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/system.h> |
30 | #include <asm/uaccess.h> | |
31 | #include <linux/types.h> | |
4fc268d2 | 32 | #include <linux/capability.h> |
1da177e4 LT |
33 | #include <linux/errno.h> |
34 | #include <linux/timer.h> | |
35 | #include <linux/mm.h> | |
36 | #include <linux/kernel.h> | |
37 | #include <linux/fcntl.h> | |
38 | #include <linux/stat.h> | |
39 | #include <linux/socket.h> | |
40 | #include <linux/in.h> | |
41 | #include <linux/inet.h> | |
42 | #include <linux/netdevice.h> | |
43 | #include <linux/inetdevice.h> | |
44 | #include <linux/igmp.h> | |
45 | #include <linux/proc_fs.h> | |
46 | #include <linux/seq_file.h> | |
47 | #include <linux/mroute.h> | |
48 | #include <linux/init.h> | |
46f25dff | 49 | #include <linux/if_ether.h> |
5a0e3ad6 | 50 | #include <linux/slab.h> |
457c4cbc | 51 | #include <net/net_namespace.h> |
1da177e4 LT |
52 | #include <net/ip.h> |
53 | #include <net/protocol.h> | |
54 | #include <linux/skbuff.h> | |
14c85021 | 55 | #include <net/route.h> |
1da177e4 LT |
56 | #include <net/sock.h> |
57 | #include <net/icmp.h> | |
58 | #include <net/udp.h> | |
59 | #include <net/raw.h> | |
60 | #include <linux/notifier.h> | |
61 | #include <linux/if_arp.h> | |
62 | #include <linux/netfilter_ipv4.h> | |
709b46e8 | 63 | #include <linux/compat.h> |
1da177e4 LT |
64 | #include <net/ipip.h> |
65 | #include <net/checksum.h> | |
dc5fc579 | 66 | #include <net/netlink.h> |
f0ad0860 | 67 | #include <net/fib_rules.h> |
1da177e4 LT |
68 | |
69 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
70 | #define CONFIG_IP_PIMSM 1 | |
71 | #endif | |
72 | ||
0c12295a | 73 | struct mr_table { |
f0ad0860 | 74 | struct list_head list; |
8de53dfb PM |
75 | #ifdef CONFIG_NET_NS |
76 | struct net *net; | |
77 | #endif | |
f0ad0860 | 78 | u32 id; |
4c968709 | 79 | struct sock __rcu *mroute_sk; |
0c12295a PM |
80 | struct timer_list ipmr_expire_timer; |
81 | struct list_head mfc_unres_queue; | |
82 | struct list_head mfc_cache_array[MFC_LINES]; | |
83 | struct vif_device vif_table[MAXVIFS]; | |
84 | int maxvif; | |
85 | atomic_t cache_resolve_queue_len; | |
86 | int mroute_do_assert; | |
87 | int mroute_do_pim; | |
88 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
89 | int mroute_reg_vif_num; | |
90 | #endif | |
91 | }; | |
92 | ||
f0ad0860 PM |
93 | struct ipmr_rule { |
94 | struct fib_rule common; | |
95 | }; | |
96 | ||
97 | struct ipmr_result { | |
98 | struct mr_table *mrt; | |
99 | }; | |
100 | ||
1da177e4 | 101 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
a8cb16dd | 102 | * Note that the changes are semaphored via rtnl_lock. |
1da177e4 LT |
103 | */ |
104 | ||
105 | static DEFINE_RWLOCK(mrt_lock); | |
106 | ||
107 | /* | |
108 | * Multicast router control variables | |
109 | */ | |
110 | ||
0c12295a | 111 | #define VIF_EXISTS(_mrt, _idx) ((_mrt)->vif_table[_idx].dev != NULL) |
1da177e4 LT |
112 | |
113 | /* Special spinlock for queue of unresolved entries */ | |
114 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
115 | ||
116 | /* We return to original Alan's scheme. Hash table of resolved | |
a8cb16dd ED |
117 | * entries is changed only in process context and protected |
118 | * with weak lock mrt_lock. Queue of unresolved entries is protected | |
119 | * with strong spinlock mfc_unres_lock. | |
120 | * | |
121 | * In this case data path is free of exclusive locks at all. | |
1da177e4 LT |
122 | */ |
123 | ||
e18b890b | 124 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 | 125 | |
f0ad0860 | 126 | static struct mr_table *ipmr_new_table(struct net *net, u32 id); |
0c12295a PM |
127 | static int ip_mr_forward(struct net *net, struct mr_table *mrt, |
128 | struct sk_buff *skb, struct mfc_cache *cache, | |
129 | int local); | |
130 | static int ipmr_cache_report(struct mr_table *mrt, | |
4feb88e5 | 131 | struct sk_buff *pkt, vifi_t vifi, int assert); |
cb6a4e46 PM |
132 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
133 | struct mfc_cache *c, struct rtmsg *rtm); | |
f0ad0860 PM |
134 | static void ipmr_expire_process(unsigned long arg); |
135 | ||
136 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
137 | #define ipmr_for_each_table(mrt, net) \ | |
138 | list_for_each_entry_rcu(mrt, &net->ipv4.mr_tables, list) | |
139 | ||
140 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
141 | { | |
142 | struct mr_table *mrt; | |
143 | ||
144 | ipmr_for_each_table(mrt, net) { | |
145 | if (mrt->id == id) | |
146 | return mrt; | |
147 | } | |
148 | return NULL; | |
149 | } | |
150 | ||
151 | static int ipmr_fib_lookup(struct net *net, struct flowi *flp, | |
152 | struct mr_table **mrt) | |
153 | { | |
154 | struct ipmr_result res; | |
155 | struct fib_lookup_arg arg = { .result = &res, }; | |
156 | int err; | |
157 | ||
158 | err = fib_rules_lookup(net->ipv4.mr_rules_ops, flp, 0, &arg); | |
159 | if (err < 0) | |
160 | return err; | |
161 | *mrt = res.mrt; | |
162 | return 0; | |
163 | } | |
164 | ||
165 | static int ipmr_rule_action(struct fib_rule *rule, struct flowi *flp, | |
166 | int flags, struct fib_lookup_arg *arg) | |
167 | { | |
168 | struct ipmr_result *res = arg->result; | |
169 | struct mr_table *mrt; | |
1da177e4 | 170 | |
f0ad0860 PM |
171 | switch (rule->action) { |
172 | case FR_ACT_TO_TBL: | |
173 | break; | |
174 | case FR_ACT_UNREACHABLE: | |
175 | return -ENETUNREACH; | |
176 | case FR_ACT_PROHIBIT: | |
177 | return -EACCES; | |
178 | case FR_ACT_BLACKHOLE: | |
179 | default: | |
180 | return -EINVAL; | |
181 | } | |
182 | ||
183 | mrt = ipmr_get_table(rule->fr_net, rule->table); | |
184 | if (mrt == NULL) | |
185 | return -EAGAIN; | |
186 | res->mrt = mrt; | |
187 | return 0; | |
188 | } | |
189 | ||
190 | static int ipmr_rule_match(struct fib_rule *rule, struct flowi *fl, int flags) | |
191 | { | |
192 | return 1; | |
193 | } | |
194 | ||
195 | static const struct nla_policy ipmr_rule_policy[FRA_MAX + 1] = { | |
196 | FRA_GENERIC_POLICY, | |
197 | }; | |
198 | ||
199 | static int ipmr_rule_configure(struct fib_rule *rule, struct sk_buff *skb, | |
200 | struct fib_rule_hdr *frh, struct nlattr **tb) | |
201 | { | |
202 | return 0; | |
203 | } | |
204 | ||
205 | static int ipmr_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh, | |
206 | struct nlattr **tb) | |
207 | { | |
208 | return 1; | |
209 | } | |
210 | ||
211 | static int ipmr_rule_fill(struct fib_rule *rule, struct sk_buff *skb, | |
212 | struct fib_rule_hdr *frh) | |
213 | { | |
214 | frh->dst_len = 0; | |
215 | frh->src_len = 0; | |
216 | frh->tos = 0; | |
217 | return 0; | |
218 | } | |
219 | ||
3d0c9c4e | 220 | static const struct fib_rules_ops __net_initdata ipmr_rules_ops_template = { |
25239cee | 221 | .family = RTNL_FAMILY_IPMR, |
f0ad0860 PM |
222 | .rule_size = sizeof(struct ipmr_rule), |
223 | .addr_size = sizeof(u32), | |
224 | .action = ipmr_rule_action, | |
225 | .match = ipmr_rule_match, | |
226 | .configure = ipmr_rule_configure, | |
227 | .compare = ipmr_rule_compare, | |
228 | .default_pref = fib_default_rule_pref, | |
229 | .fill = ipmr_rule_fill, | |
230 | .nlgroup = RTNLGRP_IPV4_RULE, | |
231 | .policy = ipmr_rule_policy, | |
232 | .owner = THIS_MODULE, | |
233 | }; | |
234 | ||
235 | static int __net_init ipmr_rules_init(struct net *net) | |
236 | { | |
237 | struct fib_rules_ops *ops; | |
238 | struct mr_table *mrt; | |
239 | int err; | |
240 | ||
241 | ops = fib_rules_register(&ipmr_rules_ops_template, net); | |
242 | if (IS_ERR(ops)) | |
243 | return PTR_ERR(ops); | |
244 | ||
245 | INIT_LIST_HEAD(&net->ipv4.mr_tables); | |
246 | ||
247 | mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
248 | if (mrt == NULL) { | |
249 | err = -ENOMEM; | |
250 | goto err1; | |
251 | } | |
252 | ||
253 | err = fib_default_rule_add(ops, 0x7fff, RT_TABLE_DEFAULT, 0); | |
254 | if (err < 0) | |
255 | goto err2; | |
256 | ||
257 | net->ipv4.mr_rules_ops = ops; | |
258 | return 0; | |
259 | ||
260 | err2: | |
261 | kfree(mrt); | |
262 | err1: | |
263 | fib_rules_unregister(ops); | |
264 | return err; | |
265 | } | |
266 | ||
267 | static void __net_exit ipmr_rules_exit(struct net *net) | |
268 | { | |
269 | struct mr_table *mrt, *next; | |
270 | ||
035320d5 ED |
271 | list_for_each_entry_safe(mrt, next, &net->ipv4.mr_tables, list) { |
272 | list_del(&mrt->list); | |
f0ad0860 | 273 | kfree(mrt); |
035320d5 | 274 | } |
f0ad0860 PM |
275 | fib_rules_unregister(net->ipv4.mr_rules_ops); |
276 | } | |
277 | #else | |
278 | #define ipmr_for_each_table(mrt, net) \ | |
279 | for (mrt = net->ipv4.mrt; mrt; mrt = NULL) | |
280 | ||
281 | static struct mr_table *ipmr_get_table(struct net *net, u32 id) | |
282 | { | |
283 | return net->ipv4.mrt; | |
284 | } | |
285 | ||
286 | static int ipmr_fib_lookup(struct net *net, struct flowi *flp, | |
287 | struct mr_table **mrt) | |
288 | { | |
289 | *mrt = net->ipv4.mrt; | |
290 | return 0; | |
291 | } | |
292 | ||
293 | static int __net_init ipmr_rules_init(struct net *net) | |
294 | { | |
295 | net->ipv4.mrt = ipmr_new_table(net, RT_TABLE_DEFAULT); | |
296 | return net->ipv4.mrt ? 0 : -ENOMEM; | |
297 | } | |
298 | ||
299 | static void __net_exit ipmr_rules_exit(struct net *net) | |
300 | { | |
301 | kfree(net->ipv4.mrt); | |
302 | } | |
303 | #endif | |
304 | ||
305 | static struct mr_table *ipmr_new_table(struct net *net, u32 id) | |
306 | { | |
307 | struct mr_table *mrt; | |
308 | unsigned int i; | |
1da177e4 | 309 | |
f0ad0860 PM |
310 | mrt = ipmr_get_table(net, id); |
311 | if (mrt != NULL) | |
312 | return mrt; | |
313 | ||
314 | mrt = kzalloc(sizeof(*mrt), GFP_KERNEL); | |
315 | if (mrt == NULL) | |
316 | return NULL; | |
8de53dfb | 317 | write_pnet(&mrt->net, net); |
f0ad0860 PM |
318 | mrt->id = id; |
319 | ||
320 | /* Forwarding cache */ | |
321 | for (i = 0; i < MFC_LINES; i++) | |
322 | INIT_LIST_HEAD(&mrt->mfc_cache_array[i]); | |
323 | ||
324 | INIT_LIST_HEAD(&mrt->mfc_unres_queue); | |
325 | ||
326 | setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process, | |
327 | (unsigned long)mrt); | |
328 | ||
329 | #ifdef CONFIG_IP_PIMSM | |
330 | mrt->mroute_reg_vif_num = -1; | |
331 | #endif | |
332 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
333 | list_add_tail_rcu(&mrt->list, &net->ipv4.mr_tables); | |
334 | #endif | |
335 | return mrt; | |
336 | } | |
1da177e4 LT |
337 | |
338 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ | |
339 | ||
d607032d WC |
340 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
341 | { | |
4feb88e5 BT |
342 | struct net *net = dev_net(dev); |
343 | ||
d607032d WC |
344 | dev_close(dev); |
345 | ||
4feb88e5 | 346 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 347 | if (dev) { |
5bc3eb7e | 348 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 349 | struct ifreq ifr; |
d607032d WC |
350 | struct ip_tunnel_parm p; |
351 | ||
352 | memset(&p, 0, sizeof(p)); | |
353 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
354 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
355 | p.iph.version = 4; | |
356 | p.iph.ihl = 5; | |
357 | p.iph.protocol = IPPROTO_IPIP; | |
358 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
359 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
360 | ||
5bc3eb7e SH |
361 | if (ops->ndo_do_ioctl) { |
362 | mm_segment_t oldfs = get_fs(); | |
363 | ||
364 | set_fs(KERNEL_DS); | |
