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 | |
25 | * Relax this requrement to work with older peers. | |
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> | |
63 | #include <net/ipip.h> | |
64 | #include <net/checksum.h> | |
dc5fc579 | 65 | #include <net/netlink.h> |
1da177e4 LT |
66 | |
67 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
68 | #define CONFIG_IP_PIMSM 1 | |
69 | #endif | |
70 | ||
0c12295a PM |
71 | struct mr_table { |
72 | struct sock *mroute_sk; | |
73 | struct timer_list ipmr_expire_timer; | |
74 | struct list_head mfc_unres_queue; | |
75 | struct list_head mfc_cache_array[MFC_LINES]; | |
76 | struct vif_device vif_table[MAXVIFS]; | |
77 | int maxvif; | |
78 | atomic_t cache_resolve_queue_len; | |
79 | int mroute_do_assert; | |
80 | int mroute_do_pim; | |
81 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
82 | int mroute_reg_vif_num; | |
83 | #endif | |
84 | }; | |
85 | ||
1da177e4 LT |
86 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
87 | Note that the changes are semaphored via rtnl_lock. | |
88 | */ | |
89 | ||
90 | static DEFINE_RWLOCK(mrt_lock); | |
91 | ||
92 | /* | |
93 | * Multicast router control variables | |
94 | */ | |
95 | ||
0c12295a | 96 | #define VIF_EXISTS(_mrt, _idx) ((_mrt)->vif_table[_idx].dev != NULL) |
1da177e4 | 97 | |
1da177e4 LT |
98 | /* Special spinlock for queue of unresolved entries */ |
99 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
100 | ||
101 | /* We return to original Alan's scheme. Hash table of resolved | |
102 | entries is changed only in process context and protected | |
103 | with weak lock mrt_lock. Queue of unresolved entries is protected | |
104 | with strong spinlock mfc_unres_lock. | |
105 | ||
106 | In this case data path is free of exclusive locks at all. | |
107 | */ | |
108 | ||
e18b890b | 109 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 | 110 | |
0c12295a PM |
111 | static int ip_mr_forward(struct net *net, struct mr_table *mrt, |
112 | struct sk_buff *skb, struct mfc_cache *cache, | |
113 | int local); | |
114 | static int ipmr_cache_report(struct mr_table *mrt, | |
4feb88e5 | 115 | struct sk_buff *pkt, vifi_t vifi, int assert); |
0c12295a | 116 | static int ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, |
d658f8a0 | 117 | struct mfc_cache *c, struct rtmsg *rtm); |
1da177e4 | 118 | |
1da177e4 LT |
119 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ |
120 | ||
d607032d WC |
121 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
122 | { | |
4feb88e5 BT |
123 | struct net *net = dev_net(dev); |
124 | ||
d607032d WC |
125 | dev_close(dev); |
126 | ||
4feb88e5 | 127 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 128 | if (dev) { |
5bc3eb7e | 129 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 130 | struct ifreq ifr; |
d607032d WC |
131 | struct ip_tunnel_parm p; |
132 | ||
133 | memset(&p, 0, sizeof(p)); | |
134 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
135 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
136 | p.iph.version = 4; | |
137 | p.iph.ihl = 5; | |
138 | p.iph.protocol = IPPROTO_IPIP; | |
139 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
140 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
141 | ||
5bc3eb7e SH |
142 | if (ops->ndo_do_ioctl) { |
143 | mm_segment_t oldfs = get_fs(); | |
144 | ||
145 | set_fs(KERNEL_DS); | |
146 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
147 | set_fs(oldfs); | |
148 | } | |
d607032d WC |
149 | } |
150 | } | |
151 | ||
1da177e4 | 152 | static |
4feb88e5 | 153 | struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
154 | { |
155 | struct net_device *dev; | |
156 | ||
4feb88e5 | 157 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
158 | |
159 | if (dev) { | |
5bc3eb7e | 160 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
161 | int err; |
162 | struct ifreq ifr; | |
1da177e4 LT |
163 | struct ip_tunnel_parm p; |
164 | struct in_device *in_dev; | |
165 | ||
166 | memset(&p, 0, sizeof(p)); | |
167 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
168 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
169 | p.iph.version = 4; | |
170 | p.iph.ihl = 5; | |
171 | p.iph.protocol = IPPROTO_IPIP; | |
172 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 173 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 174 | |
5bc3eb7e SH |
175 | if (ops->ndo_do_ioctl) { |
176 | mm_segment_t oldfs = get_fs(); | |
177 | ||
178 | set_fs(KERNEL_DS); | |
179 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
180 | set_fs(oldfs); | |
181 | } else | |
182 | err = -EOPNOTSUPP; | |
1da177e4 LT |
183 | |
184 | dev = NULL; | |
185 | ||
4feb88e5 BT |
186 | if (err == 0 && |
187 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 LT |
188 | dev->flags |= IFF_MULTICAST; |
189 | ||
e5ed6399 | 190 | in_dev = __in_dev_get_rtnl(dev); |
71e27da9 | 191 | if (in_dev == NULL) |
1da177e4 | 192 | goto failure; |
71e27da9 HX |
193 | |
194 | ipv4_devconf_setall(in_dev); | |
195 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
1da177e4 LT |
196 | |
197 | if (dev_open(dev)) | |
198 | goto failure; | |
7dc00c82 | 199 | dev_hold(dev); |
1da177e4 LT |
200 | } |
201 | } | |
202 | return dev; | |
203 | ||
204 | failure: | |
205 | /* allow the register to be completed before unregistering. */ | |
206 | rtnl_unlock(); | |
207 | rtnl_lock(); | |
208 | ||
209 | unregister_netdevice(dev); | |
210 | return NULL; | |
211 | } | |
212 | ||
213 | #ifdef CONFIG_IP_PIMSM | |
214 | ||
6fef4c0c | 215 | static netdev_tx_t reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
1da177e4 | 216 | { |
4feb88e5 | 217 | struct net *net = dev_net(dev); |
0c12295a | 218 | struct mr_table *mrt = net->ipv4.mrt; |
4feb88e5 | 219 | |
1da177e4 | 220 | read_lock(&mrt_lock); |
cf3677ae PE |
221 | dev->stats.tx_bytes += skb->len; |
222 | dev->stats.tx_packets++; | |
0c12295a | 223 | ipmr_cache_report(mrt, skb, mrt->mroute_reg_vif_num, IGMPMSG_WHOLEPKT); |
1da177e4 LT |
224 | read_unlock(&mrt_lock); |
225 | kfree_skb(skb); | |
6ed10654 | 226 | return NETDEV_TX_OK; |
1da177e4 LT |
227 | } |
228 | ||
007c3838 SH |
229 | static const struct net_device_ops reg_vif_netdev_ops = { |
230 | .ndo_start_xmit = reg_vif_xmit, | |
231 | }; | |
232 | ||
1da177e4 LT |
233 | static void reg_vif_setup(struct net_device *dev) |
234 | { | |
235 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 236 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 237 | dev->flags = IFF_NOARP; |
007c3838 | 238 | dev->netdev_ops = ®_vif_netdev_ops, |
1da177e4 | 239 | dev->destructor = free_netdev; |
403dbb97 | 240 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
241 | } |
242 | ||
403dbb97 | 243 | static struct net_device *ipmr_reg_vif(struct net *net) |
1da177e4 LT |
244 | { |
245 | struct net_device *dev; | |
246 | struct in_device *in_dev; | |
247 | ||
cf3677ae | 248 | dev = alloc_netdev(0, "pimreg", reg_vif_setup); |
1da177e4 LT |
249 | |
250 | if (dev == NULL) | |
251 | return NULL; | |
252 | ||
403dbb97 TG |
253 | dev_net_set(dev, net); |
254 | ||
1da177e4 LT |
255 | if (register_netdevice(dev)) { |
256 | free_netdev(dev); | |
257 | return NULL; | |
258 | } | |
259 | dev->iflink = 0; | |
260 | ||
71e27da9 HX |
261 | rcu_read_lock(); |
262 | if ((in_dev = __in_dev_get_rcu(dev)) == NULL) { | |
263 | rcu_read_unlock(); | |
1da177e4 | 264 | goto failure; |
71e27da9 | 265 | } |
1da177e4 | 266 | |
71e27da9 HX |
267 | ipv4_devconf_setall(in_dev); |
268 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
269 | rcu_read_unlock(); | |
1da177e4 LT |
270 | |
271 | if (dev_open(dev)) | |
272 | goto failure; | |
273 | ||
7dc00c82 WC |
274 | dev_hold(dev); |
275 | ||
1da177e4 LT |
276 | return dev; |
277 | ||
278 | failure: | |
279 | /* allow the register to be completed before unregistering. */ | |
280 | rtnl_unlock(); | |
281 | rtnl_lock(); | |
282 | ||
283 | unregister_netdevice(dev); | |
284 | return NULL; | |
285 | } | |
286 | #endif | |
287 | ||
288 | /* | |
289 | * Delete a VIF entry | |
7dc00c82 | 290 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 291 | */ |
e905a9ed | 292 | |
0c12295a | 293 | static int vif_delete(struct mr_table *mrt, int vifi, int notify, |
d17fa6fa | 294 | struct list_head *head) |
1da177e4 LT |
295 | { |
296 | struct vif_device *v; | |
297 | struct net_device *dev; | |
298 | struct in_device *in_dev; | |
299 | ||
0c12295a | 300 | if (vifi < 0 || vifi >= mrt->maxvif) |
1da177e4 LT |
301 | return -EADDRNOTAVAIL; |
302 | ||
0c12295a | 303 | v = &mrt->vif_table[vifi]; |
1da177e4 LT |
304 | |
305 | write_lock_bh(&mrt_lock); | |
306 | dev = v->dev; | |
307 | v->dev = NULL; | |
308 | ||
309 | if (!dev) { | |
310 | write_unlock_bh(&mrt_lock); | |
311 | return -EADDRNOTAVAIL; | |
312 | } | |
313 | ||
314 | #ifdef CONFIG_IP_PIMSM | |
0c12295a PM |
315 | if (vifi == mrt->mroute_reg_vif_num) |
316 | mrt->mroute_reg_vif_num = -1; | |
1da177e4 LT |
317 | #endif |
318 | ||
0c12295a | 319 | if (vifi+1 == mrt->maxvif) { |
1da177e4 LT |
320 | int tmp; |
321 | for (tmp=vifi-1; tmp>=0; tmp--) { | |
0c12295a | 322 | if (VIF_EXISTS(mrt, tmp)) |
