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> |
457c4cbc | 50 | #include <net/net_namespace.h> |
1da177e4 LT |
51 | #include <net/ip.h> |
52 | #include <net/protocol.h> | |
53 | #include <linux/skbuff.h> | |
14c85021 | 54 | #include <net/route.h> |
1da177e4 LT |
55 | #include <net/sock.h> |
56 | #include <net/icmp.h> | |
57 | #include <net/udp.h> | |
58 | #include <net/raw.h> | |
59 | #include <linux/notifier.h> | |
60 | #include <linux/if_arp.h> | |
61 | #include <linux/netfilter_ipv4.h> | |
62 | #include <net/ipip.h> | |
63 | #include <net/checksum.h> | |
dc5fc579 | 64 | #include <net/netlink.h> |
1da177e4 LT |
65 | |
66 | #if defined(CONFIG_IP_PIMSM_V1) || defined(CONFIG_IP_PIMSM_V2) | |
67 | #define CONFIG_IP_PIMSM 1 | |
68 | #endif | |
69 | ||
1da177e4 LT |
70 | /* Big lock, protecting vif table, mrt cache and mroute socket state. |
71 | Note that the changes are semaphored via rtnl_lock. | |
72 | */ | |
73 | ||
74 | static DEFINE_RWLOCK(mrt_lock); | |
75 | ||
76 | /* | |
77 | * Multicast router control variables | |
78 | */ | |
79 | ||
cf958ae3 | 80 | #define VIF_EXISTS(_net, _idx) ((_net)->ipv4.vif_table[_idx].dev != NULL) |
1da177e4 | 81 | |
1da177e4 | 82 | static struct mfc_cache *mfc_unres_queue; /* Queue of unresolved entries */ |
1da177e4 LT |
83 | |
84 | /* Special spinlock for queue of unresolved entries */ | |
85 | static DEFINE_SPINLOCK(mfc_unres_lock); | |
86 | ||
87 | /* We return to original Alan's scheme. Hash table of resolved | |
88 | entries is changed only in process context and protected | |
89 | with weak lock mrt_lock. Queue of unresolved entries is protected | |
90 | with strong spinlock mfc_unres_lock. | |
91 | ||
92 | In this case data path is free of exclusive locks at all. | |
93 | */ | |
94 | ||
e18b890b | 95 | static struct kmem_cache *mrt_cachep __read_mostly; |
1da177e4 LT |
96 | |
97 | static int ip_mr_forward(struct sk_buff *skb, struct mfc_cache *cache, int local); | |
4feb88e5 BT |
98 | static int ipmr_cache_report(struct net *net, |
99 | struct sk_buff *pkt, vifi_t vifi, int assert); | |
1da177e4 LT |
100 | static int ipmr_fill_mroute(struct sk_buff *skb, struct mfc_cache *c, struct rtmsg *rtm); |
101 | ||
102 | #ifdef CONFIG_IP_PIMSM_V2 | |
103 | static struct net_protocol pim_protocol; | |
104 | #endif | |
105 | ||
106 | static struct timer_list ipmr_expire_timer; | |
107 | ||
108 | /* Service routines creating virtual interfaces: DVMRP tunnels and PIMREG */ | |
109 | ||
d607032d WC |
110 | static void ipmr_del_tunnel(struct net_device *dev, struct vifctl *v) |
111 | { | |
4feb88e5 BT |
112 | struct net *net = dev_net(dev); |
113 | ||
d607032d WC |
114 | dev_close(dev); |
115 | ||
4feb88e5 | 116 | dev = __dev_get_by_name(net, "tunl0"); |
d607032d | 117 | if (dev) { |
5bc3eb7e | 118 | const struct net_device_ops *ops = dev->netdev_ops; |
d607032d | 119 | struct ifreq ifr; |
d607032d WC |
120 | struct ip_tunnel_parm p; |
121 | ||
122 | memset(&p, 0, sizeof(p)); | |
123 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
124 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
125 | p.iph.version = 4; | |
126 | p.iph.ihl = 5; | |
127 | p.iph.protocol = IPPROTO_IPIP; | |
128 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
129 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; | |
130 | ||
5bc3eb7e SH |
131 | if (ops->ndo_do_ioctl) { |
132 | mm_segment_t oldfs = get_fs(); | |
133 | ||
134 | set_fs(KERNEL_DS); | |
135 | ops->ndo_do_ioctl(dev, &ifr, SIOCDELTUNNEL); | |
136 | set_fs(oldfs); | |
137 | } | |
d607032d WC |
138 | } |
139 | } | |
140 | ||
1da177e4 | 141 | static |
4feb88e5 | 142 | struct net_device *ipmr_new_tunnel(struct net *net, struct vifctl *v) |
1da177e4 LT |
143 | { |
144 | struct net_device *dev; | |
145 | ||
4feb88e5 | 146 | dev = __dev_get_by_name(net, "tunl0"); |
1da177e4 LT |
147 | |
148 | if (dev) { | |
5bc3eb7e | 149 | const struct net_device_ops *ops = dev->netdev_ops; |
1da177e4 LT |
150 | int err; |
151 | struct ifreq ifr; | |
1da177e4 LT |
152 | struct ip_tunnel_parm p; |
153 | struct in_device *in_dev; | |
154 | ||
155 | memset(&p, 0, sizeof(p)); | |
156 | p.iph.daddr = v->vifc_rmt_addr.s_addr; | |
157 | p.iph.saddr = v->vifc_lcl_addr.s_addr; | |
158 | p.iph.version = 4; | |
159 | p.iph.ihl = 5; | |
160 | p.iph.protocol = IPPROTO_IPIP; | |
161 | sprintf(p.name, "dvmrp%d", v->vifc_vifi); | |
ba93ef74 | 162 | ifr.ifr_ifru.ifru_data = (__force void __user *)&p; |
1da177e4 | 163 | |
5bc3eb7e SH |
164 | if (ops->ndo_do_ioctl) { |
165 | mm_segment_t oldfs = get_fs(); | |
166 | ||
167 | set_fs(KERNEL_DS); | |
168 | err = ops->ndo_do_ioctl(dev, &ifr, SIOCADDTUNNEL); | |
169 | set_fs(oldfs); | |
170 | } else | |
171 | err = -EOPNOTSUPP; | |
1da177e4 LT |
172 | |
173 | dev = NULL; | |
174 | ||
4feb88e5 BT |
175 | if (err == 0 && |
176 | (dev = __dev_get_by_name(net, p.name)) != NULL) { | |
1da177e4 LT |
177 | dev->flags |= IFF_MULTICAST; |
178 | ||
e5ed6399 | 179 | in_dev = __in_dev_get_rtnl(dev); |
71e27da9 | 180 | if (in_dev == NULL) |
1da177e4 | 181 | goto failure; |
71e27da9 HX |
182 | |
183 | ipv4_devconf_setall(in_dev); | |
184 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
1da177e4 LT |
185 | |
186 | if (dev_open(dev)) | |
187 | goto failure; | |
7dc00c82 | 188 | dev_hold(dev); |
1da177e4 LT |
189 | } |
190 | } | |
191 | return dev; | |
192 | ||
193 | failure: | |
194 | /* allow the register to be completed before unregistering. */ | |
195 | rtnl_unlock(); | |
196 | rtnl_lock(); | |
197 | ||
198 | unregister_netdevice(dev); | |
199 | return NULL; | |
200 | } | |
201 | ||
202 | #ifdef CONFIG_IP_PIMSM | |
203 | ||
1da177e4 LT |
204 | static int reg_vif_xmit(struct sk_buff *skb, struct net_device *dev) |
