e905a9ed | 791 | case VIFF_TUNNEL: |
4feb88e5 | 792 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
793 | if (!dev) |
794 | return -ENOBUFS; | |
d607032d WC |
795 | err = dev_set_allmulti(dev, 1); |
796 | if (err) { | |
797 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 798 | dev_put(dev); |
d607032d WC |
799 | return err; |
800 | } | |
1da177e4 | 801 | break; |
ee5e81f0 | 802 | case VIFF_USE_IFINDEX: |
1da177e4 | 803 | case 0: |
ee5e81f0 I |
804 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
805 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
51456b29 | 806 | if (dev && !__in_dev_get_rtnl(dev)) { |
ee5e81f0 I |
807 | dev_put(dev); |
808 | return -EADDRNOTAVAIL; | |
809 | } | |
a8cb16dd | 810 | } else { |
ee5e81f0 | 811 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); |
a8cb16dd | 812 | } |
1da177e4 LT |
813 | if (!dev) |
814 | return -EADDRNOTAVAIL; | |
d607032d | 815 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
816 | if (err) { |
817 | dev_put(dev); | |
d607032d | 818 | return err; |
7dc00c82 | 819 | } |
1da177e4 LT |
820 | break; |
821 | default: | |
822 | return -EINVAL; | |
823 | } | |
824 | ||
a8cb16dd ED |
825 | in_dev = __in_dev_get_rtnl(dev); |
826 | if (!in_dev) { | |
d0490cfd | 827 | dev_put(dev); |
1da177e4 | 828 | return -EADDRNOTAVAIL; |
d0490cfd | 829 | } |
42f811b8 | 830 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
d67b8c61 ND |
831 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, dev->ifindex, |
832 | &in_dev->cnf); | |
1da177e4 LT |
833 | ip_rt_multicast_event(in_dev); |
834 | ||
a8cb16dd ED |
835 | /* Fill in the VIF structures */ |
836 | ||
c354e124 JK |
837 | v->rate_limit = vifc->vifc_rate_limit; |
838 | v->local = vifc->vifc_lcl_addr.s_addr; | |
839 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
840 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
841 | if (!mrtsock) |
842 | v->flags |= VIFF_STATIC; | |
c354e124 | 843 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
844 | v->bytes_in = 0; |
845 | v->bytes_out = 0; | |
846 | v->pkt_in = 0; | |
847 | v->pkt_out = 0; | |
848 | v->link = dev->ifindex; | |
a8cb16dd | 849 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER)) |
a54acb3a | 850 | v->link = dev_get_iflink(dev); |
1da177e4 LT |
851 | |
852 | /* And finish update writing critical data */ | |
853 | write_lock_bh(&mrt_lock); | |
c354e124 | 854 | v->dev = dev; |
a8cb16dd | 855 | if (v->flags & VIFF_REGISTER) |
0c12295a | 856 | mrt->mroute_reg_vif_num = vifi; |
0c12295a PM |
857 | if (vifi+1 > mrt->maxvif) |
858 | mrt->maxvif = vifi+1; | |
1da177e4 LT |
859 | write_unlock_bh(&mrt_lock); |
860 | return 0; | |
861 | } | |
862 | ||
a8c9486b | 863 | /* called with rcu_read_lock() */ |
0c12295a | 864 | static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt, |
4feb88e5 BT |
865 | __be32 origin, |
866 | __be32 mcastgrp) | |
1da177e4 | 867 | { |
c354e124 | 868 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
869 | struct mfc_cache *c; |
870 | ||
a8c9486b | 871 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) { |
862465f2 PM |
872 | if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp) |
873 | return c; | |
1da177e4 | 874 | } |
862465f2 | 875 | return NULL; |
1da177e4 LT |
876 | } |
877 | ||
660b26dc ND |
878 | /* Look for a (*,*,oif) entry */ |
879 | static struct mfc_cache *ipmr_cache_find_any_parent(struct mr_table *mrt, | |
880 | int vifi) | |
881 | { | |
360eb5da | 882 | int line = MFC_HASH(htonl(INADDR_ANY), htonl(INADDR_ANY)); |
660b26dc ND |
883 | struct mfc_cache *c; |
884 | ||
885 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) | |
360eb5da ND |
886 | if (c->mfc_origin == htonl(INADDR_ANY) && |
887 | c->mfc_mcastgrp == htonl(INADDR_ANY) && | |
660b26dc ND |
888 | c->mfc_un.res.ttls[vifi] < 255) |
889 | return c; | |
890 | ||
891 | return NULL; | |
892 | } | |
893 | ||
894 | /* Look for a (*,G) entry */ | |
895 | static struct mfc_cache *ipmr_cache_find_any(struct mr_table *mrt, | |
896 | __be32 mcastgrp, int vifi) | |
897 | { | |
360eb5da | 898 | int line = MFC_HASH(mcastgrp, htonl(INADDR_ANY)); |
660b26dc ND |
899 | struct mfc_cache *c, *proxy; |
900 | ||
360eb5da | 901 | if (mcastgrp == htonl(INADDR_ANY)) |
660b26dc ND |
902 | goto skip; |
903 | ||
904 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) | |
360eb5da | 905 | if (c->mfc_origin == htonl(INADDR_ANY) && |
660b26dc ND |
906 | c->mfc_mcastgrp == mcastgrp) { |
907 | if (c->mfc_un.res.ttls[vifi] < 255) | |
908 | return c; | |
909 | ||
910 | /* It's ok if the vifi is part of the static tree */ | |
911 | proxy = ipmr_cache_find_any_parent(mrt, | |
912 | c->mfc_parent); | |
913 | if (proxy && proxy->mfc_un.res.ttls[vifi] < 255) | |
914 | return c; | |
915 | } | |
916 | ||
917 | skip: | |
918 | return ipmr_cache_find_any_parent(mrt, vifi); | |
919 | } | |
920 | ||
7ef8f65d | 921 | /* Allocate a multicast cache entry */ |
d658f8a0 | 922 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 923 | { |
c354e124 | 924 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
a8c9486b ED |
925 | |
926 | if (c) | |
927 | c->mfc_un.res.minvif = MAXVIFS; | |
1da177e4 LT |
928 | return c; |
929 | } | |
930 | ||
d658f8a0 | 931 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 932 | { |
c354e124 | 933 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
a8c9486b ED |
934 | |
935 | if (c) { | |
936 | skb_queue_head_init(&c->mfc_un.unres.unresolved); | |
937 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
938 | } | |
1da177e4 LT |
939 | return c; |
940 | } | |
941 | ||
7ef8f65d | 942 | /* A cache entry has gone into a resolved state from queued */ |
0c12295a PM |
943 | static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt, |
944 | struct mfc_cache *uc, struct mfc_cache *c) | |
1da177e4 LT |
945 | { |
946 | struct sk_buff *skb; | |
9ef1d4c7 | 947 | struct nlmsgerr *e; |
1da177e4 | 948 | |
a8cb16dd | 949 | /* Play the pending entries through our router */ |
c354e124 | 950 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 951 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
952 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
953 | ||
573ce260 | 954 | if (__ipmr_fill_mroute(mrt, skb, c, nlmsg_data(nlh)) > 0) { |
a8cb16dd ED |
955 | nlh->nlmsg_len = skb_tail_pointer(skb) - |
956 | (u8 *)nlh; | |
1da177e4 LT |
957 | } else { |
958 | nlh->nlmsg_type = NLMSG_ERROR; | |
573ce260 | 959 | nlh->nlmsg_len = nlmsg_msg_size(sizeof(struct nlmsgerr)); |
1da177e4 | 960 | skb_trim(skb, nlh->nlmsg_len); |
573ce260 | 961 | e = nlmsg_data(nlh); |
9ef1d4c7 PM |
962 | e->error = -EMSGSIZE; |
963 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 964 | } |
2942e900 | 965 | |
15e47304 | 966 | rtnl_unicast(skb, net, NETLINK_CB(skb).portid); |
a8cb16dd | 967 | } else { |
0c12295a | 968 | ip_mr_forward(net, mrt, skb, c, 0); |
a8cb16dd | 969 | } |
1da177e4 LT |
970 | } |
971 | } | |
972 | ||
c316c629 NA |
973 | /* Bounce a cache query up to mrouted. We could use netlink for this but mrouted |
974 | * expects the following bizarre scheme. | |
1da177e4 | 975 | * |
c316c629 | 976 | * Called under mrt_lock. |
1da177e4 | 977 | */ |
0c12295a | 978 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 979 | struct sk_buff *pkt, vifi_t vifi, int assert) |
1da177e4 | 980 | { |
c9bdd4b5 | 981 | const int ihl = ip_hdrlen(pkt); |
c316c629 | 982 | struct sock *mroute_sk; |
1da177e4 LT |
983 | struct igmphdr *igmp; |
984 | struct igmpmsg *msg; | |
c316c629 | 985 | struct sk_buff *skb; |
1da177e4 LT |
986 | int ret; |
987 | ||
1da177e4 LT |
988 | if (assert == IGMPMSG_WHOLEPKT) |
989 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
990 | else | |
1da177e4 LT |
991 | skb = alloc_skb(128, GFP_ATOMIC); |
992 | ||
132adf54 | 993 | if (!skb) |
1da177e4 LT |
994 | return -ENOBUFS; |
995 | ||
1da177e4 LT |
996 | if (assert == IGMPMSG_WHOLEPKT) { |
997 | /* Ugly, but we have no choice with this interface. | |
a8cb16dd ED |
998 | * Duplicate old header, fix ihl, length etc. |
999 | * And all this only to mangle msg->im_msgtype and | |
1000 | * to set msg->im_mbz to "mbz" :-) | |
1da177e4 | 1001 | */ |
878c8145 ACM |
1002 | skb_push(skb, sizeof(struct iphdr)); |
1003 | skb_reset_network_header(skb); | |
badff6d0 | 1004 | skb_reset_transport_header(skb); |
0272ffc4 | 1005 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 1006 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
1007 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
1008 | msg->im_mbz = 0; | |
0c12295a | 1009 | msg->im_vif = mrt->mroute_reg_vif_num; |
eddc9ec5 ACM |
1010 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
1011 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
1012 | sizeof(struct iphdr)); | |
c316c629 NA |
1013 | } else { |
1014 | /* Copy the IP header */ | |
1015 | skb_set_network_header(skb, skb->len); | |
1016 | skb_put(skb, ihl); | |
1017 | skb_copy_to_linear_data(skb, pkt->data, ihl); | |
1018 | /* Flag to the kernel this is a route add */ | |
1019 | ip_hdr(skb)->protocol = 0; | |
1020 | msg = (struct igmpmsg *)skb_network_header(skb); | |