365 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
366 | set_fs(oldfs); | |
367 | } | |
d607032d WC |
368 | } |
369 | } | |
370 | ||
1da177e4 | 371 | static |
4feb88e5 | 372 | struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
373 | { |
374 | struct net_device *dev; | |
375 | ||
4feb88e5 | 376 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
377 | |
378 | if (dev) { | |
5bc3eb7e | 379 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
380 | int err; |
381 | struct ifreq ifr; | |
1da177e4 LT |
382 | struct ip_tunnel_parm p; |
383 | struct in_device *in_dev; | |
384 | ||
385 | memset(&p, 0, sizeof(p)); | |
386 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
387 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
388 | p.iph.version = 4; | |
389 | p.iph.ihl = 5; | |
390 | p.iph.protocol = IPPROTO_IPIP; | |
391 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 392 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 393 | |
5bc3eb7e SH |
394 | if (ops->ndo_do_ioctl) { |
395 | mm_segment_t oldfs = get_fs(); | |
396 | ||
397 | set_fs(KERNEL_DS); | |
398 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
399 | set_fs(oldfs); | |
a8cb16dd | 400 | } else { |
5bc3eb7e | 401 | err = -EOPNOTSUPP; |
a8cb16dd | 402 | } |
1da177e4 LT |
403 | dev = NULL; |
404 | ||
4feb88e5 BT |
405 | if (err == 0 && |
406 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 LT |
407 | dev->flags |= IFF_MULTICAST; |
408 | ||
e5ed6399 | 409 | in_dev = __in_dev_get_rtnl(dev); |
71e27da9 | 410 | if (in_dev == NULL) |
1da177e4 | 411 | goto failure; |
71e27da9 HX |
412 | |
413 | ipv4_devconf_setall(in_dev); | |
414 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
1da177e4 LT |
415 | |
416 | if (dev_open(dev)) | |
417 | goto failure; | |
7dc00c82 | 418 | dev_hold(dev); |
1da177e4 LT |
419 | } |
420 | } | |
421 | return dev; | |
422 | ||
423 | failure: | |
424 | /* allow the register to be completed before unregistering. */ | |
425 | rtnl_unlock(); | |
426 | rtnl_lock(); | |
427 | ||
428 | unregister_netdevice(dev); | |
429 | return NULL; | |
430 | } | |
431 | ||
432 | #ifdef CONFIG_IP_PIMSM | |
433 | ||
6fef4c0c | 434 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 435 | { |
4feb88e5 | 436 | struct net *net = dev_net(dev); |
f0ad0860 PM |
437 | struct mr_table *mrt; |
438 | struct flowi fl = { | |
439 | .oif = dev->ifindex, | |
440 | .iif = skb->skb_iif, | |
441 | .mark = skb->mark, | |
442 | }; | |
443 | int err; | |
444 | ||
445 | err = ipmr_fib_lookup(net, &fl, &mrt); | |
e40dbc51 BG |
446 | if (err < 0) { |
447 | kfree_skb(skb); | |
f0ad0860 | 448 | return err; |
e40dbc51 | 449 | } |
4feb88e5 | 450 | |
1da177e4 | 451 | read_lock(&mrt_lock); |
cf3677ae PE |
452 | dev->stats.tx_bytes += skb->len; |
453 | dev->stats.tx_packets++; | |
0c12295a | 454 | ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT); |
1da177e4 LT |
455 | read_unlock(&mrt_lock); |
456 | kfree_skb(skb); | |
6ed10654 | 457 | return NETDEV_TX_OK; |
1da177e4 LT |
458 | } |
459 | ||
007c3838 SH |
460 | static const struct net_device_ops reg_vif_netdev_ops = { |
461 | .ndo_start_xmit = reg_vif_xmit, | |
462 | }; | |
463 | ||
1da177e4 LT |
464 | static void reg_vif_setup(struct net_device *dev) |
465 | { | |
466 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 467 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 468 | dev->flags = IFF_NOARP; |
007c3838 | 469 | dev->netdev_ops = ®_vif_netdev_ops, |
1da177e4 | 470 | dev->destructor = free_netdev; |
403dbb97 | 471 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
472 | } |
473 | ||
f0ad0860 | 474 | static struct net_device *ipmr_reg_vif(struct net *net, struct mr_table *mrt) |
1da177e4 LT |
475 | { |
476 | struct net_device *dev; | |
477 | struct in_device *in_dev; | |
f0ad0860 | 478 | char name[IFNAMSIZ]; |
1da177e4 | 479 | |
f0ad0860 PM |
480 | if (mrt->id == RT_TABLE_DEFAULT) |
481 | sprintf(name, "pimreg"); | |
482 | else | |
483 | sprintf(name, "pimreg%u", mrt->id); | |
1da177e4 | 484 | |
f0ad0860 | 485 | dev = alloc_netdev(0, name, reg_vif_setup); |
1da177e4 LT |
486 | |
487 | if (dev == NULL) | |
488 | return NULL; | |
489 | ||
403dbb97 TG |
490 | dev_net_set(dev, net); |
491 | ||
1da177e4 LT |
492 | if (register_netdevice(dev)) { |
493 | free_netdev(dev); | |
494 | return NULL; | |
495 | } | |
496 | dev->iflink = 0; | |
497 | ||
71e27da9 | 498 | rcu_read_lock(); |
a8cb16dd ED |
499 | in_dev = __in_dev_get_rcu(dev); |
500 | if (!in_dev) { | |
71e27da9 | 501 | rcu_read_unlock(); |
1da177e4 | 502 | goto failure; |
71e27da9 | 503 | } |
1da177e4 | 504 | |
71e27da9 HX |
505 | ipv4_devconf_setall(in_dev); |
506 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
507 | rcu_read_unlock(); | |
1da177e4 LT |
508 | |
509 | if (dev_open(dev)) | |
510 | goto failure; | |
511 | ||
7dc00c82 WC |
512 | dev_hold(dev); |
513 | ||
1da177e4 LT |
514 | return dev; |
515 | ||
516 | failure: | |
517 | /* allow the register to be completed before unregistering. */ | |
518 | rtnl_unlock(); | |
519 | rtnl_lock(); | |
520 | ||
521 | unregister_netdevice(dev); | |
522 | return NULL; | |
523 | } | |
524 | #endif | |
525 | ||
526 | /* | |
527 | * Delete a VIF entry | |
7dc00c82 | 528 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 529 | */ |
e905a9ed | 530 | |
0c12295a | 531 | static int vif_delete(struct mr_table *mrt, int vifi, int notify, |
d17fa6fa | 532 | struct list_head *head) |
1da177e4 LT |
533 | { |
534 | struct vif_device *v; | |
535 | struct net_device *dev; | |
536 | struct in_device *in_dev; | |
537 | ||
0c12295a | 538 | if (vifi < 0 || vifi >= mrt->maxvif) |
1da177e4 LT |
539 | return -EADDRNOTAVAIL; |
540 | ||
0c12295a | 541 | v = &mrt->vif_table[vifi]; |
1da177e4 LT |
542 | |
543 | write_lock_bh(&mrt_lock); | |
544 | dev = v->dev; | |
545 | v->dev = NULL; | |
546 | ||
547 | if (!dev) { | |
548 | write_unlock_bh(&mrt_lock); | |
549 | return -EADDRNOTAVAIL; | |
550 | } | |
551 | ||
552 | #ifdef CONFIG_IP_PIMSM | |
0c12295a PM |
553 | if (vifi == mrt->mroute_reg_vif_num) |
554 | mrt->mroute_reg_vif_num = -1; | |
1da177e4 LT |
555 | #endif |
556 | ||
a8cb16dd | 557 | if (vifi + 1 == mrt->maxvif) { |
1da177e4 | 558 | int tmp; |
a8cb16dd ED |
559 | |
560 | for (tmp = vifi - 1; tmp >= 0; tmp--) { | |
0c12295a | 561 | if (VIF_EXISTS(mrt, tmp)) |
1da177e4 LT |
562 | break; |
563 | } | |
0c12295a | 564 | mrt->maxvif = tmp+1; |
1da177e4 LT |
565 | } |
566 | ||
567 | write_unlock_bh(&mrt_lock); | |
568 | ||
569 | dev_set_allmulti(dev, -1); | |
570 | ||
a8cb16dd ED |
571 | in_dev = __in_dev_get_rtnl(dev); |
572 | if (in_dev) { | |
42f811b8 | 573 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
1da177e4 LT |
574 | ip_rt_multicast_event(in_dev); |
575 | } | |
576 | ||
a8cb16dd | 577 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER) && !notify) |
d17fa6fa | 578 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
579 | |
580 | dev_put(dev); | |
581 | return 0; | |
582 | } | |
583 | ||
a8c9486b | 584 | static void ipmr_cache_free_rcu(struct rcu_head *head) |
5c0a66f5 | 585 | { |
a8c9486b ED |
586 | struct mfc_cache *c = container_of(head, struct mfc_cache, rcu); |
587 | ||
5c0a66f5 BT |
588 | kmem_cache_free(mrt_cachep, c); |
589 | } | |
590 | ||
a8c9486b ED |
591 | static inline void ipmr_cache_free(struct mfc_cache *c) |
592 | { | |
593 | call_rcu(&c->rcu, ipmr_cache_free_rcu); | |
594 | } | |
595 | ||
1da177e4 | 596 | /* Destroy an unresolved cache entry, killing queued skbs |
a8cb16dd | 597 | * and reporting error to netlink readers. |
1da177e4 LT |
598 | */ |
599 | ||
0c12295a | 600 | static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c) |
1da177e4 | 601 | { |
8de53dfb | 602 | struct net *net = read_pnet(&mrt->net); |
1da177e4 | 603 | struct sk_buff *skb; |
9ef1d4c7 | 604 | struct nlmsgerr *e; |
1da177e4 | 605 | |
0c12295a | 606 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 | 607 | |
c354e124 | 608 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 609 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
610 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
611 | nlh->nlmsg_type = NLMSG_ERROR; | |
612 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
613 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
614 | e = NLMSG_DATA(nlh); |
615 | e->error = -ETIMEDOUT; | |
616 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 617 | |
4feb88e5 | 618 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
a8cb16dd | 619 | } else { |
1da177e4 | 620 | kfree_skb(skb); |
a8cb16dd | 621 | } |
1da177e4 LT |
622 | } |
623 | ||
5c0a66f5 | 624 | ipmr_cache_free(c); |
1da177e4 LT |
625 | } |
626 | ||
627 | ||
e258beb2 | 628 | /* Timer process for the unresolved queue. */ |
1da177e4 | 629 | |
e258beb2 | 630 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 631 | { |
0c12295a | 632 | struct mr_table *mrt = (struct mr_table *)arg; |
1da177e4 LT |
633 | unsigned long now; |
634 | unsigned long expires; | |
862465f2 | 635 | struct mfc_cache *c, *next; |
1da177e4 LT |
636 | |
637 | if (!spin_trylock(&mfc_unres_lock)) { | |
0c12295a | 638 | mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
639 | return; |
640 | } | |
641 | ||
0c12295a | 642 | if (list_empty(&mrt->mfc_unres_queue)) |
1da177e4 LT |
643 | goto out; |
644 | ||
645 | now = jiffies; | |
646 | expires = 10*HZ; | |
1da177e4 | 647 | |
0c12295a | 648 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
1da177e4 LT |
649 | if (time_after(c->mfc_un.unres.expires, now)) { |
650 | unsigned long interval = c->mfc_un.unres.expires - now; | |
651 | if (interval < expires) | |
652 | expires = interval; | |
1da177e4 LT |
653 | continue; |
654 | } | |
655 | ||
862465f2 | 656 | list_del(&c->list); |
0c12295a | 657 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
658 | } |
659 | ||
0c12295a PM |
660 | if (!list_empty(&mrt->mfc_unres_queue)) |
661 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
662 | |
663 | out: | |
664 | spin_unlock(&mfc_unres_lock); | |
665 | } | |
666 | ||
667 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
668 | ||
0c12295a | 669 | static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache, |
d658f8a0 | 670 | unsigned char *ttls) |
1da177e4 LT |
671 | { |
672 | int vifi; | |
673 | ||
674 | cache->mfc_un.res.minvif = MAXVIFS; | |
675 | cache->mfc_un.res.maxvif = 0; | |
676 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
677 | ||
0c12295a PM |
678 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
679 | if (VIF_EXISTS(mrt, vifi) && | |
cf958ae3 | 680 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
681 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
682 | if (cache->mfc_un.res.minvif > vifi) | |
683 | cache->mfc_un.res.minvif = vifi; | |
684 | if (cache->mfc_un.res.maxvif <= vifi) | |
685 | cache->mfc_un.res.maxvif = vifi + 1; | |
686 | } | |
687 | } | |
688 | } | |
689 | ||
0c12295a PM |
690 | static int vif_add(struct net *net, struct mr_table *mrt, |
691 | struct vifctl *vifc, int mrtsock) | |
1da177e4 LT |
692 | { |
693 | int vifi = vifc->vifc_vifi; | |
0c12295a | 694 | struct vif_device *v = &mrt->vif_table[vifi]; |
1da177e4 LT |
695 | struct net_device *dev; |
696 | struct in_device *in_dev; | |
d607032d | 697 | int err; |
1da177e4 LT |
698 | |
699 | /* Is vif busy ? */ | |
0c12295a | 700 | if (VIF_EXISTS(mrt, vifi)) |