205 | { | |
4feb88e5 BT |
206 | struct net *net = dev_net(dev); |
207 | ||
1da177e4 | 208 | read_lock(&mrt_lock); |
cf3677ae PE |
209 | dev->stats.tx_bytes += skb->len; |
210 | dev->stats.tx_packets++; | |
4feb88e5 BT |
211 | ipmr_cache_report(net, skb, net->ipv4.mroute_reg_vif_num, |
212 | IGMPMSG_WHOLEPKT); | |
1da177e4 LT |
213 | read_unlock(&mrt_lock); |
214 | kfree_skb(skb); | |
6ed10654 | 215 | return NETDEV_TX_OK; |
1da177e4 LT |
216 | } |
217 | ||
007c3838 SH |
218 | static const struct net_device_ops reg_vif_netdev_ops = { |
219 | .ndo_start_xmit = reg_vif_xmit, | |
220 | }; | |
221 | ||
1da177e4 LT |
222 | static void reg_vif_setup(struct net_device *dev) |
223 | { | |
224 | dev->type = ARPHRD_PIMREG; | |
46f25dff | 225 | dev->mtu = ETH_DATA_LEN - sizeof(struct iphdr) - 8; |
1da177e4 | 226 | dev->flags = IFF_NOARP; |
007c3838 | 227 | dev->netdev_ops = ®_vif_netdev_ops, |
1da177e4 | 228 | dev->destructor = free_netdev; |
403dbb97 | 229 | dev->features |= NETIF_F_NETNS_LOCAL; |
1da177e4 LT |
230 | } |
231 | ||
403dbb97 | 232 | static struct net_device *ipmr_reg_vif(struct net *net) |
1da177e4 LT |
233 | { |
234 | struct net_device *dev; | |
235 | struct in_device *in_dev; | |
236 | ||
cf3677ae | 237 | dev = alloc_netdev(0, "pimreg", reg_vif_setup); |
1da177e4 LT |
238 | |
239 | if (dev == NULL) | |
240 | return NULL; | |
241 | ||
403dbb97 TG |
242 | dev_net_set(dev, net); |
243 | ||
1da177e4 LT |
244 | if (register_netdevice(dev)) { |
245 | free_netdev(dev); | |
246 | return NULL; | |
247 | } | |
248 | dev->iflink = 0; | |
249 | ||
71e27da9 HX |
250 | rcu_read_lock(); |
251 | if ((in_dev = __in_dev_get_rcu(dev)) == NULL) { | |
252 | rcu_read_unlock(); | |
1da177e4 | 253 | goto failure; |
71e27da9 | 254 | } |
1da177e4 | 255 | |
71e27da9 HX |
256 | ipv4_devconf_setall(in_dev); |
257 | IPV4_DEVCONF(in_dev->cnf, RP_FILTER) = 0; | |
258 | rcu_read_unlock(); | |
1da177e4 LT |
259 | |
260 | if (dev_open(dev)) | |
261 | goto failure; | |
262 | ||
7dc00c82 WC |
263 | dev_hold(dev); |
264 | ||
1da177e4 LT |
265 | return dev; |
266 | ||
267 | failure: | |
268 | /* allow the register to be completed before unregistering. */ | |
269 | rtnl_unlock(); | |
270 | rtnl_lock(); | |
271 | ||
272 | unregister_netdevice(dev); | |
273 | return NULL; | |
274 | } | |
275 | #endif | |
276 | ||
277 | /* | |
278 | * Delete a VIF entry | |
7dc00c82 | 279 | * @notify: Set to 1, if the caller is a notifier_call |
1da177e4 | 280 | */ |
e905a9ed | 281 | |
4feb88e5 | 282 | static int vif_delete(struct net *net, int vifi, int notify) |
1da177e4 LT |
283 | { |
284 | struct vif_device *v; | |
285 | struct net_device *dev; | |
286 | struct in_device *in_dev; | |
287 | ||
4feb88e5 | 288 | if (vifi < 0 || vifi >= net->ipv4.maxvif) |
1da177e4 LT |
289 | return -EADDRNOTAVAIL; |
290 | ||
4feb88e5 | 291 | v = &net->ipv4.vif_table[vifi]; |
1da177e4 LT |
292 | |
293 | write_lock_bh(&mrt_lock); | |
294 | dev = v->dev; | |
295 | v->dev = NULL; | |
296 | ||
297 | if (!dev) { | |
298 | write_unlock_bh(&mrt_lock); | |
299 | return -EADDRNOTAVAIL; | |
300 | } | |
301 | ||
302 | #ifdef CONFIG_IP_PIMSM | |
4feb88e5 BT |
303 | if (vifi == net->ipv4.mroute_reg_vif_num) |
304 | net->ipv4.mroute_reg_vif_num = -1; | |
1da177e4 LT |
305 | #endif |
306 | ||
4feb88e5 | 307 | if (vifi+1 == net->ipv4.maxvif) { |
1da177e4 LT |
308 | int tmp; |
309 | for (tmp=vifi-1; tmp>=0; tmp--) { | |
4feb88e5 | 310 | if (VIF_EXISTS(net, tmp)) |
1da177e4 LT |
311 | break; |
312 | } | |
4feb88e5 | 313 | net->ipv4.maxvif = tmp+1; |
1da177e4 LT |
314 | } |
315 | ||
316 | write_unlock_bh(&mrt_lock); | |
317 | ||
318 | dev_set_allmulti(dev, -1); | |
319 | ||
e5ed6399 | 320 | if ((in_dev = __in_dev_get_rtnl(dev)) != NULL) { |
42f811b8 | 321 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)--; |
1da177e4 LT |
322 | ip_rt_multicast_event(in_dev); |
323 | } | |
324 | ||
7dc00c82 | 325 | if (v->flags&(VIFF_TUNNEL|VIFF_REGISTER) && !notify) |
1da177e4 LT |
326 | unregister_netdevice(dev); |
327 | ||
328 | dev_put(dev); | |
329 | return 0; | |
330 | } | |
331 | ||
5c0a66f5 BT |
332 | static inline void ipmr_cache_free(struct mfc_cache *c) |
333 | { | |
334 | release_net(mfc_net(c)); | |
335 | kmem_cache_free(mrt_cachep, c); | |
336 | } | |
337 | ||
1da177e4 LT |
338 | /* Destroy an unresolved cache entry, killing queued skbs |
339 | and reporting error to netlink readers. | |
340 | */ | |
341 | ||
342 | static void ipmr_destroy_unres(struct mfc_cache *c) | |
343 | { | |
344 | struct sk_buff *skb; | |
9ef1d4c7 | 345 | struct nlmsgerr *e; |
4feb88e5 | 346 | struct net *net = mfc_net(c); |
1da177e4 | 347 | |
4feb88e5 | 348 | atomic_dec(&net->ipv4.cache_resolve_queue_len); |
1da177e4 | 349 | |
c354e124 | 350 | while ((skb = skb_dequeue(&c->mfc_un.unres.unresolved))) { |
eddc9ec5 | 351 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
352 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
353 | nlh->nlmsg_type = NLMSG_ERROR; | |
354 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
355 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
356 | e = NLMSG_DATA(nlh); |
357 | e->error = -ETIMEDOUT; | |
358 | memset(&e->msg, 0, sizeof(e->msg)); | |
2942e900 | 359 | |
4feb88e5 | 360 | rtnl_unicast(skb, net, NETLINK_CB(skb).pid); |
1da177e4 LT |
361 | } else |
362 | kfree_skb(skb); | |
363 | } | |
364 | ||
5c0a66f5 | 365 | ipmr_cache_free(c); |
1da177e4 LT |
366 | } |
367 | ||
368 | ||
369 | /* Single timer process for all the unresolved queue. */ | |
370 | ||
371 | static void ipmr_expire_process(unsigned long dummy) | |
372 | { | |
373 | unsigned long now; | |
374 | unsigned long expires; | |
375 | struct mfc_cache *c, **cp; | |
376 | ||