1da177e4 LT |
323 | break; |
324 | } | |
0c12295a | 325 | mrt->maxvif = tmp+1; |
1da177e4 LT |
326 | } |
327 | ||
328 | write_unlock_bh(&mrt_lock); | |
329 | ||
330 | dev_set_allmulti(dev, -1); | |
331 | ||
e5ed6399 | 332 | if ((in_dev = __in_dev_get_rtnl(dev)) != NULL) { |
42f811b8 | 333 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
1da177e4 LT |
334 | ip_rt_multicast_event(in_dev); |
335 | } | |
336 | ||
7dc00c82 | 337 | if (v->flags&(VIFF_TUNNEL|VIFF_REGISTER) && !notify) |
d17fa6fa | 338 | unregister_netdevice_queue(dev, head); |
1da177e4 LT |
339 | |
340 | dev_put(dev); | |
341 | return 0; | |
342 | } | |
343 | ||
5c0a66f5 BT |
344 | static inline void ipmr_cache_free(struct mfc_cache *c) |
345 | { | |
5c0a66f5 BT |
346 | kmem_cache_free(mrt_cachep, c); |
347 | } | |
348 | ||
1da177e4 LT |
349 | /* Destroy an unresolved cache entry, killing queued skbs |
350 | and reporting error to netlink readers. | |
351 | */ | |
352 | ||
0c12295a | 353 | static void ipmr_destroy_unres(struct mr_table *mrt, struct mfc_cache *c) |
1da177e4 | 354 | { |
0c12295a | 355 | struct net *net = NULL; //mrt->net; |
1da177e4 | 356 | struct sk_buff *skb; |
9ef1d4c7 | 357 | struct nlmsgerr *e; |
1da177e4 | 358 | |
0c12295a | 359 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 | 360 | |
c354e124 | 361 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 362 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
363 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
364 | nlh->nlmsg_type = NLMSG_ERROR; | |
365 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
366 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
367 | e = NLMSG_DATA(nlh); |
368 | e->error = -ETIMEDOUT; | |
369 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 370 | |
4feb88e5 | 371 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
1da177e4 LT |
372 | } else |
373 | kfree_skb(skb); | |
374 | } | |
375 | ||
5c0a66f5 | 376 | ipmr_cache_free(c); |
1da177e4 LT |
377 | } |
378 | ||
379 | ||
e258beb2 | 380 | /* Timer process for the unresolved queue. */ |
1da177e4 | 381 | |
e258beb2 | 382 | static void ipmr_expire_process(unsigned long arg) |
1da177e4 | 383 | { |
0c12295a | 384 | struct mr_table *mrt = (struct mr_table *)arg; |
1da177e4 LT |
385 | unsigned long now; |
386 | unsigned long expires; | |
862465f2 | 387 | struct mfc_cache *c, *next; |
1da177e4 LT |
388 | |
389 | if (!spin_trylock(&mfc_unres_lock)) { | |
0c12295a | 390 | mod_timer(&mrt->ipmr_expire_timer, jiffies+HZ/10); |
1da177e4 LT |
391 | return; |
392 | } | |
393 | ||
0c12295a | 394 | if (list_empty(&mrt->mfc_unres_queue)) |
1da177e4 LT |
395 | goto out; |
396 | ||
397 | now = jiffies; | |
398 | expires = 10*HZ; | |
1da177e4 | 399 | |
0c12295a | 400 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
1da177e4 LT |
401 | if (time_after(c->mfc_un.unres.expires, now)) { |
402 | unsigned long interval = c->mfc_un.unres.expires - now; | |
403 | if (interval < expires) | |
404 | expires = interval; | |
1da177e4 LT |
405 | continue; |
406 | } | |
407 | ||
862465f2 | 408 | list_del(&c->list); |
0c12295a | 409 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
410 | } |
411 | ||
0c12295a PM |
412 | if (!list_empty(&mrt->mfc_unres_queue)) |
413 | mod_timer(&mrt->ipmr_expire_timer, jiffies + expires); | |
1da177e4 LT |
414 | |
415 | out: | |
416 | spin_unlock(&mfc_unres_lock); | |
417 | } | |
418 | ||
419 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
420 | ||
0c12295a | 421 | static void ipmr_update_thresholds(struct mr_table *mrt, struct mfc_cache *cache, |
d658f8a0 | 422 | unsigned char *ttls) |
1da177e4 LT |
423 | { |
424 | int vifi; | |
425 | ||
426 | cache->mfc_un.res.minvif = MAXVIFS; | |
427 | cache->mfc_un.res.maxvif = 0; | |
428 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
429 | ||
0c12295a PM |
430 | for (vifi = 0; vifi < mrt->maxvif; vifi++) { |
431 | if (VIF_EXISTS(mrt, vifi) && | |
cf958ae3 | 432 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
433 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
434 | if (cache->mfc_un.res.minvif > vifi) | |
435 | cache->mfc_un.res.minvif = vifi; | |
436 | if (cache->mfc_un.res.maxvif <= vifi) | |
437 | cache->mfc_un.res.maxvif = vifi + 1; | |
438 | } | |
439 | } | |
440 | } | |
441 | ||
0c12295a PM |
442 | static int vif_add(struct net *net, struct mr_table *mrt, |
443 | struct vifctl *vifc, int mrtsock) | |
1da177e4 LT |
444 | { |
445 | int vifi = vifc->vifc_vifi; | |
0c12295a | 446 | struct vif_device *v = &mrt->vif_table[vifi]; |
1da177e4 LT |
447 | struct net_device *dev; |
448 | struct in_device *in_dev; | |
d607032d | 449 | int err; |
1da177e4 LT |
450 | |
451 | /* Is vif busy ? */ | |
0c12295a | 452 | if (VIF_EXISTS(mrt, vifi)) |
1da177e4 LT |
453 | return -EADDRINUSE; |
454 | ||
455 | switch (vifc->vifc_flags) { | |
456 | #ifdef CONFIG_IP_PIMSM | |
457 | case VIFF_REGISTER: | |
458 | /* | |
459 | * Special Purpose VIF in PIM | |
460 | * All the packets will be sent to the daemon | |
461 | */ | |
0c12295a | 462 | if (mrt->mroute_reg_vif_num >= 0) |
1da177e4 | 463 | return -EADDRINUSE; |
403dbb97 | 464 | dev = ipmr_reg_vif(net); |
1da177e4 LT |
465 | if (!dev) |
466 | return -ENOBUFS; | |
d607032d WC |
467 | err = dev_set_allmulti(dev, 1); |
468 | if (err) { | |
469 | unregister_netdevice(dev); | |
7dc00c82 | 470 | dev_put(dev); |
d607032d WC |
471 | return err; |
472 | } | |
1da177e4 LT |
473 | break; |
474 | #endif | |
e905a9ed | 475 | case VIFF_TUNNEL: |
4feb88e5 | 476 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
477 | if (!dev) |
478 | return -ENOBUFS; | |
d607032d WC |
479 | err = dev_set_allmulti(dev, 1); |
480 | if (err) { | |
481 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 482 | dev_put(dev); |
d607032d WC |
483 | return err; |
484 | } | |
1da177e4 | 485 | break; |
ee5e81f0 I |
486 | |
487 | case VIFF_USE_IFINDEX: | |
1da177e4 | 488 | case 0: |
ee5e81f0 I |
489 | if (vifc->vifc_flags == VIFF_USE_IFINDEX) { |
490 | dev = dev_get_by_index(net, vifc->vifc_lcl_ifindex); | |
491 | if (dev && dev->ip_ptr == NULL) { | |
492 | dev_put(dev); | |
493 | return -EADDRNOTAVAIL; | |
494 | } | |
495 | } else | |
496 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); | |
497 | ||
1da177e4 LT |
498 | if (!dev) |
499 | return -EADDRNOTAVAIL; | |
d607032d | 500 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
501 | if (err) { |
502 | dev_put(dev); | |
d607032d | 503 | return err; |
7dc00c82 | 504 | } |
1da177e4 LT |
505 | break; |
506 | default: | |
507 | return -EINVAL; | |
508 | } | |
509 | ||
d0490cfd DC |
510 | if ((in_dev = __in_dev_get_rtnl(dev)) == NULL) { |
511 | dev_put(dev); | |
1da177e4 | 512 | return -EADDRNOTAVAIL; |
d0490cfd | 513 | } |
42f811b8 | 514 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
1da177e4 LT |
515 | ip_rt_multicast_event(in_dev); |
516 | ||
517 | /* | |
518 | * Fill in the VIF structures | |
519 | */ | |
c354e124 JK |
520 | v->rate_limit = vifc->vifc_rate_limit; |
521 | v->local = vifc->vifc_lcl_addr.s_addr; | |
522 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
523 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
524 | if (!mrtsock) |
525 | v->flags |= VIFF_STATIC; | |
c354e124 | 526 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
527 | v->bytes_in = 0; |
528 | v->bytes_out = 0; | |
529 | v->pkt_in = 0; | |
530 | v->pkt_out = 0; | |
531 | v->link = dev->ifindex; | |
532 | if (v->flags&(VIFF_TUNNEL|VIFF_REGISTER)) | |
533 | v->link = dev->iflink; | |
534 | ||
535 | /* And finish update writing critical data */ | |
536 | write_lock_bh(&mrt_lock); | |
c354e124 | 537 | v->dev = dev; |
1da177e4 LT |
538 | #ifdef CONFIG_IP_PIMSM |
539 | if (v->flags&VIFF_REGISTER) | |
0c12295a | 540 | mrt->mroute_reg_vif_num = vifi; |
1da177e4 | 541 | #endif |
0c12295a PM |
542 | if (vifi+1 > mrt->maxvif) |
543 | mrt->maxvif = vifi+1; | |
1da177e4 LT |
544 | write_unlock_bh(&mrt_lock); |
545 | return 0; | |
546 | } | |
547 | ||
0c12295a | 548 | static struct mfc_cache *ipmr_cache_find(struct mr_table *mrt, |
4feb88e5 BT |
549 | __be32 origin, |
550 | __be32 mcastgrp) | |
1da177e4 | 551 | { |
c354e124 | 552 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
553 | struct mfc_cache *c; |
554 | ||
0c12295a | 555 | list_for_each_entry(c, &mrt->mfc_cache_array[line], list) { |
862465f2 PM |
556 | if (c->mfc_origin == origin && c->mfc_mcastgrp == mcastgrp) |
557 | return c; | |
1da177e4 | 558 | } |
862465f2 | 559 | return NULL; |
1da177e4 LT |
560 | } |
561 | ||
562 | /* | |
563 | * Allocate a multicast cache entry | |
564 | */ | |
d658f8a0 | 565 | static struct mfc_cache *ipmr_cache_alloc(void) |
1da177e4 | 566 | { |
c354e124 JK |
567 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
568 | if (c == NULL) | |
1da177e4 | 569 | return NULL; |
1da177e4 LT |
570 | c->mfc_un.res.minvif = MAXVIFS; |
571 | return c; | |
572 | } | |
573 | ||
d658f8a0 | 574 | static struct mfc_cache *ipmr_cache_alloc_unres(void) |
1da177e4 | 575 | { |
c354e124 JK |
576 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
577 | if (c == NULL) | |