1da177e4 LT |
701 | return -EADDRINUSE; |
702 | ||
703 | switch (vifc->vifc_flags) { | |
704 | #ifdef CONFIG_IP_PIMSM | |
705 | case VIFF_REGISTER: | |
706 | /* | |
707 | * Special Purpose VIF in PIM | |
708 | * All the packets will be sent to the daemon | |
709 | */ | |
0c12295a | 710 | if (mrt->mroute_reg_vif_num >= 0) |
1da177e4 | 711 | return -EADDRINUSE; |
f0ad0860 | 712 | dev = ipmr_reg_vif(net, mrt); |
1da177e4 LT |
713 | if (!dev) |
714 | return -ENOBUFS; | |
d607032d WC |
715 | err = dev_set_allmulti(dev, 1); |
716 | if (err) { | |
717 | unregister_netdevice(dev); | |
7dc00c82 | 718 | dev_put(dev); |
d607032d WC |
719 | return err; |
720 | } | |
1da177e4 LT |
721 | break; |
722 | #endif | |
e905a9ed | 723 | case VIFF_TUNNEL: |
4feb88e5 | 724 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
725 | if (!dev) |
726 | return -ENOBUFS; | |
d607032d WC |
727 | err = dev_set_allmulti(dev, 1); |
728 | if (err) { | |
729 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 730 | dev_put(dev); |
d607032d WC |
731 | return err; |
732 | } | |
1da177e4 | 733 | break; |
ee5e81f0 I |
734 | |
735 | case VIFF_USE_IFINDEX: | |
1da177e4 | 736 | case 0: |
ee5e81f0 I |
737 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
738 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
95ae6b22 | 739 | if (dev && __in_dev_get_rtnl(dev) == NULL) { |
ee5e81f0 I |
740 | dev_put(dev); |
741 | return -EADDRNOTAVAIL; | |
742 | } | |
a8cb16dd | 743 | } else { |
ee5e81f0 | 744 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); |
a8cb16dd | 745 | } |
1da177e4 LT |
746 | if (!dev) |
747 | return -EADDRNOTAVAIL; | |
d607032d | 748 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
749 | if (err) { |
750 | dev_put(dev); | |
d607032d | 751 | return err; |
7dc00c82 | 752 | } |
1da177e4 LT |
753 | break; |
754 | default: | |
755 | return -EINVAL; | |
756 | } | |
757 | ||
a8cb16dd ED |
758 | in_dev = __in_dev_get_rtnl(dev); |
759 | if (!in_dev) { | |
d0490cfd | 760 | dev_put(dev); |
1da177e4 | 761 | return -EADDRNOTAVAIL; |
d0490cfd | 762 | } |
42f811b8 | 763 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
1da177e4 LT |
764 | ip_rt_multicast_event(in_dev); |
765 | ||
a8cb16dd ED |
766 | /* Fill in the VIF structures */ |
767 | ||
c354e124 JK |
768 | v->rate_limit = vifc->vifc_rate_limit; |
769 | v->local = vifc->vifc_lcl_addr.s_addr; | |
770 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
771 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
772 | if (!mrtsock) |
773 | v->flags |= VIFF_STATIC; | |
c354e124 | 774 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
775 | v->bytes_in = 0; |
776 | v->bytes_out = 0; | |
777 | v->pkt_in = 0; | |
778 | v->pkt_out = 0; | |
779 | v->link = dev->ifindex; | |
a8cb16dd | 780 | if (v->flags & (VIFF_TUNNEL | VIFF_REGISTER)) |
1da177e4 LT |
781 | v->link = dev->iflink; |
782 | ||
783 | /* And finish update writing critical data */ | |
784 | write_lock_bh(&mrt_lock); | |
c354e124 | 785 | v->dev = dev; |
1da177e4 | 786 | #ifdef CONFIG_IP_PIMSM |
a8cb16dd | 787 | if (v->flags & VIFF_REGISTER) |
0c12295a | 788 | mrt->mroute_reg_vif_num = vifi; |
1da177e4 | 789 | #endif |
0c12295a PM |
790 | if (vifi+1 > mrt->maxvif) |
791 | mrt->maxvif = vifi+1; | |
1da177e4 LT |
792 | write_unlock_bh(&mrt_lock); |
793 | return 0; | |
794 | } | |
795 | ||
a8c9486b | 796 | /* called with rcu_read_lock() */ |
0c12295a | 797 | static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt, |
4feb88e5 BT |
798 | __be32 origin, |
799 | __be32 mcastgrp) | |
1da177e4 | 800 | { |
c354e124 | 801 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
802 | struct mfc_cache *c; |
803 | ||
a8c9486b | 804 | list_for_each_entry_rcu(c, &mrt->mfc_cache_array[line], list) { |
862465f2 PM |
805 | if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp) |
806 | return c; | |
1da177e4 | 807 | } |
862465f2 | 808 | return NULL; |
1da177e4 LT |
809 | } |
810 | ||
811 | /* | |
812 | * Allocate a multicast cache entry | |
813 | */ | |
d658f8a0 | 814 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 815 | { |
c354e124 | 816 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
a8c9486b ED |
817 | |
818 | if (c) | |
819 | c->mfc_un.res.minvif = MAXVIFS; | |
1da177e4 LT |
820 | return c; |
821 | } | |
822 | ||
d658f8a0 | 823 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 824 | { |
c354e124 | 825 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
a8c9486b ED |
826 | |
827 | if (c) { | |
828 | skb_queue_head_init(&c->mfc_un.unres.unresolved); | |
829 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
830 | } | |
1da177e4 LT |
831 | return c; |
832 | } | |
833 | ||
834 | /* | |
835 | * A cache entry has gone into a resolved state from queued | |
836 | */ | |
e905a9ed | 837 | |
0c12295a PM |
838 | static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt, |
839 | struct mfc_cache *uc, struct mfc_cache *c) | |
1da177e4 LT |
840 | { |
841 | struct sk_buff *skb; | |
9ef1d4c7 | 842 | struct nlmsgerr *e; |
1da177e4 | 843 | |
a8cb16dd | 844 | /* Play the pending entries through our router */ |
1da177e4 | 845 | |
c354e124 | 846 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 847 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
848 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
849 | ||
cb6a4e46 | 850 | if (__ipmr_fill_mroute(mrt, skb, c, NLMSG_DATA(nlh)) > 0) { |
a8cb16dd ED |
851 | nlh->nlmsg_len = skb_tail_pointer(skb) - |
852 | (u8 *)nlh; | |
1da177e4 LT |
853 | } else { |
854 | nlh->nlmsg_type = NLMSG_ERROR; | |
855 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
856 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
857 | e = NLMSG_DATA(nlh); |
858 | e->error = -EMSGSIZE; | |
859 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 860 | } |
2942e900 | 861 | |
d658f8a0 | 862 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
a8cb16dd | 863 | } else { |
0c12295a | 864 | ip_mr_forward(net, mrt, skb, c, 0); |
a8cb16dd | 865 | } |
1da177e4 LT |
866 | } |
867 | } | |
868 | ||
869 | /* | |
870 | * Bounce a cache query up to mrouted. We could use netlink for this but mrouted | |
871 | * expects the following bizarre scheme. | |
872 | * | |
873 | * Called under mrt_lock. | |
874 | */ | |
e905a9ed | 875 | |
0c12295a | 876 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 877 | struct sk_buff *pkt, vifi_t vifi, int assert) |
1da177e4 LT |
878 | { |
879 | struct sk_buff *skb; | |
c9bdd4b5 | 880 | const int ihl = ip_hdrlen(pkt); |
1da177e4 LT |
881 | struct igmphdr *igmp; |
882 | struct igmpmsg *msg; | |
4c968709 | 883 | struct sock *mroute_sk; |
1da177e4 LT |
884 | int ret; |
885 | ||
886 | #ifdef CONFIG_IP_PIMSM | |
887 | if (assert == IGMPMSG_WHOLEPKT) | |
888 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
889 | else | |
890 | #endif | |
891 | skb = alloc_skb(128, GFP_ATOMIC); | |
892 | ||
132adf54 | 893 | if (!skb) |
1da177e4 LT |
894 | return -ENOBUFS; |
895 | ||
896 | #ifdef CONFIG_IP_PIMSM | |
897 | if (assert == IGMPMSG_WHOLEPKT) { | |
898 | /* Ugly, but we have no choice with this interface. | |
a8cb16dd ED |
899 | * Duplicate old header, fix ihl, length etc. |
900 | * And all this only to mangle msg->im_msgtype and | |
901 | * to set msg->im_mbz to "mbz" :-) | |
1da177e4 | 902 | */ |
878c8145 ACM |
903 | skb_push(skb, sizeof(struct iphdr)); |
904 | skb_reset_network_header(skb); | |
badff6d0 | 905 | skb_reset_transport_header(skb); |
0272ffc4 | 906 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 907 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
908 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
909 | msg->im_mbz = 0; | |
0c12295a | 910 | msg->im_vif = mrt->mroute_reg_vif_num; |
eddc9ec5 ACM |
911 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
912 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
913 | sizeof(struct iphdr)); | |
e905a9ed | 914 | } else |
1da177e4 | 915 | #endif |
e905a9ed YH |
916 | { |
917 | ||
a8cb16dd | 918 | /* Copy the IP header */ |
1da177e4 | 919 | |
27a884dc | 920 | skb->network_header = skb->tail; |
ddc7b8e3 | 921 | skb_put(skb, ihl); |
27d7ff46 | 922 | skb_copy_to_linear_data(skb, pkt->data, ihl); |
a8cb16dd | 923 | ip_hdr(skb)->protocol = 0; /* Flag to the kernel this is a route add */ |
eddc9ec5 | 924 | msg = (struct igmpmsg *)skb_network_header(skb); |
1da177e4 | 925 | msg->im_vif = vifi; |
adf30907 | 926 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); |
1da177e4 | 927 | |
a8cb16dd | 928 | /* Add our header */ |
1da177e4 | 929 | |
a8cb16dd | 930 | igmp = (struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); |
1da177e4 LT |
931 | igmp->type = |
932 | msg->im_msgtype = assert; | |
a8cb16dd ED |
933 | igmp->code = 0; |
934 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ | |
b0e380b1 | 935 | skb->transport_header = skb->network_header; |
e905a9ed | 936 | } |
1da177e4 | 937 | |
4c968709 ED |
938 | rcu_read_lock(); |
939 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
940 | if (mroute_sk == NULL) { | |
941 | rcu_read_unlock(); | |
1da177e4 LT |
942 | kfree_skb(skb); |
943 | return -EINVAL; | |
944 | } | |
945 | ||
a8cb16dd ED |
946 | /* Deliver to mrouted */ |
947 | ||
4c968709 ED |
948 | ret = sock_queue_rcv_skb(mroute_sk, skb); |
949 | rcu_read_unlock(); | |
70a269e6 | 950 | if (ret < 0) { |
1da177e4 LT |
951 | if (net_ratelimit()) |
952 | printk(KERN_WARNING "mroute: pending queue full, dropping entries.\n"); | |
953 | kfree_skb(skb); | |
954 | } | |
955 | ||
956 | return ret; | |
957 | } | |
958 | ||
959 | /* | |
960 | * Queue a packet for resolution. It gets locked cache entry! | |
961 | */ | |
e905a9ed | 962 | |
1da177e4 | 963 | static int |
0c12295a | 964 | ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, struct sk_buff *skb) |
1da177e4 | 965 | { |
862465f2 | 966 | bool found = false; |
1da177e4 LT |
967 | int err; |
968 | struct mfc_cache *c; | |
eddc9ec5 | 969 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
970 | |
971 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 972 | list_for_each_entry(c, &mrt->mfc_unres_queue, list) { |
e258beb2 | 973 | if (c->mfc_mcastgrp == iph->daddr && |
862465f2 PM |
974 | c->mfc_origin == iph->saddr) { |
975 | found = true; | |
1da177e4 | 976 | break; |
862465f2 | 977 | } |
1da177e4 LT |
978 | } |
979 | ||
862465f2 | 980 | if (!found) { |
a8cb16dd | 981 | /* Create a new entry if allowable */ |
1da177e4 | 982 | |
0c12295a | 983 | if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 || |
d658f8a0 | 984 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
985 | spin_unlock_bh(&mfc_unres_lock); |
986 | ||
987 | kfree_skb(skb); | |
988 | return -ENOBUFS; | |
989 | } | |
990 | ||
a8cb16dd ED |
991 | /* Fill in the new cache entry */ |
992 | ||
eddc9ec5 ACM |
993 | c->mfc_parent = -1; |
994 | c->mfc_origin = iph->saddr; | |
995 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 | 996 | |
a8cb16dd ED |
997 | /* Reflect first query at mrouted. */ |
998 | ||
0c12295a | 999 | err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE); |
4feb88e5 | 1000 | if (err < 0) { |
e905a9ed | 1001 | /* If the report failed throw the cache entry |
1da177e4 LT |
1002 | out - Brad Parker |
1003 | */ | |
1004 | spin_unlock_bh(&mfc_unres_lock); | |
1005 | ||
5c0a66f5 | 1006 | ipmr_cache_free(c); |