377 | if (!spin_trylock(&mfc_unres_lock)) { | |
378 | mod_timer(&ipmr_expire_timer, jiffies+HZ/10); | |
379 | return; | |
380 | } | |
381 | ||
1e8fb3b6 | 382 | if (mfc_unres_queue == NULL) |
1da177e4 LT |
383 | goto out; |
384 | ||
385 | now = jiffies; | |
386 | expires = 10*HZ; | |
387 | cp = &mfc_unres_queue; | |
388 | ||
389 | while ((c=*cp) != NULL) { | |
390 | if (time_after(c->mfc_un.unres.expires, now)) { | |
391 | unsigned long interval = c->mfc_un.unres.expires - now; | |
392 | if (interval < expires) | |
393 | expires = interval; | |
394 | cp = &c->next; | |
395 | continue; | |
396 | } | |
397 | ||
398 | *cp = c->next; | |
399 | ||
400 | ipmr_destroy_unres(c); | |
401 | } | |
402 | ||
1e8fb3b6 | 403 | if (mfc_unres_queue != NULL) |
1da177e4 LT |
404 | mod_timer(&ipmr_expire_timer, jiffies + expires); |
405 | ||
406 | out: | |
407 | spin_unlock(&mfc_unres_lock); | |
408 | } | |
409 | ||
410 | /* Fill oifs list. It is called under write locked mrt_lock. */ | |
411 | ||
d1b04c08 | 412 | static void ipmr_update_thresholds(struct mfc_cache *cache, unsigned char *ttls) |
1da177e4 LT |
413 | { |
414 | int vifi; | |
4feb88e5 | 415 | struct net *net = mfc_net(cache); |
1da177e4 LT |
416 | |
417 | cache->mfc_un.res.minvif = MAXVIFS; | |
418 | cache->mfc_un.res.maxvif = 0; | |
419 | memset(cache->mfc_un.res.ttls, 255, MAXVIFS); | |
420 | ||
4feb88e5 BT |
421 | for (vifi = 0; vifi < net->ipv4.maxvif; vifi++) { |
422 | if (VIF_EXISTS(net, vifi) && | |
cf958ae3 | 423 | ttls[vifi] && ttls[vifi] < 255) { |
1da177e4 LT |
424 | cache->mfc_un.res.ttls[vifi] = ttls[vifi]; |
425 | if (cache->mfc_un.res.minvif > vifi) | |
426 | cache->mfc_un.res.minvif = vifi; | |
427 | if (cache->mfc_un.res.maxvif <= vifi) | |
428 | cache->mfc_un.res.maxvif = vifi + 1; | |
429 | } | |
430 | } | |
431 | } | |
432 | ||
4feb88e5 | 433 | static int vif_add(struct net *net, struct vifctl *vifc, int mrtsock) |
1da177e4 LT |
434 | { |
435 | int vifi = vifc->vifc_vifi; | |
4feb88e5 | 436 | struct vif_device *v = &net->ipv4.vif_table[vifi]; |
1da177e4 LT |
437 | struct net_device *dev; |
438 | struct in_device *in_dev; | |
d607032d | 439 | int err; |
1da177e4 LT |
440 | |
441 | /* Is vif busy ? */ | |
4feb88e5 | 442 | if (VIF_EXISTS(net, vifi)) |
1da177e4 LT |
443 | return -EADDRINUSE; |
444 | ||
445 | switch (vifc->vifc_flags) { | |
446 | #ifdef CONFIG_IP_PIMSM | |
447 | case VIFF_REGISTER: | |
448 | /* | |
449 | * Special Purpose VIF in PIM | |
450 | * All the packets will be sent to the daemon | |
451 | */ | |
4feb88e5 | 452 | if (net->ipv4.mroute_reg_vif_num >= 0) |
1da177e4 | 453 | return -EADDRINUSE; |
403dbb97 | 454 | dev = ipmr_reg_vif(net); |
1da177e4 LT |
455 | if (!dev) |
456 | return -ENOBUFS; | |
d607032d WC |
457 | err = dev_set_allmulti(dev, 1); |
458 | if (err) { | |
459 | unregister_netdevice(dev); | |
7dc00c82 | 460 | dev_put(dev); |
d607032d WC |
461 | return err; |
462 | } | |
1da177e4 LT |
463 | break; |
464 | #endif | |
e905a9ed | 465 | case VIFF_TUNNEL: |
4feb88e5 | 466 | dev = ipmr_new_tunnel(net, vifc); |
1da177e4 LT |
467 | if (!dev) |
468 | return -ENOBUFS; | |
d607032d WC |
469 | err = dev_set_allmulti(dev, 1); |
470 | if (err) { | |
471 | ipmr_del_tunnel(dev, vifc); | |
7dc00c82 | 472 | dev_put(dev); |
d607032d WC |
473 | return err; |
474 | } | |
1da177e4 LT |
475 | break; |
476 | case 0: | |
4feb88e5 | 477 | dev = ip_dev_find(net, vifc->vifc_lcl_addr.s_addr); |
1da177e4 LT |
478 | if (!dev) |
479 | return -EADDRNOTAVAIL; | |
d607032d | 480 | err = dev_set_allmulti(dev, 1); |
7dc00c82 WC |
481 | if (err) { |
482 | dev_put(dev); | |
d607032d | 483 | return err; |
7dc00c82 | 484 | } |
1da177e4 LT |
485 | break; |
486 | default: | |
487 | return -EINVAL; | |
488 | } | |
489 | ||
e5ed6399 | 490 | if ((in_dev = __in_dev_get_rtnl(dev)) == NULL) |
1da177e4 | 491 | return -EADDRNOTAVAIL; |
42f811b8 | 492 | IPV4_DEVCONF(in_dev->cnf, MC_FORWARDING)++; |
1da177e4 LT |
493 | ip_rt_multicast_event(in_dev); |
494 | ||
495 | /* | |
496 | * Fill in the VIF structures | |
497 | */ | |
c354e124 JK |
498 | v->rate_limit = vifc->vifc_rate_limit; |
499 | v->local = vifc->vifc_lcl_addr.s_addr; | |
500 | v->remote = vifc->vifc_rmt_addr.s_addr; | |
501 | v->flags = vifc->vifc_flags; | |
1da177e4 LT |
502 | if (!mrtsock) |
503 | v->flags |= VIFF_STATIC; | |
c354e124 | 504 | v->threshold = vifc->vifc_threshold; |
1da177e4 LT |
505 | v->bytes_in = 0; |
506 | v->bytes_out = 0; | |
507 | v->pkt_in = 0; | |
508 | v->pkt_out = 0; | |
509 | v->link = dev->ifindex; | |
510 | if (v->flags&(VIFF_TUNNEL|VIFF_REGISTER)) | |
511 | v->link = dev->iflink; | |
512 | ||
513 | /* And finish update writing critical data */ | |
514 | write_lock_bh(&mrt_lock); | |
c354e124 | 515 | v->dev = dev; |
1da177e4 LT |
516 | #ifdef CONFIG_IP_PIMSM |
517 | if (v->flags&VIFF_REGISTER) | |
4feb88e5 | 518 | net->ipv4.mroute_reg_vif_num = vifi; |
1da177e4 | 519 | #endif |
4feb88e5 BT |
520 | if (vifi+1 > net->ipv4.maxvif) |
521 | net->ipv4.maxvif = vifi+1; | |
1da177e4 LT |
522 | write_unlock_bh(&mrt_lock); |
523 | return 0; | |
524 | } | |
525 | ||
4feb88e5 BT |
526 | static struct mfc_cache *ipmr_cache_find(struct net *net, |
527 | __be32 origin, | |
528 | __be32 mcastgrp) | |
1da177e4 | 529 | { |
c354e124 | 530 | int line = MFC_HASH(mcastgrp, origin); |
1da177e4 LT |
531 | struct mfc_cache *c; |
532 | ||
4feb88e5 | 533 | for (c = net->ipv4.mfc_cache_array[line]; c; c = c->next) { |
1da177e4 LT |
534 | if (c->mfc_origin==origin && c->mfc_mcastgrp==mcastgrp) |
535 | break; | |
536 | } | |
537 | return c; | |
538 | } | |
539 | ||
540 | /* | |
541 | * Allocate a multicast cache entry | |
542 | */ | |
5c0a66f5 | 543 | static struct mfc_cache *ipmr_cache_alloc(struct net *net) |