1021 | msg->im_vif = vifi; | |
1022 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); | |
1023 | /* Add our header */ | |
1024 | igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); | |
1025 | igmp->type = assert; | |
1026 | msg->im_msgtype = assert; | |
1027 | igmp->code = 0; | |
1028 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ | |
1029 | skb->transport_header = skb->network_header; | |
e905a9ed | 1030 | } |
1da177e4 | 1031 | |
4c968709 ED |
1032 | rcu_read_lock(); |
1033 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
51456b29 | 1034 | if (!mroute_sk) { |
4c968709 | 1035 | rcu_read_unlock(); |
1da177e4 LT |
1036 | kfree_skb(skb); |
1037 | return -EINVAL; | |
1038 | } | |
1039 | ||
a8cb16dd | 1040 | /* Deliver to mrouted */ |
4c968709 ED |
1041 | ret = sock_queue_rcv_skb(mroute_sk, skb); |
1042 | rcu_read_unlock(); | |
70a269e6 | 1043 | if (ret < 0) { |
e87cc472 | 1044 | net_warn_ratelimited("mroute: pending queue full, dropping entries\n"); |
1da177e4 LT |
1045 | kfree_skb(skb); |
1046 | } | |
1047 | ||
1048 | return ret; | |
1049 | } | |
1050 | ||
7ef8f65d NA |
1051 | /* Queue a packet for resolution. It gets locked cache entry! */ |
1052 | static int ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, | |
1053 | struct sk_buff *skb) | |
1da177e4 | 1054 | { |
862465f2 | 1055 | bool found = false; |
1da177e4 LT |
1056 | int err; |
1057 | struct mfc_cache *c; | |
eddc9ec5 | 1058 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
1059 | |
1060 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 1061 | list_for_each_entry(c, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1062 | if (c->mfc_mcastgrp == iph->daddr && |
862465f2 PM |
1063 | c->mfc_origin == iph->saddr) { |
1064 | found = true; | |
1da177e4 | 1065 | break; |
862465f2 | 1066 | } |
1da177e4 LT |
1067 | } |
1068 | ||
862465f2 | 1069 | if (!found) { |
a8cb16dd | 1070 | /* Create a new entry if allowable */ |
0c12295a | 1071 | if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 || |
d658f8a0 | 1072 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
1073 | spin_unlock_bh(&mfc_unres_lock); |
1074 | ||
1075 | kfree_skb(skb); | |
1076 | return -ENOBUFS; | |
1077 | } | |
1078 | ||
a8cb16dd | 1079 | /* Fill in the new cache entry */ |
eddc9ec5 ACM |
1080 | c->mfc_parent = -1; |
1081 | c->mfc_origin = iph->saddr; | |
1082 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 | 1083 | |
a8cb16dd | 1084 | /* Reflect first query at mrouted. */ |
0c12295a | 1085 | err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE); |
4feb88e5 | 1086 | if (err < 0) { |
e905a9ed | 1087 | /* If the report failed throw the cache entry |
1da177e4 LT |
1088 | out - Brad Parker |
1089 | */ | |
1090 | spin_unlock_bh(&mfc_unres_lock); | |
1091 | ||
5c0a66f5 | 1092 | ipmr_cache_free(c); |
1da177e4 LT |
1093 | kfree_skb(skb); |
1094 | return err; | |
1095 | } | |
1096 | ||
0c12295a PM |
1097 | atomic_inc(&mrt->cache_resolve_queue_len); |
1098 | list_add(&c->list, &mrt->mfc_unres_queue); | |
8cd3ac9f | 1099 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 | 1100 | |
278554bd DM |
1101 | if (atomic_read(&mrt->cache_resolve_queue_len) == 1) |
1102 | mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires); | |
1da177e4 LT |
1103 | } |
1104 | ||
a8cb16dd | 1105 | /* See if we can append the packet */ |
a8cb16dd | 1106 | if (c->mfc_un.unres.unresolved.qlen > 3) { |
1da177e4 LT |
1107 | kfree_skb(skb); |
1108 | err = -ENOBUFS; | |
1109 | } else { | |
c354e124 | 1110 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
1111 | err = 0; |
1112 | } | |
1113 | ||
1114 | spin_unlock_bh(&mfc_unres_lock); | |
1115 | return err; | |
1116 | } | |
1117 | ||
7ef8f65d | 1118 | /* MFC cache manipulation by user space mroute daemon */ |
1da177e4 | 1119 | |
660b26dc | 1120 | static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc, int parent) |
1da177e4 LT |
1121 | { |
1122 | int line; | |
862465f2 | 1123 | struct mfc_cache *c, *next; |
1da177e4 | 1124 | |
c354e124 | 1125 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1126 | |
0c12295a | 1127 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 1128 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
660b26dc ND |
1129 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr && |
1130 | (parent == -1 || parent == c->mfc_parent)) { | |
a8c9486b | 1131 | list_del_rcu(&c->list); |
8cd3ac9f | 1132 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
5c0a66f5 | 1133 | ipmr_cache_free(c); |
1da177e4 LT |
1134 | return 0; |
1135 | } | |
1136 | } | |
1137 | return -ENOENT; | |
1138 | } | |
1139 | ||
0c12295a | 1140 | static int ipmr_mfc_add(struct net *net, struct mr_table *mrt, |
660b26dc | 1141 | struct mfcctl *mfc, int mrtsock, int parent) |
1da177e4 | 1142 | { |
862465f2 | 1143 | bool found = false; |
1da177e4 | 1144 | int line; |
862465f2 | 1145 | struct mfc_cache *uc, *c; |
1da177e4 | 1146 | |
a50436f2 PM |
1147 | if (mfc->mfcc_parent >= MAXVIFS) |
1148 | return -ENFILE; | |
1149 | ||
c354e124 | 1150 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1151 | |
0c12295a | 1152 | list_for_each_entry(c, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 1153 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
660b26dc ND |
1154 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr && |
1155 | (parent == -1 || parent == c->mfc_parent)) { | |
862465f2 | 1156 | found = true; |
1da177e4 | 1157 | break; |
862465f2 | 1158 | } |
1da177e4 LT |
1159 | } |
1160 | ||
862465f2 | 1161 | if (found) { |
1da177e4 LT |
1162 | write_lock_bh(&mrt_lock); |
1163 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1164 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1165 | if (!mrtsock) |
1166 | c->mfc_flags |= MFC_STATIC; | |
1167 | write_unlock_bh(&mrt_lock); | |
8cd3ac9f | 1168 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 LT |
1169 | return 0; |
1170 | } | |
1171 | ||
360eb5da | 1172 | if (mfc->mfcc_mcastgrp.s_addr != htonl(INADDR_ANY) && |
660b26dc | 1173 | !ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
1174 | return -EINVAL; |
1175 | ||
d658f8a0 | 1176 | c = ipmr_cache_alloc(); |
51456b29 | 1177 | if (!c) |
1da177e4 LT |
1178 | return -ENOMEM; |
1179 | ||
c354e124 JK |
1180 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
1181 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
1182 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1183 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1184 | if (!mrtsock) |
1185 | c->mfc_flags |= MFC_STATIC; | |
1186 | ||
a8c9486b | 1187 | list_add_rcu(&c->list, &mrt->mfc_cache_array[line]); |
1da177e4 | 1188 | |
7ef8f65d NA |
1189 | /* Check to see if we resolved a queued list. If so we |
1190 | * need to send on the frames and tidy up. | |
1da177e4 | 1191 | */ |
b0ebb739 | 1192 | found = false; |
1da177e4 | 1193 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1194 | list_for_each_entry(uc, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1195 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 | 1196 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
862465f2 | 1197 | list_del(&uc->list); |
0c12295a | 1198 | atomic_dec(&mrt->cache_resolve_queue_len); |
b0ebb739 | 1199 | found = true; |
1da177e4 LT |
1200 | break; |
1201 | } | |
1202 | } | |
0c12295a PM |
1203 | if (list_empty(&mrt->mfc_unres_queue)) |
1204 | del_timer(&mrt->ipmr_expire_timer); | |
1da177e4 LT |
1205 | spin_unlock_bh(&mfc_unres_lock); |
1206 | ||
b0ebb739 | 1207 | if (found) { |
0c12295a | 1208 | ipmr_cache_resolve(net, mrt, uc, c); |
5c0a66f5 | 1209 | ipmr_cache_free(uc); |
1da177e4 | 1210 | } |
8cd3ac9f | 1211 | mroute_netlink_event(mrt, c, RTM_NEWROUTE); |
1da177e4 LT |
1212 | return 0; |
1213 | } | |
1214 | ||
7ef8f65d | 1215 | /* Close the multicast socket, and clear the vif tables etc */ |
0c12295a | 1216 | static void mroute_clean_tables(struct mr_table *mrt) |
1da177e4 LT |
1217 | { |
1218 | int i; | |
d17fa6fa | 1219 | LIST_HEAD(list); |
862465f2 | 1220 | struct mfc_cache *c, *next; |
e905a9ed | 1221 | |
a8cb16dd | 1222 | /* Shut down all active vif entries */ |
0c12295a | 1223 | for (i = 0; i < mrt->maxvif; i++) { |
a8cb16dd | 1224 | if (!(mrt->vif_table[i].flags & VIFF_STATIC)) |
0c12295a | 1225 | vif_delete(mrt, i, 0, &list); |
1da177e4 | 1226 | } |
d17fa6fa | 1227 | unregister_netdevice_many(&list); |
1da177e4 | 1228 | |
a8cb16dd | 1229 | /* Wipe the cache */ |
862465f2 | 1230 | for (i = 0; i < MFC_LINES; i++) { |
0c12295a | 1231 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) { |
a8c9486b | 1232 | if (c->mfc_flags & MFC_STATIC) |
1da177e4 | 1233 | continue; |
a8c9486b | 1234 | list_del_rcu(&c->list); |
8cd3ac9f | 1235 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
5c0a66f5 | 1236 | ipmr_cache_free(c); |
1da177e4 LT |
1237 | } |
1238 | } | |
1239 | ||
0c12295a | 1240 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { |
1da177e4 | 1241 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1242 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
862465f2 | 1243 | list_del(&c->list); |