1da177e4 LT |
1007 | kfree_skb(skb); |
1008 | return err; | |
1009 | } | |
1010 | ||
0c12295a PM |
1011 | atomic_inc(&mrt->cache_resolve_queue_len); |
1012 | list_add(&c->list, &mrt->mfc_unres_queue); | |
1da177e4 | 1013 | |
278554bd DM |
1014 | if (atomic_read(&mrt->cache_resolve_queue_len) == 1) |
1015 | mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires); | |
1da177e4 LT |
1016 | } |
1017 | ||
a8cb16dd ED |
1018 | /* See if we can append the packet */ |
1019 | ||
1020 | if (c->mfc_un.unres.unresolved.qlen > 3) { | |
1da177e4 LT |
1021 | kfree_skb(skb); |
1022 | err = -ENOBUFS; | |
1023 | } else { | |
c354e124 | 1024 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
1025 | err = 0; |
1026 | } | |
1027 | ||
1028 | spin_unlock_bh(&mfc_unres_lock); | |
1029 | return err; | |
1030 | } | |
1031 | ||
1032 | /* | |
1033 | * MFC cache manipulation by user space mroute daemon | |
1034 | */ | |
1035 | ||
0c12295a | 1036 | static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc) |
1da177e4 LT |
1037 | { |
1038 | int line; | |
862465f2 | 1039 | struct mfc_cache *c, *next; |
1da177e4 | 1040 | |
c354e124 | 1041 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1042 | |
0c12295a | 1043 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) { |
1da177e4 LT |
1044 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
1045 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { | |
a8c9486b | 1046 | list_del_rcu(&c->list); |
1da177e4 | 1047 | |
5c0a66f5 | 1048 | ipmr_cache_free(c); |
1da177e4 LT |
1049 | return 0; |
1050 | } | |
1051 | } | |
1052 | return -ENOENT; | |
1053 | } | |
1054 | ||
0c12295a PM |
1055 | static int ipmr_mfc_add(struct net *net, struct mr_table *mrt, |
1056 | struct mfcctl *mfc, int mrtsock) | |
1da177e4 | 1057 | { |
862465f2 | 1058 | bool found = false; |
1da177e4 | 1059 | int line; |
862465f2 | 1060 | struct mfc_cache *uc, *c; |
1da177e4 | 1061 | |
a50436f2 PM |
1062 | if (mfc->mfcc_parent >= MAXVIFS) |
1063 | return -ENFILE; | |
1064 | ||
c354e124 | 1065 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 1066 | |
0c12295a | 1067 | list_for_each_entry(c, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 1068 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
862465f2 PM |
1069 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { |
1070 | found = true; | |
1da177e4 | 1071 | break; |
862465f2 | 1072 | } |
1da177e4 LT |
1073 | } |
1074 | ||
862465f2 | 1075 | if (found) { |
1da177e4 LT |
1076 | write_lock_bh(&mrt_lock); |
1077 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1078 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1079 | if (!mrtsock) |
1080 | c->mfc_flags |= MFC_STATIC; | |
1081 | write_unlock_bh(&mrt_lock); | |
1082 | return 0; | |
1083 | } | |
1084 | ||
f97c1e0c | 1085 | if (!ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
1086 | return -EINVAL; |
1087 | ||
d658f8a0 | 1088 | c = ipmr_cache_alloc(); |
c354e124 | 1089 | if (c == NULL) |
1da177e4 LT |
1090 | return -ENOMEM; |
1091 | ||
c354e124 JK |
1092 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
1093 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
1094 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 1095 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
1096 | if (!mrtsock) |
1097 | c->mfc_flags |= MFC_STATIC; | |
1098 | ||
a8c9486b | 1099 | list_add_rcu(&c->list, &mrt->mfc_cache_array[line]); |
1da177e4 LT |
1100 | |
1101 | /* | |
1102 | * Check to see if we resolved a queued list. If so we | |
1103 | * need to send on the frames and tidy up. | |
1104 | */ | |
b0ebb739 | 1105 | found = false; |
1da177e4 | 1106 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1107 | list_for_each_entry(uc, &mrt->mfc_unres_queue, list) { |
e258beb2 | 1108 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 | 1109 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
862465f2 | 1110 | list_del(&uc->list); |
0c12295a | 1111 | atomic_dec(&mrt->cache_resolve_queue_len); |
b0ebb739 | 1112 | found = true; |
1da177e4 LT |
1113 | break; |
1114 | } | |
1115 | } | |
0c12295a PM |
1116 | if (list_empty(&mrt->mfc_unres_queue)) |
1117 | del_timer(&mrt->ipmr_expire_timer); | |
1da177e4 LT |
1118 | spin_unlock_bh(&mfc_unres_lock); |
1119 | ||
b0ebb739 | 1120 | if (found) { |
0c12295a | 1121 | ipmr_cache_resolve(net, mrt, uc, c); |
5c0a66f5 | 1122 | ipmr_cache_free(uc); |
1da177e4 LT |
1123 | } |
1124 | return 0; | |
1125 | } | |
1126 | ||
1127 | /* | |
1128 | * Close the multicast socket, and clear the vif tables etc | |
1129 | */ | |
e905a9ed | 1130 | |
0c12295a | 1131 | static void mroute_clean_tables(struct mr_table *mrt) |
1da177e4 LT |
1132 | { |
1133 | int i; | |
d17fa6fa | 1134 | LIST_HEAD(list); |
862465f2 | 1135 | struct mfc_cache *c, *next; |
e905a9ed | 1136 | |
a8cb16dd ED |
1137 | /* Shut down all active vif entries */ |
1138 | ||
0c12295a | 1139 | for (i = 0; i < mrt->maxvif; i++) { |
a8cb16dd | 1140 | if (!(mrt->vif_table[i].flags & VIFF_STATIC)) |
0c12295a | 1141 | vif_delete(mrt, i, 0, &list); |
1da177e4 | 1142 | } |
d17fa6fa | 1143 | unregister_netdevice_many(&list); |
1da177e4 | 1144 | |
a8cb16dd ED |
1145 | /* Wipe the cache */ |
1146 | ||
862465f2 | 1147 | for (i = 0; i < MFC_LINES; i++) { |
0c12295a | 1148 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) { |
a8c9486b | 1149 | if (c->mfc_flags & MFC_STATIC) |
1da177e4 | 1150 | continue; |
a8c9486b | 1151 | list_del_rcu(&c->list); |
5c0a66f5 | 1152 | ipmr_cache_free(c); |
1da177e4 LT |
1153 | } |
1154 | } | |
1155 | ||
0c12295a | 1156 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { |
1da177e4 | 1157 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1158 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
862465f2 | 1159 | list_del(&c->list); |
0c12295a | 1160 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
1161 | } |
1162 | spin_unlock_bh(&mfc_unres_lock); | |
1163 | } | |
1164 | } | |
1165 | ||
4c968709 ED |
1166 | /* called from ip_ra_control(), before an RCU grace period, |
1167 | * we dont need to call synchronize_rcu() here | |
1168 | */ | |
1da177e4 LT |
1169 | static void mrtsock_destruct(struct sock *sk) |
1170 | { | |
4feb88e5 | 1171 | struct net *net = sock_net(sk); |
f0ad0860 | 1172 | struct mr_table *mrt; |
4feb88e5 | 1173 | |
1da177e4 | 1174 | rtnl_lock(); |
f0ad0860 | 1175 | ipmr_for_each_table(mrt, net) { |
4c968709 | 1176 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
f0ad0860 | 1177 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; |
4c968709 | 1178 | rcu_assign_pointer(mrt->mroute_sk, NULL); |
f0ad0860 PM |
1179 | mroute_clean_tables(mrt); |
1180 | } | |
1da177e4 LT |
1181 | } |
1182 | rtnl_unlock(); | |
1183 | } | |
1184 | ||
1185 | /* | |
1186 | * Socket options and virtual interface manipulation. The whole | |
1187 | * virtual interface system is a complete heap, but unfortunately | |
1188 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
1189 | * MOSPF/PIM router set up we can clean this up. | |
1190 | */ | |
e905a9ed | 1191 | |
b7058842 | 1192 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen) |
1da177e4 LT |
1193 | { |
1194 | int ret; | |
1195 | struct vifctl vif; | |
1196 | struct mfcctl mfc; | |
4feb88e5 | 1197 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1198 | struct mr_table *mrt; |
1199 | ||
1200 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
1201 | if (mrt == NULL) | |
1202 | return -ENOENT; | |
e905a9ed | 1203 | |
132adf54 | 1204 | if (optname != MRT_INIT) { |
4c968709 ED |
1205 | if (sk != rcu_dereference_raw(mrt->mroute_sk) && |
1206 | !capable(CAP_NET_ADMIN)) | |
1da177e4 LT |
1207 | return -EACCES; |
1208 | } | |
1209 | ||
132adf54 SH |
1210 | switch (optname) { |
1211 | case MRT_INIT: | |
1212 | if (sk->sk_type != SOCK_RAW || | |
c720c7e8 | 1213 | inet_sk(sk)->inet_num != IPPROTO_IGMP) |
132adf54 | 1214 | return -EOPNOTSUPP; |
c354e124 | 1215 | if (optlen != sizeof(int)) |
132adf54 | 1216 | return -ENOPROTOOPT; |
1da177e4 | 1217 | |
132adf54 | 1218 | rtnl_lock(); |
4c968709 | 1219 | if (rtnl_dereference(mrt->mroute_sk)) { |
1da177e4 | 1220 | rtnl_unlock(); |
132adf54 SH |
1221 | return -EADDRINUSE; |
1222 | } | |
1223 | ||
1224 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
1225 | if (ret == 0) { | |
4c968709 | 1226 | rcu_assign_pointer(mrt->mroute_sk, sk); |
4feb88e5 | 1227 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
132adf54 SH |
1228 | } |
1229 | rtnl_unlock(); | |
1230 | return ret; | |
1231 | case MRT_DONE: | |
4c968709 | 1232 | if (sk != rcu_dereference_raw(mrt->mroute_sk)) |
132adf54 SH |
1233 | return -EACCES; |
1234 | return ip_ra_control(sk, 0, NULL); | |
1235 | case MRT_ADD_VIF: | |
1236 | case MRT_DEL_VIF: | |
c354e124 | 1237 | if (optlen != sizeof(vif)) |
132adf54 | 1238 | return -EINVAL; |
c354e124 | 1239 | if (copy_from_user(&vif, optval, sizeof(vif))) |
132adf54 SH |
1240 | return -EFAULT; |
1241 | if (vif.vifc_vifi >= MAXVIFS) | |
1242 | return -ENFILE; | |
1243 | rtnl_lock(); | |
c354e124 | 1244 | if (optname == MRT_ADD_VIF) { |
4c968709 ED |
1245 | ret = vif_add(net, mrt, &vif, |
1246 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 | 1247 | } else { |
0c12295a | 1248 | ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL); |
132adf54 SH |
1249 | } |
1250 | rtnl_unlock(); | |
1251 | return ret; | |
1da177e4 LT |
1252 | |
1253 | /* | |
1254 | * Manipulate the forwarding caches. These live | |
1255 | * in a sort of kernel/user symbiosis. | |
1256 | */ | |
132adf54 SH |
1257 | case MRT_ADD_MFC: |
1258 | case MRT_DEL_MFC: | |
c354e124 | 1259 | if (optlen != sizeof(mfc)) |
132adf54 | 1260 | return -EINVAL; |
c354e124 | 1261 | if (copy_from_user(&mfc, optval, sizeof(mfc))) |
132adf54 SH |
1262 | return -EFAULT; |
1263 | rtnl_lock(); | |
c354e124 | 1264 | if (optname == MRT_DEL_MFC) |
0c12295a | 1265 | ret = ipmr_mfc_delete(mrt, &mfc); |
132adf54 | 1266 | else |
4c968709 ED |
1267 | ret = ipmr_mfc_add(net, mrt, &mfc, |
1268 | sk == rtnl_dereference(mrt->mroute_sk)); | |
132adf54 SH |
1269 | rtnl_unlock(); |
1270 | return ret; | |
1da177e4 LT |
1271 | /* |
1272 | * Control PIM assert. | |
1273 | */ | |
132adf54 SH |
1274 | case MRT_ASSERT: |
1275 | { | |
1276 | int v; | |
a8cb16dd | 1277 | if (get_user(v, (int __user *)optval)) |
132adf54 | 1278 | return -EFAULT; |
0c12295a | 1279 | mrt->mroute_do_assert = (v) ? 1 : 0; |
132adf54 SH |
1280 | return 0; |
1281 | } | |
1da177e4 | 1282 | #ifdef CONFIG_IP_PIMSM |
132adf54 SH |
1283 | case MRT_PIM: |
1284 | { | |
ba93ef74 SH |
1285 | int v; |
1286 | ||
a8cb16dd | 1287 | if (get_user(v, (int __user *)optval)) |
132adf54 | 1288 | return -EFAULT; |
ba93ef74 SH |
1289 | v = (v) ? 1 : 0; |
1290 | ||
132adf54 SH |
1291 | rtnl_lock(); |
1292 | ret = 0; | |
0c12295a PM |
1293 | if (v != mrt->mroute_do_pim) { |
1294 | mrt->mroute_do_pim = v; | |
1295 | mrt->mroute_do_assert = v; | |
1da177e4 | 1296 | } |
132adf54 SH |
1297 | rtnl_unlock(); |
1298 | return ret; | |
1299 | } | |
f0ad0860 PM |
1300 | #endif |
1301 | #ifdef CONFIG_IP_MROUTE_MULTIPLE_TABLES | |
1302 | case MRT_TABLE: | |
1303 | { | |
1304 | u32 v; | |
1305 | ||
1306 | if (optlen != sizeof(u32)) | |
1307 | return -EINVAL; | |
1308 | if (get_user(v, (u32 __user *)optval)) | |
1309 | return -EFAULT; | |
f0ad0860 PM |
1310 | |
1311 | rtnl_lock(); | |
1312 | ret = 0; | |