1da177e4 | 544 | { |
c354e124 JK |
545 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_KERNEL); |
546 | if (c == NULL) | |
1da177e4 | 547 | return NULL; |
1da177e4 | 548 | c->mfc_un.res.minvif = MAXVIFS; |
5c0a66f5 | 549 | mfc_net_set(c, net); |
1da177e4 LT |
550 | return c; |
551 | } | |
552 | ||
5c0a66f5 | 553 | static struct mfc_cache *ipmr_cache_alloc_unres(struct net *net) |
1da177e4 | 554 | { |
c354e124 JK |
555 | struct mfc_cache *c = kmem_cache_zalloc(mrt_cachep, GFP_ATOMIC); |
556 | if (c == NULL) | |
1da177e4 | 557 | return NULL; |
1da177e4 LT |
558 | skb_queue_head_init(&c->mfc_un.unres.unresolved); |
559 | c->mfc_un.unres.expires = jiffies + 10*HZ; | |
5c0a66f5 | 560 | mfc_net_set(c, net); |
1da177e4 LT |
561 | return c; |
562 | } | |
563 | ||
564 | /* | |
565 | * A cache entry has gone into a resolved state from queued | |
566 | */ | |
e905a9ed | 567 | |
1da177e4 LT |
568 | static void ipmr_cache_resolve(struct mfc_cache *uc, struct mfc_cache *c) |
569 | { | |
570 | struct sk_buff *skb; | |
9ef1d4c7 | 571 | struct nlmsgerr *e; |
1da177e4 LT |
572 | |
573 | /* | |
574 | * Play the pending entries through our router | |
575 | */ | |
576 | ||
c354e124 | 577 | while ((skb = __skb_dequeue(&uc->mfc_un.unres.unresolved))) { |
eddc9ec5 | 578 | if (ip_hdr(skb)->version == 0) { |
1da177e4 LT |
579 | struct nlmsghdr *nlh = (struct nlmsghdr *)skb_pull(skb, sizeof(struct iphdr)); |
580 | ||
581 | if (ipmr_fill_mroute(skb, c, NLMSG_DATA(nlh)) > 0) { | |
27a884dc ACM |
582 | nlh->nlmsg_len = (skb_tail_pointer(skb) - |
583 | (u8 *)nlh); | |
1da177e4 LT |
584 | } else { |
585 | nlh->nlmsg_type = NLMSG_ERROR; | |
586 | nlh->nlmsg_len = NLMSG_LENGTH(sizeof(struct nlmsgerr)); | |
587 | skb_trim(skb, nlh->nlmsg_len); | |
9ef1d4c7 PM |
588 | e = NLMSG_DATA(nlh); |
589 | e->error = -EMSGSIZE; | |
590 | memset(&e->msg, 0, sizeof(e->msg)); | |
1da177e4 | 591 | } |
2942e900 | 592 | |
4feb88e5 | 593 | rtnl_unicast(skb, mfc_net(c), NETLINK_CB(skb).pid); |
1da177e4 LT |
594 | } else |
595 | ip_mr_forward(skb, c, 0); | |
596 | } | |
597 | } | |
598 | ||
599 | /* | |
600 | * Bounce a cache query up to mrouted. We could use netlink for this but mrouted | |
601 | * expects the following bizarre scheme. | |
602 | * | |
603 | * Called under mrt_lock. | |
604 | */ | |
e905a9ed | 605 | |
4feb88e5 BT |
606 | static int ipmr_cache_report(struct net *net, |
607 | struct sk_buff *pkt, vifi_t vifi, int assert) | |
1da177e4 LT |
608 | { |
609 | struct sk_buff *skb; | |
c9bdd4b5 | 610 | const int ihl = ip_hdrlen(pkt); |
1da177e4 LT |
611 | struct igmphdr *igmp; |
612 | struct igmpmsg *msg; | |
613 | int ret; | |
614 | ||
615 | #ifdef CONFIG_IP_PIMSM | |
616 | if (assert == IGMPMSG_WHOLEPKT) | |
617 | skb = skb_realloc_headroom(pkt, sizeof(struct iphdr)); | |
618 | else | |
619 | #endif | |
620 | skb = alloc_skb(128, GFP_ATOMIC); | |
621 | ||
132adf54 | 622 | if (!skb) |
1da177e4 LT |
623 | return -ENOBUFS; |
624 | ||
625 | #ifdef CONFIG_IP_PIMSM | |
626 | if (assert == IGMPMSG_WHOLEPKT) { | |
627 | /* Ugly, but we have no choice with this interface. | |
628 | Duplicate old header, fix ihl, length etc. | |
629 | And all this only to mangle msg->im_msgtype and | |
630 | to set msg->im_mbz to "mbz" :-) | |
631 | */ | |
878c8145 ACM |
632 | skb_push(skb, sizeof(struct iphdr)); |
633 | skb_reset_network_header(skb); | |
badff6d0 | 634 | skb_reset_transport_header(skb); |
0272ffc4 | 635 | msg = (struct igmpmsg *)skb_network_header(skb); |
d56f90a7 | 636 | memcpy(msg, skb_network_header(pkt), sizeof(struct iphdr)); |
1da177e4 LT |
637 | msg->im_msgtype = IGMPMSG_WHOLEPKT; |
638 | msg->im_mbz = 0; | |
4feb88e5 | 639 | msg->im_vif = net->ipv4.mroute_reg_vif_num; |
eddc9ec5 ACM |
640 | ip_hdr(skb)->ihl = sizeof(struct iphdr) >> 2; |
641 | ip_hdr(skb)->tot_len = htons(ntohs(ip_hdr(pkt)->tot_len) + | |
642 | sizeof(struct iphdr)); | |
e905a9ed | 643 | } else |
1da177e4 | 644 | #endif |
e905a9ed YH |
645 | { |
646 | ||
1da177e4 LT |
647 | /* |
648 | * Copy the IP header | |
649 | */ | |
650 | ||
27a884dc | 651 | skb->network_header = skb->tail; |
ddc7b8e3 | 652 | skb_put(skb, ihl); |
27d7ff46 | 653 | skb_copy_to_linear_data(skb, pkt->data, ihl); |
eddc9ec5 ACM |
654 | ip_hdr(skb)->protocol = 0; /* Flag to the kernel this is a route add */ |
655 | msg = (struct igmpmsg *)skb_network_header(skb); | |
1da177e4 | 656 | msg->im_vif = vifi; |
adf30907 | 657 | skb_dst_set(skb, dst_clone(skb_dst(pkt))); |
1da177e4 LT |
658 | |
659 | /* | |
660 | * Add our header | |
661 | */ | |
662 | ||
c354e124 | 663 | igmp=(struct igmphdr *)skb_put(skb, sizeof(struct igmphdr)); |
1da177e4 LT |
664 | igmp->type = |
665 | msg->im_msgtype = assert; | |
666 | igmp->code = 0; | |
eddc9ec5 | 667 | ip_hdr(skb)->tot_len = htons(skb->len); /* Fix the length */ |
b0e380b1 | 668 | skb->transport_header = skb->network_header; |
e905a9ed | 669 | } |
1da177e4 | 670 | |
4feb88e5 | 671 | if (net->ipv4.mroute_sk == NULL) { |
1da177e4 LT |
672 | kfree_skb(skb); |
673 | return -EINVAL; | |
674 | } | |
675 | ||
676 | /* | |
677 | * Deliver to mrouted | |
678 | */ | |
4feb88e5 | 679 | ret = sock_queue_rcv_skb(net->ipv4.mroute_sk, skb); |
70a269e6 | 680 | if (ret < 0) { |
1da177e4 LT |
681 | if (net_ratelimit()) |
682 | printk(KERN_WARNING "mroute: pending queue full, dropping entries.\n"); | |
683 | kfree_skb(skb); | |
684 | } | |
685 | ||
686 | return ret; | |
687 | } | |
688 | ||
689 | /* | |
690 | * Queue a packet for resolution. It gets locked cache entry! | |
691 | */ | |
e905a9ed | 692 | |
1da177e4 | 693 | static int |
4feb88e5 | 694 | ipmr_cache_unresolved(struct net *net, vifi_t vifi, struct sk_buff *skb) |
1da177e4 LT |
695 | { |
696 | int err; | |