8cd3ac9f | 1244 | mroute_netlink_event(mrt, c, RTM_DELROUTE); |
0c12295a | 1245 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
1246 | } |
1247 | spin_unlock_bh(&mfc_unres_lock); | |
1248 | } | |
1249 | } | |
1250 | ||
4c968709 ED |
1251 | /* called from ip_ra_control(), before an RCU grace period, |
1252 | * we dont need to call synchronize_rcu() here | |
1253 | */ | |
1da177e4 LT |
1254 | static void mrtsock_destruct(struct sock *sk) |
1255 | { | |
4feb88e5 | 1256 | struct net *net = sock_net(sk); |
f0ad0860 | 1257 | struct mr_table *mrt; |
4feb88e5 | 1258 | |
1da177e4 | 1259 | rtnl_lock(); |
f0ad0860 | 1260 | ipmr_for_each_table(mrt, net) { |
4c968709 | 1261 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
f0ad0860 | 1262 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; |
d67b8c61 ND |
1263 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1264 | NETCONFA_IFINDEX_ALL, | |
1265 | net->ipv4.devconf_all); | |
a9b3cd7f | 1266 | RCU_INIT_POINTER(mrt->mroute_sk, NULL); |
f0ad0860 PM |
1267 | mroute_clean_tables(mrt); |
1268 | } | |
1da177e4 LT |
1269 | } |
1270 | rtnl_unlock(); | |
1271 | } | |
1272 | ||
7ef8f65d NA |
1273 | /* Socket options and virtual interface manipulation. The whole |
1274 | * virtual interface system is a complete heap, but unfortunately | |
1275 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
1276 | * MOSPF/PIM router set up we can clean this up. | |
1da177e4 | 1277 | */ |
e905a9ed | 1278 | |
b7058842 | 1279 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen) |
1da177e4 | 1280 | { |
660b26dc | 1281 | int ret, parent = 0; |
1da177e4 LT |
1282 | struct vifctl vif; |
1283 | struct mfcctl mfc; | |
4feb88e5 | 1284 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1285 | struct mr_table *mrt; |
1286 | ||
5e1859fb ED |
1287 | if (sk->sk_type != SOCK_RAW || |
1288 | inet_sk(sk)->inet_num != IPPROTO_IGMP) | |
1289 | return -EOPNOTSUPP; | |
1290 | ||
f0ad0860 | 1291 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
51456b29 | 1292 | if (!mrt) |
f0ad0860 | 1293 | return -ENOENT; |
e905a9ed | 1294 | |
132adf54 | 1295 | if (optname != MRT_INIT) { |
33d480ce | 1296 | if (sk != rcu_access_pointer(mrt->mroute_sk) && |
52e804c6 | 1297 | !ns_capable(net->user_ns, CAP_NET_ADMIN)) |
1da177e4 LT |
1298 | return -EACCES; |
1299 | } | |
1300 | ||
132adf54 SH |
1301 | switch (optname) { |
1302 | case MRT_INIT: | |
c354e124 | 1303 | if (optlen != sizeof(int)) |
5e1859fb | 1304 | return -EINVAL; |
1da177e4 | 1305 | |
132adf54 | 1306 | rtnl_lock(); |
4c968709 | 1307 | if (rtnl_dereference(mrt->mroute_sk)) { |
1da177e4 | 1308 | rtnl_unlock(); |
132adf54 SH |
1309 | return -EADDRINUSE; |
1310 | } | |
1311 | ||
1312 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
1313 | if (ret == 0) { | |
cf778b00 | 1314 | rcu_assign_pointer(mrt->mroute_sk, sk); |
4feb88e5 | 1315 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
d67b8c61 ND |
1316 | inet_netconf_notify_devconf(net, NETCONFA_MC_FORWARDING, |
1317 | NETCONFA_IFINDEX_ALL, | |
1318 | net->ipv4.devconf_all); | |
132adf54 SH |
1319 | } |
1320 | rtnl_unlock(); | |
1321 | return ret; | |
1322 | case MRT_DONE: | |
33d480ce | 1323 | if (sk != rcu_access_pointer(mrt->mroute_sk)) |
132adf54 SH |
1324 | return -EACCES; |
1325 | return ip_ra_control(sk, 0, NULL); | |
1326 | case MRT_ADD_VIF: | |
1327 | case MRT_DEL_VIF: | |
c354e124 | 1328 | if (optlen != sizeof(vif)) |
132adf54 | 1329 | return -EINVAL; |
c354e124 | 1330 | if (copy_from_user(&vif, optval, sizeof(vif))) |
132adf54 SH |
1331 | return -EFAULT; |
1332 | if (vif.vifc_vifi >= MAXVIFS) | |
1333 | return -ENFILE; | |
1334 | rtnl_lock(); | |
c354e124 | 1335 | if (optname == MRT_ADD_VIF) { |
4c968709 ED |
1336 | ret = vif_add(net, mrt, &vif, |
1337 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 | 1338 | } else { |
0c12295a | 1339 | ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL); |
132adf54 SH |
1340 | } |
1341 | rtnl_unlock(); | |
1342 | return ret; | |
1da177e4 | 1343 | |
7ef8f65d NA |
1344 | /* Manipulate the forwarding caches. These live |
1345 | * in a sort of kernel/user symbiosis. | |
1346 | */ | |
132adf54 SH |
1347 | case MRT_ADD_MFC: |
1348 | case MRT_DEL_MFC: | |
660b26dc ND |
1349 | parent = -1; |
1350 | case MRT_ADD_MFC_PROXY: | |
1351 | case MRT_DEL_MFC_PROXY: | |
c354e124 | 1352 | if (optlen != sizeof(mfc)) |
132adf54 | 1353 | return -EINVAL; |
c354e124 | 1354 | if (copy_from_user(&mfc, optval, sizeof(mfc))) |
132adf54 | 1355 | return -EFAULT; |
660b26dc ND |
1356 | if (parent == 0) |
1357 | parent = mfc.mfcc_parent; | |
132adf54 | 1358 | rtnl_lock(); |
660b26dc ND |
1359 | if (optname == MRT_DEL_MFC || optname == MRT_DEL_MFC_PROXY) |
1360 | ret = ipmr_mfc_delete(mrt, &mfc, parent); | |
132adf54 | 1361 | else |
4c968709 | 1362 | ret = ipmr_mfc_add(net, mrt, &mfc, |
660b26dc ND |
1363 | sk == rtnl_dereference(mrt->mroute_sk), |
1364 | parent); | |
132adf54 SH |
1365 | rtnl_unlock(); |
1366 | return ret; | |
7ef8f65d | 1367 | /* Control PIM assert. */ |
132adf54 SH |
1368 | case MRT_ASSERT: |
1369 | { | |
1370 | int v; | |
5e1859fb ED |
1371 | if (optlen != sizeof(v)) |
1372 | return -EINVAL; | |
a8cb16dd | 1373 | if (get_user(v, (int __user *)optval)) |
132adf54 | 1374 | return -EFAULT; |
53d6841d | 1375 | mrt->mroute_do_assert = v; |
132adf54 SH |
1376 | return 0; |
1377 | } | |
132adf54 SH |
1378 | case MRT_PIM: |
1379 | { | |
ba93ef74 SH |
1380 | int v; |
1381 | ||
c316c629 NA |
1382 | if (!pimsm_enabled()) |
1383 | return -ENOPROTOOPT; | |
5e1859fb ED |
1384 | if (optlen != sizeof(v)) |
1385 | return -EINVAL; | |
a8cb16dd | 1386 | if (get_user(v, (int __user *)optval)) |
132adf54 | 1387 | return -EFAULT; |
5e1859fb | 1388 | v = !!v; |
ba93ef74 | 1389 | |
132adf54 SH |
1390 | rtnl_lock(); |
1391 | ret = 0; | |
0c12295a PM |
1392 | if (v != mrt->mroute_do_pim) { |
1393 | mrt->mroute_do_pim = v; | |
1394 | mrt->mroute_do_assert = v; | |
1da177e4 | 1395 | } |
132adf54 SH |
1396 | rtnl_unlock(); |
1397 | return ret; | |
1398 | } | |
f0ad0860 PM |
1399 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES |
1400 | case MRT_TABLE: | |
1401 | { | |
1402 | u32 v; | |
1403 | ||
1404 | if (optlen != sizeof(u32)) | |
1405 | return -EINVAL; | |
1406 | if (get_user(v, (u32 __user *)optval)) | |
1407 | return -EFAULT; | |
f0ad0860 PM |
1408 | |
1409 | rtnl_lock(); | |
1410 | ret = 0; | |
4c968709 ED |
1411 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
1412 | ret = -EBUSY; | |
1413 | } else { | |
1113ebbc NA |
1414 | mrt = ipmr_new_table(net, v); |
1415 | if (IS_ERR(mrt)) | |
1416 | ret = PTR_ERR(mrt); | |
5e1859fb ED |
1417 | else |
1418 | raw_sk(sk)->ipmr_table = v; | |
4c968709 | 1419 | } |
f0ad0860 PM |
1420 | rtnl_unlock(); |
1421 | return ret; | |
1422 | } | |
1da177e4 | 1423 | #endif |
7ef8f65d | 1424 | /* Spurious command, or MRT_VERSION which you cannot set. */ |
132adf54 SH |
1425 | default: |
1426 | return -ENOPROTOOPT; | |
1da177e4 LT |
1427 | } |
1428 | } | |
1429 | ||
7ef8f65d | 1430 | /* Getsock opt support for the multicast routing system. */ |
c354e124 | 1431 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1432 | { |
1433 | int olr; | |
1434 | int val; | |
4feb88e5 | 1435 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1436 | struct mr_table *mrt; |
1437 | ||
5e1859fb ED |
1438 | if (sk->sk_type != SOCK_RAW || |
1439 | inet_sk(sk)->inet_num != IPPROTO_IGMP) | |
1440 | return -EOPNOTSUPP; | |
1441 | ||
f0ad0860 | 1442 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); |
51456b29 | 1443 | if (!mrt) |
f0ad0860 | 1444 | return -ENOENT; |
1da177e4 | 1445 | |
c354e124 | 1446 | if (optname != MRT_VERSION && |
a8cb16dd | 1447 | optname != MRT_PIM && |
a8cb16dd | 1448 | optname != MRT_ASSERT) |
1da177e4 LT |
1449 | return -ENOPROTOOPT; |
1450 | ||
1451 | if (get_user(olr, optlen)) | |
1452 | return -EFAULT; | |
1453 | ||
1454 | olr = min_t(unsigned int, olr, sizeof(int)); | |
1455 | if (olr < 0) | |
1456 | return -EINVAL; | |
e905a9ed | 1457 | |
c354e124 | 1458 | if (put_user(olr, optlen)) |
1da177e4 | 1459 | return -EFAULT; |
c316c629 | 1460 | if (optname == MRT_VERSION) { |
c354e124 | 1461 | val = 0x0305; |
c316c629 NA |
1462 | } else if (optname == MRT_PIM) { |
1463 | if (!pimsm_enabled()) | |
1464 | return -ENOPROTOOPT; | |
0c12295a | 1465 | val = mrt->mroute_do_pim; |
c316c629 | 1466 | } else { |
0c12295a | 1467 | val = mrt->mroute_do_assert; |
c316c629 | 1468 | } |
c354e124 | 1469 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1470 | return -EFAULT; |
1471 | return 0; | |
1472 | } | |
1473 | ||
7ef8f65d | 1474 | /* The IP multicast ioctl support routines. */ |
1da177e4 LT |
1475 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1476 | { | |
1477 | struct sioc_sg_req sr; | |
1478 | struct sioc_vif_req vr; | |
1479 | struct vif_device *vif; | |
1480 | struct mfc_cache *c; | |
4feb88e5 | 1481 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1482 | struct mr_table *mrt; |
1483 | ||
1484 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
51456b29 | 1485 | if (!mrt) |
f0ad0860 | 1486 | return -ENOENT; |
e905a9ed | 1487 | |
132adf54 SH |