1da177e4 | 578 | return NULL; |
1da177e4 LT |
579 | skb_queue_head_init(&c->mfc_un.unres.unresolved); |
580 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
581 | return c; | |
582 | } | |
583 | ||
584 | /* | |
585 | * A cache entry has gone into a resolved state from queued | |
586 | */ | |
e905a9ed | 587 | |
0c12295a PM |
588 | static void ipmr_cache_resolve(struct net *net, struct mr_table *mrt, |
589 | struct mfc_cache *uc, struct mfc_cache *c) | |
1da177e4 LT |
590 | { |
591 | struct sk_buff *skb; | |
9ef1d4c7 | 592 | struct nlmsgerr *e; |
1da177e4 LT |
593 | |
594 | /* | |
595 | * Play the pending entries through our router | |
596 | */ | |
597 | ||
c354e124 | 598 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 599 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
600 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
601 | ||
0c12295a | 602 | if (ipmr_fill_mroute(mrt, skb, c, NLMSG_DATA(nlh)) > 0) { |
27a884dc ACM |
603 | nlh->nlmsg_len = (skb_tail_pointer(skb) - |
604 | (u8 *)nlh); | |
1da177e4 LT |
605 | } else { |
606 | nlh->nlmsg_type = NLMSG_ERROR; | |
607 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
608 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
609 | e = NLMSG_DATA(nlh); |
610 | e->error = -EMSGSIZE; | |
611 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 612 | } |
2942e900 | 613 | |
d658f8a0 | 614 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
1da177e4 | 615 | } else |
0c12295a | 616 | ip_mr_forward(net, mrt, skb, c, 0); |
1da177e4 LT |
617 | } |
618 | } | |
619 | ||
620 | /* | |
621 | * Bounce a cache query up to mrouted. We could use netlink for this but mrouted | |
622 | * expects the following bizarre scheme. | |
623 | * | |
624 | * Called under mrt_lock. | |
625 | */ | |
e905a9ed | 626 | |
0c12295a | 627 | static int ipmr_cache_report(struct mr_table *mrt, |
4feb88e5 | 628 | struct sk_buff *pkt, vifi_t vifi, int assert) |
1da177e4 LT |
629 | { |
630 | struct sk_buff *skb; | |
c9bdd4b5 | 631 | const int ihl = ip_hdrlen(pkt); |
1da177e4 LT |
632 | struct igmphdr *igmp; |
633 | struct igmpmsg *msg; | |
634 | int ret; | |
635 | ||
636 | #ifdef CONFIG_IP_PIMSM | |
637 | if (assert == IGMPMSG_WHOLEPKT) | |
638 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
639 | else | |
640 | #endif | |
641 | skb = alloc_skb(128, GFP_ATOMIC); | |
642 | ||
132adf54 | 643 | if (!skb) |
1da177e4 LT |
644 | return -ENOBUFS; |
645 | ||
646 | #ifdef CONFIG_IP_PIMSM | |
647 | if (assert == IGMPMSG_WHOLEPKT) { | |
648 | /* Ugly, but we have no choice with this interface. | |
649 | Duplicate old header, fix ihl, length etc. | |
650 | And all this only to mangle msg->im_msgtype and | |
651 | to set msg->im_mbz to "mbz" :-) | |
652 | */ | |
878c8145 ACM |
653 | skb_push(skb, sizeof(struct iphdr)); |
654 | skb_reset_network_header(skb); | |
badff6d0 | 655 | skb_reset_transport_header(skb); |
0272ffc4 | 656 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 657 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
658 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
659 | msg->im_mbz = 0; | |
0c12295a | 660 | msg->im_vif = mrt->mroute_reg_vif_num; |
eddc9ec5 ACM |
661 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
662 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
663 | sizeof(struct iphdr)); | |
e905a9ed | 664 | } else |
1da177e4 | 665 | #endif |
e905a9ed YH |
666 | { |
667 | ||
1da177e4 LT |
668 | /* |
669 | * Copy the IP header | |
670 | */ | |
671 | ||
27a884dc | 672 | skb->network_header = skb->tail; |
ddc7b8e3 | 673 | skb_put(skb, ihl); |
27d7ff46 | 674 | skb_copy_to_linear_data(skb, pkt->data, ihl); |
eddc9ec5 ACM |
675 | ip_hdr(skb)->protocol = 0; /* Flag to the kernel this is a route add */ |
676 | msg = (struct igmpmsg *)skb_network_header(skb); | |
1da177e4 | 677 | msg->im_vif = vifi; |
adf30907 | 678 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); |
1da177e4 LT |
679 | |
680 | /* | |
681 | * Add our header | |
682 | */ | |
683 | ||
c354e124 | 684 | igmp=(struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); |
1da177e4 LT |
685 | igmp->type = |
686 | msg->im_msgtype = assert; | |
687 | igmp->code = 0; | |
eddc9ec5 | 688 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ |
b0e380b1 | 689 | skb->transport_header = skb->network_header; |
e905a9ed | 690 | } |
1da177e4 | 691 | |
0c12295a | 692 | if (mrt->mroute_sk == NULL) { |
1da177e4 LT |
693 | kfree_skb(skb); |
694 | return -EINVAL; | |
695 | } | |
696 | ||
697 | /* | |
698 | * Deliver to mrouted | |
699 | */ | |
0c12295a | 700 | ret = sock_queue_rcv_skb(mrt->mroute_sk, skb); |
70a269e6 | 701 | if (ret < 0) { |
1da177e4 LT |
702 | if (net_ratelimit()) |
703 | printk(KERN_WARNING "mroute: pending queue full, dropping entries.\n"); | |
704 | kfree_skb(skb); | |
705 | } | |
706 | ||
707 | return ret; | |
708 | } | |
709 | ||
710 | /* | |
711 | * Queue a packet for resolution. It gets locked cache entry! | |
712 | */ | |
e905a9ed | 713 | |
1da177e4 | 714 | static int |
0c12295a | 715 | ipmr_cache_unresolved(struct mr_table *mrt, vifi_t vifi, struct sk_buff *skb) |
1da177e4 | 716 | { |
862465f2 | 717 | bool found = false; |
1da177e4 LT |
718 | int err; |
719 | struct mfc_cache *c; | |
eddc9ec5 | 720 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
721 | |
722 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 723 | list_for_each_entry(c, &mrt->mfc_unres_queue, list) { |
e258beb2 | 724 | if (c->mfc_mcastgrp == iph->daddr && |
862465f2 PM |
725 | c->mfc_origin == iph->saddr) { |
726 | found = true; | |
1da177e4 | 727 | break; |
862465f2 | 728 | } |
1da177e4 LT |
729 | } |
730 | ||
862465f2 | 731 | if (!found) { |
1da177e4 LT |
732 | /* |
733 | * Create a new entry if allowable | |
734 | */ | |
735 | ||
0c12295a | 736 | if (atomic_read(&mrt->cache_resolve_queue_len) >= 10 || |
d658f8a0 | 737 | (c = ipmr_cache_alloc_unres()) == NULL) { |
1da177e4 LT |
738 | spin_unlock_bh(&mfc_unres_lock); |
739 | ||
740 | kfree_skb(skb); | |
741 | return -ENOBUFS; | |
742 | } | |
743 | ||
744 | /* | |
745 | * Fill in the new cache entry | |
746 | */ | |
eddc9ec5 ACM |
747 | c->mfc_parent = -1; |
748 | c->mfc_origin = iph->saddr; | |
749 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 LT |
750 | |
751 | /* | |
752 | * Reflect first query at mrouted. | |
753 | */ | |
0c12295a | 754 | err = ipmr_cache_report(mrt, skb, vifi, IGMPMSG_NOCACHE); |
4feb88e5 | 755 | if (err < 0) { |
e905a9ed | 756 | /* If the report failed throw the cache entry |
1da177e4 LT |
757 | out - Brad Parker |
758 | */ | |
759 | spin_unlock_bh(&mfc_unres_lock); | |
760 | ||
5c0a66f5 | 761 | ipmr_cache_free(c); |
1da177e4 LT |
762 | kfree_skb(skb); |
763 | return err; | |
764 | } | |
765 | ||
0c12295a PM |
766 | atomic_inc(&mrt->cache_resolve_queue_len); |
767 | list_add(&c->list, &mrt->mfc_unres_queue); | |
1da177e4 | 768 | |
0c12295a | 769 | mod_timer(&mrt->ipmr_expire_timer, c->mfc_un.unres.expires); |
1da177e4 LT |
770 | } |
771 | ||
772 | /* | |
773 | * See if we can append the packet | |
774 | */ | |
775 | if (c->mfc_un.unres.unresolved.qlen>3) { | |
776 | kfree_skb(skb); | |
777 | err = -ENOBUFS; | |
778 | } else { | |
c354e124 | 779 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
780 | err = 0; |
781 | } | |
782 | ||
783 | spin_unlock_bh(&mfc_unres_lock); | |
784 | return err; | |
785 | } | |
786 | ||
787 | /* | |
788 | * MFC cache manipulation by user space mroute daemon | |
789 | */ | |
790 | ||
0c12295a | 791 | static int ipmr_mfc_delete(struct mr_table *mrt, struct mfcctl *mfc) |
1da177e4 LT |
792 | { |
793 | int line; | |
862465f2 | 794 | struct mfc_cache *c, *next; |
1da177e4 | 795 | |
c354e124 | 796 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 797 | |
0c12295a | 798 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[line], list) { |
1da177e4 LT |
799 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
800 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { | |
801 | write_lock_bh(&mrt_lock); | |
862465f2 | 802 | list_del(&c->list); |
1da177e4 LT |
803 | write_unlock_bh(&mrt_lock); |
804 | ||
5c0a66f5 | 805 | ipmr_cache_free(c); |
1da177e4 LT |
806 | return 0; |
807 | } | |
808 | } | |
809 | return -ENOENT; | |
810 | } | |
811 | ||
0c12295a PM |
812 | static int ipmr_mfc_add(struct net *net, struct mr_table *mrt, |
813 | struct mfcctl *mfc, int mrtsock) | |
1da177e4 | 814 | { |
862465f2 | 815 | bool found = false; |
1da177e4 | 816 | int line; |
862465f2 | 817 | struct mfc_cache *uc, *c; |
1da177e4 | 818 | |
a50436f2 PM |
819 | if (mfc->mfcc_parent >= MAXVIFS) |
820 | return -ENFILE; | |
821 | ||
c354e124 | 822 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 823 | |
0c12295a | 824 | list_for_each_entry(c, &mrt->mfc_cache_array[line], list) { |
1da177e4 | 825 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
862465f2 PM |
826 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { |
827 | found = true; | |
1da177e4 | 828 | break; |
862465f2 | 829 | } |
1da177e4 LT |
830 | } |
831 | ||
862465f2 | 832 | if (found) { |
1da177e4 LT |
833 | write_lock_bh(&mrt_lock); |