697 | struct mfc_cache *c; | |
eddc9ec5 | 698 | const struct iphdr *iph = ip_hdr(skb); |
1da177e4 LT |
699 | |
700 | spin_lock_bh(&mfc_unres_lock); | |
701 | for (c=mfc_unres_queue; c; c=c->next) { | |
4feb88e5 | 702 | if (net_eq(mfc_net(c), net) && |
1e8fb3b6 | 703 | c->mfc_mcastgrp == iph->daddr && |
eddc9ec5 | 704 | c->mfc_origin == iph->saddr) |
1da177e4 LT |
705 | break; |
706 | } | |
707 | ||
708 | if (c == NULL) { | |
709 | /* | |
710 | * Create a new entry if allowable | |
711 | */ | |
712 | ||
4feb88e5 BT |
713 | if (atomic_read(&net->ipv4.cache_resolve_queue_len) >= 10 || |
714 | (c = ipmr_cache_alloc_unres(net)) == NULL) { | |
1da177e4 LT |
715 | spin_unlock_bh(&mfc_unres_lock); |
716 | ||
717 | kfree_skb(skb); | |
718 | return -ENOBUFS; | |
719 | } | |
720 | ||
721 | /* | |
722 | * Fill in the new cache entry | |
723 | */ | |
eddc9ec5 ACM |
724 | c->mfc_parent = -1; |
725 | c->mfc_origin = iph->saddr; | |
726 | c->mfc_mcastgrp = iph->daddr; | |
1da177e4 LT |
727 | |
728 | /* | |
729 | * Reflect first query at mrouted. | |
730 | */ | |
4feb88e5 BT |
731 | err = ipmr_cache_report(net, skb, vifi, IGMPMSG_NOCACHE); |
732 | if (err < 0) { | |
e905a9ed | 733 | /* If the report failed throw the cache entry |
1da177e4 LT |
734 | out - Brad Parker |
735 | */ | |
736 | spin_unlock_bh(&mfc_unres_lock); | |
737 | ||
5c0a66f5 | 738 | ipmr_cache_free(c); |
1da177e4 LT |
739 | kfree_skb(skb); |
740 | return err; | |
741 | } | |
742 | ||
4feb88e5 | 743 | atomic_inc(&net->ipv4.cache_resolve_queue_len); |
1da177e4 LT |
744 | c->next = mfc_unres_queue; |
745 | mfc_unres_queue = c; | |
746 | ||
747 | mod_timer(&ipmr_expire_timer, c->mfc_un.unres.expires); | |
748 | } | |
749 | ||
750 | /* | |
751 | * See if we can append the packet | |
752 | */ | |
753 | if (c->mfc_un.unres.unresolved.qlen>3) { | |
754 | kfree_skb(skb); | |
755 | err = -ENOBUFS; | |
756 | } else { | |
c354e124 | 757 | skb_queue_tail(&c->mfc_un.unres.unresolved, skb); |
1da177e4 LT |
758 | err = 0; |
759 | } | |
760 | ||
761 | spin_unlock_bh(&mfc_unres_lock); | |
762 | return err; | |
763 | } | |
764 | ||
765 | /* | |
766 | * MFC cache manipulation by user space mroute daemon | |
767 | */ | |
768 | ||
4feb88e5 | 769 | static int ipmr_mfc_delete(struct net *net, struct mfcctl *mfc) |
1da177e4 LT |
770 | { |
771 | int line; | |
772 | struct mfc_cache *c, **cp; | |
773 | ||
c354e124 | 774 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 775 | |
4feb88e5 | 776 | for (cp = &net->ipv4.mfc_cache_array[line]; |
2bb8b26c | 777 | (c = *cp) != NULL; cp = &c->next) { |
1da177e4 LT |
778 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
779 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) { | |
780 | write_lock_bh(&mrt_lock); | |
781 | *cp = c->next; | |
782 | write_unlock_bh(&mrt_lock); | |
783 | ||
5c0a66f5 | 784 | ipmr_cache_free(c); |
1da177e4 LT |
785 | return 0; |
786 | } | |
787 | } | |
788 | return -ENOENT; | |
789 | } | |
790 | ||
4feb88e5 | 791 | static int ipmr_mfc_add(struct net *net, struct mfcctl *mfc, int mrtsock) |
1da177e4 LT |
792 | { |
793 | int line; | |
794 | struct mfc_cache *uc, *c, **cp; | |
795 | ||
c354e124 | 796 | line = MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr); |
1da177e4 | 797 | |
4feb88e5 | 798 | for (cp = &net->ipv4.mfc_cache_array[line]; |
2bb8b26c | 799 | (c = *cp) != NULL; cp = &c->next) { |
1da177e4 LT |
800 | if (c->mfc_origin == mfc->mfcc_origin.s_addr && |
801 | c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) | |
802 | break; | |
803 | } | |
804 | ||
805 | if (c != NULL) { | |
806 | write_lock_bh(&mrt_lock); | |
807 | c->mfc_parent = mfc->mfcc_parent; | |
d1b04c08 | 808 | ipmr_update_thresholds(c, mfc->mfcc_ttls); |
1da177e4 LT |
809 | if (!mrtsock) |
810 | c->mfc_flags |= MFC_STATIC; | |
811 | write_unlock_bh(&mrt_lock); | |
812 | return 0; | |
813 | } | |
814 | ||
f97c1e0c | 815 | if (!ipv4_is_multicast(mfc->mfcc_mcastgrp.s_addr)) |
1da177e4 LT |
816 | return -EINVAL; |
817 | ||
4feb88e5 | 818 | c = ipmr_cache_alloc(net); |
c354e124 | 819 | if (c == NULL) |
1da177e4 LT |
820 | return -ENOMEM; |
821 | ||
c354e124 JK |
822 | c->mfc_origin = mfc->mfcc_origin.s_addr; |
823 | c->mfc_mcastgrp = mfc->mfcc_mcastgrp.s_addr; | |
824 | c->mfc_parent = mfc->mfcc_parent; | |
d1b04c08 | 825 | ipmr_update_thresholds(c, mfc->mfcc_ttls); |
1da177e4 LT |
826 | if (!mrtsock) |
827 | c->mfc_flags |= MFC_STATIC; | |
828 | ||
829 | write_lock_bh(&mrt_lock); | |
4feb88e5 BT |
830 | c->next = net->ipv4.mfc_cache_array[line]; |
831 | net->ipv4.mfc_cache_array[line] = c; | |
1da177e4 LT |
832 | write_unlock_bh(&mrt_lock); |
833 | ||
834 | /* | |
835 | * Check to see if we resolved a queued list. If so we | |
836 | * need to send on the frames and tidy up. | |
837 | */ | |
838 | spin_lock_bh(&mfc_unres_lock); | |
839 | for (cp = &mfc_unres_queue; (uc=*cp) != NULL; | |
840 | cp = &uc->next) { | |
4feb88e5 | 841 | if (net_eq(mfc_net(uc), net) && |
1e8fb3b6 | 842 | uc->mfc_origin == c->mfc_origin && |
1da177e4 LT |
843 | uc->mfc_mcastgrp == c->mfc_mcastgrp) { |
844 | *cp = uc->next; | |
4feb88e5 | 845 | atomic_dec(&net->ipv4.cache_resolve_queue_len); |
1da177e4 LT |
846 | break; |
847 | } | |
848 | } | |
1e8fb3b6 BT |
849 | if (mfc_unres_queue == NULL) |
850 | del_timer(&ipmr_expire_timer); | |
1da177e4 LT |
851 | spin_unlock_bh(&mfc_unres_lock); |
852 | ||
853 | if (uc) { | |
854 | ipmr_cache_resolve(uc, c); | |
5c0a66f5 | 855 | ipmr_cache_free(uc); |
1da177e4 LT |
856 | } |
857 | return 0; | |
858 | } | |
859 | ||
860 | /* | |
861 | * Close the multicast socket, and clear the vif tables etc | |
862 | */ | |
e905a9ed | 863 | |
4feb88e5 | 864 | static void mroute_clean_tables(struct net *net) |