1488 | switch (cmd) { |
1489 | case SIOCGETVIFCNT: | |
c354e124 | 1490 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1491 | return -EFAULT; |
0c12295a | 1492 | if (vr.vifi >= mrt->maxvif) |
132adf54 SH |
1493 | return -EINVAL; |
1494 | read_lock(&mrt_lock); | |
0c12295a PM |
1495 | vif = &mrt->vif_table[vr.vifi]; |
1496 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
c354e124 JK |
1497 | vr.icount = vif->pkt_in; |
1498 | vr.ocount = vif->pkt_out; | |
1499 | vr.ibytes = vif->bytes_in; | |
1500 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1501 | read_unlock(&mrt_lock); |
1da177e4 | 1502 | |
c354e124 | 1503 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1504 | return -EFAULT; |
1505 | return 0; | |
1506 | } | |
1507 | read_unlock(&mrt_lock); | |
1508 | return -EADDRNOTAVAIL; | |
1509 | case SIOCGETSGCNT: | |
c354e124 | 1510 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1511 | return -EFAULT; |
1512 | ||
a8c9486b | 1513 | rcu_read_lock(); |
0c12295a | 1514 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
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; | |
a8c9486b | 1519 | rcu_read_unlock(); |
132adf54 | 1520 | |
c354e124 | 1521 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1522 | return -EFAULT; |
1523 | return 0; | |
1524 | } | |
a8c9486b | 1525 | rcu_read_unlock(); |
132adf54 SH |
1526 | return -EADDRNOTAVAIL; |
1527 | default: | |
1528 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1529 | } |
1530 | } | |
1531 | ||
709b46e8 EB |
1532 | #ifdef CONFIG_COMPAT |
1533 | struct compat_sioc_sg_req { | |
1534 | struct in_addr src; | |
1535 | struct in_addr grp; | |
1536 | compat_ulong_t pktcnt; | |
1537 | compat_ulong_t bytecnt; | |
1538 | compat_ulong_t wrong_if; | |
1539 | }; | |
1540 | ||
ca6b8bb0 DM |
1541 | struct compat_sioc_vif_req { |
1542 | vifi_t vifi; /* Which iface */ | |
1543 | compat_ulong_t icount; | |
1544 | compat_ulong_t ocount; | |
1545 | compat_ulong_t ibytes; | |
1546 | compat_ulong_t obytes; | |
1547 | }; | |
1548 | ||
709b46e8 EB |
1549 | int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg) |
1550 | { | |
0033d5ad | 1551 | struct compat_sioc_sg_req sr; |
ca6b8bb0 DM |
1552 | struct compat_sioc_vif_req vr; |
1553 | struct vif_device *vif; | |
709b46e8 EB |
1554 | struct mfc_cache *c; |
1555 | struct net *net = sock_net(sk); | |
1556 | struct mr_table *mrt; | |
1557 | ||
1558 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
51456b29 | 1559 | if (!mrt) |
709b46e8 EB |
1560 | return -ENOENT; |
1561 | ||
1562 | switch (cmd) { | |
ca6b8bb0 DM |
1563 | case SIOCGETVIFCNT: |
1564 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1565 | return -EFAULT; | |
1566 | if (vr.vifi >= mrt->maxvif) | |
1567 | return -EINVAL; | |
1568 | read_lock(&mrt_lock); | |
1569 | vif = &mrt->vif_table[vr.vifi]; | |
1570 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
1571 | vr.icount = vif->pkt_in; | |
1572 | vr.ocount = vif->pkt_out; | |
1573 | vr.ibytes = vif->bytes_in; | |
1574 | vr.obytes = vif->bytes_out; | |
1575 | read_unlock(&mrt_lock); | |
1576 | ||
1577 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1578 | return -EFAULT; | |
1579 | return 0; | |
1580 | } | |
1581 | read_unlock(&mrt_lock); | |
1582 | return -EADDRNOTAVAIL; | |
709b46e8 EB |
1583 | case SIOCGETSGCNT: |
1584 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1585 | return -EFAULT; | |
1586 | ||
1587 | rcu_read_lock(); | |
1588 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); | |
1589 | if (c) { | |
1590 | sr.pktcnt = c->mfc_un.res.pkt; | |
1591 | sr.bytecnt = c->mfc_un.res.bytes; | |
1592 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1593 | rcu_read_unlock(); | |
1594 | ||
1595 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
1596 | return -EFAULT; | |
1597 | return 0; | |
1598 | } | |
1599 | rcu_read_unlock(); | |
1600 | return -EADDRNOTAVAIL; | |
1601 | default: | |
1602 | return -ENOIOCTLCMD; | |
1603 | } | |
1604 | } | |
1605 | #endif | |
1606 | ||
1da177e4 LT |
1607 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) |
1608 | { | |
351638e7 | 1609 | struct net_device *dev = netdev_notifier_info_to_dev(ptr); |
4feb88e5 | 1610 | struct net *net = dev_net(dev); |
f0ad0860 | 1611 | struct mr_table *mrt; |
1da177e4 LT |
1612 | struct vif_device *v; |
1613 | int ct; | |
e9dc8653 | 1614 | |
1da177e4 LT |
1615 | if (event != NETDEV_UNREGISTER) |
1616 | return NOTIFY_DONE; | |
f0ad0860 PM |
1617 | |
1618 | ipmr_for_each_table(mrt, net) { | |
1619 | v = &mrt->vif_table[0]; | |
1620 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { | |
1621 | if (v->dev == dev) | |
e92036a6 | 1622 | vif_delete(mrt, ct, 1, NULL); |
f0ad0860 | 1623 | } |
1da177e4 LT |
1624 | } |
1625 | return NOTIFY_DONE; | |
1626 | } | |
1627 | ||
c354e124 | 1628 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1629 | .notifier_call = ipmr_device_event, |
1630 | }; | |
1631 | ||
7ef8f65d NA |
1632 | /* Encapsulate a packet by attaching a valid IPIP header to it. |
1633 | * This avoids tunnel drivers and other mess and gives us the speed so | |
1634 | * important for multicast video. | |
1da177e4 | 1635 | */ |
b6a7719a HFS |
1636 | static void ip_encap(struct net *net, struct sk_buff *skb, |
1637 | __be32 saddr, __be32 daddr) | |
1da177e4 | 1638 | { |
8856dfa3 | 1639 | struct iphdr *iph; |
b71d1d42 | 1640 | const struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1641 | |
1642 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1643 | skb->transport_header = skb->network_header; |
8856dfa3 | 1644 | skb_reset_network_header(skb); |
eddc9ec5 | 1645 | iph = ip_hdr(skb); |
1da177e4 | 1646 | |
a8cb16dd | 1647 | iph->version = 4; |
e023dd64 ACM |
1648 | iph->tos = old_iph->tos; |
1649 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1650 | iph->frag_off = 0; |
1651 | iph->daddr = daddr; | |
1652 | iph->saddr = saddr; | |
1653 | iph->protocol = IPPROTO_IPIP; | |
1654 | iph->ihl = 5; | |
1655 | iph->tot_len = htons(skb->len); | |
b6a7719a | 1656 | ip_select_ident(net, skb, NULL); |
1da177e4 LT |
1657 | ip_send_check(iph); |
1658 | ||
1da177e4 LT |
1659 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1660 | nf_reset(skb); | |
1661 | } | |
1662 | ||
0c4b51f0 EB |
1663 | static inline int ipmr_forward_finish(struct net *net, struct sock *sk, |
1664 | struct sk_buff *skb) | |
1da177e4 | 1665 | { |
a8cb16dd | 1666 | struct ip_options *opt = &(IPCB(skb)->opt); |
1da177e4 | 1667 | |
73186df8 DM |
1668 | IP_INC_STATS(net, IPSTATS_MIB_OUTFORWDATAGRAMS); |
1669 | IP_ADD_STATS(net, IPSTATS_MIB_OUTOCTETS, skb->len); | |
1da177e4 LT |
1670 | |
1671 | if (unlikely(opt->optlen)) | |
1672 | ip_forward_options(skb); | |
1673 | ||
13206b6b | 1674 | return dst_output(net, sk, skb); |
1da177e4 LT |
1675 | } |
1676 | ||
7ef8f65d | 1677 | /* Processing handlers for ipmr_forward */ |
1da177e4 | 1678 | |
0c12295a PM |
1679 | static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt, |
1680 | struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1da177e4 | 1681 | { |
eddc9ec5 | 1682 | const struct iphdr *iph = ip_hdr(skb); |
0c12295a | 1683 | struct vif_device *vif = &mrt->vif_table[vifi]; |
1da177e4 LT |
1684 | struct net_device *dev; |
1685 | struct rtable *rt; | |
31e4543d | 1686 | struct flowi4 fl4; |
1da177e4 LT |
1687 | int encap = 0; |
1688 | ||
51456b29 | 1689 | if (!vif->dev) |
1da177e4 LT |
1690 | goto out_free; |
1691 | ||
1da177e4 LT |
1692 | if (vif->flags & VIFF_REGISTER) { |
1693 | vif->pkt_out++; | |
c354e124 | 1694 | vif->bytes_out += skb->len; |
cf3677ae PE |
1695 | vif->dev->stats.tx_bytes += skb->len; |
1696 | vif->dev->stats.tx_packets++; | |
0c12295a | 1697 | ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1698 | goto out_free; |
1da177e4 | 1699 | } |
1da177e4 | 1700 | |
a8cb16dd | 1701 | if (vif->flags & VIFF_TUNNEL) { |
31e4543d | 1702 | rt = ip_route_output_ports(net, &fl4, NULL, |
78fbfd8a DM |
1703 | vif->remote, vif->local, |
1704 | 0, 0, | |
1705 | IPPROTO_IPIP, | |
1706 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1707 | if (IS_ERR(rt)) |
1da177e4 LT |
1708 | goto out_free; |
1709 | encap = sizeof(struct iphdr); | |
1710 | } else { | |
31e4543d | 1711 | rt = ip_route_output_ports(net, &fl4, NULL, iph->daddr, 0, |
78fbfd8a DM |
1712 | 0, 0, |
1713 | IPPROTO_IPIP, | |
1714 | RT_TOS(iph->tos), vif->link); | |
b23dd4fe | 1715 | if (IS_ERR(rt)) |
1da177e4 LT |
1716 | goto out_free; |
1717 | } | |
1718 | ||
d8d1f30b | 1719 | dev = rt->dst.dev; |
1da177e4 | 1720 | |
d8d1f30b | 1721 | if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) { |
1da177e4 | 1722 | /* Do not fragment multicasts. Alas, IPv4 does not |
a8cb16dd ED |
1723 | * allow to send ICMP, so that packets will disappear |
1724 | * to blackhole. | |
1da177e4 | 1725 | */ |
73186df8 | 1726 | IP_INC_STATS(net, IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1727 | ip_rt_put(rt); |
1728 | goto out_free; | |
1729 | } | |
1730 | ||
d8d1f30b | 1731 | encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len; |
1da177e4 LT |
1732 | |