834 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 835 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
836 | if (!mrtsock) |
837 | c->mfc_flags |= MFC_STATIC; | |
838 | write_unlock_bh(&mrt_lock); | |
839 | return 0; | |
840 | } | |
841 | ||
f97c1e0c | 842 | if (!ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
843 | return -EINVAL; |
844 | ||
d658f8a0 | 845 | c = ipmr_cache_alloc(); |
c354e124 | 846 | if (c == NULL) |
1da177e4 LT |
847 | return -ENOMEM; |
848 | ||
c354e124 JK |
849 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
850 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
851 | c->mfc_parent = mfc->mfcc_parent; | |
0c12295a | 852 | ipmr_update_thresholds(mrt, c, mfc->mfcc_ttls); |
1da177e4 LT |
853 | if (!mrtsock) |
854 | c->mfc_flags |= MFC_STATIC; | |
855 | ||
856 | write_lock_bh(&mrt_lock); | |
0c12295a | 857 | list_add(&c->list, &mrt->mfc_cache_array[line]); |
1da177e4 LT |
858 | write_unlock_bh(&mrt_lock); |
859 | ||
860 | /* | |
861 | * Check to see if we resolved a queued list. If so we | |
862 | * need to send on the frames and tidy up. | |
863 | */ | |
864 | spin_lock_bh(&mfc_unres_lock); | |
0c12295a | 865 | list_for_each_entry(uc, &mrt->mfc_unres_queue, list) { |
e258beb2 | 866 | if (uc->mfc_origin == c->mfc_origin && |
1da177e4 | 867 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
862465f2 | 868 | list_del(&uc->list); |
0c12295a | 869 | atomic_dec(&mrt->cache_resolve_queue_len); |
1da177e4 LT |
870 | break; |
871 | } | |
872 | } | |
0c12295a PM |
873 | if (list_empty(&mrt->mfc_unres_queue)) |
874 | del_timer(&mrt->ipmr_expire_timer); | |
1da177e4 LT |
875 | spin_unlock_bh(&mfc_unres_lock); |
876 | ||
877 | if (uc) { | |
0c12295a | 878 | ipmr_cache_resolve(net, mrt, uc, c); |
5c0a66f5 | 879 | ipmr_cache_free(uc); |
1da177e4 LT |
880 | } |
881 | return 0; | |
882 | } | |
883 | ||
884 | /* | |
885 | * Close the multicast socket, and clear the vif tables etc | |
886 | */ | |
e905a9ed | 887 | |
0c12295a | 888 | static void mroute_clean_tables(struct mr_table *mrt) |
1da177e4 LT |
889 | { |
890 | int i; | |
d17fa6fa | 891 | LIST_HEAD(list); |
862465f2 | 892 | struct mfc_cache *c, *next; |
e905a9ed | 893 | |
1da177e4 LT |
894 | /* |
895 | * Shut down all active vif entries | |
896 | */ | |
0c12295a PM |
897 | for (i = 0; i < mrt->maxvif; i++) { |
898 | if (!(mrt->vif_table[i].flags&VIFF_STATIC)) | |
899 | vif_delete(mrt, i, 0, &list); | |
1da177e4 | 900 | } |
d17fa6fa | 901 | unregister_netdevice_many(&list); |
1da177e4 LT |
902 | |
903 | /* | |
904 | * Wipe the cache | |
905 | */ | |
862465f2 | 906 | for (i = 0; i < MFC_LINES; i++) { |
0c12295a | 907 | list_for_each_entry_safe(c, next, &mrt->mfc_cache_array[i], list) { |
862465f2 | 908 | if (c->mfc_flags&MFC_STATIC) |
1da177e4 | 909 | continue; |
1da177e4 | 910 | write_lock_bh(&mrt_lock); |
862465f2 | 911 | list_del(&c->list); |
1da177e4 LT |
912 | write_unlock_bh(&mrt_lock); |
913 | ||
5c0a66f5 | 914 | ipmr_cache_free(c); |
1da177e4 LT |
915 | } |
916 | } | |
917 | ||
0c12295a | 918 | if (atomic_read(&mrt->cache_resolve_queue_len) != 0) { |
1da177e4 | 919 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 920 | list_for_each_entry_safe(c, next, &mrt->mfc_unres_queue, list) { |
862465f2 | 921 | list_del(&c->list); |
0c12295a | 922 | ipmr_destroy_unres(mrt, c); |
1da177e4 LT |
923 | } |
924 | spin_unlock_bh(&mfc_unres_lock); | |
925 | } | |
926 | } | |
927 | ||
928 | static void mrtsock_destruct(struct sock *sk) | |
929 | { | |
4feb88e5 | 930 | struct net *net = sock_net(sk); |
0c12295a | 931 | struct mr_table *mrt = net->ipv4.mrt; |
4feb88e5 | 932 | |
1da177e4 | 933 | rtnl_lock(); |
0c12295a | 934 | if (sk == mrt->mroute_sk) { |
4feb88e5 | 935 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; |
1da177e4 LT |
936 | |
937 | write_lock_bh(&mrt_lock); | |
0c12295a | 938 | mrt->mroute_sk = NULL; |
1da177e4 LT |
939 | write_unlock_bh(&mrt_lock); |
940 | ||
0c12295a | 941 | mroute_clean_tables(mrt); |
1da177e4 LT |
942 | } |
943 | rtnl_unlock(); | |
944 | } | |
945 | ||
946 | /* | |
947 | * Socket options and virtual interface manipulation. The whole | |
948 | * virtual interface system is a complete heap, but unfortunately | |
949 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
950 | * MOSPF/PIM router set up we can clean this up. | |
951 | */ | |
e905a9ed | 952 | |
b7058842 | 953 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, unsigned int optlen) |
1da177e4 LT |
954 | { |
955 | int ret; | |
956 | struct vifctl vif; | |
957 | struct mfcctl mfc; | |
4feb88e5 | 958 | struct net *net = sock_net(sk); |
0c12295a | 959 | struct mr_table *mrt = net->ipv4.mrt; |
e905a9ed | 960 | |
132adf54 | 961 | if (optname != MRT_INIT) { |
0c12295a | 962 | if (sk != mrt->mroute_sk && !capable(CAP_NET_ADMIN)) |
1da177e4 LT |
963 | return -EACCES; |
964 | } | |
965 | ||
132adf54 SH |
966 | switch (optname) { |
967 | case MRT_INIT: | |
968 | if (sk->sk_type != SOCK_RAW || | |
c720c7e8 | 969 | inet_sk(sk)->inet_num != IPPROTO_IGMP) |
132adf54 | 970 | return -EOPNOTSUPP; |
c354e124 | 971 | if (optlen != sizeof(int)) |
132adf54 | 972 | return -ENOPROTOOPT; |
1da177e4 | 973 | |
132adf54 | 974 | rtnl_lock(); |
0c12295a | 975 | if (mrt->mroute_sk) { |
1da177e4 | 976 | rtnl_unlock(); |
132adf54 SH |
977 | return -EADDRINUSE; |
978 | } | |
979 | ||
980 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
981 | if (ret == 0) { | |
982 | write_lock_bh(&mrt_lock); | |
0c12295a | 983 | mrt->mroute_sk = sk; |
132adf54 SH |
984 | write_unlock_bh(&mrt_lock); |
985 | ||
4feb88e5 | 986 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
132adf54 SH |
987 | } |
988 | rtnl_unlock(); | |
989 | return ret; | |
990 | case MRT_DONE: | |
0c12295a | 991 | if (sk != mrt->mroute_sk) |
132adf54 SH |
992 | return -EACCES; |
993 | return ip_ra_control(sk, 0, NULL); | |
994 | case MRT_ADD_VIF: | |
995 | case MRT_DEL_VIF: | |
c354e124 | 996 | if (optlen != sizeof(vif)) |
132adf54 | 997 | return -EINVAL; |
c354e124 | 998 | if (copy_from_user(&vif, optval, sizeof(vif))) |
132adf54 SH |
999 | return -EFAULT; |
1000 | if (vif.vifc_vifi >= MAXVIFS) | |
1001 | return -ENFILE; | |
1002 | rtnl_lock(); | |
c354e124 | 1003 | if (optname == MRT_ADD_VIF) { |
0c12295a | 1004 | ret = vif_add(net, mrt, &vif, sk == mrt->mroute_sk); |
132adf54 | 1005 | } else { |
0c12295a | 1006 | ret = vif_delete(mrt, vif.vifc_vifi, 0, NULL); |
132adf54 SH |
1007 | } |
1008 | rtnl_unlock(); | |
1009 | return ret; | |
1da177e4 LT |
1010 | |
1011 | /* | |
1012 | * Manipulate the forwarding caches. These live | |
1013 | * in a sort of kernel/user symbiosis. | |
1014 | */ | |
132adf54 SH |
1015 | case MRT_ADD_MFC: |
1016 | case MRT_DEL_MFC: | |
c354e124 | 1017 | if (optlen != sizeof(mfc)) |
132adf54 | 1018 | return -EINVAL; |
c354e124 | 1019 | if (copy_from_user(&mfc, optval, sizeof(mfc))) |
132adf54 SH |
1020 | return -EFAULT; |
1021 | rtnl_lock(); | |
c354e124 | 1022 | if (optname == MRT_DEL_MFC) |
0c12295a | 1023 | ret = ipmr_mfc_delete(mrt, &mfc); |
132adf54 | 1024 | else |
0c12295a | 1025 | ret = ipmr_mfc_add(net, mrt, &mfc, sk == mrt->mroute_sk); |
132adf54 SH |
1026 | rtnl_unlock(); |
1027 | return ret; | |
1da177e4 LT |
1028 | /* |
1029 | * Control PIM assert. | |
1030 | */ | |
132adf54 SH |
1031 | case MRT_ASSERT: |
1032 | { | |
1033 | int v; | |
1034 | if (get_user(v,(int __user *)optval)) | |
1035 | return -EFAULT; | |
0c12295a | 1036 | mrt->mroute_do_assert = (v) ? 1 : 0; |
132adf54 SH |
1037 | return 0; |
1038 | } | |
1da177e4 | 1039 | #ifdef CONFIG_IP_PIMSM |
132adf54 SH |
1040 | case MRT_PIM: |
1041 | { | |
ba93ef74 SH |
1042 | int v; |
1043 | ||
132adf54 SH |
1044 | if (get_user(v,(int __user *)optval)) |
1045 | return -EFAULT; | |
ba93ef74 SH |
1046 | v = (v) ? 1 : 0; |
1047 | ||
132adf54 SH |
1048 | rtnl_lock(); |
1049 | ret = 0; | |
0c12295a PM |
1050 | if (v != mrt->mroute_do_pim) { |
1051 | mrt->mroute_do_pim = v; | |
1052 | mrt->mroute_do_assert = v; | |
1da177e4 | 1053 | } |
132adf54 SH |
1054 | rtnl_unlock(); |
1055 | return ret; | |
1056 | } | |
1da177e4 | 1057 | #endif |
132adf54 SH |
1058 | /* |
1059 | * Spurious command, or MRT_VERSION which you cannot | |
1060 | * set. | |
1061 | */ | |
1062 | default: | |
1063 | return -ENOPROTOOPT; | |
1da177e4 LT |
1064 | } |
1065 | } | |
1066 | ||
1067 | /* | |
1068 | * Getsock opt support for the multicast routing system. | |
1069 | */ | |
e905a9ed | 1070 | |
c354e124 | 1071 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1072 | { |
1073 | int olr; | |
1074 | int val; | |
4feb88e5 | 1075 | struct net *net = sock_net(sk); |
0c12295a | 1076 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 | 1077 | |
c354e124 | 1078 | if (optname != MRT_VERSION && |
1da177e4 LT |
1079 | #ifdef CONFIG_IP_PIMSM |
1080 | optname!=MRT_PIM && | |
1081 | #endif | |
1082 | optname!=MRT_ASSERT) | |
1083 | return -ENOPROTOOPT; | |
1084 | ||
1085 | if (get_user(olr, optlen)) | |
1086 | return -EFAULT; | |
1087 | ||
1088 | olr = min_t(unsigned int, olr, sizeof(int)); | |
1089 | if (olr < 0) | |
1090 | return -EINVAL; | |