4c968709 ED |
1313 | if (sk == rtnl_dereference(mrt->mroute_sk)) { |
1314 | ret = -EBUSY; | |
1315 | } else { | |
1316 | if (!ipmr_new_table(net, v)) | |
1317 | ret = -ENOMEM; | |
1318 | raw_sk(sk)->ipmr_table = v; | |
1319 | } | |
f0ad0860 PM |
1320 | rtnl_unlock(); |
1321 | return ret; | |
1322 | } | |
1da177e4 | 1323 | #endif |
132adf54 SH |
1324 | /* |
1325 | * Spurious command, or MRT_VERSION which you cannot | |
1326 | * set. | |
1327 | */ | |
1328 | default: | |
1329 | return -ENOPROTOOPT; | |
1da177e4 LT |
1330 | } |
1331 | } | |
1332 | ||
1333 | /* | |
1334 | * Getsock opt support for the multicast routing system. | |
1335 | */ | |
e905a9ed | 1336 | |
c354e124 | 1337 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1338 | { |
1339 | int olr; | |
1340 | int val; | |
4feb88e5 | 1341 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1342 | struct mr_table *mrt; |
1343 | ||
1344 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
1345 | if (mrt == NULL) | |
1346 | return -ENOENT; | |
1da177e4 | 1347 | |
c354e124 | 1348 | if (optname != MRT_VERSION && |
1da177e4 | 1349 | #ifdef CONFIG_IP_PIMSM |
a8cb16dd | 1350 | optname != MRT_PIM && |
1da177e4 | 1351 | #endif |
a8cb16dd | 1352 | optname != MRT_ASSERT) |
1da177e4 LT |
1353 | return -ENOPROTOOPT; |
1354 | ||
1355 | if (get_user(olr, optlen)) | |
1356 | return -EFAULT; | |
1357 | ||
1358 | olr = min_t(unsigned int, olr, sizeof(int)); | |
1359 | if (olr < 0) | |
1360 | return -EINVAL; | |
e905a9ed | 1361 | |
c354e124 | 1362 | if (put_user(olr, optlen)) |
1da177e4 | 1363 | return -EFAULT; |
c354e124 JK |
1364 | if (optname == MRT_VERSION) |
1365 | val = 0x0305; | |
1da177e4 | 1366 | #ifdef CONFIG_IP_PIMSM |
c354e124 | 1367 | else if (optname == MRT_PIM) |
0c12295a | 1368 | val = mrt->mroute_do_pim; |
1da177e4 LT |
1369 | #endif |
1370 | else | |
0c12295a | 1371 | val = mrt->mroute_do_assert; |
c354e124 | 1372 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1373 | return -EFAULT; |
1374 | return 0; | |
1375 | } | |
1376 | ||
1377 | /* | |
1378 | * The IP multicast ioctl support routines. | |
1379 | */ | |
e905a9ed | 1380 | |
1da177e4 LT |
1381 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1382 | { | |
1383 | struct sioc_sg_req sr; | |
1384 | struct sioc_vif_req vr; | |
1385 | struct vif_device *vif; | |
1386 | struct mfc_cache *c; | |
4feb88e5 | 1387 | struct net *net = sock_net(sk); |
f0ad0860 PM |
1388 | struct mr_table *mrt; |
1389 | ||
1390 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
1391 | if (mrt == NULL) | |
1392 | return -ENOENT; | |
e905a9ed | 1393 | |
132adf54 SH |
1394 | switch (cmd) { |
1395 | case SIOCGETVIFCNT: | |
c354e124 | 1396 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1397 | return -EFAULT; |
0c12295a | 1398 | if (vr.vifi >= mrt->maxvif) |
132adf54 SH |
1399 | return -EINVAL; |
1400 | read_lock(&mrt_lock); | |
0c12295a PM |
1401 | vif = &mrt->vif_table[vr.vifi]; |
1402 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
c354e124 JK |
1403 | vr.icount = vif->pkt_in; |
1404 | vr.ocount = vif->pkt_out; | |
1405 | vr.ibytes = vif->bytes_in; | |
1406 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1407 | read_unlock(&mrt_lock); |
1da177e4 | 1408 | |
c354e124 | 1409 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1410 | return -EFAULT; |
1411 | return 0; | |
1412 | } | |
1413 | read_unlock(&mrt_lock); | |
1414 | return -EADDRNOTAVAIL; | |
1415 | case SIOCGETSGCNT: | |
c354e124 | 1416 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1417 | return -EFAULT; |
1418 | ||
a8c9486b | 1419 | rcu_read_lock(); |
0c12295a | 1420 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1421 | if (c) { |
1422 | sr.pktcnt = c->mfc_un.res.pkt; | |
1423 | sr.bytecnt = c->mfc_un.res.bytes; | |
1424 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
a8c9486b | 1425 | rcu_read_unlock(); |
132adf54 | 1426 | |
c354e124 | 1427 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1428 | return -EFAULT; |
1429 | return 0; | |
1430 | } | |
a8c9486b | 1431 | rcu_read_unlock(); |
132adf54 SH |
1432 | return -EADDRNOTAVAIL; |
1433 | default: | |
1434 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1435 | } |
1436 | } | |
1437 | ||
709b46e8 EB |
1438 | #ifdef CONFIG_COMPAT |
1439 | struct compat_sioc_sg_req { | |
1440 | struct in_addr src; | |
1441 | struct in_addr grp; | |
1442 | compat_ulong_t pktcnt; | |
1443 | compat_ulong_t bytecnt; | |
1444 | compat_ulong_t wrong_if; | |
1445 | }; | |
1446 | ||
ca6b8bb0 DM |
1447 | struct compat_sioc_vif_req { |
1448 | vifi_t vifi; /* Which iface */ | |
1449 | compat_ulong_t icount; | |
1450 | compat_ulong_t ocount; | |
1451 | compat_ulong_t ibytes; | |
1452 | compat_ulong_t obytes; | |
1453 | }; | |
1454 | ||
709b46e8 EB |
1455 | int ipmr_compat_ioctl(struct sock *sk, unsigned int cmd, void __user *arg) |
1456 | { | |
0033d5ad | 1457 | struct compat_sioc_sg_req sr; |
ca6b8bb0 DM |
1458 | struct compat_sioc_vif_req vr; |
1459 | struct vif_device *vif; | |
709b46e8 EB |
1460 | struct mfc_cache *c; |
1461 | struct net *net = sock_net(sk); | |
1462 | struct mr_table *mrt; | |
1463 | ||
1464 | mrt = ipmr_get_table(net, raw_sk(sk)->ipmr_table ? : RT_TABLE_DEFAULT); | |
1465 | if (mrt == NULL) | |
1466 | return -ENOENT; | |
1467 | ||
1468 | switch (cmd) { | |
ca6b8bb0 DM |
1469 | case SIOCGETVIFCNT: |
1470 | if (copy_from_user(&vr, arg, sizeof(vr))) | |
1471 | return -EFAULT; | |
1472 | if (vr.vifi >= mrt->maxvif) | |
1473 | return -EINVAL; | |
1474 | read_lock(&mrt_lock); | |
1475 | vif = &mrt->vif_table[vr.vifi]; | |
1476 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
1477 | vr.icount = vif->pkt_in; | |
1478 | vr.ocount = vif->pkt_out; | |
1479 | vr.ibytes = vif->bytes_in; | |
1480 | vr.obytes = vif->bytes_out; | |
1481 | read_unlock(&mrt_lock); | |
1482 | ||
1483 | if (copy_to_user(arg, &vr, sizeof(vr))) | |
1484 | return -EFAULT; | |
1485 | return 0; | |
1486 | } | |
1487 | read_unlock(&mrt_lock); | |
1488 | return -EADDRNOTAVAIL; | |
709b46e8 EB |
1489 | case SIOCGETSGCNT: |
1490 | if (copy_from_user(&sr, arg, sizeof(sr))) | |
1491 | return -EFAULT; | |
1492 | ||
1493 | rcu_read_lock(); | |
1494 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); | |
1495 | if (c) { | |
1496 | sr.pktcnt = c->mfc_un.res.pkt; | |
1497 | sr.bytecnt = c->mfc_un.res.bytes; | |
1498 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1499 | rcu_read_unlock(); | |
1500 | ||
1501 | if (copy_to_user(arg, &sr, sizeof(sr))) | |
1502 | return -EFAULT; | |
1503 | return 0; | |
1504 | } | |
1505 | rcu_read_unlock(); | |
1506 | return -EADDRNOTAVAIL; | |
1507 | default: | |
1508 | return -ENOIOCTLCMD; | |
1509 | } | |
1510 | } | |
1511 | #endif | |
1512 | ||
1da177e4 LT |
1513 | |
1514 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) | |
1515 | { | |
e9dc8653 | 1516 | struct net_device *dev = ptr; |
4feb88e5 | 1517 | struct net *net = dev_net(dev); |
f0ad0860 | 1518 | struct mr_table *mrt; |
1da177e4 LT |
1519 | struct vif_device *v; |
1520 | int ct; | |
d17fa6fa | 1521 | LIST_HEAD(list); |
e9dc8653 | 1522 | |
1da177e4 LT |
1523 | if (event != NETDEV_UNREGISTER) |
1524 | return NOTIFY_DONE; | |
f0ad0860 PM |
1525 | |
1526 | ipmr_for_each_table(mrt, net) { | |
1527 | v = &mrt->vif_table[0]; | |
1528 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { | |
1529 | if (v->dev == dev) | |
1530 | vif_delete(mrt, ct, 1, &list); | |
1531 | } | |
1da177e4 | 1532 | } |
d17fa6fa | 1533 | unregister_netdevice_many(&list); |
1da177e4 LT |
1534 | return NOTIFY_DONE; |
1535 | } | |
1536 | ||
1537 | ||
c354e124 | 1538 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1539 | .notifier_call = ipmr_device_event, |
1540 | }; | |
1541 | ||
1542 | /* | |
a8cb16dd | 1543 | * Encapsulate a packet by attaching a valid IPIP header to it. |
1da177e4 LT |
1544 | * This avoids tunnel drivers and other mess and gives us the speed so |
1545 | * important for multicast video. | |
1546 | */ | |
e905a9ed | 1547 | |
114c7844 | 1548 | static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr) |
1da177e4 | 1549 | { |
8856dfa3 | 1550 | struct iphdr *iph; |
eddc9ec5 | 1551 | struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1552 | |
1553 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1554 | skb->transport_header = skb->network_header; |
8856dfa3 | 1555 | skb_reset_network_header(skb); |
eddc9ec5 | 1556 | iph = ip_hdr(skb); |
1da177e4 | 1557 | |
a8cb16dd | 1558 | iph->version = 4; |
e023dd64 ACM |
1559 | iph->tos = old_iph->tos; |
1560 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1561 | iph->frag_off = 0; |
1562 | iph->daddr = daddr; | |
1563 | iph->saddr = saddr; | |
1564 | iph->protocol = IPPROTO_IPIP; | |
1565 | iph->ihl = 5; | |
1566 | iph->tot_len = htons(skb->len); | |
adf30907 | 1567 | ip_select_ident(iph, skb_dst(skb), NULL); |
1da177e4 LT |
1568 | ip_send_check(iph); |
1569 | ||
1da177e4 LT |
1570 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1571 | nf_reset(skb); | |
1572 | } | |
1573 | ||
1574 | static inline int ipmr_forward_finish(struct sk_buff *skb) | |
1575 | { | |
a8cb16dd | 1576 | struct ip_options *opt = &(IPCB(skb)->opt); |
1da177e4 | 1577 | |
adf30907 | 1578 | IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS); |
1da177e4 LT |
1579 | |
1580 | if (unlikely(opt->optlen)) | |
1581 | ip_forward_options(skb); | |
1582 | ||
1583 | return dst_output(skb); | |
1584 | } | |
1585 | ||
1586 | /* | |
1587 | * Processing handlers for ipmr_forward | |
1588 | */ | |
1589 | ||
0c12295a PM |
1590 | static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt, |
1591 | struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1da177e4 | 1592 | { |
eddc9ec5 | 1593 | const struct iphdr *iph = ip_hdr(skb); |
0c12295a | 1594 | struct vif_device *vif = &mrt->vif_table[vifi]; |
1da177e4 LT |
1595 | struct net_device *dev; |
1596 | struct rtable *rt; | |
1597 | int encap = 0; | |
1598 | ||
1599 | if (vif->dev == NULL) | |
1600 | goto out_free; | |
1601 | ||
1602 | #ifdef CONFIG_IP_PIMSM | |
1603 | if (vif->flags & VIFF_REGISTER) { | |
1604 | vif->pkt_out++; | |
c354e124 | 1605 | vif->bytes_out += skb->len; |
cf3677ae PE |
1606 | vif->dev->stats.tx_bytes += skb->len; |
1607 | vif->dev->stats.tx_packets++; | |
0c12295a | 1608 | ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1609 | goto out_free; |
1da177e4 LT |
1610 | } |
1611 | #endif | |
1612 | ||
a8cb16dd ED |
1613 | if (vif->flags & VIFF_TUNNEL) { |
1614 | struct flowi fl = { | |
1615 | .oif = vif->link, | |
5811662b CG |
1616 | .fl4_dst = vif->remote, |
1617 | .fl4_src = vif->local, | |
1618 | .fl4_tos = RT_TOS(iph->tos), | |
a8cb16dd ED |
1619 | .proto = IPPROTO_IPIP |
1620 | }; | |
b23dd4fe DM |
1621 | rt = ip_route_output_key(net, &fl); |
1622 | if (IS_ERR(rt)) | |
1da177e4 LT |
1623 | goto out_free; |
1624 | encap = sizeof(struct iphdr); | |
1625 | } else { | |
a8cb16dd ED |
1626 | struct flowi fl = { |
1627 | .oif = vif->link, | |
5811662b CG |
1628 | .fl4_dst = iph->daddr, |
1629 | .fl4_tos = RT_TOS(iph->tos), | |
a8cb16dd ED |
1630 | .proto = IPPROTO_IPIP |
1631 | }; | |
b23dd4fe DM |
1632 | rt = ip_route_output_key(net, &fl); |
1633 | if (IS_ERR(rt)) | |
1da177e4 LT |
1634 | goto out_free; |
1635 | } | |
1636 | ||
d8d1f30b | 1637 | dev = rt->dst.dev; |
1da177e4 | 1638 | |