1733 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1734 | ip_rt_put(rt); |
1da177e4 LT |
1735 | goto out_free; |
1736 | } | |
1737 | ||
1738 | vif->pkt_out++; | |
c354e124 | 1739 | vif->bytes_out += skb->len; |
1da177e4 | 1740 | |
adf30907 | 1741 | skb_dst_drop(skb); |
d8d1f30b | 1742 | skb_dst_set(skb, &rt->dst); |
eddc9ec5 | 1743 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1744 | |
1745 | /* FIXME: forward and output firewalls used to be called here. | |
a8cb16dd ED |
1746 | * What do we do with netfilter? -- RR |
1747 | */ | |
1da177e4 | 1748 | if (vif->flags & VIFF_TUNNEL) { |
b6a7719a | 1749 | ip_encap(net, skb, vif->local, vif->remote); |
1da177e4 | 1750 | /* FIXME: extra output firewall step used to be here. --RR */ |
2f4c02d4 PE |
1751 | vif->dev->stats.tx_packets++; |
1752 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1753 | } |
1754 | ||
1755 | IPCB(skb)->flags |= IPSKB_FORWARDED; | |
1756 | ||
7ef8f65d | 1757 | /* RFC1584 teaches, that DVMRP/PIM router must deliver packets locally |
1da177e4 LT |
1758 | * not only before forwarding, but after forwarding on all output |
1759 | * interfaces. It is clear, if mrouter runs a multicasting | |
1760 | * program, it should receive packets not depending to what interface | |
1761 | * program is joined. | |
1762 | * If we will not make it, the program will have to join on all | |
1763 | * interfaces. On the other hand, multihoming host (or router, but | |
1764 | * not mrouter) cannot join to more than one interface - it will | |
1765 | * result in receiving multiple packets. | |
1766 | */ | |
29a26a56 EB |
1767 | NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, |
1768 | net, NULL, skb, skb->dev, dev, | |
1da177e4 LT |
1769 | ipmr_forward_finish); |
1770 | return; | |
1771 | ||
1772 | out_free: | |
1773 | kfree_skb(skb); | |
1da177e4 LT |
1774 | } |
1775 | ||
0c12295a | 1776 | static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev) |
1da177e4 LT |
1777 | { |
1778 | int ct; | |
0c12295a PM |
1779 | |
1780 | for (ct = mrt->maxvif-1; ct >= 0; ct--) { | |
1781 | if (mrt->vif_table[ct].dev == dev) | |
1da177e4 LT |
1782 | break; |
1783 | } | |
1784 | return ct; | |
1785 | } | |
1786 | ||
1787 | /* "local" means that we should preserve one skb (for local delivery) */ | |
c4854ec8 RR |
1788 | static void ip_mr_forward(struct net *net, struct mr_table *mrt, |
1789 | struct sk_buff *skb, struct mfc_cache *cache, | |
1790 | int local) | |
1da177e4 LT |
1791 | { |
1792 | int psend = -1; | |
1793 | int vif, ct; | |
660b26dc | 1794 | int true_vifi = ipmr_find_vif(mrt, skb->dev); |
1da177e4 LT |
1795 | |
1796 | vif = cache->mfc_parent; | |
1797 | cache->mfc_un.res.pkt++; | |
1798 | cache->mfc_un.res.bytes += skb->len; | |
1799 | ||
360eb5da | 1800 | if (cache->mfc_origin == htonl(INADDR_ANY) && true_vifi >= 0) { |
660b26dc ND |
1801 | struct mfc_cache *cache_proxy; |
1802 | ||
1803 | /* For an (*,G) entry, we only check that the incomming | |
1804 | * interface is part of the static tree. | |
1805 | */ | |
1806 | cache_proxy = ipmr_cache_find_any_parent(mrt, vif); | |
1807 | if (cache_proxy && | |
1808 | cache_proxy->mfc_un.res.ttls[true_vifi] < 255) | |
1809 | goto forward; | |
1810 | } | |
1811 | ||
7ef8f65d | 1812 | /* Wrong interface: drop packet and (maybe) send PIM assert. */ |
0c12295a | 1813 | if (mrt->vif_table[vif].dev != skb->dev) { |
c7537967 | 1814 | if (rt_is_output_route(skb_rtable(skb))) { |
1da177e4 | 1815 | /* It is our own packet, looped back. |
a8cb16dd ED |
1816 | * Very complicated situation... |
1817 | * | |
1818 | * The best workaround until routing daemons will be | |
1819 | * fixed is not to redistribute packet, if it was | |
1820 | * send through wrong interface. It means, that | |
1821 | * multicast applications WILL NOT work for | |
1822 | * (S,G), which have default multicast route pointing | |
1823 | * to wrong oif. In any case, it is not a good | |
1824 | * idea to use multicasting applications on router. | |
1da177e4 LT |
1825 | */ |
1826 | goto dont_forward; | |
1827 | } | |
1828 | ||
1829 | cache->mfc_un.res.wrong_if++; | |
1da177e4 | 1830 | |
0c12295a | 1831 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
1da177e4 | 1832 | /* pimsm uses asserts, when switching from RPT to SPT, |
a8cb16dd ED |
1833 | * so that we cannot check that packet arrived on an oif. |
1834 | * It is bad, but otherwise we would need to move pretty | |
1835 | * large chunk of pimd to kernel. Ough... --ANK | |
1da177e4 | 1836 | */ |
0c12295a | 1837 | (mrt->mroute_do_pim || |
6f9374a9 | 1838 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1839 | time_after(jiffies, |
1da177e4 LT |
1840 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1841 | cache->mfc_un.res.last_assert = jiffies; | |
0c12295a | 1842 | ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1843 | } |
1844 | goto dont_forward; | |
1845 | } | |
1846 | ||
660b26dc | 1847 | forward: |
0c12295a PM |
1848 | mrt->vif_table[vif].pkt_in++; |
1849 | mrt->vif_table[vif].bytes_in += skb->len; | |
1da177e4 | 1850 | |
7ef8f65d | 1851 | /* Forward the frame */ |
360eb5da ND |
1852 | if (cache->mfc_origin == htonl(INADDR_ANY) && |
1853 | cache->mfc_mcastgrp == htonl(INADDR_ANY)) { | |
660b26dc ND |
1854 | if (true_vifi >= 0 && |
1855 | true_vifi != cache->mfc_parent && | |
1856 | ip_hdr(skb)->ttl > | |
1857 | cache->mfc_un.res.ttls[cache->mfc_parent]) { | |
1858 | /* It's an (*,*) entry and the packet is not coming from | |
1859 | * the upstream: forward the packet to the upstream | |
1860 | * only. | |
1861 | */ | |
1862 | psend = cache->mfc_parent; | |
1863 | goto last_forward; | |
1864 | } | |
1865 | goto dont_forward; | |
1866 | } | |
a8cb16dd ED |
1867 | for (ct = cache->mfc_un.res.maxvif - 1; |
1868 | ct >= cache->mfc_un.res.minvif; ct--) { | |
660b26dc | 1869 | /* For (*,G) entry, don't forward to the incoming interface */ |
360eb5da ND |
1870 | if ((cache->mfc_origin != htonl(INADDR_ANY) || |
1871 | ct != true_vifi) && | |
660b26dc | 1872 | ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1873 | if (psend != -1) { |
1874 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1875 | |
1da177e4 | 1876 | if (skb2) |
0c12295a PM |
1877 | ipmr_queue_xmit(net, mrt, skb2, cache, |
1878 | psend); | |
1da177e4 | 1879 | } |
c354e124 | 1880 | psend = ct; |
1da177e4 LT |
1881 | } |
1882 | } | |
660b26dc | 1883 | last_forward: |
1da177e4 LT |
1884 | if (psend != -1) { |
1885 | if (local) { | |
1886 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1887 | |
1da177e4 | 1888 | if (skb2) |
0c12295a | 1889 | ipmr_queue_xmit(net, mrt, skb2, cache, psend); |
1da177e4 | 1890 | } else { |
0c12295a | 1891 | ipmr_queue_xmit(net, mrt, skb, cache, psend); |
c4854ec8 | 1892 | return; |
1da177e4 LT |
1893 | } |
1894 | } | |
1895 | ||
1896 | dont_forward: | |
1897 | if (!local) | |
1898 | kfree_skb(skb); | |
1da177e4 LT |
1899 | } |
1900 | ||
417da66f | 1901 | static struct mr_table *ipmr_rt_fib_lookup(struct net *net, struct sk_buff *skb) |
ee3f1aaf | 1902 | { |
417da66f DM |
1903 | struct rtable *rt = skb_rtable(skb); |
1904 | struct iphdr *iph = ip_hdr(skb); | |
da91981b | 1905 | struct flowi4 fl4 = { |
417da66f DM |
1906 | .daddr = iph->daddr, |
1907 | .saddr = iph->saddr, | |
b0fe4a31 | 1908 | .flowi4_tos = RT_TOS(iph->tos), |
4fd551d7 DM |
1909 | .flowi4_oif = (rt_is_output_route(rt) ? |
1910 | skb->dev->ifindex : 0), | |
1911 | .flowi4_iif = (rt_is_output_route(rt) ? | |
1fb9489b | 1912 | LOOPBACK_IFINDEX : |
4fd551d7 | 1913 | skb->dev->ifindex), |
b4869889 | 1914 | .flowi4_mark = skb->mark, |
ee3f1aaf DM |
1915 | }; |
1916 | struct mr_table *mrt; | |
1917 | int err; | |
1918 | ||
da91981b | 1919 | err = ipmr_fib_lookup(net, &fl4, &mrt); |
ee3f1aaf DM |
1920 | if (err) |
1921 | return ERR_PTR(err); | |
1922 | return mrt; | |
1923 | } | |
1da177e4 | 1924 | |
7ef8f65d NA |
1925 | /* Multicast packets for forwarding arrive here |
1926 | * Called with rcu_read_lock(); | |
1da177e4 | 1927 | */ |
1da177e4 LT |
1928 | int ip_mr_input(struct sk_buff *skb) |
1929 | { | |
1930 | struct mfc_cache *cache; | |
4feb88e5 | 1931 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1932 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
f0ad0860 | 1933 | struct mr_table *mrt; |
1da177e4 LT |
1934 | |
1935 | /* Packet is looped back after forward, it should not be | |
a8cb16dd | 1936 | * forwarded second time, but still can be delivered locally. |
1da177e4 | 1937 | */ |
4c968709 | 1938 | if (IPCB(skb)->flags & IPSKB_FORWARDED) |
1da177e4 LT |
1939 | goto dont_forward; |
1940 | ||
417da66f | 1941 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
1942 | if (IS_ERR(mrt)) { |
1943 | kfree_skb(skb); | |
1944 | return PTR_ERR(mrt); | |
e40dbc51 | 1945 | } |
1da177e4 | 1946 | if (!local) { |
4c968709 ED |
1947 | if (IPCB(skb)->opt.router_alert) { |
1948 | if (ip_call_ra_chain(skb)) | |
1949 | return 0; | |
1950 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) { | |
1951 | /* IGMPv1 (and broken IGMPv2 implementations sort of | |
1952 | * Cisco IOS <= 11.2(8)) do not put router alert | |
1953 | * option to IGMP packets destined to routable | |