e905a9ed | 1091 | |
c354e124 | 1092 | if (put_user(olr, optlen)) |
1da177e4 | 1093 | return -EFAULT; |
c354e124 JK |
1094 | if (optname == MRT_VERSION) |
1095 | val = 0x0305; | |
1da177e4 | 1096 | #ifdef CONFIG_IP_PIMSM |
c354e124 | 1097 | else if (optname == MRT_PIM) |
0c12295a | 1098 | val = mrt->mroute_do_pim; |
1da177e4 LT |
1099 | #endif |
1100 | else | |
0c12295a | 1101 | val = mrt->mroute_do_assert; |
c354e124 | 1102 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1103 | return -EFAULT; |
1104 | return 0; | |
1105 | } | |
1106 | ||
1107 | /* | |
1108 | * The IP multicast ioctl support routines. | |
1109 | */ | |
e905a9ed | 1110 | |
1da177e4 LT |
1111 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1112 | { | |
1113 | struct sioc_sg_req sr; | |
1114 | struct sioc_vif_req vr; | |
1115 | struct vif_device *vif; | |
1116 | struct mfc_cache *c; | |
4feb88e5 | 1117 | struct net *net = sock_net(sk); |
0c12295a | 1118 | struct mr_table *mrt = net->ipv4.mrt; |
e905a9ed | 1119 | |
132adf54 SH |
1120 | switch (cmd) { |
1121 | case SIOCGETVIFCNT: | |
c354e124 | 1122 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1123 | return -EFAULT; |
0c12295a | 1124 | if (vr.vifi >= mrt->maxvif) |
132adf54 SH |
1125 | return -EINVAL; |
1126 | read_lock(&mrt_lock); | |
0c12295a PM |
1127 | vif = &mrt->vif_table[vr.vifi]; |
1128 | if (VIF_EXISTS(mrt, vr.vifi)) { | |
c354e124 JK |
1129 | vr.icount = vif->pkt_in; |
1130 | vr.ocount = vif->pkt_out; | |
1131 | vr.ibytes = vif->bytes_in; | |
1132 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1133 | read_unlock(&mrt_lock); |
1da177e4 | 1134 | |
c354e124 | 1135 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1136 | return -EFAULT; |
1137 | return 0; | |
1138 | } | |
1139 | read_unlock(&mrt_lock); | |
1140 | return -EADDRNOTAVAIL; | |
1141 | case SIOCGETSGCNT: | |
c354e124 | 1142 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1143 | return -EFAULT; |
1144 | ||
1145 | read_lock(&mrt_lock); | |
0c12295a | 1146 | c = ipmr_cache_find(mrt, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1147 | if (c) { |
1148 | sr.pktcnt = c->mfc_un.res.pkt; | |
1149 | sr.bytecnt = c->mfc_un.res.bytes; | |
1150 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1da177e4 | 1151 | read_unlock(&mrt_lock); |
132adf54 | 1152 | |
c354e124 | 1153 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1154 | return -EFAULT; |
1155 | return 0; | |
1156 | } | |
1157 | read_unlock(&mrt_lock); | |
1158 | return -EADDRNOTAVAIL; | |
1159 | default: | |
1160 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1161 | } |
1162 | } | |
1163 | ||
1164 | ||
1165 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) | |
1166 | { | |
e9dc8653 | 1167 | struct net_device *dev = ptr; |
4feb88e5 | 1168 | struct net *net = dev_net(dev); |
0c12295a | 1169 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 LT |
1170 | struct vif_device *v; |
1171 | int ct; | |
d17fa6fa | 1172 | LIST_HEAD(list); |
e9dc8653 | 1173 | |
1da177e4 LT |
1174 | if (event != NETDEV_UNREGISTER) |
1175 | return NOTIFY_DONE; | |
0c12295a PM |
1176 | v = &mrt->vif_table[0]; |
1177 | for (ct = 0; ct < mrt->maxvif; ct++, v++) { | |
c354e124 | 1178 | if (v->dev == dev) |
0c12295a | 1179 | vif_delete(mrt, ct, 1, &list); |
1da177e4 | 1180 | } |
d17fa6fa | 1181 | unregister_netdevice_many(&list); |
1da177e4 LT |
1182 | return NOTIFY_DONE; |
1183 | } | |
1184 | ||
1185 | ||
c354e124 | 1186 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1187 | .notifier_call = ipmr_device_event, |
1188 | }; | |
1189 | ||
1190 | /* | |
1191 | * Encapsulate a packet by attaching a valid IPIP header to it. | |
1192 | * This avoids tunnel drivers and other mess and gives us the speed so | |
1193 | * important for multicast video. | |
1194 | */ | |
e905a9ed | 1195 | |
114c7844 | 1196 | static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr) |
1da177e4 | 1197 | { |
8856dfa3 | 1198 | struct iphdr *iph; |
eddc9ec5 | 1199 | struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1200 | |
1201 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1202 | skb->transport_header = skb->network_header; |
8856dfa3 | 1203 | skb_reset_network_header(skb); |
eddc9ec5 | 1204 | iph = ip_hdr(skb); |
1da177e4 LT |
1205 | |
1206 | iph->version = 4; | |
e023dd64 ACM |
1207 | iph->tos = old_iph->tos; |
1208 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1209 | iph->frag_off = 0; |
1210 | iph->daddr = daddr; | |
1211 | iph->saddr = saddr; | |
1212 | iph->protocol = IPPROTO_IPIP; | |
1213 | iph->ihl = 5; | |
1214 | iph->tot_len = htons(skb->len); | |
adf30907 | 1215 | ip_select_ident(iph, skb_dst(skb), NULL); |
1da177e4 LT |
1216 | ip_send_check(iph); |
1217 | ||
1da177e4 LT |
1218 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1219 | nf_reset(skb); | |
1220 | } | |
1221 | ||
1222 | static inline int ipmr_forward_finish(struct sk_buff *skb) | |
1223 | { | |
1224 | struct ip_options * opt = &(IPCB(skb)->opt); | |
1225 | ||
adf30907 | 1226 | IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS); |
1da177e4 LT |
1227 | |
1228 | if (unlikely(opt->optlen)) | |
1229 | ip_forward_options(skb); | |
1230 | ||
1231 | return dst_output(skb); | |
1232 | } | |
1233 | ||
1234 | /* | |
1235 | * Processing handlers for ipmr_forward | |
1236 | */ | |
1237 | ||
0c12295a PM |
1238 | static void ipmr_queue_xmit(struct net *net, struct mr_table *mrt, |
1239 | struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1da177e4 | 1240 | { |
eddc9ec5 | 1241 | const struct iphdr *iph = ip_hdr(skb); |
0c12295a | 1242 | struct vif_device *vif = &mrt->vif_table[vifi]; |
1da177e4 LT |
1243 | struct net_device *dev; |
1244 | struct rtable *rt; | |
1245 | int encap = 0; | |
1246 | ||
1247 | if (vif->dev == NULL) | |
1248 | goto out_free; | |
1249 | ||
1250 | #ifdef CONFIG_IP_PIMSM | |
1251 | if (vif->flags & VIFF_REGISTER) { | |
1252 | vif->pkt_out++; | |
c354e124 | 1253 | vif->bytes_out += skb->len; |
cf3677ae PE |
1254 | vif->dev->stats.tx_bytes += skb->len; |
1255 | vif->dev->stats.tx_packets++; | |
0c12295a | 1256 | ipmr_cache_report(mrt, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1257 | goto out_free; |
1da177e4 LT |
1258 | } |
1259 | #endif | |
1260 | ||
1261 | if (vif->flags&VIFF_TUNNEL) { | |
1262 | struct flowi fl = { .oif = vif->link, | |
1263 | .nl_u = { .ip4_u = | |
1264 | { .daddr = vif->remote, | |
1265 | .saddr = vif->local, | |
1266 | .tos = RT_TOS(iph->tos) } }, | |
1267 | .proto = IPPROTO_IPIP }; | |
4feb88e5 | 1268 | if (ip_route_output_key(net, &rt, &fl)) |
1da177e4 LT |
1269 | goto out_free; |
1270 | encap = sizeof(struct iphdr); | |
1271 | } else { | |
1272 | struct flowi fl = { .oif = vif->link, | |
1273 | .nl_u = { .ip4_u = | |
1274 | { .daddr = iph->daddr, | |
1275 | .tos = RT_TOS(iph->tos) } }, | |
1276 | .proto = IPPROTO_IPIP }; | |
4feb88e5 | 1277 | if (ip_route_output_key(net, &rt, &fl)) |
1da177e4 LT |
1278 | goto out_free; |
1279 | } | |
1280 | ||
1281 | dev = rt->u.dst.dev; | |
1282 | ||
1283 | if (skb->len+encap > dst_mtu(&rt->u.dst) && (ntohs(iph->frag_off) & IP_DF)) { | |
1284 | /* Do not fragment multicasts. Alas, IPv4 does not | |
1285 | allow to send ICMP, so that packets will disappear | |
1286 | to blackhole. | |
1287 | */ | |
1288 | ||
7c73a6fa | 1289 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1290 | ip_rt_put(rt); |
1291 | goto out_free; | |
1292 | } | |
1293 | ||
1294 | encap += LL_RESERVED_SPACE(dev) + rt->u.dst.header_len; | |
1295 | ||
1296 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1297 | ip_rt_put(rt); |
1da177e4 LT |
1298 | goto out_free; |
1299 | } | |
1300 | ||
1301 | vif->pkt_out++; | |
c354e124 | 1302 | vif->bytes_out += skb->len; |
1da177e4 | 1303 | |
adf30907 ED |
1304 | skb_dst_drop(skb); |
1305 | skb_dst_set(skb, &rt->u.dst); | |
eddc9ec5 | 1306 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1307 | |
1308 | /* FIXME: forward and output firewalls used to be called here. | |
1309 | * What do we do with netfilter? -- RR */ | |
1310 | if (vif->flags & VIFF_TUNNEL) { | |
1311 | ip_encap(skb, vif->local, vif->remote); | |
1312 | /* FIXME: extra output firewall step used to be here. --RR */ | |
2f4c02d4 PE |
1313 | vif->dev->stats.tx_packets++; |
1314 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1315 | } |
1316 | ||
1317 | IPCB(skb)->flags |= IPSKB_FORWARDED; | |
1318 | ||
1319 | /* | |
1320 | * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally | |
1321 | * not only before forwarding, but after forwarding on all output | |
1322 | * interfaces. It is clear, if mrouter runs a multicasting | |
1323 | * program, it should receive packets not depending to what interface | |
1324 | * program is joined. | |
1325 | * If we will not make it, the program will have to join on all | |
1326 | * interfaces. On the other hand, multihoming host (or router, but | |
1327 | * not mrouter) cannot join to more than one interface - it will | |
1328 | * result in receiving multiple packets. | |
1329 | */ | |
6e23ae2a | 1330 | NF_HOOK(PF_INET, NF_INET_FORWARD, skb, skb->dev, dev, |
1da177e4 LT |