d8d1f30b | 1639 | if (skb->len+encap > dst_mtu(&rt->dst) && (ntohs(iph->frag_off) & IP_DF)) { |
1da177e4 | 1640 | /* Do not fragment multicasts. Alas, IPv4 does not |
a8cb16dd ED |
1641 | * allow to send ICMP, so that packets will disappear |
1642 | * to blackhole. | |
1da177e4 LT |
1643 | */ |
1644 | ||
7c73a6fa | 1645 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1646 | ip_rt_put(rt); |
1647 | goto out_free; | |
1648 | } | |
1649 | ||
d8d1f30b | 1650 | encap += LL_RESERVED_SPACE(dev) + rt->dst.header_len; |
1da177e4 LT |
1651 | |
1652 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1653 | ip_rt_put(rt); |
1da177e4 LT |
1654 | goto out_free; |
1655 | } | |
1656 | ||
1657 | vif->pkt_out++; | |
c354e124 | 1658 | vif->bytes_out += skb->len; |
1da177e4 | 1659 | |
adf30907 | 1660 | skb_dst_drop(skb); |
d8d1f30b | 1661 | skb_dst_set(skb, &rt->dst); |
eddc9ec5 | 1662 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1663 | |
1664 | /* FIXME: forward and output firewalls used to be called here. | |
a8cb16dd ED |
1665 | * What do we do with netfilter? -- RR |
1666 | */ | |
1da177e4 LT |
1667 | if (vif->flags & VIFF_TUNNEL) { |
1668 | ip_encap(skb, vif->local, vif->remote); | |
1669 | /* FIXME: extra output firewall step used to be here. --RR */ | |
2f4c02d4 PE |
1670 | vif->dev->stats.tx_packets++; |
1671 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1672 | } |
1673 | ||
1674 | IPCB(skb)->flags |= IPSKB_FORWARDED; | |
1675 | ||
1676 | /* | |
1677 | * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally | |
1678 | * not only before forwarding, but after forwarding on all output | |
1679 | * interfaces. It is clear, if mrouter runs a multicasting | |
1680 | * program, it should receive packets not depending to what interface | |
1681 | * program is joined. | |
1682 | * If we will not make it, the program will have to join on all | |
1683 | * interfaces. On the other hand, multihoming host (or router, but | |
1684 | * not mrouter) cannot join to more than one interface - it will | |
1685 | * result in receiving multiple packets. | |
1686 | */ | |
9bbc768a | 1687 | NF_HOOK(NFPROTO_IPV4, NF_INET_FORWARD, skb, skb->dev, dev, |
1da177e4 LT |
1688 | ipmr_forward_finish); |
1689 | return; | |
1690 | ||
1691 | out_free: | |
1692 | kfree_skb(skb); | |
1da177e4 LT |
1693 | } |
1694 | ||
0c12295a | 1695 | static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev) |
1da177e4 LT |
1696 | { |
1697 | int ct; | |
0c12295a PM |
1698 | |
1699 | for (ct = mrt->maxvif-1; ct >= 0; ct--) { | |
1700 | if (mrt->vif_table[ct].dev == dev) | |
1da177e4 LT |
1701 | break; |
1702 | } | |
1703 | return ct; | |
1704 | } | |
1705 | ||
1706 | /* "local" means that we should preserve one skb (for local delivery) */ | |
1707 | ||
0c12295a PM |
1708 | static int ip_mr_forward(struct net *net, struct mr_table *mrt, |
1709 | struct sk_buff *skb, struct mfc_cache *cache, | |
1710 | int local) | |
1da177e4 LT |
1711 | { |
1712 | int psend = -1; | |
1713 | int vif, ct; | |
1714 | ||
1715 | vif = cache->mfc_parent; | |
1716 | cache->mfc_un.res.pkt++; | |
1717 | cache->mfc_un.res.bytes += skb->len; | |
1718 | ||
1719 | /* | |
1720 | * Wrong interface: drop packet and (maybe) send PIM assert. | |
1721 | */ | |
0c12295a | 1722 | if (mrt->vif_table[vif].dev != skb->dev) { |
1da177e4 LT |
1723 | int true_vifi; |
1724 | ||
c7537967 | 1725 | if (rt_is_output_route(skb_rtable(skb))) { |
1da177e4 | 1726 | /* It is our own packet, looped back. |
a8cb16dd ED |
1727 | * Very complicated situation... |
1728 | * | |
1729 | * The best workaround until routing daemons will be | |
1730 | * fixed is not to redistribute packet, if it was | |
1731 | * send through wrong interface. It means, that | |
1732 | * multicast applications WILL NOT work for | |
1733 | * (S,G), which have default multicast route pointing | |
1734 | * to wrong oif. In any case, it is not a good | |
1735 | * idea to use multicasting applications on router. | |
1da177e4 LT |
1736 | */ |
1737 | goto dont_forward; | |
1738 | } | |
1739 | ||
1740 | cache->mfc_un.res.wrong_if++; | |
0c12295a | 1741 | true_vifi = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1742 | |
0c12295a | 1743 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
1da177e4 | 1744 | /* pimsm uses asserts, when switching from RPT to SPT, |
a8cb16dd ED |
1745 | * so that we cannot check that packet arrived on an oif. |
1746 | * It is bad, but otherwise we would need to move pretty | |
1747 | * large chunk of pimd to kernel. Ough... --ANK | |
1da177e4 | 1748 | */ |
0c12295a | 1749 | (mrt->mroute_do_pim || |
6f9374a9 | 1750 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1751 | time_after(jiffies, |
1da177e4 LT |
1752 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1753 | cache->mfc_un.res.last_assert = jiffies; | |
0c12295a | 1754 | ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1755 | } |
1756 | goto dont_forward; | |
1757 | } | |
1758 | ||
0c12295a PM |
1759 | mrt->vif_table[vif].pkt_in++; |
1760 | mrt->vif_table[vif].bytes_in += skb->len; | |
1da177e4 LT |
1761 | |
1762 | /* | |
1763 | * Forward the frame | |
1764 | */ | |
a8cb16dd ED |
1765 | for (ct = cache->mfc_un.res.maxvif - 1; |
1766 | ct >= cache->mfc_un.res.minvif; ct--) { | |
eddc9ec5 | 1767 | if (ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1768 | if (psend != -1) { |
1769 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1770 | |
1da177e4 | 1771 | if (skb2) |
0c12295a PM |
1772 | ipmr_queue_xmit(net, mrt, skb2, cache, |
1773 | psend); | |
1da177e4 | 1774 | } |
c354e124 | 1775 | psend = ct; |
1da177e4 LT |
1776 | } |
1777 | } | |
1778 | if (psend != -1) { | |
1779 | if (local) { | |
1780 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
a8cb16dd | 1781 | |
1da177e4 | 1782 | if (skb2) |
0c12295a | 1783 | ipmr_queue_xmit(net, mrt, skb2, cache, psend); |
1da177e4 | 1784 | } else { |
0c12295a | 1785 | ipmr_queue_xmit(net, mrt, skb, cache, psend); |
1da177e4 LT |
1786 | return 0; |
1787 | } | |
1788 | } | |
1789 | ||
1790 | dont_forward: | |
1791 | if (!local) | |
1792 | kfree_skb(skb); | |
1793 | return 0; | |
1794 | } | |
1795 | ||
1796 | ||
1797 | /* | |
1798 | * Multicast packets for forwarding arrive here | |
4c968709 | 1799 | * Called with rcu_read_lock(); |
1da177e4 LT |
1800 | */ |
1801 | ||
1802 | int ip_mr_input(struct sk_buff *skb) | |
1803 | { | |
1804 | struct mfc_cache *cache; | |
4feb88e5 | 1805 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1806 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
f0ad0860 PM |
1807 | struct mr_table *mrt; |
1808 | int err; | |
1da177e4 LT |
1809 | |
1810 | /* Packet is looped back after forward, it should not be | |
a8cb16dd | 1811 | * forwarded second time, but still can be delivered locally. |
1da177e4 | 1812 | */ |
4c968709 | 1813 | if (IPCB(skb)->flags & IPSKB_FORWARDED) |
1da177e4 LT |
1814 | goto dont_forward; |
1815 | ||
5e2b61f7 DM |
1816 | { |
1817 | struct rtable *rt = skb_rtable(skb); | |
1818 | struct flowi fl = { | |
1819 | .fl4_dst = rt->rt_key_dst, | |
1820 | .fl4_src = rt->rt_key_src, | |
1821 | .fl4_tos = rt->rt_tos, | |
1822 | .oif = rt->rt_oif, | |
1823 | .iif = rt->rt_iif, | |
1824 | .mark = rt->rt_mark, | |
1825 | }; | |
1826 | err = ipmr_fib_lookup(net, &fl, &mrt); | |
1827 | if (err < 0) { | |
1828 | kfree_skb(skb); | |
1829 | return err; | |
1830 | } | |
e40dbc51 | 1831 | } |
1da177e4 | 1832 | if (!local) { |
4c968709 ED |
1833 | if (IPCB(skb)->opt.router_alert) { |
1834 | if (ip_call_ra_chain(skb)) | |
1835 | return 0; | |
1836 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP) { | |
1837 | /* IGMPv1 (and broken IGMPv2 implementations sort of | |
1838 | * Cisco IOS <= 11.2(8)) do not put router alert | |
1839 | * option to IGMP packets destined to routable | |
1840 | * groups. It is very bad, because it means | |
1841 | * that we can forward NO IGMP messages. | |
1842 | */ | |
1843 | struct sock *mroute_sk; | |
1844 | ||
1845 | mroute_sk = rcu_dereference(mrt->mroute_sk); | |
1846 | if (mroute_sk) { | |
1847 | nf_reset(skb); | |
1848 | raw_rcv(mroute_sk, skb); | |
1849 | return 0; | |
1850 | } | |
1da177e4 LT |
1851 | } |
1852 | } | |
1853 | ||
a8c9486b | 1854 | /* already under rcu_read_lock() */ |
0c12295a | 1855 | cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
1da177e4 LT |
1856 | |
1857 | /* | |
1858 | * No usable cache entry | |
1859 | */ | |
c354e124 | 1860 | if (cache == NULL) { |
1da177e4 LT |
1861 | int vif; |
1862 | ||
1863 | if (local) { | |
1864 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1865 | ip_local_deliver(skb); | |
a8c9486b | 1866 | if (skb2 == NULL) |
1da177e4 | 1867 | return -ENOBUFS; |
1da177e4 LT |
1868 | skb = skb2; |
1869 | } | |
1870 | ||
a8c9486b | 1871 | read_lock(&mrt_lock); |
0c12295a | 1872 | vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1873 | if (vif >= 0) { |
0eae88f3 | 1874 | int err2 = ipmr_cache_unresolved(mrt, vif, skb); |
1da177e4 LT |
1875 | read_unlock(&mrt_lock); |
1876 | ||
0eae88f3 | 1877 | return err2; |
1da177e4 LT |
1878 | } |
1879 | read_unlock(&mrt_lock); | |
1880 | kfree_skb(skb); | |
1881 | return -ENODEV; | |
1882 | } | |
1883 | ||
a8c9486b | 1884 | read_lock(&mrt_lock); |
0c12295a | 1885 | ip_mr_forward(net, mrt, skb, cache, local); |
1da177e4 LT |
1886 | read_unlock(&mrt_lock); |
1887 | ||
1888 | if (local) | |
1889 | return ip_local_deliver(skb); | |
1890 | ||
1891 | return 0; | |
1892 | ||
1893 | dont_forward: | |
1894 | if (local) | |
1895 | return ip_local_deliver(skb); | |
1896 | kfree_skb(skb); | |
1897 | return 0; | |
1898 | } | |
1899 | ||
b1879204 | 1900 | #ifdef CONFIG_IP_PIMSM |
55747a0a | 1901 | /* called with rcu_read_lock() */ |
f0ad0860 PM |
1902 | static int __pim_rcv(struct mr_table *mrt, struct sk_buff *skb, |
1903 | unsigned int pimlen) | |
1da177e4 | 1904 | { |
b1879204 IJ |
1905 | struct net_device *reg_dev = NULL; |
1906 | struct iphdr *encap; | |
1da177e4 | 1907 | |
b1879204 | 1908 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); |
1da177e4 | 1909 | /* |
a8cb16dd ED |
1910 | * Check that: |
1911 | * a. packet is really sent to a multicast group | |
1912 | * b. packet is not a NULL-REGISTER | |
1913 | * c. packet is not truncated | |
1da177e4 | 1914 | */ |
f97c1e0c | 1915 | if (!ipv4_is_multicast(encap->daddr) || |
1da177e4 | 1916 | encap->tot_len == 0 || |
b1879204 IJ |
1917 | ntohs(encap->tot_len) + pimlen > skb->len) |
1918 | return 1; | |
1da177e4 LT |
1919 | |
1920 | read_lock(&mrt_lock); | |
0c12295a PM |
1921 | if (mrt->mroute_reg_vif_num >= 0) |
1922 | reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev; | |
1da177e4 LT |
1923 | read_unlock(&mrt_lock); |
1924 | ||
e905a9ed | 1925 | if (reg_dev == NULL) |
b1879204 | 1926 | return 1; |
1da177e4 | 1927 | |
b0e380b1 | 1928 | skb->mac_header = skb->network_header; |
55747a0a | 1929 | skb_pull(skb, (u8 *)encap - skb->data); |
31c7711b | 1930 | skb_reset_network_header(skb); |
1da177e4 | 1931 | skb->protocol = htons(ETH_P_IP); |
55747a0a | 1932 | skb->ip_summed = CHECKSUM_NONE; |
1da177e4 | 1933 | skb->pkt_type = PACKET_HOST; |
d19d56dd ED |
1934 | |
1935 | skb_tunnel_rx(skb, reg_dev); | |
1936 | ||
1da177e4 | 1937 | netif_rx(skb); |
b1879204 | 1938 | |
55747a0a | 1939 | return NET_RX_SUCCESS; |
b1879204 IJ |
1940 | } |
1941 | #endif | |
1942 | ||