1954 | * groups. It is very bad, because it means | |
1955 | * that we can forward NO IGMP messages. | |
1956 | */ | |
1957 | struct sock *mroute_sk; | |
1958 | ||
1959 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
1960 | if (mroute_sk) { | |
1961 | nf_reset(skb); | |
1962 | raw_rcv(mroute_sk, skb); | |
1963 | return 0; | |
1964 | } | |
1da177e4 LT |
1965 | } |
1966 | } | |
1967 | ||
a8c9486b | 1968 | /* already under rcu_read_lock() */ |
0c12295a | 1969 | cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
51456b29 | 1970 | if (!cache) { |
660b26dc ND |
1971 | int vif = ipmr_find_vif(mrt, skb->dev); |
1972 | ||
1973 | if (vif >= 0) | |
1974 | cache = ipmr_cache_find_any(mrt, ip_hdr(skb)->daddr, | |
1975 | vif); | |
1976 | } | |
1da177e4 | 1977 | |
7ef8f65d | 1978 | /* No usable cache entry */ |
51456b29 | 1979 | if (!cache) { |
1da177e4 LT |
1980 | int vif; |
1981 | ||
1982 | if (local) { | |
1983 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1984 | ip_local_deliver(skb); | |
51456b29 | 1985 | if (!skb2) |
1da177e4 | 1986 | return -ENOBUFS; |
1da177e4 LT |
1987 | skb = skb2; |
1988 | } | |
1989 | ||
a8c9486b | 1990 | read_lock(&mrt_lock); |
0c12295a | 1991 | vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1992 | if (vif >= 0) { |
0eae88f3 | 1993 | int err2 = ipmr_cache_unresolved(mrt, vif, skb); |
1da177e4 LT |
1994 | read_unlock(&mrt_lock); |
1995 | ||
0eae88f3 | 1996 | return err2; |
1da177e4 LT |
1997 | } |
1998 | read_unlock(&mrt_lock); | |
1999 | kfree_skb(skb); | |
2000 | return -ENODEV; | |
2001 | } | |
2002 | ||
a8c9486b | 2003 | read_lock(&mrt_lock); |
0c12295a | 2004 | ip_mr_forward(net, mrt, skb, cache, local); |
1da177e4 LT |
2005 | read_unlock(&mrt_lock); |
2006 | ||
2007 | if (local) | |
2008 | return ip_local_deliver(skb); | |
2009 | ||
2010 | return 0; | |
2011 | ||
2012 | dont_forward: | |
2013 | if (local) | |
2014 | return ip_local_deliver(skb); | |
2015 | kfree_skb(skb); | |
2016 | return 0; | |
2017 | } | |
2018 | ||
b1879204 | 2019 | #ifdef CONFIG_IP_PIMSM_V1 |
7ef8f65d | 2020 | /* Handle IGMP messages of PIMv1 */ |
a8cb16dd | 2021 | int pim_rcv_v1(struct sk_buff *skb) |
b1879204 IJ |
2022 | { |
2023 | struct igmphdr *pim; | |
4feb88e5 | 2024 | struct net *net = dev_net(skb->dev); |
f0ad0860 | 2025 | struct mr_table *mrt; |
b1879204 IJ |
2026 | |
2027 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
2028 | goto drop; | |
2029 | ||
2030 | pim = igmp_hdr(skb); | |
2031 | ||
417da66f | 2032 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2033 | if (IS_ERR(mrt)) |
2034 | goto drop; | |
0c12295a | 2035 | if (!mrt->mroute_do_pim || |
b1879204 IJ |
2036 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
2037 | goto drop; | |
2038 | ||
f0ad0860 | 2039 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2040 | drop: |
2041 | kfree_skb(skb); | |
2042 | } | |
1da177e4 LT |
2043 | return 0; |
2044 | } | |
2045 | #endif | |
2046 | ||
2047 | #ifdef CONFIG_IP_PIMSM_V2 | |
a8cb16dd | 2048 | static int pim_rcv(struct sk_buff *skb) |
1da177e4 LT |
2049 | { |
2050 | struct pimreghdr *pim; | |
f0ad0860 PM |
2051 | struct net *net = dev_net(skb->dev); |
2052 | struct mr_table *mrt; | |
1da177e4 | 2053 | |
b1879204 | 2054 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
2055 | goto drop; |
2056 | ||
9c70220b | 2057 | pim = (struct pimreghdr *)skb_transport_header(skb); |
a8cb16dd ED |
2058 | if (pim->type != ((PIM_VERSION << 4) | (PIM_REGISTER)) || |
2059 | (pim->flags & PIM_NULL_REGISTER) || | |
e905a9ed | 2060 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 2061 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
2062 | goto drop; |
2063 | ||
417da66f | 2064 | mrt = ipmr_rt_fib_lookup(net, skb); |
ee3f1aaf DM |
2065 | if (IS_ERR(mrt)) |
2066 | goto drop; | |
f0ad0860 | 2067 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2068 | drop: |
2069 | kfree_skb(skb); | |
2070 | } | |
1da177e4 LT |
2071 | return 0; |
2072 | } | |
2073 | #endif | |
2074 | ||
cb6a4e46 PM |
2075 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2076 | struct mfc_cache *c, struct rtmsg *rtm) | |
1da177e4 LT |
2077 | { |
2078 | int ct; | |
2079 | struct rtnexthop *nhp; | |
92a395e5 | 2080 | struct nlattr *mp_attr; |
adfa85e4 | 2081 | struct rta_mfc_stats mfcs; |
1da177e4 | 2082 | |
7438189b | 2083 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
ed0f160a | 2084 | if (c->mfc_parent >= MAXVIFS) |
7438189b ND |
2085 | return -ENOENT; |
2086 | ||
92a395e5 TG |
2087 | if (VIF_EXISTS(mrt, c->mfc_parent) && |
2088 | nla_put_u32(skb, RTA_IIF, mrt->vif_table[c->mfc_parent].dev->ifindex) < 0) | |
2089 | return -EMSGSIZE; | |
1da177e4 | 2090 | |
92a395e5 TG |
2091 | if (!(mp_attr = nla_nest_start(skb, RTA_MULTIPATH))) |
2092 | return -EMSGSIZE; | |
1da177e4 LT |
2093 | |
2094 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
0c12295a | 2095 | if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) { |
92a395e5 TG |
2096 | if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp)))) { |
2097 | nla_nest_cancel(skb, mp_attr); | |
2098 | return -EMSGSIZE; | |
2099 | } | |
2100 | ||
1da177e4 LT |
2101 | nhp->rtnh_flags = 0; |
2102 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
0c12295a | 2103 | nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex; |
1da177e4 LT |
2104 | nhp->rtnh_len = sizeof(*nhp); |
2105 | } | |
2106 | } | |
92a395e5 TG |
2107 | |
2108 | nla_nest_end(skb, mp_attr); | |
2109 | ||
adfa85e4 ND |
2110 | mfcs.mfcs_packets = c->mfc_un.res.pkt; |
2111 | mfcs.mfcs_bytes = c->mfc_un.res.bytes; | |
2112 | mfcs.mfcs_wrong_if = c->mfc_un.res.wrong_if; | |
2113 | if (nla_put(skb, RTA_MFC_STATS, sizeof(mfcs), &mfcs) < 0) | |
2114 | return -EMSGSIZE; | |
2115 | ||
1da177e4 LT |
2116 | rtm->rtm_type = RTN_MULTICAST; |
2117 | return 1; | |
1da177e4 LT |
2118 | } |
2119 | ||
9a1b9496 DM |
2120 | int ipmr_get_route(struct net *net, struct sk_buff *skb, |
2121 | __be32 saddr, __be32 daddr, | |
2122 | struct rtmsg *rtm, int nowait) | |
1da177e4 | 2123 | { |
1da177e4 | 2124 | struct mfc_cache *cache; |
9a1b9496 DM |
2125 | struct mr_table *mrt; |
2126 | int err; | |
1da177e4 | 2127 | |
f0ad0860 | 2128 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
51456b29 | 2129 | if (!mrt) |
f0ad0860 PM |
2130 | return -ENOENT; |
2131 | ||
a8c9486b | 2132 | rcu_read_lock(); |
9a1b9496 | 2133 | cache = ipmr_cache_find(mrt, saddr, daddr); |
51456b29 | 2134 | if (!cache && skb->dev) { |
660b26dc | 2135 | int vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 2136 | |
660b26dc ND |
2137 | if (vif >= 0) |
2138 | cache = ipmr_cache_find_any(mrt, daddr, vif); | |
2139 | } | |
51456b29 | 2140 | if (!cache) { |
72287490 | 2141 | struct sk_buff *skb2; |
eddc9ec5 | 2142 | struct iphdr *iph; |
1da177e4 | 2143 | struct net_device *dev; |
a8cb16dd | 2144 | int vif = -1; |
1da177e4 LT |
2145 | |
2146 | if (nowait) { | |
a8c9486b | 2147 | rcu_read_unlock(); |
1da177e4 LT |
2148 | return -EAGAIN; |
2149 | } | |
2150 | ||
2151 | dev = skb->dev; | |
a8c9486b | 2152 | read_lock(&mrt_lock); |
a8cb16dd ED |
2153 | if (dev) |
2154 | vif = ipmr_find_vif(mrt, dev); | |
2155 | if (vif < 0) { | |
1da177e4 | 2156 | read_unlock(&mrt_lock); |
a8c9486b | 2157 | rcu_read_unlock(); |
1da177e4 LT |
2158 | return -ENODEV; |
2159 | } | |
72287490 AK |
2160 | skb2 = skb_clone(skb, GFP_ATOMIC); |
2161 | if (!skb2) { | |
2162 | read_unlock(&mrt_lock); | |
a8c9486b | 2163 | rcu_read_unlock(); |
72287490 AK |
2164 | return -ENOMEM; |
2165 | } | |
2166 | ||
e2d1bca7 ACM |
2167 | skb_push(skb2, sizeof(struct iphdr)); |
2168 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
2169 | iph = ip_hdr(skb2); |
2170 | iph->ihl = sizeof(struct iphdr) >> 2; | |
9a1b9496 DM |
2171 | iph->saddr = saddr; |
2172 | iph->daddr = daddr; | |
eddc9ec5 | 2173 | iph->version = 0; |
0c12295a | 2174 | err = ipmr_cache_unresolved(mrt, vif, skb2); |
1da177e4 | 2175 | read_unlock(&mrt_lock); |
a8c9486b | 2176 | rcu_read_unlock(); |
1da177e4 LT |
2177 | return err; |
2178 | } | |
2179 | ||
a8c9486b ED |
2180 | read_lock(&mrt_lock); |
2181 | if (!nowait && (rtm->rtm_flags & RTM_F_NOTIFY)) | |
1da177e4 | 2182 | cache->mfc_flags |= MFC_NOTIFY; |
cb6a4e46 | 2183 | err = __ipmr_fill_mroute(mrt, skb, cache, rtm); |
1da177e4 | 2184 | read_unlock(&mrt_lock); |
a8c9486b | 2185 | rcu_read_unlock(); |
1da177e4 LT |
2186 | return err; |
2187 | } | |
2188 | ||
cb6a4e46 | 2189 | static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
65886f43 ND |
2190 | u32 portid, u32 seq, struct mfc_cache *c, int cmd, |
2191 | int flags) | |
cb6a4e46 PM |
2192 | { |
2193 | struct nlmsghdr *nlh; | |
2194 | struct rtmsg *rtm; | |
1eb99af5 | 2195 | int err; |
cb6a4e46 | 2196 | |
65886f43 | 2197 | nlh = nlmsg_put(skb, portid, seq, cmd, sizeof(*rtm), flags); |
51456b29 | 2198 | if (!nlh) |
cb6a4e46 PM |
2199 | return -EMSGSIZE; |
2200 | ||
2201 | rtm = nlmsg_data(nlh); | |
2202 | rtm->rtm_family = RTNL_FAMILY_IPMR; | |
2203 | rtm->rtm_dst_len = 32; | |
2204 | rtm->rtm_src_len = 32; | |
2205 | rtm->rtm_tos = 0; | |
2206 | rtm->rtm_table = mrt->id; | |