1331 | ipmr_forward_finish); |
1332 | return; | |
1333 | ||
1334 | out_free: | |
1335 | kfree_skb(skb); | |
1336 | return; | |
1337 | } | |
1338 | ||
0c12295a | 1339 | static int ipmr_find_vif(struct mr_table *mrt, struct net_device *dev) |
1da177e4 LT |
1340 | { |
1341 | int ct; | |
0c12295a PM |
1342 | |
1343 | for (ct = mrt->maxvif-1; ct >= 0; ct--) { | |
1344 | if (mrt->vif_table[ct].dev == dev) | |
1da177e4 LT |
1345 | break; |
1346 | } | |
1347 | return ct; | |
1348 | } | |
1349 | ||
1350 | /* "local" means that we should preserve one skb (for local delivery) */ | |
1351 | ||
0c12295a PM |
1352 | static int ip_mr_forward(struct net *net, struct mr_table *mrt, |
1353 | struct sk_buff *skb, struct mfc_cache *cache, | |
1354 | int local) | |
1da177e4 LT |
1355 | { |
1356 | int psend = -1; | |
1357 | int vif, ct; | |
1358 | ||
1359 | vif = cache->mfc_parent; | |
1360 | cache->mfc_un.res.pkt++; | |
1361 | cache->mfc_un.res.bytes += skb->len; | |
1362 | ||
1363 | /* | |
1364 | * Wrong interface: drop packet and (maybe) send PIM assert. | |
1365 | */ | |
0c12295a | 1366 | if (mrt->vif_table[vif].dev != skb->dev) { |
1da177e4 LT |
1367 | int true_vifi; |
1368 | ||
511c3f92 | 1369 | if (skb_rtable(skb)->fl.iif == 0) { |
1da177e4 LT |
1370 | /* It is our own packet, looped back. |
1371 | Very complicated situation... | |
1372 | ||
1373 | The best workaround until routing daemons will be | |
1374 | fixed is not to redistribute packet, if it was | |
1375 | send through wrong interface. It means, that | |
1376 | multicast applications WILL NOT work for | |
1377 | (S,G), which have default multicast route pointing | |
1378 | to wrong oif. In any case, it is not a good | |
1379 | idea to use multicasting applications on router. | |
1380 | */ | |
1381 | goto dont_forward; | |
1382 | } | |
1383 | ||
1384 | cache->mfc_un.res.wrong_if++; | |
0c12295a | 1385 | true_vifi = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1386 | |
0c12295a | 1387 | if (true_vifi >= 0 && mrt->mroute_do_assert && |
1da177e4 LT |
1388 | /* pimsm uses asserts, when switching from RPT to SPT, |
1389 | so that we cannot check that packet arrived on an oif. | |
1390 | It is bad, but otherwise we would need to move pretty | |
1391 | large chunk of pimd to kernel. Ough... --ANK | |
1392 | */ | |
0c12295a | 1393 | (mrt->mroute_do_pim || |
6f9374a9 | 1394 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1395 | time_after(jiffies, |
1da177e4 LT |
1396 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1397 | cache->mfc_un.res.last_assert = jiffies; | |
0c12295a | 1398 | ipmr_cache_report(mrt, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1399 | } |
1400 | goto dont_forward; | |
1401 | } | |
1402 | ||
0c12295a PM |
1403 | mrt->vif_table[vif].pkt_in++; |
1404 | mrt->vif_table[vif].bytes_in += skb->len; | |
1da177e4 LT |
1405 | |
1406 | /* | |
1407 | * Forward the frame | |
1408 | */ | |
1409 | for (ct = cache->mfc_un.res.maxvif-1; ct >= cache->mfc_un.res.minvif; ct--) { | |
eddc9ec5 | 1410 | if (ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1411 | if (psend != -1) { |
1412 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1413 | if (skb2) | |
0c12295a PM |
1414 | ipmr_queue_xmit(net, mrt, skb2, cache, |
1415 | psend); | |
1da177e4 | 1416 | } |
c354e124 | 1417 | psend = ct; |
1da177e4 LT |
1418 | } |
1419 | } | |
1420 | if (psend != -1) { | |
1421 | if (local) { | |
1422 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1423 | if (skb2) | |
0c12295a | 1424 | ipmr_queue_xmit(net, mrt, skb2, cache, psend); |
1da177e4 | 1425 | } else { |
0c12295a | 1426 | ipmr_queue_xmit(net, mrt, skb, cache, psend); |
1da177e4 LT |
1427 | return 0; |
1428 | } | |
1429 | } | |
1430 | ||
1431 | dont_forward: | |
1432 | if (!local) | |
1433 | kfree_skb(skb); | |
1434 | return 0; | |
1435 | } | |
1436 | ||
1437 | ||
1438 | /* | |
1439 | * Multicast packets for forwarding arrive here | |
1440 | */ | |
1441 | ||
1442 | int ip_mr_input(struct sk_buff *skb) | |
1443 | { | |
1444 | struct mfc_cache *cache; | |
4feb88e5 | 1445 | struct net *net = dev_net(skb->dev); |
0c12295a | 1446 | struct mr_table *mrt = net->ipv4.mrt; |
511c3f92 | 1447 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
1da177e4 LT |
1448 | |
1449 | /* Packet is looped back after forward, it should not be | |
1450 | forwarded second time, but still can be delivered locally. | |
1451 | */ | |
1452 | if (IPCB(skb)->flags&IPSKB_FORWARDED) | |
1453 | goto dont_forward; | |
1454 | ||
1455 | if (!local) { | |
1456 | if (IPCB(skb)->opt.router_alert) { | |
1457 | if (ip_call_ra_chain(skb)) | |
1458 | return 0; | |
eddc9ec5 | 1459 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP){ |
1da177e4 LT |
1460 | /* IGMPv1 (and broken IGMPv2 implementations sort of |
1461 | Cisco IOS <= 11.2(8)) do not put router alert | |
1462 | option to IGMP packets destined to routable | |
1463 | groups. It is very bad, because it means | |
1464 | that we can forward NO IGMP messages. | |
1465 | */ | |
1466 | read_lock(&mrt_lock); | |
0c12295a | 1467 | if (mrt->mroute_sk) { |
2715bcf9 | 1468 | nf_reset(skb); |
0c12295a | 1469 | raw_rcv(mrt->mroute_sk, skb); |
1da177e4 LT |
1470 | read_unlock(&mrt_lock); |
1471 | return 0; | |
1472 | } | |
1473 | read_unlock(&mrt_lock); | |
1474 | } | |
1475 | } | |
1476 | ||
1477 | read_lock(&mrt_lock); | |
0c12295a | 1478 | cache = ipmr_cache_find(mrt, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
1da177e4 LT |
1479 | |
1480 | /* | |
1481 | * No usable cache entry | |
1482 | */ | |
c354e124 | 1483 | if (cache == NULL) { |
1da177e4 LT |
1484 | int vif; |
1485 | ||
1486 | if (local) { | |
1487 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1488 | ip_local_deliver(skb); | |
1489 | if (skb2 == NULL) { | |
1490 | read_unlock(&mrt_lock); | |
1491 | return -ENOBUFS; | |
1492 | } | |
1493 | skb = skb2; | |
1494 | } | |
1495 | ||
0c12295a | 1496 | vif = ipmr_find_vif(mrt, skb->dev); |
1da177e4 | 1497 | if (vif >= 0) { |
0c12295a | 1498 | int err = ipmr_cache_unresolved(mrt, vif, skb); |
1da177e4 LT |
1499 | read_unlock(&mrt_lock); |
1500 | ||
1501 | return err; | |
1502 | } | |
1503 | read_unlock(&mrt_lock); | |
1504 | kfree_skb(skb); | |
1505 | return -ENODEV; | |
1506 | } | |
1507 | ||
0c12295a | 1508 | ip_mr_forward(net, mrt, skb, cache, local); |
1da177e4 LT |
1509 | |
1510 | read_unlock(&mrt_lock); | |
1511 | ||
1512 | if (local) | |
1513 | return ip_local_deliver(skb); | |
1514 | ||
1515 | return 0; | |
1516 | ||
1517 | dont_forward: | |
1518 | if (local) | |
1519 | return ip_local_deliver(skb); | |
1520 | kfree_skb(skb); | |
1521 | return 0; | |
1522 | } | |
1523 | ||
b1879204 IJ |
1524 | #ifdef CONFIG_IP_PIMSM |
1525 | static int __pim_rcv(struct sk_buff *skb, unsigned int pimlen) | |
1da177e4 | 1526 | { |
b1879204 IJ |
1527 | struct net_device *reg_dev = NULL; |
1528 | struct iphdr *encap; | |
4feb88e5 | 1529 | struct net *net = dev_net(skb->dev); |
0c12295a | 1530 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 | 1531 | |
b1879204 | 1532 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); |
1da177e4 LT |
1533 | /* |
1534 | Check that: | |
1535 | a. packet is really destinted to a multicast group | |
1536 | b. packet is not a NULL-REGISTER | |
1537 | c. packet is not truncated | |
1538 | */ | |
f97c1e0c | 1539 | if (!ipv4_is_multicast(encap->daddr) || |
1da177e4 | 1540 | encap->tot_len == 0 || |
b1879204 IJ |
1541 | ntohs(encap->tot_len) + pimlen > skb->len) |
1542 | return 1; | |
1da177e4 LT |
1543 | |
1544 | read_lock(&mrt_lock); | |
0c12295a PM |
1545 | if (mrt->mroute_reg_vif_num >= 0) |
1546 | reg_dev = mrt->vif_table[mrt->mroute_reg_vif_num].dev; | |
1da177e4 LT |
1547 | if (reg_dev) |
1548 | dev_hold(reg_dev); | |
1549 | read_unlock(&mrt_lock); | |
1550 | ||
e905a9ed | 1551 | if (reg_dev == NULL) |
b1879204 | 1552 | return 1; |
1da177e4 | 1553 | |
b0e380b1 | 1554 | skb->mac_header = skb->network_header; |
1da177e4 | 1555 | skb_pull(skb, (u8*)encap - skb->data); |
31c7711b | 1556 | skb_reset_network_header(skb); |
1da177e4 | 1557 | skb->dev = reg_dev; |
1da177e4 LT |
1558 | skb->protocol = htons(ETH_P_IP); |
1559 | skb->ip_summed = 0; | |
1560 | skb->pkt_type = PACKET_HOST; | |
adf30907 | 1561 | skb_dst_drop(skb); |
cf3677ae PE |
1562 | reg_dev->stats.rx_bytes += skb->len; |
1563 | reg_dev->stats.rx_packets++; | |
1da177e4 LT |
1564 | nf_reset(skb); |
1565 | netif_rx(skb); | |
1566 | dev_put(reg_dev); | |
b1879204 | 1567 | |
1da177e4 | 1568 | return 0; |
b1879204 IJ |
1569 | } |
1570 | #endif | |
1571 | ||
1572 | #ifdef CONFIG_IP_PIMSM_V1 | |
1573 | /* | |
1574 | * Handle IGMP messages of PIMv1 | |
1575 | */ | |
1576 | ||
1577 | int pim_rcv_v1(struct sk_buff * skb) | |
1578 | { | |
1579 | struct igmphdr *pim; | |
4feb88e5 | 1580 | struct net *net = dev_net(skb->dev); |
0c12295a | 1581 | struct mr_table *mrt = net->ipv4.mrt; |
b1879204 IJ |
1582 | |
1583 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
1584 | goto drop; | |
1585 | ||
1586 | pim = igmp_hdr(skb); | |
1587 | ||
0c12295a | 1588 | if (!mrt->mroute_do_pim || |
b1879204 IJ |