1da177e4 LT |
865 | { |
866 | int i; | |
e905a9ed | 867 | |
1da177e4 LT |
868 | /* |
869 | * Shut down all active vif entries | |
870 | */ | |
4feb88e5 BT |
871 | for (i = 0; i < net->ipv4.maxvif; i++) { |
872 | if (!(net->ipv4.vif_table[i].flags&VIFF_STATIC)) | |
873 | vif_delete(net, i, 0); | |
1da177e4 LT |
874 | } |
875 | ||
876 | /* | |
877 | * Wipe the cache | |
878 | */ | |
c354e124 | 879 | for (i=0; i<MFC_LINES; i++) { |
1da177e4 LT |
880 | struct mfc_cache *c, **cp; |
881 | ||
4feb88e5 | 882 | cp = &net->ipv4.mfc_cache_array[i]; |
1da177e4 LT |
883 | while ((c = *cp) != NULL) { |
884 | if (c->mfc_flags&MFC_STATIC) { | |
885 | cp = &c->next; | |
886 | continue; | |
887 | } | |
888 | write_lock_bh(&mrt_lock); | |
889 | *cp = c->next; | |
890 | write_unlock_bh(&mrt_lock); | |
891 | ||
5c0a66f5 | 892 | ipmr_cache_free(c); |
1da177e4 LT |
893 | } |
894 | } | |
895 | ||
4feb88e5 | 896 | if (atomic_read(&net->ipv4.cache_resolve_queue_len) != 0) { |
1e8fb3b6 | 897 | struct mfc_cache *c, **cp; |
1da177e4 LT |
898 | |
899 | spin_lock_bh(&mfc_unres_lock); | |
1e8fb3b6 BT |
900 | cp = &mfc_unres_queue; |
901 | while ((c = *cp) != NULL) { | |
4feb88e5 | 902 | if (!net_eq(mfc_net(c), net)) { |
1e8fb3b6 BT |
903 | cp = &c->next; |
904 | continue; | |
905 | } | |
906 | *cp = c->next; | |
1da177e4 LT |
907 | |
908 | ipmr_destroy_unres(c); | |
1da177e4 LT |
909 | } |
910 | spin_unlock_bh(&mfc_unres_lock); | |
911 | } | |
912 | } | |
913 | ||
914 | static void mrtsock_destruct(struct sock *sk) | |
915 | { | |
4feb88e5 BT |
916 | struct net *net = sock_net(sk); |
917 | ||
1da177e4 | 918 | rtnl_lock(); |
4feb88e5 BT |
919 | if (sk == net->ipv4.mroute_sk) { |
920 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)--; | |
1da177e4 LT |
921 | |
922 | write_lock_bh(&mrt_lock); | |
4feb88e5 | 923 | net->ipv4.mroute_sk = NULL; |
1da177e4 LT |
924 | write_unlock_bh(&mrt_lock); |
925 | ||
4feb88e5 | 926 | mroute_clean_tables(net); |
1da177e4 LT |
927 | } |
928 | rtnl_unlock(); | |
929 | } | |
930 | ||
931 | /* | |
932 | * Socket options and virtual interface manipulation. The whole | |
933 | * virtual interface system is a complete heap, but unfortunately | |
934 | * that's how BSD mrouted happens to think. Maybe one day with a proper | |
935 | * MOSPF/PIM router set up we can clean this up. | |
936 | */ | |
e905a9ed | 937 | |
c354e124 | 938 | int ip_mroute_setsockopt(struct sock *sk, int optname, char __user *optval, int optlen) |
1da177e4 LT |
939 | { |
940 | int ret; | |
941 | struct vifctl vif; | |
942 | struct mfcctl mfc; | |
4feb88e5 | 943 | struct net *net = sock_net(sk); |
e905a9ed | 944 | |
132adf54 | 945 | if (optname != MRT_INIT) { |
4feb88e5 | 946 | if (sk != net->ipv4.mroute_sk && !capable(CAP_NET_ADMIN)) |
1da177e4 LT |
947 | return -EACCES; |
948 | } | |
949 | ||
132adf54 SH |
950 | switch (optname) { |
951 | case MRT_INIT: | |
952 | if (sk->sk_type != SOCK_RAW || | |
953 | inet_sk(sk)->num != IPPROTO_IGMP) | |
954 | return -EOPNOTSUPP; | |
c354e124 | 955 | if (optlen != sizeof(int)) |
132adf54 | 956 | return -ENOPROTOOPT; |
1da177e4 | 957 | |
132adf54 | 958 | rtnl_lock(); |
4feb88e5 | 959 | if (net->ipv4.mroute_sk) { |
1da177e4 | 960 | rtnl_unlock(); |
132adf54 SH |
961 | return -EADDRINUSE; |
962 | } | |
963 | ||
964 | ret = ip_ra_control(sk, 1, mrtsock_destruct); | |
965 | if (ret == 0) { | |
966 | write_lock_bh(&mrt_lock); | |
4feb88e5 | 967 | net->ipv4.mroute_sk = sk; |
132adf54 SH |
968 | write_unlock_bh(&mrt_lock); |
969 | ||
4feb88e5 | 970 | IPV4_DEVCONF_ALL(net, MC_FORWARDING)++; |
132adf54 SH |
971 | } |
972 | rtnl_unlock(); | |
973 | return ret; | |
974 | case MRT_DONE: | |
4feb88e5 | 975 | if (sk != net->ipv4.mroute_sk) |
132adf54 SH |
976 | return -EACCES; |
977 | return ip_ra_control(sk, 0, NULL); | |
978 | case MRT_ADD_VIF: | |
979 | case MRT_DEL_VIF: | |
c354e124 | 980 | if (optlen != sizeof(vif)) |
132adf54 | 981 | return -EINVAL; |
c354e124 | 982 | if (copy_from_user(&vif, optval, sizeof(vif))) |
132adf54 SH |
983 | return -EFAULT; |
984 | if (vif.vifc_vifi >= MAXVIFS) | |
985 | return -ENFILE; | |
986 | rtnl_lock(); | |
c354e124 | 987 | if (optname == MRT_ADD_VIF) { |
4feb88e5 | 988 | ret = vif_add(net, &vif, sk == net->ipv4.mroute_sk); |
132adf54 | 989 | } else { |
4feb88e5 | 990 | ret = vif_delete(net, vif.vifc_vifi, 0); |
132adf54 SH |
991 | } |
992 | rtnl_unlock(); | |
993 | return ret; | |
1da177e4 LT |
994 | |
995 | /* | |
996 | * Manipulate the forwarding caches. These live | |
997 | * in a sort of kernel/user symbiosis. | |
998 | */ | |
132adf54 SH |
999 | case MRT_ADD_MFC: |
1000 | case MRT_DEL_MFC: | |
c354e124 | 1001 | if (optlen != sizeof(mfc)) |
132adf54 | 1002 | return -EINVAL; |
c354e124 | 1003 | if (copy_from_user(&mfc, optval, sizeof(mfc))) |
132adf54 SH |
1004 | return -EFAULT; |
1005 | rtnl_lock(); | |
c354e124 | 1006 | if (optname == MRT_DEL_MFC) |
4feb88e5 | 1007 | ret = ipmr_mfc_delete(net, &mfc); |
132adf54 | 1008 | else |
4feb88e5 | 1009 | ret = ipmr_mfc_add(net, &mfc, sk == net->ipv4.mroute_sk); |
132adf54 SH |
1010 | rtnl_unlock(); |
1011 | return ret; | |
1da177e4 LT |
1012 | /* |
1013 | * Control PIM assert. | |
1014 | */ | |
132adf54 SH |
1015 | case MRT_ASSERT: |
1016 | { | |
1017 | int v; | |
1018 | if (get_user(v,(int __user *)optval)) | |
1019 | return -EFAULT; | |
4feb88e5 | 1020 | net->ipv4.mroute_do_assert = (v) ? 1 : 0; |
132adf54 SH |
1021 | return 0; |
1022 | } | |
1da177e4 | 1023 | #ifdef CONFIG_IP_PIMSM |
132adf54 SH |
1024 | case MRT_PIM: |
1025 | { | |
ba93ef74 SH |
1026 | int v; |
1027 | ||
132adf54 SH |
1028 | if (get_user(v,(int __user *)optval)) |
1029 | return -EFAULT; | |
ba93ef74 SH |
1030 | v = (v) ? 1 : 0; |