1943 | #ifdef CONFIG_IP_PIMSM_V1 | |
1944 | /* | |
1945 | * Handle IGMP messages of PIMv1 | |
1946 | */ | |
1947 | ||
a8cb16dd | 1948 | int pim_rcv_v1(struct sk_buff *skb) |
b1879204 IJ |
1949 | { |
1950 | struct igmphdr *pim; | |
4feb88e5 | 1951 | struct net *net = dev_net(skb->dev); |
f0ad0860 | 1952 | struct mr_table *mrt; |
b1879204 IJ |
1953 | |
1954 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
1955 | goto drop; | |
1956 | ||
1957 | pim = igmp_hdr(skb); | |
1958 | ||
5e2b61f7 DM |
1959 | { |
1960 | struct rtable *rt = skb_rtable(skb); | |
1961 | struct flowi fl = { | |
1962 | .fl4_dst = rt->rt_key_dst, | |
1963 | .fl4_src = rt->rt_key_src, | |
1964 | .fl4_tos = rt->rt_tos, | |
1965 | .oif = rt->rt_oif, | |
1966 | .iif = rt->rt_iif, | |
1967 | .mark = rt->rt_mark, | |
1968 | }; | |
1969 | if (ipmr_fib_lookup(net, &fl, &mrt) < 0) | |
1970 | goto drop; | |
1971 | } | |
0c12295a | 1972 | if (!mrt->mroute_do_pim || |
b1879204 IJ |
1973 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
1974 | goto drop; | |
1975 | ||
f0ad0860 | 1976 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
1977 | drop: |
1978 | kfree_skb(skb); | |
1979 | } | |
1da177e4 LT |
1980 | return 0; |
1981 | } | |
1982 | #endif | |
1983 | ||
1984 | #ifdef CONFIG_IP_PIMSM_V2 | |
a8cb16dd | 1985 | static int pim_rcv(struct sk_buff *skb) |
1da177e4 LT |
1986 | { |
1987 | struct pimreghdr *pim; | |
f0ad0860 PM |
1988 | struct net *net = dev_net(skb->dev); |
1989 | struct mr_table *mrt; | |
1da177e4 | 1990 | |
b1879204 | 1991 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
1992 | goto drop; |
1993 | ||
9c70220b | 1994 | pim = (struct pimreghdr *)skb_transport_header(skb); |
a8cb16dd ED |
1995 | if (pim->type != ((PIM_VERSION << 4) | (PIM_REGISTER)) || |
1996 | (pim->flags & PIM_NULL_REGISTER) || | |
e905a9ed | 1997 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 1998 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
1999 | goto drop; |
2000 | ||
5e2b61f7 DM |
2001 | { |
2002 | struct rtable *rt = skb_rtable(skb); | |
2003 | struct flowi fl = { | |
2004 | .fl4_dst = rt->rt_key_dst, | |
2005 | .fl4_src = rt->rt_key_src, | |
2006 | .fl4_tos = rt->rt_tos, | |
2007 | .oif = rt->rt_oif, | |
2008 | .iif = rt->rt_iif, | |
2009 | .mark = rt->rt_mark, | |
2010 | }; | |
2011 | if (ipmr_fib_lookup(net, &fl, &mrt) < 0) | |
2012 | goto drop; | |
2013 | } | |
f0ad0860 | 2014 | if (__pim_rcv(mrt, skb, sizeof(*pim))) { |
b1879204 IJ |
2015 | drop: |
2016 | kfree_skb(skb); | |
2017 | } | |
1da177e4 LT |
2018 | return 0; |
2019 | } | |
2020 | #endif | |
2021 | ||
cb6a4e46 PM |
2022 | static int __ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2023 | struct mfc_cache *c, struct rtmsg *rtm) | |
1da177e4 LT |
2024 | { |
2025 | int ct; | |
2026 | struct rtnexthop *nhp; | |
27a884dc | 2027 | u8 *b = skb_tail_pointer(skb); |
1da177e4 LT |
2028 | struct rtattr *mp_head; |
2029 | ||
7438189b | 2030 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
ed0f160a | 2031 | if (c->mfc_parent >= MAXVIFS) |
7438189b ND |
2032 | return -ENOENT; |
2033 | ||
0c12295a PM |
2034 | if (VIF_EXISTS(mrt, c->mfc_parent)) |
2035 | RTA_PUT(skb, RTA_IIF, 4, &mrt->vif_table[c->mfc_parent].dev->ifindex); | |
1da177e4 | 2036 | |
c354e124 | 2037 | mp_head = (struct rtattr *)skb_put(skb, RTA_LENGTH(0)); |
1da177e4 LT |
2038 | |
2039 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
0c12295a | 2040 | if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) { |
1da177e4 LT |
2041 | if (skb_tailroom(skb) < RTA_ALIGN(RTA_ALIGN(sizeof(*nhp)) + 4)) |
2042 | goto rtattr_failure; | |
c354e124 | 2043 | nhp = (struct rtnexthop *)skb_put(skb, RTA_ALIGN(sizeof(*nhp))); |
1da177e4 LT |
2044 | nhp->rtnh_flags = 0; |
2045 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
0c12295a | 2046 | nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex; |
1da177e4 LT |
2047 | nhp->rtnh_len = sizeof(*nhp); |
2048 | } | |
2049 | } | |
2050 | mp_head->rta_type = RTA_MULTIPATH; | |
27a884dc | 2051 | mp_head->rta_len = skb_tail_pointer(skb) - (u8 *)mp_head; |
1da177e4 LT |
2052 | rtm->rtm_type = RTN_MULTICAST; |
2053 | return 1; | |
2054 | ||
2055 | rtattr_failure: | |
dc5fc579 | 2056 | nlmsg_trim(skb, b); |
1da177e4 LT |
2057 | return -EMSGSIZE; |
2058 | } | |
2059 | ||
4feb88e5 BT |
2060 | int ipmr_get_route(struct net *net, |
2061 | struct sk_buff *skb, struct rtmsg *rtm, int nowait) | |
1da177e4 LT |
2062 | { |
2063 | int err; | |
f0ad0860 | 2064 | struct mr_table *mrt; |
1da177e4 | 2065 | struct mfc_cache *cache; |
511c3f92 | 2066 | struct rtable *rt = skb_rtable(skb); |
1da177e4 | 2067 | |
f0ad0860 PM |
2068 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
2069 | if (mrt == NULL) | |
2070 | return -ENOENT; | |
2071 | ||
a8c9486b | 2072 | rcu_read_lock(); |
0c12295a | 2073 | cache = ipmr_cache_find(mrt, rt->rt_src, rt->rt_dst); |
1da177e4 | 2074 | |
c354e124 | 2075 | if (cache == NULL) { |
72287490 | 2076 | struct sk_buff *skb2; |
eddc9ec5 | 2077 | struct iphdr *iph; |
1da177e4 | 2078 | struct net_device *dev; |
a8cb16dd | 2079 | int vif = -1; |
1da177e4 LT |
2080 | |
2081 | if (nowait) { | |
a8c9486b | 2082 | rcu_read_unlock(); |
1da177e4 LT |
2083 | return -EAGAIN; |
2084 | } | |
2085 | ||
2086 | dev = skb->dev; | |
a8c9486b | 2087 | read_lock(&mrt_lock); |
a8cb16dd ED |
2088 | if (dev) |
2089 | vif = ipmr_find_vif(mrt, dev); | |
2090 | if (vif < 0) { | |
1da177e4 | 2091 | read_unlock(&mrt_lock); |
a8c9486b | 2092 | rcu_read_unlock(); |
1da177e4 LT |
2093 | return -ENODEV; |
2094 | } | |
72287490 AK |
2095 | skb2 = skb_clone(skb, GFP_ATOMIC); |
2096 | if (!skb2) { | |
2097 | read_unlock(&mrt_lock); | |
a8c9486b | 2098 | rcu_read_unlock(); |
72287490 AK |
2099 | return -ENOMEM; |
2100 | } | |
2101 | ||
e2d1bca7 ACM |
2102 | skb_push(skb2, sizeof(struct iphdr)); |
2103 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
2104 | iph = ip_hdr(skb2); |
2105 | iph->ihl = sizeof(struct iphdr) >> 2; | |
2106 | iph->saddr = rt->rt_src; | |
2107 | iph->daddr = rt->rt_dst; | |
2108 | iph->version = 0; | |
0c12295a | 2109 | err = ipmr_cache_unresolved(mrt, vif, skb2); |
1da177e4 | 2110 | read_unlock(&mrt_lock); |
a8c9486b | 2111 | rcu_read_unlock(); |
1da177e4 LT |
2112 | return err; |
2113 | } | |
2114 | ||
a8c9486b ED |
2115 | read_lock(&mrt_lock); |
2116 | if (!nowait && (rtm->rtm_flags & RTM_F_NOTIFY)) | |
1da177e4 | 2117 | cache->mfc_flags |= MFC_NOTIFY; |
cb6a4e46 | 2118 | err = __ipmr_fill_mroute(mrt, skb, cache, rtm); |
1da177e4 | 2119 | read_unlock(&mrt_lock); |
a8c9486b | 2120 | rcu_read_unlock(); |
1da177e4 LT |
2121 | return err; |
2122 | } | |
2123 | ||
cb6a4e46 PM |
2124 | static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
2125 | u32 pid, u32 seq, struct mfc_cache *c) | |
2126 | { | |
2127 | struct nlmsghdr *nlh; | |
2128 | struct rtmsg *rtm; | |
2129 | ||
2130 | nlh = nlmsg_put(skb, pid, seq, RTM_NEWROUTE, sizeof(*rtm), NLM_F_MULTI); | |
2131 | if (nlh == NULL) | |
2132 | return -EMSGSIZE; | |
2133 | ||
2134 | rtm = nlmsg_data(nlh); | |
2135 | rtm->rtm_family = RTNL_FAMILY_IPMR; | |
2136 | rtm->rtm_dst_len = 32; | |
2137 | rtm->rtm_src_len = 32; | |
2138 | rtm->rtm_tos = 0; | |
2139 | rtm->rtm_table = mrt->id; | |
2140 | NLA_PUT_U32(skb, RTA_TABLE, mrt->id); | |
2141 | rtm->rtm_type = RTN_MULTICAST; | |
2142 | rtm->rtm_scope = RT_SCOPE_UNIVERSE; | |
2143 | rtm->rtm_protocol = RTPROT_UNSPEC; | |
2144 | rtm->rtm_flags = 0; | |
2145 | ||
2146 | NLA_PUT_BE32(skb, RTA_SRC, c->mfc_origin); | |
2147 | NLA_PUT_BE32(skb, RTA_DST, c->mfc_mcastgrp); | |
2148 | ||
2149 | if (__ipmr_fill_mroute(mrt, skb, c, rtm) < 0) | |
2150 | goto nla_put_failure; | |
2151 | ||
2152 | return nlmsg_end(skb, nlh); | |
2153 | ||
2154 | nla_put_failure: | |
2155 | nlmsg_cancel(skb, nlh); | |
2156 | return -EMSGSIZE; | |
2157 | } | |
2158 | ||
2159 | static int ipmr_rtm_dumproute(struct sk_buff *skb, struct netlink_callback *cb) | |
2160 | { | |
2161 | struct net *net = sock_net(skb->sk); | |
2162 | struct mr_table *mrt; | |
2163 | struct mfc_cache *mfc; | |
2164 | unsigned int t = 0, s_t; | |
2165 | unsigned int h = 0, s_h; | |
2166 | unsigned int e = 0, s_e; | |
2167 | ||
2168 | s_t = cb->args[0]; | |
2169 | s_h = cb->args[1]; | |
2170 | s_e = cb->args[2]; | |
2171 | ||
a8c9486b | 2172 | rcu_read_lock(); |
cb6a4e46 PM |
2173 | ipmr_for_each_table(mrt, net) { |
2174 | if (t < s_t) | |
2175 | goto next_table; | |
2176 | if (t > s_t) | |
2177 | s_h = 0; | |
2178 | for (h = s_h; h < MFC_LINES; h++) { | |
a8c9486b | 2179 | list_for_each_entry_rcu(mfc, &mrt->mfc_cache_array[h], list) { |
cb6a4e46 PM |
2180 | if (e < s_e) |
2181 | goto next_entry; | |
2182 | if (ipmr_fill_mroute(mrt, skb, | |
2183 | NETLINK_CB(cb->skb).pid, | |
2184 | cb->nlh->nlmsg_seq, | |
2185 | mfc) < 0) | |
2186 | goto done; | |
2187 | next_entry: | |
2188 | e++; | |
2189 | } | |
2190 | e = s_e = 0; | |
2191 | } | |
2192 | s_h = 0; | |
2193 | next_table: | |
2194 | t++; | |
2195 | } | |
2196 | done: | |
a8c9486b | 2197 | rcu_read_unlock(); |
cb6a4e46 PM |
2198 | |
2199 | cb->args[2] = e; | |
2200 | cb->args[1] = h; | |
2201 | cb->args[0] = t; | |
2202 | ||
2203 | return skb->len; | |
2204 | } | |
2205 | ||
e905a9ed | 2206 | #ifdef CONFIG_PROC_FS |
1da177e4 | 2207 | /* |
a8cb16dd ED |
2208 | * The /proc interfaces to multicast routing : |
2209 | * /proc/net/ip_mr_cache & /proc/net/ip_mr_vif | |
1da177e4 LT |
2210 | */ |
2211 | struct ipmr_vif_iter { | |
f6bb4514 | 2212 | struct seq_net_private p; |
f0ad0860 | 2213 | struct mr_table *mrt; |
1da177e4 LT |
2214 | int ct; |
2215 | }; | |
2216 | ||
f6bb4514 BT |
2217 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
2218 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
2219 | loff_t pos) |
2220 | { | |
f0ad0860 | 2221 | struct mr_table *mrt = iter->mrt; |
0c12295a PM |
2222 | |
2223 | for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) { | |
2224 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2225 | continue; |
e905a9ed | 2226 | if (pos-- == 0) |
0c12295a | 2227 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2228 | } |
2229 | return NULL; | |
2230 | } | |
2231 | ||
2232 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 2233 | __acquires(mrt_lock) |
1da177e4 | 2234 | { |
f0ad0860 | 2235 | struct ipmr_vif_iter *iter = seq->private; |
f6bb4514 | 2236 | struct net *net = seq_file_net(seq); |
f0ad0860 PM |
2237 | struct mr_table *mrt; |
2238 | ||
2239 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); | |
2240 | if (mrt == NULL) | |
2241 | return ERR_PTR(-ENOENT); | |
2242 | ||
2243 | iter->mrt = mrt; | |
f6bb4514 | 2244 | |
1da177e4 | 2245 | read_lock(&mrt_lock); |
f6bb4514 | 2246 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2247 | : SEQ_START_TOKEN; |
2248 | } | |
2249 | ||
2250 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2251 | { | |
2252 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 2253 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2254 | struct mr_table *mrt = iter->mrt; |
1da177e4 LT |
2255 | |
2256 | ++*pos; | |
2257 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2258 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 2259 | |
0c12295a PM |
2260 | while (++iter->ct < mrt->maxvif) { |
2261 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 2262 | continue; |