1589 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
1590 | goto drop; | |
1591 | ||
1592 | if (__pim_rcv(skb, sizeof(*pim))) { | |
1593 | drop: | |
1594 | kfree_skb(skb); | |
1595 | } | |
1da177e4 LT |
1596 | return 0; |
1597 | } | |
1598 | #endif | |
1599 | ||
1600 | #ifdef CONFIG_IP_PIMSM_V2 | |
1601 | static int pim_rcv(struct sk_buff * skb) | |
1602 | { | |
1603 | struct pimreghdr *pim; | |
1da177e4 | 1604 | |
b1879204 | 1605 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
1606 | goto drop; |
1607 | ||
9c70220b | 1608 | pim = (struct pimreghdr *)skb_transport_header(skb); |
e905a9ed | 1609 | if (pim->type != ((PIM_VERSION<<4)|(PIM_REGISTER)) || |
1da177e4 | 1610 | (pim->flags&PIM_NULL_REGISTER) || |
e905a9ed | 1611 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 1612 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
1613 | goto drop; |
1614 | ||
b1879204 IJ |
1615 | if (__pim_rcv(skb, sizeof(*pim))) { |
1616 | drop: | |
1617 | kfree_skb(skb); | |
1618 | } | |
1da177e4 LT |
1619 | return 0; |
1620 | } | |
1621 | #endif | |
1622 | ||
1623 | static int | |
0c12295a | 1624 | ipmr_fill_mroute(struct mr_table *mrt, struct sk_buff *skb, struct mfc_cache *c, |
d658f8a0 | 1625 | struct rtmsg *rtm) |
1da177e4 LT |
1626 | { |
1627 | int ct; | |
1628 | struct rtnexthop *nhp; | |
27a884dc | 1629 | u8 *b = skb_tail_pointer(skb); |
1da177e4 LT |
1630 | struct rtattr *mp_head; |
1631 | ||
7438189b ND |
1632 | /* If cache is unresolved, don't try to parse IIF and OIF */ |
1633 | if (c->mfc_parent > MAXVIFS) | |
1634 | return -ENOENT; | |
1635 | ||
0c12295a PM |
1636 | if (VIF_EXISTS(mrt, c->mfc_parent)) |
1637 | RTA_PUT(skb, RTA_IIF, 4, &mrt->vif_table[c->mfc_parent].dev->ifindex); | |
1da177e4 | 1638 | |
c354e124 | 1639 | mp_head = (struct rtattr *)skb_put(skb, RTA_LENGTH(0)); |
1da177e4 LT |
1640 | |
1641 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
0c12295a | 1642 | if (VIF_EXISTS(mrt, ct) && c->mfc_un.res.ttls[ct] < 255) { |
1da177e4 LT |
1643 | if (skb_tailroom(skb) < RTA_ALIGN(RTA_ALIGN(sizeof(*nhp)) + 4)) |
1644 | goto rtattr_failure; | |
c354e124 | 1645 | nhp = (struct rtnexthop *)skb_put(skb, RTA_ALIGN(sizeof(*nhp))); |
1da177e4 LT |
1646 | nhp->rtnh_flags = 0; |
1647 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
0c12295a | 1648 | nhp->rtnh_ifindex = mrt->vif_table[ct].dev->ifindex; |
1da177e4 LT |
1649 | nhp->rtnh_len = sizeof(*nhp); |
1650 | } | |
1651 | } | |
1652 | mp_head->rta_type = RTA_MULTIPATH; | |
27a884dc | 1653 | mp_head->rta_len = skb_tail_pointer(skb) - (u8 *)mp_head; |
1da177e4 LT |
1654 | rtm->rtm_type = RTN_MULTICAST; |
1655 | return 1; | |
1656 | ||
1657 | rtattr_failure: | |
dc5fc579 | 1658 | nlmsg_trim(skb, b); |
1da177e4 LT |
1659 | return -EMSGSIZE; |
1660 | } | |
1661 | ||
4feb88e5 BT |
1662 | int ipmr_get_route(struct net *net, |
1663 | struct sk_buff *skb, struct rtmsg *rtm, int nowait) | |
1da177e4 LT |
1664 | { |
1665 | int err; | |
0c12295a | 1666 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 | 1667 | struct mfc_cache *cache; |
511c3f92 | 1668 | struct rtable *rt = skb_rtable(skb); |
1da177e4 LT |
1669 | |
1670 | read_lock(&mrt_lock); | |
0c12295a | 1671 | cache = ipmr_cache_find(mrt, rt->rt_src, rt->rt_dst); |
1da177e4 | 1672 | |
c354e124 | 1673 | if (cache == NULL) { |
72287490 | 1674 | struct sk_buff *skb2; |
eddc9ec5 | 1675 | struct iphdr *iph; |
1da177e4 LT |
1676 | struct net_device *dev; |
1677 | int vif; | |
1678 | ||
1679 | if (nowait) { | |
1680 | read_unlock(&mrt_lock); | |
1681 | return -EAGAIN; | |
1682 | } | |
1683 | ||
1684 | dev = skb->dev; | |
0c12295a | 1685 | if (dev == NULL || (vif = ipmr_find_vif(mrt, dev)) < 0) { |
1da177e4 LT |
1686 | read_unlock(&mrt_lock); |
1687 | return -ENODEV; | |
1688 | } | |
72287490 AK |
1689 | skb2 = skb_clone(skb, GFP_ATOMIC); |
1690 | if (!skb2) { | |
1691 | read_unlock(&mrt_lock); | |
1692 | return -ENOMEM; | |
1693 | } | |
1694 | ||
e2d1bca7 ACM |
1695 | skb_push(skb2, sizeof(struct iphdr)); |
1696 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
1697 | iph = ip_hdr(skb2); |
1698 | iph->ihl = sizeof(struct iphdr) >> 2; | |
1699 | iph->saddr = rt->rt_src; | |
1700 | iph->daddr = rt->rt_dst; | |
1701 | iph->version = 0; | |
0c12295a | 1702 | err = ipmr_cache_unresolved(mrt, vif, skb2); |
1da177e4 LT |
1703 | read_unlock(&mrt_lock); |
1704 | return err; | |
1705 | } | |
1706 | ||
1707 | if (!nowait && (rtm->rtm_flags&RTM_F_NOTIFY)) | |
1708 | cache->mfc_flags |= MFC_NOTIFY; | |
0c12295a | 1709 | err = ipmr_fill_mroute(mrt, skb, cache, rtm); |
1da177e4 LT |
1710 | read_unlock(&mrt_lock); |
1711 | return err; | |
1712 | } | |
1713 | ||
e905a9ed | 1714 | #ifdef CONFIG_PROC_FS |
1da177e4 LT |
1715 | /* |
1716 | * The /proc interfaces to multicast routing /proc/ip_mr_cache /proc/ip_mr_vif | |
1717 | */ | |
1718 | struct ipmr_vif_iter { | |
f6bb4514 | 1719 | struct seq_net_private p; |
1da177e4 LT |
1720 | int ct; |
1721 | }; | |
1722 | ||
f6bb4514 BT |
1723 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
1724 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
1725 | loff_t pos) |
1726 | { | |
0c12295a PM |
1727 | struct mr_table *mrt = net->ipv4.mrt; |
1728 | ||
1729 | for (iter->ct = 0; iter->ct < mrt->maxvif; ++iter->ct) { | |
1730 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 1731 | continue; |
e905a9ed | 1732 | if (pos-- == 0) |
0c12295a | 1733 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
1734 | } |
1735 | return NULL; | |
1736 | } | |
1737 | ||
1738 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 1739 | __acquires(mrt_lock) |
1da177e4 | 1740 | { |
f6bb4514 BT |
1741 | struct net *net = seq_file_net(seq); |
1742 | ||
1da177e4 | 1743 | read_lock(&mrt_lock); |
f6bb4514 | 1744 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
1745 | : SEQ_START_TOKEN; |
1746 | } | |
1747 | ||
1748 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
1749 | { | |
1750 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 1751 | struct net *net = seq_file_net(seq); |
0c12295a | 1752 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 LT |
1753 | |
1754 | ++*pos; | |
1755 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 1756 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 1757 | |
0c12295a PM |
1758 | while (++iter->ct < mrt->maxvif) { |
1759 | if (!VIF_EXISTS(mrt, iter->ct)) | |
1da177e4 | 1760 | continue; |
0c12295a | 1761 | return &mrt->vif_table[iter->ct]; |
1da177e4 LT |
1762 | } |
1763 | return NULL; | |
1764 | } | |
1765 | ||
1766 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 1767 | __releases(mrt_lock) |
1da177e4 LT |
1768 | { |
1769 | read_unlock(&mrt_lock); | |
1770 | } | |
1771 | ||
1772 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
1773 | { | |
f6bb4514 | 1774 | struct net *net = seq_file_net(seq); |
0c12295a | 1775 | struct mr_table *mrt = net->ipv4.mrt; |
f6bb4514 | 1776 | |
1da177e4 | 1777 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 1778 | seq_puts(seq, |
1da177e4 LT |
1779 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
1780 | } else { | |
1781 | const struct vif_device *vif = v; | |
1782 | const char *name = vif->dev ? vif->dev->name : "none"; | |
1783 | ||
1784 | seq_printf(seq, | |
1785 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
0c12295a | 1786 | vif - mrt->vif_table, |
e905a9ed | 1787 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
1788 | vif->bytes_out, vif->pkt_out, |
1789 | vif->flags, vif->local, vif->remote); | |
1790 | } | |
1791 | return 0; | |
1792 | } | |
1793 | ||
f690808e | 1794 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
1795 | .start = ipmr_vif_seq_start, |
1796 | .next = ipmr_vif_seq_next, | |
1797 | .stop = ipmr_vif_seq_stop, | |
1798 | .show = ipmr_vif_seq_show, | |
1799 | }; | |
1800 | ||
1801 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
1802 | { | |
f6bb4514 BT |
1803 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
1804 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
1805 | } |
1806 | ||
9a32144e | 1807 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
1808 | .owner = THIS_MODULE, |
1809 | .open = ipmr_vif_open, | |
1810 | .read = seq_read, | |
1811 | .llseek = seq_lseek, | |
f6bb4514 | 1812 | .release = seq_release_net, |
1da177e4 LT |
1813 | }; |
1814 | ||
1815 | struct ipmr_mfc_iter { | |
f6bb4514 | 1816 | struct seq_net_private p; |
862465f2 | 1817 | struct list_head *cache; |
1da177e4 LT |
1818 | int ct; |
1819 | }; | |
1820 | ||
1821 | ||
f6bb4514 BT |
1822 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
1823 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 | 1824 | { |
0c12295a | 1825 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 LT |
1826 | struct mfc_cache *mfc; |
1827 | ||
1da177e4 | 1828 | read_lock(&mrt_lock); |
862465f2 | 1829 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) { |
0c12295a | 1830 | it->cache = &mrt->mfc_cache_array[it->ct]; |
862465f2 | 1831 | list_for_each_entry(mfc, it->cache, list) |
e905a9ed | 1832 | if (pos-- == 0) |
1da177e4 | 1833 | return mfc; |
862465f2 | 1834 | } |
1da177e4 LT |