f3756b79 DM |
2207 | if (nla_put_u32(skb, RTA_TABLE, mrt->id)) |
2208 | goto nla_put_failure; | |
cb6a4e46 PM |
2209 | rtm->rtm_type = RTN_MULTICAST; |
2210 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
9a68ac72 ND |
2211 | if (c->mfc_flags & MFC_STATIC) |
2212 | rtm->rtm_protocol = RTPROT_STATIC; | |
2213 | else | |
2214 | rtm->rtm_protocol = RTPROT_MROUTED; | |
cb6a4e46 PM |
2215 | rtm->rtm_flags = 0; |
2216 | ||
930345ea JB |
2217 | if (nla_put_in_addr(skb, RTA_SRC, c->mfc_origin) || |
2218 | nla_put_in_addr(skb, RTA_DST, c->mfc_mcastgrp)) | |
f3756b79 | 2219 | goto nla_put_failure; |
1eb99af5 ND |
2220 | err = __ipmr_fill_mroute(mrt, skb, c, rtm); |
2221 | /* do not break the dump if cache is unresolved */ | |
2222 | if (err < 0 && err != -ENOENT) | |
cb6a4e46 PM |
2223 | goto nla_put_failure; |
2224 | ||
053c095a JB |
2225 | nlmsg_end(skb, nlh); |
2226 | return 0; | |
cb6a4e46 PM |
2227 | |
2228 | nla_put_failure: | |
2229 | nlmsg_cancel(skb, nlh); | |
2230 | return -EMSGSIZE; | |
2231 | } | |
2232 | ||
8cd3ac9f ND |
2233 | static size_t mroute_msgsize(bool unresolved, int maxvif) |
2234 | { | |
2235 | size_t len = | |
2236 | NLMSG_ALIGN(sizeof(struct rtmsg)) | |
2237 | + nla_total_size(4) /* RTA_TABLE */ | |
2238 | + nla_total_size(4) /* RTA_SRC */ | |
2239 | + nla_total_size(4) /* RTA_DST */ | |
2240 | ; | |
2241 | ||
2242 | if (!unresolved) | |
2243 | len = len | |
2244 | + nla_total_size(4) /* RTA_IIF */ | |
2245 | + nla_total_size(0) /* RTA_MULTIPATH */ | |
2246 | + maxvif * NLA_ALIGN(sizeof(struct rtnexthop)) | |
2247 | /* RTA_MFC_STATS */ | |
2248 | + nla_total_size(sizeof(struct rta_mfc_stats)) | |
2249 | ; | |
2250 | ||
2251 | return len; | |
2252 | } | |
2253 | ||
2254 | static void mroute_netlink_event(struct mr_table *mrt, struct mfc_cache *mfc, | |
2255 | int cmd) | |
2256 | { | |
2257 | struct net *net = read_pnet(&mrt->net); | |
2258 | struct sk_buff *skb; | |
2259 | int err = -ENOBUFS; | |
2260 | ||
2261 | skb = nlmsg_new(mroute_msgsize(mfc->mfc_parent >= MAXVIFS, mrt->maxvif), | |
2262 | GFP_ATOMIC); | |
51456b29 | 2263 | if (!skb) |
8cd3ac9f ND |
2264 | goto errout; |
2265 | ||
65886f43 | 2266 | err = ipmr_fill_mroute(mrt, skb, 0, 0, mfc, cmd, 0); |
8cd3ac9f ND |
2267 | if (err < 0) |
2268 | goto errout; | |
2269 | ||
2270 | rtnl_notify(skb, net, 0, RTNLGRP_IPV4_MROUTE, NULL, GFP_ATOMIC); | |
2271 | return; | |
2272 | ||
2273 | errout: | |
2274 | kfree_skb(skb); | |
2275 | if (err < 0) | |
2276 | rtnl_set_sk_err(net, RTNLGRP_IPV4_MROUTE, err); | |
2277 | } | |
2278 | ||
cb6a4e46 PM |
2279 | static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb) |
2280 | { | |
2281 | struct net *net = sock_net(skb->sk); | |
2282 | struct mr_table *mrt; | |
2283 | struct mfc_cache *mfc; | |
2284 | unsigned int t = 0, s_t; | |
2285 | unsigned int h = 0, s_h; | |
2286 | unsigned int e = 0, s_e; | |
2287 | ||
2288 | s_t = cb->args[0]; | |
2289 | s_h = cb->args[1]; | |
2290 | s_e = cb->args[2]; | |
2291 | ||
a8c9486b | 2292 | rcu_read_lock(); |
cb6a4e46 PM |
2293 | ipmr_for_each_table(mrt, net) { |
2294 | if (t < s_t) | |
2295 | goto next_table; | |
2296 | if (t > s_t) | |
2297 | s_h = 0; | |
2298 | for (h = s_h; h < MFC_LINES; h++) { | |
a8c9486b | 2299 | list_for_each_entry_rcu(mfc, &mrt->mfc_cache_array[h], list) { |
cb6a4e46 PM |
2300 | if (e < s_e) |
2301 | goto next_entry; | |
2302 | if (ipmr_fill_mroute(mrt, skb, | |
15e47304 | 2303 | NETLINK_CB(cb->skb).portid, |
cb6a4e46 | 2304 | cb->nlh->nlmsg_seq, |
65886f43 ND |
2305 | mfc, RTM_NEWROUTE, |
2306 | NLM_F_MULTI) < 0) | |
cb6a4e46 PM |
2307 | goto done; |
2308 | next_entry: | |
2309 | e++; | |
2310 | } | |
2311 | e = s_e = 0; | |
2312 | } | |
1eb99af5 ND |
2313 | spin_lock_bh(&mfc_unres_lock); |
2314 | list_for_each_entry(mfc, &mrt->mfc_unres_queue, list) { | |
2315 | if (e < s_e) | |
2316 | goto next_entry2; | |
2317 | if (ipmr_fill_mroute(mrt, skb, | |
2318 | NETLINK_CB(cb->skb).portid, | |
2319 | cb->nlh->nlmsg_seq, | |
65886f43 ND |
2320 | mfc, RTM_NEWROUTE, |
2321 | NLM_F_MULTI) < 0) { | |
1eb99af5 ND |
2322 | spin_unlock_bh(&mfc_unres_lock); |
2323 | goto done; | |
2324 | } | |
2325 | next_entry2: | |
2326 | e++; | |
2327 | } | |
2328 | spin_unlock_bh(&mfc_unres_lock); | |
2329 | e = s_e = 0; | |
cb6a4e46 PM |
2330 | s_h = 0; |
2331 | next_table: | |
2332 | t++; | |
2333 | } | |
2334 | done: | |
a8c9486b | 2335 | rcu_read_unlock(); |
cb6a4e46 PM |
2336 | |
2337 | cb->args[2] = e; | |
2338 | cb->args[1] = h; | |
2339 | cb->args[0] = t; | |
2340 | ||
2341 | return skb->len; | |
2342 | } | |
2343 | ||
e905a9ed | 2344 | #ifdef CONFIG_PROC_FS |
7ef8f65d NA |
2345 | /* The /proc interfaces to multicast routing : |
2346 | * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif | |
1da177e4 LT |
2347 | */ |
2348 | struct ipmr_vif_iter { | |
f6bb4514 | 2349 | struct seq_net_private p; |
f0ad0860 | 2350 | struct mr_table *mrt; |
1da177e4 LT |
2351 | int ct; |
2352 | }; | |
2353 | ||
f6bb4514 BT |
2354 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
2355 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
2356 | loff_t pos) |
2357 | { | |
f0ad0860 | 2358 | struct mr_table *mrt = iter->mrt; |
0c12295a PM |
2359 | |
2360 | for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) { | |
2361 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2362 | continue; |
e905a9ed | 2363 | if (pos-- == 0) |
0c12295a | 2364 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2365 | } |
2366 | return NULL; | |
2367 | } | |
2368 | ||
2369 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 2370 | __acquires(mrt_lock) |
1da177e4 | 2371 | { |
f0ad0860 | 2372 | struct ipmr_vif_iter *iter = seq->private; |
f6bb4514 | 2373 | struct net *net = seq_file_net(seq); |
f0ad0860 PM |
2374 | struct mr_table *mrt; |
2375 | ||
2376 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); | |
51456b29 | 2377 | if (!mrt) |
f0ad0860 PM |
2378 | return ERR_PTR(-ENOENT); |
2379 | ||
2380 | iter->mrt = mrt; | |
f6bb4514 | 2381 | |
1da177e4 | 2382 | read_lock(&mrt_lock); |
f6bb4514 | 2383 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2384 | : SEQ_START_TOKEN; |
2385 | } | |
2386 | ||
2387 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2388 | { | |
2389 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 2390 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2391 | struct mr_table *mrt = iter->mrt; |
1da177e4 LT |
2392 | |
2393 | ++*pos; | |
2394 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2395 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 2396 | |
0c12295a PM |
2397 | while (++iter->ct < mrt->maxvif) { |
2398 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2399 | continue; |
0c12295a | 2400 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2401 | } |
2402 | return NULL; | |
2403 | } | |
2404 | ||
2405 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 2406 | __releases(mrt_lock) |
1da177e4 LT |
2407 | { |
2408 | read_unlock(&mrt_lock); | |
2409 | } | |
2410 | ||
2411 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
2412 | { | |
f0ad0860 PM |
2413 | struct ipmr_vif_iter *iter = seq->private; |
2414 | struct mr_table *mrt = iter->mrt; | |
f6bb4514 | 2415 | |
1da177e4 | 2416 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 2417 | seq_puts(seq, |
1da177e4 LT |
2418 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
2419 | } else { | |
2420 | const struct vif_device *vif = v; | |
2421 | const char *name = vif->dev ? vif->dev->name : "none"; | |
2422 | ||
2423 | seq_printf(seq, | |
2424 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
0c12295a | 2425 | vif - mrt->vif_table, |
e905a9ed | 2426 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
2427 | vif->bytes_out, vif->pkt_out, |
2428 | vif->flags, vif->local, vif->remote); | |
2429 | } | |
2430 | return 0; | |
2431 | } | |
2432 | ||
f690808e | 2433 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
2434 | .start = ipmr_vif_seq_start, |
2435 | .next = ipmr_vif_seq_next, | |
2436 | .stop = ipmr_vif_seq_stop, | |
2437 | .show = ipmr_vif_seq_show, | |
2438 | }; | |
2439 | ||
2440 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
2441 | { | |
f6bb4514 BT |
2442 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
2443 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
2444 | } |
2445 | ||
9a32144e | 2446 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
2447 | .owner = THIS_MODULE, |
2448 | .open = ipmr_vif_open, | |
2449 | .read = seq_read, | |
2450 | .llseek = seq_lseek, | |
f6bb4514 | 2451 | .release = seq_release_net, |
1da177e4 LT |
2452 | }; |
2453 | ||
2454 | struct ipmr_mfc_iter { | |
f6bb4514 | 2455 | struct seq_net_private p; |
f0ad0860 | 2456 | struct mr_table *mrt; |
862465f2 | 2457 | struct list_head *cache; |
1da177e4 LT |
2458 | int ct; |
2459 | }; | |
2460 | ||
2461 | ||
f6bb4514 BT |
2462 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
2463 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 | 2464 | { |
f0ad0860 | 2465 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2466 | struct mfc_cache *mfc; |