1031 | ||
132adf54 SH |
1032 | rtnl_lock(); |
1033 | ret = 0; | |
4feb88e5 BT |
1034 | if (v != net->ipv4.mroute_do_pim) { |
1035 | net->ipv4.mroute_do_pim = v; | |
1036 | net->ipv4.mroute_do_assert = v; | |
1da177e4 | 1037 | } |
132adf54 SH |
1038 | rtnl_unlock(); |
1039 | return ret; | |
1040 | } | |
1da177e4 | 1041 | #endif |
132adf54 SH |
1042 | /* |
1043 | * Spurious command, or MRT_VERSION which you cannot | |
1044 | * set. | |
1045 | */ | |
1046 | default: | |
1047 | return -ENOPROTOOPT; | |
1da177e4 LT |
1048 | } |
1049 | } | |
1050 | ||
1051 | /* | |
1052 | * Getsock opt support for the multicast routing system. | |
1053 | */ | |
e905a9ed | 1054 | |
c354e124 | 1055 | int ip_mroute_getsockopt(struct sock *sk, int optname, char __user *optval, int __user *optlen) |
1da177e4 LT |
1056 | { |
1057 | int olr; | |
1058 | int val; | |
4feb88e5 | 1059 | struct net *net = sock_net(sk); |
1da177e4 | 1060 | |
c354e124 | 1061 | if (optname != MRT_VERSION && |
1da177e4 LT |
1062 | #ifdef CONFIG_IP_PIMSM |
1063 | optname!=MRT_PIM && | |
1064 | #endif | |
1065 | optname!=MRT_ASSERT) | |
1066 | return -ENOPROTOOPT; | |
1067 | ||
1068 | if (get_user(olr, optlen)) | |
1069 | return -EFAULT; | |
1070 | ||
1071 | olr = min_t(unsigned int, olr, sizeof(int)); | |
1072 | if (olr < 0) | |
1073 | return -EINVAL; | |
e905a9ed | 1074 | |
c354e124 | 1075 | if (put_user(olr, optlen)) |
1da177e4 | 1076 | return -EFAULT; |
c354e124 JK |
1077 | if (optname == MRT_VERSION) |
1078 | val = 0x0305; | |
1da177e4 | 1079 | #ifdef CONFIG_IP_PIMSM |
c354e124 | 1080 | else if (optname == MRT_PIM) |
4feb88e5 | 1081 | val = net->ipv4.mroute_do_pim; |
1da177e4 LT |
1082 | #endif |
1083 | else | |
4feb88e5 | 1084 | val = net->ipv4.mroute_do_assert; |
c354e124 | 1085 | if (copy_to_user(optval, &val, olr)) |
1da177e4 LT |
1086 | return -EFAULT; |
1087 | return 0; | |
1088 | } | |
1089 | ||
1090 | /* | |
1091 | * The IP multicast ioctl support routines. | |
1092 | */ | |
e905a9ed | 1093 | |
1da177e4 LT |
1094 | int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg) |
1095 | { | |
1096 | struct sioc_sg_req sr; | |
1097 | struct sioc_vif_req vr; | |
1098 | struct vif_device *vif; | |
1099 | struct mfc_cache *c; | |
4feb88e5 | 1100 | struct net *net = sock_net(sk); |
e905a9ed | 1101 | |
132adf54 SH |
1102 | switch (cmd) { |
1103 | case SIOCGETVIFCNT: | |
c354e124 | 1104 | if (copy_from_user(&vr, arg, sizeof(vr))) |
132adf54 | 1105 | return -EFAULT; |
4feb88e5 | 1106 | if (vr.vifi >= net->ipv4.maxvif) |
132adf54 SH |
1107 | return -EINVAL; |
1108 | read_lock(&mrt_lock); | |
4feb88e5 BT |
1109 | vif = &net->ipv4.vif_table[vr.vifi]; |
1110 | if (VIF_EXISTS(net, vr.vifi)) { | |
c354e124 JK |
1111 | vr.icount = vif->pkt_in; |
1112 | vr.ocount = vif->pkt_out; | |
1113 | vr.ibytes = vif->bytes_in; | |
1114 | vr.obytes = vif->bytes_out; | |
1da177e4 | 1115 | read_unlock(&mrt_lock); |
1da177e4 | 1116 | |
c354e124 | 1117 | if (copy_to_user(arg, &vr, sizeof(vr))) |
132adf54 SH |
1118 | return -EFAULT; |
1119 | return 0; | |
1120 | } | |
1121 | read_unlock(&mrt_lock); | |
1122 | return -EADDRNOTAVAIL; | |
1123 | case SIOCGETSGCNT: | |
c354e124 | 1124 | if (copy_from_user(&sr, arg, sizeof(sr))) |
132adf54 SH |
1125 | return -EFAULT; |
1126 | ||
1127 | read_lock(&mrt_lock); | |
4feb88e5 | 1128 | c = ipmr_cache_find(net, sr.src.s_addr, sr.grp.s_addr); |
132adf54 SH |
1129 | if (c) { |
1130 | sr.pktcnt = c->mfc_un.res.pkt; | |
1131 | sr.bytecnt = c->mfc_un.res.bytes; | |
1132 | sr.wrong_if = c->mfc_un.res.wrong_if; | |
1da177e4 | 1133 | read_unlock(&mrt_lock); |
132adf54 | 1134 | |
c354e124 | 1135 | if (copy_to_user(arg, &sr, sizeof(sr))) |
132adf54 SH |
1136 | return -EFAULT; |
1137 | return 0; | |
1138 | } | |
1139 | read_unlock(&mrt_lock); | |
1140 | return -EADDRNOTAVAIL; | |
1141 | default: | |
1142 | return -ENOIOCTLCMD; | |
1da177e4 LT |
1143 | } |
1144 | } | |
1145 | ||
1146 | ||
1147 | static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr) | |
1148 | { | |
e9dc8653 | 1149 | struct net_device *dev = ptr; |
4feb88e5 | 1150 | struct net *net = dev_net(dev); |
1da177e4 LT |
1151 | struct vif_device *v; |
1152 | int ct; | |
e9dc8653 | 1153 | |
4feb88e5 | 1154 | if (!net_eq(dev_net(dev), net)) |
e9dc8653 EB |
1155 | return NOTIFY_DONE; |
1156 | ||
1da177e4 LT |
1157 | if (event != NETDEV_UNREGISTER) |
1158 | return NOTIFY_DONE; | |
4feb88e5 BT |
1159 | v = &net->ipv4.vif_table[0]; |
1160 | for (ct = 0; ct < net->ipv4.maxvif; ct++, v++) { | |
c354e124 | 1161 | if (v->dev == dev) |
4feb88e5 | 1162 | vif_delete(net, ct, 1); |
1da177e4 LT |
1163 | } |
1164 | return NOTIFY_DONE; | |
1165 | } | |
1166 | ||
1167 | ||
c354e124 | 1168 | static struct notifier_block ip_mr_notifier = { |
1da177e4 LT |
1169 | .notifier_call = ipmr_device_event, |
1170 | }; | |
1171 | ||
1172 | /* | |
1173 | * Encapsulate a packet by attaching a valid IPIP header to it. | |
1174 | * This avoids tunnel drivers and other mess and gives us the speed so | |
1175 | * important for multicast video. | |
1176 | */ | |
e905a9ed | 1177 | |
114c7844 | 1178 | static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr) |
1da177e4 | 1179 | { |
8856dfa3 | 1180 | struct iphdr *iph; |
eddc9ec5 | 1181 | struct iphdr *old_iph = ip_hdr(skb); |
8856dfa3 ACM |
1182 | |
1183 | skb_push(skb, sizeof(struct iphdr)); | |
b0e380b1 | 1184 | skb->transport_header = skb->network_header; |
8856dfa3 | 1185 | skb_reset_network_header(skb); |
eddc9ec5 | 1186 | iph = ip_hdr(skb); |
1da177e4 LT |
1187 | |
1188 | iph->version = 4; | |
e023dd64 ACM |
1189 | iph->tos = old_iph->tos; |
1190 | iph->ttl = old_iph->ttl; | |
1da177e4 LT |
1191 | iph->frag_off = 0; |
1192 | iph->daddr = daddr; | |