1835 | read_unlock(&mrt_lock); |
1836 | ||
1da177e4 | 1837 | spin_lock_bh(&mfc_unres_lock); |
0c12295a | 1838 | it->cache = &mrt->mfc_unres_queue; |
862465f2 | 1839 | list_for_each_entry(mfc, it->cache, list) |
e258beb2 | 1840 | if (pos-- == 0) |
1da177e4 LT |
1841 | return mfc; |
1842 | spin_unlock_bh(&mfc_unres_lock); | |
1843 | ||
1844 | it->cache = NULL; | |
1845 | return NULL; | |
1846 | } | |
1847 | ||
1848 | ||
1849 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
1850 | { | |
1851 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 BT |
1852 | struct net *net = seq_file_net(seq); |
1853 | ||
1da177e4 LT |
1854 | it->cache = NULL; |
1855 | it->ct = 0; | |
f6bb4514 | 1856 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
1857 | : SEQ_START_TOKEN; |
1858 | } | |
1859 | ||
1860 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
1861 | { | |
1862 | struct mfc_cache *mfc = v; | |
1863 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 1864 | struct net *net = seq_file_net(seq); |
0c12295a | 1865 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 LT |
1866 | |
1867 | ++*pos; | |
1868 | ||
1869 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 1870 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 | 1871 | |
862465f2 PM |
1872 | if (mfc->list.next != it->cache) |
1873 | return list_entry(mfc->list.next, struct mfc_cache, list); | |
e905a9ed | 1874 | |
0c12295a | 1875 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 LT |
1876 | goto end_of_list; |
1877 | ||
0c12295a | 1878 | BUG_ON(it->cache != &mrt->mfc_cache_array[it->ct]); |
1da177e4 LT |
1879 | |
1880 | while (++it->ct < MFC_LINES) { | |
0c12295a | 1881 | it->cache = &mrt->mfc_cache_array[it->ct]; |
862465f2 PM |
1882 | if (list_empty(it->cache)) |
1883 | continue; | |
1884 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 LT |
1885 | } |
1886 | ||
1887 | /* exhausted cache_array, show unresolved */ | |
1888 | read_unlock(&mrt_lock); | |
0c12295a | 1889 | it->cache = &mrt->mfc_unres_queue; |
1da177e4 | 1890 | it->ct = 0; |
e905a9ed | 1891 | |
1da177e4 | 1892 | spin_lock_bh(&mfc_unres_lock); |
862465f2 PM |
1893 | if (!list_empty(it->cache)) |
1894 | return list_first_entry(it->cache, struct mfc_cache, list); | |
1da177e4 LT |
1895 | |
1896 | end_of_list: | |
1897 | spin_unlock_bh(&mfc_unres_lock); | |
1898 | it->cache = NULL; | |
1899 | ||
1900 | return NULL; | |
1901 | } | |
1902 | ||
1903 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
1904 | { | |
1905 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 1906 | struct net *net = seq_file_net(seq); |
0c12295a | 1907 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 | 1908 | |
0c12295a | 1909 | if (it->cache == &mrt->mfc_unres_queue) |
1da177e4 | 1910 | spin_unlock_bh(&mfc_unres_lock); |
0c12295a | 1911 | else if (it->cache == &mrt->mfc_cache_array[it->ct]) |
1da177e4 LT |
1912 | read_unlock(&mrt_lock); |
1913 | } | |
1914 | ||
1915 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
1916 | { | |
1917 | int n; | |
f6bb4514 | 1918 | struct net *net = seq_file_net(seq); |
0c12295a | 1919 | struct mr_table *mrt = net->ipv4.mrt; |
1da177e4 LT |
1920 | |
1921 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 1922 | seq_puts(seq, |
1da177e4 LT |
1923 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
1924 | } else { | |
1925 | const struct mfc_cache *mfc = v; | |
1926 | const struct ipmr_mfc_iter *it = seq->private; | |
e905a9ed | 1927 | |
999890b2 | 1928 | seq_printf(seq, "%08lX %08lX %-3hd", |
1da177e4 LT |
1929 | (unsigned long) mfc->mfc_mcastgrp, |
1930 | (unsigned long) mfc->mfc_origin, | |
1ea472e2 | 1931 | mfc->mfc_parent); |
1da177e4 | 1932 | |
0c12295a | 1933 | if (it->cache != &mrt->mfc_unres_queue) { |
1ea472e2 BT |
1934 | seq_printf(seq, " %8lu %8lu %8lu", |
1935 | mfc->mfc_un.res.pkt, | |
1936 | mfc->mfc_un.res.bytes, | |
1937 | mfc->mfc_un.res.wrong_if); | |
132adf54 SH |
1938 | for (n = mfc->mfc_un.res.minvif; |
1939 | n < mfc->mfc_un.res.maxvif; n++ ) { | |
0c12295a | 1940 | if (VIF_EXISTS(mrt, n) && |
cf958ae3 BT |
1941 | mfc->mfc_un.res.ttls[n] < 255) |
1942 | seq_printf(seq, | |
e905a9ed | 1943 | " %2d:%-3d", |
1da177e4 LT |
1944 | n, mfc->mfc_un.res.ttls[n]); |
1945 | } | |
1ea472e2 BT |
1946 | } else { |
1947 | /* unresolved mfc_caches don't contain | |
1948 | * pkt, bytes and wrong_if values | |
1949 | */ | |
1950 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
1951 | } |
1952 | seq_putc(seq, '\n'); | |
1953 | } | |
1954 | return 0; | |
1955 | } | |
1956 | ||
f690808e | 1957 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
1958 | .start = ipmr_mfc_seq_start, |
1959 | .next = ipmr_mfc_seq_next, | |
1960 | .stop = ipmr_mfc_seq_stop, | |
1961 | .show = ipmr_mfc_seq_show, | |
1962 | }; | |
1963 | ||
1964 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
1965 | { | |
f6bb4514 BT |
1966 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
1967 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
1968 | } |
1969 | ||
9a32144e | 1970 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
1971 | .owner = THIS_MODULE, |
1972 | .open = ipmr_mfc_open, | |
1973 | .read = seq_read, | |
1974 | .llseek = seq_lseek, | |
f6bb4514 | 1975 | .release = seq_release_net, |
1da177e4 | 1976 | }; |
e905a9ed | 1977 | #endif |
1da177e4 LT |
1978 | |
1979 | #ifdef CONFIG_IP_PIMSM_V2 | |
32613090 | 1980 | static const struct net_protocol pim_protocol = { |
1da177e4 | 1981 | .handler = pim_rcv, |
403dbb97 | 1982 | .netns_ok = 1, |
1da177e4 LT |
1983 | }; |
1984 | #endif | |
1985 | ||
1986 | ||
1987 | /* | |
1988 | * Setup for IP multicast routing | |
1989 | */ | |
cf958ae3 BT |
1990 | static int __net_init ipmr_net_init(struct net *net) |
1991 | { | |
0c12295a | 1992 | struct mr_table *mrt; |
862465f2 | 1993 | unsigned int i; |
cf958ae3 BT |
1994 | int err = 0; |
1995 | ||
0c12295a PM |
1996 | mrt = kzalloc(sizeof(*mrt), GFP_KERNEL); |
1997 | if (mrt == NULL) { | |
cf958ae3 BT |
1998 | err = -ENOMEM; |
1999 | goto fail; | |
2000 | } | |
2bb8b26c BT |
2001 | |
2002 | /* Forwarding cache */ | |
862465f2 | 2003 | for (i = 0; i < MFC_LINES; i++) |
0c12295a | 2004 | INIT_LIST_HEAD(&mrt->mfc_cache_array[i]); |
862465f2 | 2005 | |
0c12295a | 2006 | INIT_LIST_HEAD(&mrt->mfc_unres_queue); |
862465f2 | 2007 | |
0c12295a | 2008 | setup_timer(&mrt->ipmr_expire_timer, ipmr_expire_process, |
e258beb2 PM |
2009 | (unsigned long)net); |
2010 | ||
6c5143db | 2011 | #ifdef CONFIG_IP_PIMSM |
0c12295a | 2012 | mrt->mroute_reg_vif_num = -1; |
6c5143db | 2013 | #endif |
f6bb4514 BT |
2014 | |
2015 | #ifdef CONFIG_PROC_FS | |
2016 | err = -ENOMEM; | |
2017 | if (!proc_net_fops_create(net, "ip_mr_vif", 0, &ipmr_vif_fops)) | |
2018 | goto proc_vif_fail; | |
2019 | if (!proc_net_fops_create(net, "ip_mr_cache", 0, &ipmr_mfc_fops)) | |
2020 | goto proc_cache_fail; | |
2021 | #endif | |
0c12295a PM |
2022 | |
2023 | net->ipv4.mrt = mrt; | |
2bb8b26c BT |
2024 | return 0; |
2025 | ||
f6bb4514 BT |
2026 | #ifdef CONFIG_PROC_FS |
2027 | proc_cache_fail: | |
2028 | proc_net_remove(net, "ip_mr_vif"); | |
2029 | proc_vif_fail: | |
0c12295a | 2030 | kfree(mrt); |
f6bb4514 | 2031 | #endif |
cf958ae3 BT |
2032 | fail: |
2033 | return err; | |
2034 | } | |
2035 | ||
2036 | static void __net_exit ipmr_net_exit(struct net *net) | |
2037 | { | |
f6bb4514 BT |
2038 | #ifdef CONFIG_PROC_FS |
2039 | proc_net_remove(net, "ip_mr_cache"); | |
2040 | proc_net_remove(net, "ip_mr_vif"); | |
2041 | #endif | |
0c12295a | 2042 | kfree(net->ipv4.mrt); |
cf958ae3 BT |
2043 | } |
2044 | ||
2045 | static struct pernet_operations ipmr_net_ops = { | |
2046 | .init = ipmr_net_init, | |
2047 | .exit = ipmr_net_exit, | |
2048 | }; | |
e905a9ed | 2049 | |
03d2f897 | 2050 | int __init ip_mr_init(void) |
1da177e4 | 2051 | { |
03d2f897 WC |
2052 | int err; |
2053 | ||
1da177e4 LT |
2054 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2055 | sizeof(struct mfc_cache), | |
e5d679f3 | 2056 | 0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, |
20c2df83 | 2057 | NULL); |
03d2f897 WC |
2058 | if (!mrt_cachep) |
2059 | return -ENOMEM; | |
2060 | ||
cf958ae3 BT |
2061 | err = register_pernet_subsys(&ipmr_net_ops); |
2062 | if (err) | |
2063 | goto reg_pernet_fail; | |
2064 | ||
03d2f897 WC |
2065 | err = register_netdevice_notifier(&ip_mr_notifier); |
2066 | if (err) | |
2067 | goto reg_notif_fail; | |
403dbb97 TG |
2068 | #ifdef CONFIG_IP_PIMSM_V2 |
2069 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
2070 | printk(KERN_ERR "ip_mr_init: can't add PIM protocol\n"); | |
2071 | err = -EAGAIN; | |
2072 | goto add_proto_fail; | |
2073 | } | |
2074 | #endif | |
03d2f897 | 2075 | return 0; |
f6bb4514 | 2076 | |
403dbb97 TG |
2077 | #ifdef CONFIG_IP_PIMSM_V2 |
2078 | add_proto_fail: | |
2079 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2080 | #endif | |
c3e38896 | 2081 | reg_notif_fail: |
cf958ae3 BT |
2082 | unregister_pernet_subsys(&ipmr_net_ops); |
2083 | reg_pernet_fail: | |
c3e38896 | 2084 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2085 | return err; |
1da177e4 | 2086 | } |