2467 | ||
a8c9486b | 2468 | rcu_read_lock(); |
862465f2 | 2469 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) { |
0c12295a | 2470 | it->cache = &mrt->mfc_cache_array[it->ct]; |
a8c9486b | 2471 | list_for_each_entry_rcu(mfc, it->cache, list) |
e905a9ed | 2472 | if (pos-- == 0) |
1da177e4 | 2473 | return mfc; |
862465f2 | 2474 | } |
a8c9486b | 2475 | rcu_read_unlock(); |
1da177e4 | 2476 | |
1da177e4 | 2477 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 2478 | it->cache = &mrt->mfc_unres_queue; |
862465f2 | 2479 | list_for_each_entry(mfc, it->cache, list) |
e258beb2 | 2480 | if (pos-- == 0) |
1da177e4 LT |
2481 | return mfc; |
2482 | spin_unlock_bh(&mfc_unres_lock); | |
2483 | ||
2484 | it->cache = NULL; | |
2485 | return NULL; | |
2486 | } | |
2487 | ||
2488 | ||
2489 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
2490 | { | |
2491 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2492 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2493 | struct mr_table *mrt; |
f6bb4514 | 2494 | |
f0ad0860 | 2495 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
51456b29 | 2496 | if (!mrt) |
f0ad0860 | 2497 | return ERR_PTR(-ENOENT); |
f6bb4514 | 2498 | |
f0ad0860 | 2499 | it->mrt = mrt; |
1da177e4 LT |
2500 | it->cache = NULL; |
2501 | it->ct = 0; | |
f6bb4514 | 2502 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2503 | : SEQ_START_TOKEN; |
2504 | } | |
2505 | ||
2506 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2507 | { | |
2508 | struct mfc_cache *mfc = v; | |
2509 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2510 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2511 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2512 | |
2513 | ++*pos; | |
2514 | ||
2515 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2516 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 | 2517 | |
862465f2 PM |
2518 | if (mfc->list.next != it->cache) |
2519 | return list_entry(mfc->list.next, struct mfc_cache, list); | |
e905a9ed | 2520 | |
0c12295a | 2521 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 LT |
2522 | goto end_of_list; |
2523 | ||
0c12295a | 2524 | BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]); |
1da177e4 LT |
2525 | |
2526 | while (++it->ct < MFC_LINES) { | |
0c12295a | 2527 | it->cache = &mrt->mfc_cache_array[it->ct]; |
862465f2 PM |
2528 | if (list_empty(it->cache)) |
2529 | continue; | |
2530 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 LT |
2531 | } |
2532 | ||
2533 | /* exhausted cache_array, show unresolved */ | |
a8c9486b | 2534 | rcu_read_unlock(); |
0c12295a | 2535 | it->cache = &mrt->mfc_unres_queue; |
1da177e4 | 2536 | it->ct = 0; |
e905a9ed | 2537 | |
1da177e4 | 2538 | spin_lock_bh(&mfc_unres_lock); |
862465f2 PM |
2539 | if (!list_empty(it->cache)) |
2540 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 | 2541 | |
a8cb16dd | 2542 | end_of_list: |
1da177e4 LT |
2543 | spin_unlock_bh(&mfc_unres_lock); |
2544 | it->cache = NULL; | |
2545 | ||
2546 | return NULL; | |
2547 | } | |
2548 | ||
2549 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
2550 | { | |
2551 | struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2552 | struct mr_table *mrt = it->mrt; |
1da177e4 | 2553 | |
0c12295a | 2554 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 | 2555 | spin_unlock_bh(&mfc_unres_lock); |
0c12295a | 2556 | else if (it->cache == &mrt->mfc_cache_array[it->ct]) |
a8c9486b | 2557 | rcu_read_unlock(); |
1da177e4 LT |
2558 | } |
2559 | ||
2560 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
2561 | { | |
2562 | int n; | |
2563 | ||
2564 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 2565 | seq_puts(seq, |
1da177e4 LT |
2566 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
2567 | } else { | |
2568 | const struct mfc_cache *mfc = v; | |
2569 | const struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2570 | const struct mr_table *mrt = it->mrt; |
e905a9ed | 2571 | |
0eae88f3 ED |
2572 | seq_printf(seq, "%08X %08X %-3hd", |
2573 | (__force u32) mfc->mfc_mcastgrp, | |
2574 | (__force u32) mfc->mfc_origin, | |
1ea472e2 | 2575 | mfc->mfc_parent); |
1da177e4 | 2576 | |
0c12295a | 2577 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 BT |
2578 | seq_printf(seq, " %8lu %8lu %8lu", |
2579 | mfc->mfc_un.res.pkt, | |
2580 | mfc->mfc_un.res.bytes, | |
2581 | mfc->mfc_un.res.wrong_if); | |
132adf54 | 2582 | for (n = mfc->mfc_un.res.minvif; |
a8cb16dd | 2583 | n < mfc->mfc_un.res.maxvif; n++) { |
0c12295a | 2584 | if (VIF_EXISTS(mrt, n) && |
cf958ae3 BT |
2585 | mfc->mfc_un.res.ttls[n] < 255) |
2586 | seq_printf(seq, | |
e905a9ed | 2587 | " %2d:%-3d", |
1da177e4 LT |
2588 | n, mfc->mfc_un.res.ttls[n]); |
2589 | } | |
1ea472e2 BT |
2590 | } else { |
2591 | /* unresolved mfc_caches don't contain | |
2592 | * pkt, bytes and wrong_if values | |
2593 | */ | |
2594 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
2595 | } |
2596 | seq_putc(seq, '\n'); | |
2597 | } | |
2598 | return 0; | |
2599 | } | |
2600 | ||
f690808e | 2601 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
2602 | .start = ipmr_mfc_seq_start, |
2603 | .next = ipmr_mfc_seq_next, | |
2604 | .stop = ipmr_mfc_seq_stop, | |
2605 | .show = ipmr_mfc_seq_show, | |
2606 | }; | |
2607 | ||
2608 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
2609 | { | |
f6bb4514 BT |
2610 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
2611 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
2612 | } |
2613 | ||
9a32144e | 2614 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
2615 | .owner = THIS_MODULE, |
2616 | .open = ipmr_mfc_open, | |
2617 | .read = seq_read, | |
2618 | .llseek = seq_lseek, | |
f6bb4514 | 2619 | .release = seq_release_net, |
1da177e4 | 2620 | }; |
e905a9ed | 2621 | #endif |
1da177e4 LT |
2622 | |
2623 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 2624 | static const struct net_protocol pim_protocol = { |
1da177e4 | 2625 | .handler = pim_rcv, |
403dbb97 | 2626 | .netns_ok = 1, |
1da177e4 LT |
2627 | }; |
2628 | #endif | |
2629 | ||
7ef8f65d | 2630 | /* Setup for IP multicast routing */ |
cf958ae3 BT |
2631 | static int __net_init ipmr_net_init(struct net *net) |
2632 | { | |
f0ad0860 | 2633 | int err; |
cf958ae3 | 2634 | |
f0ad0860 PM |
2635 | err = ipmr_rules_init(net); |
2636 | if (err < 0) | |
cf958ae3 | 2637 | goto fail; |
f6bb4514 BT |
2638 | |
2639 | #ifdef CONFIG_PROC_FS | |
2640 | err = -ENOMEM; | |
d4beaa66 | 2641 | if (!proc_create("ip_mr_vif", 0, net->proc_net, &ipmr_vif_fops)) |
f6bb4514 | 2642 | goto proc_vif_fail; |
d4beaa66 | 2643 | if (!proc_create("ip_mr_cache", 0, net->proc_net, &ipmr_mfc_fops)) |
f6bb4514 BT |
2644 | goto proc_cache_fail; |
2645 | #endif | |
2bb8b26c BT |
2646 | return 0; |
2647 | ||
f6bb4514 BT |
2648 | #ifdef CONFIG_PROC_FS |
2649 | proc_cache_fail: | |
ece31ffd | 2650 | remove_proc_entry("ip_mr_vif", net->proc_net); |
f6bb4514 | 2651 | proc_vif_fail: |
f0ad0860 | 2652 | ipmr_rules_exit(net); |
f6bb4514 | 2653 | #endif |
cf958ae3 BT |
2654 | fail: |
2655 | return err; | |
2656 | } | |
2657 | ||
2658 | static void __net_exit ipmr_net_exit(struct net *net) | |
2659 | { | |
f6bb4514 | 2660 | #ifdef CONFIG_PROC_FS |
ece31ffd G |
2661 | remove_proc_entry("ip_mr_cache", net->proc_net); |
2662 | remove_proc_entry("ip_mr_vif", net->proc_net); | |
f6bb4514 | 2663 | #endif |
f0ad0860 | 2664 | ipmr_rules_exit(net); |
cf958ae3 BT |
2665 | } |
2666 | ||
2667 | static struct pernet_operations ipmr_net_ops = { | |
2668 | .init = ipmr_net_init, | |
2669 | .exit = ipmr_net_exit, | |
2670 | }; | |
e905a9ed | 2671 | |
03d2f897 | 2672 | int __init ip_mr_init(void) |
1da177e4 | 2673 | { |
03d2f897 WC |
2674 | int err; |
2675 | ||
1da177e4 LT |
2676 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2677 | sizeof(struct mfc_cache), | |
a8c9486b | 2678 | 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, |
20c2df83 | 2679 | NULL); |
03d2f897 WC |
2680 | if (!mrt_cachep) |
2681 | return -ENOMEM; | |
2682 | ||
cf958ae3 BT |
2683 | err = register_pernet_subsys(&ipmr_net_ops); |
2684 | if (err) | |
2685 | goto reg_pernet_fail; | |
2686 | ||
03d2f897 WC |
2687 | err = register_netdevice_notifier(&ip_mr_notifier); |
2688 | if (err) | |
2689 | goto reg_notif_fail; | |
403dbb97 TG |
2690 | #ifdef CONFIG_IP_PIMSM_V2 |
2691 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
058bd4d2 | 2692 | pr_err("%s: can't add PIM protocol\n", __func__); |
403dbb97 TG |
2693 | err = -EAGAIN; |
2694 | goto add_proto_fail; | |
2695 | } | |
2696 | #endif | |
c7ac8679 GR |
2697 | rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE, |
2698 | NULL, ipmr_rtm_dumproute, NULL); | |
03d2f897 | 2699 | return 0; |
f6bb4514 | 2700 | |
403dbb97 TG |
2701 | #ifdef CONFIG_IP_PIMSM_V2 |
2702 | add_proto_fail: | |
2703 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2704 | #endif | |
c3e38896 | 2705 | reg_notif_fail: |
cf958ae3 BT |
2706 | unregister_pernet_subsys(&ipmr_net_ops); |
2707 | reg_pernet_fail: | |
c3e38896 | 2708 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2709 | return err; |
1da177e4 | 2710 | } |