0c12295a | 2263 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
2264 | } |
2265 | return NULL; | |
2266 | } | |
2267 | ||
2268 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 2269 | __releases(mrt_lock) |
1da177e4 LT |
2270 | { |
2271 | read_unlock(&mrt_lock); | |
2272 | } | |
2273 | ||
2274 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
2275 | { | |
f0ad0860 PM |
2276 | struct ipmr_vif_iter *iter = seq->private; |
2277 | struct mr_table *mrt = iter->mrt; | |
f6bb4514 | 2278 | |
1da177e4 | 2279 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 2280 | seq_puts(seq, |
1da177e4 LT |
2281 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
2282 | } else { | |
2283 | const struct vif_device *vif = v; | |
2284 | const char *name = vif->dev ? vif->dev->name : "none"; | |
2285 | ||
2286 | seq_printf(seq, | |
2287 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
0c12295a | 2288 | vif - mrt->vif_table, |
e905a9ed | 2289 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
2290 | vif->bytes_out, vif->pkt_out, |
2291 | vif->flags, vif->local, vif->remote); | |
2292 | } | |
2293 | return 0; | |
2294 | } | |
2295 | ||
f690808e | 2296 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
2297 | .start = ipmr_vif_seq_start, |
2298 | .next = ipmr_vif_seq_next, | |
2299 | .stop = ipmr_vif_seq_stop, | |
2300 | .show = ipmr_vif_seq_show, | |
2301 | }; | |
2302 | ||
2303 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
2304 | { | |
f6bb4514 BT |
2305 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
2306 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
2307 | } |
2308 | ||
9a32144e | 2309 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
2310 | .owner = THIS_MODULE, |
2311 | .open = ipmr_vif_open, | |
2312 | .read = seq_read, | |
2313 | .llseek = seq_lseek, | |
f6bb4514 | 2314 | .release = seq_release_net, |
1da177e4 LT |
2315 | }; |
2316 | ||
2317 | struct ipmr_mfc_iter { | |
f6bb4514 | 2318 | struct seq_net_private p; |
f0ad0860 | 2319 | struct mr_table *mrt; |
862465f2 | 2320 | struct list_head *cache; |
1da177e4 LT |
2321 | int ct; |
2322 | }; | |
2323 | ||
2324 | ||
f6bb4514 BT |
2325 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
2326 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 | 2327 | { |
f0ad0860 | 2328 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2329 | struct mfc_cache *mfc; |
2330 | ||
a8c9486b | 2331 | rcu_read_lock(); |
862465f2 | 2332 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) { |
0c12295a | 2333 | it->cache = &mrt->mfc_cache_array[it->ct]; |
a8c9486b | 2334 | list_for_each_entry_rcu(mfc, it->cache, list) |
e905a9ed | 2335 | if (pos-- == 0) |
1da177e4 | 2336 | return mfc; |
862465f2 | 2337 | } |
a8c9486b | 2338 | rcu_read_unlock(); |
1da177e4 | 2339 | |
1da177e4 | 2340 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 2341 | it->cache = &mrt->mfc_unres_queue; |
862465f2 | 2342 | list_for_each_entry(mfc, it->cache, list) |
e258beb2 | 2343 | if (pos-- == 0) |
1da177e4 LT |
2344 | return mfc; |
2345 | spin_unlock_bh(&mfc_unres_lock); | |
2346 | ||
2347 | it->cache = NULL; | |
2348 | return NULL; | |
2349 | } | |
2350 | ||
2351 | ||
2352 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
2353 | { | |
2354 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2355 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2356 | struct mr_table *mrt; |
f6bb4514 | 2357 | |
f0ad0860 PM |
2358 | mrt = ipmr_get_table(net, RT_TABLE_DEFAULT); |
2359 | if (mrt == NULL) | |
2360 | return ERR_PTR(-ENOENT); | |
f6bb4514 | 2361 | |
f0ad0860 | 2362 | it->mrt = mrt; |
1da177e4 LT |
2363 | it->cache = NULL; |
2364 | it->ct = 0; | |
f6bb4514 | 2365 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
2366 | : SEQ_START_TOKEN; |
2367 | } | |
2368 | ||
2369 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
2370 | { | |
2371 | struct mfc_cache *mfc = v; | |
2372 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 2373 | struct net *net = seq_file_net(seq); |
f0ad0860 | 2374 | struct mr_table *mrt = it->mrt; |
1da177e4 LT |
2375 | |
2376 | ++*pos; | |
2377 | ||
2378 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 2379 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 | 2380 | |
862465f2 PM |
2381 | if (mfc->list.next != it->cache) |
2382 | return list_entry(mfc->list.next, struct mfc_cache, list); | |
e905a9ed | 2383 | |
0c12295a | 2384 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 LT |
2385 | goto end_of_list; |
2386 | ||
0c12295a | 2387 | BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]); |
1da177e4 LT |
2388 | |
2389 | while (++it->ct < MFC_LINES) { | |
0c12295a | 2390 | it->cache = &mrt->mfc_cache_array[it->ct]; |
862465f2 PM |
2391 | if (list_empty(it->cache)) |
2392 | continue; | |
2393 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 LT |
2394 | } |
2395 | ||
2396 | /* exhausted cache_array, show unresolved */ | |
a8c9486b | 2397 | rcu_read_unlock(); |
0c12295a | 2398 | it->cache = &mrt->mfc_unres_queue; |
1da177e4 | 2399 | it->ct = 0; |
e905a9ed | 2400 | |
1da177e4 | 2401 | spin_lock_bh(&mfc_unres_lock); |
862465f2 PM |
2402 | if (!list_empty(it->cache)) |
2403 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 | 2404 | |
a8cb16dd | 2405 | end_of_list: |
1da177e4 LT |
2406 | spin_unlock_bh(&mfc_unres_lock); |
2407 | it->cache = NULL; | |
2408 | ||
2409 | return NULL; | |
2410 | } | |
2411 | ||
2412 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
2413 | { | |
2414 | struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2415 | struct mr_table *mrt = it->mrt; |
1da177e4 | 2416 | |
0c12295a | 2417 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 | 2418 | spin_unlock_bh(&mfc_unres_lock); |
0c12295a | 2419 | else if (it->cache == &mrt->mfc_cache_array[it->ct]) |
a8c9486b | 2420 | rcu_read_unlock(); |
1da177e4 LT |
2421 | } |
2422 | ||
2423 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
2424 | { | |
2425 | int n; | |
2426 | ||
2427 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 2428 | seq_puts(seq, |
1da177e4 LT |
2429 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
2430 | } else { | |
2431 | const struct mfc_cache *mfc = v; | |
2432 | const struct ipmr_mfc_iter *it = seq->private; | |
f0ad0860 | 2433 | const struct mr_table *mrt = it->mrt; |
e905a9ed | 2434 | |
0eae88f3 ED |
2435 | seq_printf(seq, "%08X %08X %-3hd", |
2436 | (__force u32) mfc->mfc_mcastgrp, | |
2437 | (__force u32) mfc->mfc_origin, | |
1ea472e2 | 2438 | mfc->mfc_parent); |
1da177e4 | 2439 | |
0c12295a | 2440 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 BT |
2441 | seq_printf(seq, " %8lu %8lu %8lu", |
2442 | mfc->mfc_un.res.pkt, | |
2443 | mfc->mfc_un.res.bytes, | |
2444 | mfc->mfc_un.res.wrong_if); | |
132adf54 | 2445 | for (n = mfc->mfc_un.res.minvif; |
a8cb16dd | 2446 | n < mfc->mfc_un.res.maxvif; n++) { |
0c12295a | 2447 | if (VIF_EXISTS(mrt, n) && |
cf958ae3 BT |
2448 | mfc->mfc_un.res.ttls[n] < 255) |
2449 | seq_printf(seq, | |
e905a9ed | 2450 | " %2d:%-3d", |
1da177e4 LT |
2451 | n, mfc->mfc_un.res.ttls[n]); |
2452 | } | |
1ea472e2 BT |
2453 | } else { |
2454 | /* unresolved mfc_caches don't contain | |
2455 | * pkt, bytes and wrong_if values | |
2456 | */ | |
2457 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
2458 | } |
2459 | seq_putc(seq, '\n'); | |
2460 | } | |
2461 | return 0; | |
2462 | } | |
2463 | ||
f690808e | 2464 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
2465 | .start = ipmr_mfc_seq_start, |
2466 | .next = ipmr_mfc_seq_next, | |
2467 | .stop = ipmr_mfc_seq_stop, | |
2468 | .show = ipmr_mfc_seq_show, | |
2469 | }; | |
2470 | ||
2471 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
2472 | { | |
f6bb4514 BT |
2473 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
2474 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
2475 | } |
2476 | ||
9a32144e | 2477 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
2478 | .owner = THIS_MODULE, |
2479 | .open = ipmr_mfc_open, | |
2480 | .read = seq_read, | |
2481 | .llseek = seq_lseek, | |
f6bb4514 | 2482 | .release = seq_release_net, |
1da177e4 | 2483 | }; |
e905a9ed | 2484 | #endif |
1da177e4 LT |
2485 | |
2486 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 2487 | static const struct net_protocol pim_protocol = { |
1da177e4 | 2488 | .handler = pim_rcv, |
403dbb97 | 2489 | .netns_ok = 1, |
1da177e4 LT |
2490 | }; |
2491 | #endif | |
2492 | ||
2493 | ||
2494 | /* | |
2495 | * Setup for IP multicast routing | |
2496 | */ | |
cf958ae3 BT |
2497 | static int __net_init ipmr_net_init(struct net *net) |
2498 | { | |
f0ad0860 | 2499 | int err; |
cf958ae3 | 2500 | |
f0ad0860 PM |
2501 | err = ipmr_rules_init(net); |
2502 | if (err < 0) | |
cf958ae3 | 2503 | goto fail; |
f6bb4514 BT |
2504 | |
2505 | #ifdef CONFIG_PROC_FS | |
2506 | err = -ENOMEM; | |
2507 | if (!proc_net_fops_create(net, "ip_mr_vif", 0, &ipmr_vif_fops)) | |
2508 | goto proc_vif_fail; | |
2509 | if (!proc_net_fops_create(net, "ip_mr_cache", 0, &ipmr_mfc_fops)) | |
2510 | goto proc_cache_fail; | |
2511 | #endif | |
2bb8b26c BT |
2512 | return 0; |
2513 | ||
f6bb4514 BT |
2514 | #ifdef CONFIG_PROC_FS |
2515 | proc_cache_fail: | |
2516 | proc_net_remove(net, "ip_mr_vif"); | |
2517 | proc_vif_fail: | |
f0ad0860 | 2518 | ipmr_rules_exit(net); |
f6bb4514 | 2519 | #endif |
cf958ae3 BT |
2520 | fail: |
2521 | return err; | |
2522 | } | |
2523 | ||
2524 | static void __net_exit ipmr_net_exit(struct net *net) | |
2525 | { | |
f6bb4514 BT |
2526 | #ifdef CONFIG_PROC_FS |
2527 | proc_net_remove(net, "ip_mr_cache"); | |
2528 | proc_net_remove(net, "ip_mr_vif"); | |
2529 | #endif | |
f0ad0860 | 2530 | ipmr_rules_exit(net); |
cf958ae3 BT |
2531 | } |
2532 | ||
2533 | static struct pernet_operations ipmr_net_ops = { | |
2534 | .init = ipmr_net_init, | |
2535 | .exit = ipmr_net_exit, | |
2536 | }; | |
e905a9ed | 2537 | |
03d2f897 | 2538 | int __init ip_mr_init(void) |
1da177e4 | 2539 | { |
03d2f897 WC |
2540 | int err; |
2541 | ||
1da177e4 LT |
2542 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2543 | sizeof(struct mfc_cache), | |
a8c9486b | 2544 | 0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, |
20c2df83 | 2545 | NULL); |
03d2f897 WC |
2546 | if (!mrt_cachep) |
2547 | return -ENOMEM; | |
2548 | ||
cf958ae3 BT |
2549 | err = register_pernet_subsys(&ipmr_net_ops); |
2550 | if (err) | |
2551 | goto reg_pernet_fail; | |
2552 | ||
03d2f897 WC |
2553 | err = register_netdevice_notifier(&ip_mr_notifier); |
2554 | if (err) | |
2555 | goto reg_notif_fail; | |
403dbb97 TG |
2556 | #ifdef CONFIG_IP_PIMSM_V2 |
2557 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
2558 | printk(KERN_ERR "ip_mr_init: can't add PIM protocol\n"); | |
2559 | err = -EAGAIN; | |
2560 | goto add_proto_fail; | |
2561 | } | |
2562 | #endif | |
cb6a4e46 | 2563 | rtnl_register(RTNL_FAMILY_IPMR, RTM_GETROUTE, NULL, ipmr_rtm_dumproute); |
03d2f897 | 2564 | return 0; |
f6bb4514 | 2565 | |
403dbb97 TG |
2566 | #ifdef CONFIG_IP_PIMSM_V2 |
2567 | add_proto_fail: | |
2568 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2569 | #endif | |
c3e38896 | 2570 | reg_notif_fail: |
cf958ae3 BT |
2571 | unregister_pernet_subsys(&ipmr_net_ops); |
2572 | reg_pernet_fail: | |
c3e38896 | 2573 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2574 | return err; |
1da177e4 | 2575 | } |