1193 | iph->saddr = saddr; | |
1194 | iph->protocol = IPPROTO_IPIP; | |
1195 | iph->ihl = 5; | |
1196 | iph->tot_len = htons(skb->len); | |
adf30907 | 1197 | ip_select_ident(iph, skb_dst(skb), NULL); |
1da177e4 LT |
1198 | ip_send_check(iph); |
1199 | ||
1da177e4 LT |
1200 | memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt)); |
1201 | nf_reset(skb); | |
1202 | } | |
1203 | ||
1204 | static inline int ipmr_forward_finish(struct sk_buff *skb) | |
1205 | { | |
1206 | struct ip_options * opt = &(IPCB(skb)->opt); | |
1207 | ||
adf30907 | 1208 | IP_INC_STATS_BH(dev_net(skb_dst(skb)->dev), IPSTATS_MIB_OUTFORWDATAGRAMS); |
1da177e4 LT |
1209 | |
1210 | if (unlikely(opt->optlen)) | |
1211 | ip_forward_options(skb); | |
1212 | ||
1213 | return dst_output(skb); | |
1214 | } | |
1215 | ||
1216 | /* | |
1217 | * Processing handlers for ipmr_forward | |
1218 | */ | |
1219 | ||
1220 | static void ipmr_queue_xmit(struct sk_buff *skb, struct mfc_cache *c, int vifi) | |
1221 | { | |
4feb88e5 | 1222 | struct net *net = mfc_net(c); |
eddc9ec5 | 1223 | const struct iphdr *iph = ip_hdr(skb); |
4feb88e5 | 1224 | struct vif_device *vif = &net->ipv4.vif_table[vifi]; |
1da177e4 LT |
1225 | struct net_device *dev; |
1226 | struct rtable *rt; | |
1227 | int encap = 0; | |
1228 | ||
1229 | if (vif->dev == NULL) | |
1230 | goto out_free; | |
1231 | ||
1232 | #ifdef CONFIG_IP_PIMSM | |
1233 | if (vif->flags & VIFF_REGISTER) { | |
1234 | vif->pkt_out++; | |
c354e124 | 1235 | vif->bytes_out += skb->len; |
cf3677ae PE |
1236 | vif->dev->stats.tx_bytes += skb->len; |
1237 | vif->dev->stats.tx_packets++; | |
4feb88e5 | 1238 | ipmr_cache_report(net, skb, vifi, IGMPMSG_WHOLEPKT); |
69ebbf58 | 1239 | goto out_free; |
1da177e4 LT |
1240 | } |
1241 | #endif | |
1242 | ||
1243 | if (vif->flags&VIFF_TUNNEL) { | |
1244 | struct flowi fl = { .oif = vif->link, | |
1245 | .nl_u = { .ip4_u = | |
1246 | { .daddr = vif->remote, | |
1247 | .saddr = vif->local, | |
1248 | .tos = RT_TOS(iph->tos) } }, | |
1249 | .proto = IPPROTO_IPIP }; | |
4feb88e5 | 1250 | if (ip_route_output_key(net, &rt, &fl)) |
1da177e4 LT |
1251 | goto out_free; |
1252 | encap = sizeof(struct iphdr); | |
1253 | } else { | |
1254 | struct flowi fl = { .oif = vif->link, | |
1255 | .nl_u = { .ip4_u = | |
1256 | { .daddr = iph->daddr, | |
1257 | .tos = RT_TOS(iph->tos) } }, | |
1258 | .proto = IPPROTO_IPIP }; | |
4feb88e5 | 1259 | if (ip_route_output_key(net, &rt, &fl)) |
1da177e4 LT |
1260 | goto out_free; |
1261 | } | |
1262 | ||
1263 | dev = rt->u.dst.dev; | |
1264 | ||
1265 | if (skb->len+encap > dst_mtu(&rt->u.dst) && (ntohs(iph->frag_off) & IP_DF)) { | |
1266 | /* Do not fragment multicasts. Alas, IPv4 does not | |
1267 | allow to send ICMP, so that packets will disappear | |
1268 | to blackhole. | |
1269 | */ | |
1270 | ||
7c73a6fa | 1271 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_FRAGFAILS); |
1da177e4 LT |
1272 | ip_rt_put(rt); |
1273 | goto out_free; | |
1274 | } | |
1275 | ||
1276 | encap += LL_RESERVED_SPACE(dev) + rt->u.dst.header_len; | |
1277 | ||
1278 | if (skb_cow(skb, encap)) { | |
e905a9ed | 1279 | ip_rt_put(rt); |
1da177e4 LT |
1280 | goto out_free; |
1281 | } | |
1282 | ||
1283 | vif->pkt_out++; | |
c354e124 | 1284 | vif->bytes_out += skb->len; |
1da177e4 | 1285 | |
adf30907 ED |
1286 | skb_dst_drop(skb); |
1287 | skb_dst_set(skb, &rt->u.dst); | |
eddc9ec5 | 1288 | ip_decrease_ttl(ip_hdr(skb)); |
1da177e4 LT |
1289 | |
1290 | /* FIXME: forward and output firewalls used to be called here. | |
1291 | * What do we do with netfilter? -- RR */ | |
1292 | if (vif->flags & VIFF_TUNNEL) { | |
1293 | ip_encap(skb, vif->local, vif->remote); | |
1294 | /* FIXME: extra output firewall step used to be here. --RR */ | |
2f4c02d4 PE |
1295 | vif->dev->stats.tx_packets++; |
1296 | vif->dev->stats.tx_bytes += skb->len; | |
1da177e4 LT |
1297 | } |
1298 | ||
1299 | IPCB(skb)->flags |= IPSKB_FORWARDED; | |
1300 | ||
1301 | /* | |
1302 | * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally | |
1303 | * not only before forwarding, but after forwarding on all output | |
1304 | * interfaces. It is clear, if mrouter runs a multicasting | |
1305 | * program, it should receive packets not depending to what interface | |
1306 | * program is joined. | |
1307 | * If we will not make it, the program will have to join on all | |
1308 | * interfaces. On the other hand, multihoming host (or router, but | |
1309 | * not mrouter) cannot join to more than one interface - it will | |
1310 | * result in receiving multiple packets. | |
1311 | */ | |
6e23ae2a | 1312 | NF_HOOK(PF_INET, NF_INET_FORWARD, skb, skb->dev, dev, |
1da177e4 LT |
1313 | ipmr_forward_finish); |
1314 | return; | |
1315 | ||
1316 | out_free: | |
1317 | kfree_skb(skb); | |
1318 | return; | |
1319 | } | |
1320 | ||
1321 | static int ipmr_find_vif(struct net_device *dev) | |
1322 | { | |
4feb88e5 | 1323 | struct net *net = dev_net(dev); |
1da177e4 | 1324 | int ct; |
4feb88e5 BT |
1325 | for (ct = net->ipv4.maxvif-1; ct >= 0; ct--) { |
1326 | if (net->ipv4.vif_table[ct].dev == dev) | |
1da177e4 LT |
1327 | break; |
1328 | } | |
1329 | return ct; | |
1330 | } | |
1331 | ||
1332 | /* "local" means that we should preserve one skb (for local delivery) */ | |
1333 | ||
1334 | static int ip_mr_forward(struct sk_buff *skb, struct mfc_cache *cache, int local) | |
1335 | { | |
1336 | int psend = -1; | |
1337 | int vif, ct; | |
4feb88e5 | 1338 | struct net *net = mfc_net(cache); |
1da177e4 LT |
1339 | |
1340 | vif = cache->mfc_parent; | |
1341 | cache->mfc_un.res.pkt++; | |
1342 | cache->mfc_un.res.bytes += skb->len; | |
1343 | ||
1344 | /* | |
1345 | * Wrong interface: drop packet and (maybe) send PIM assert. | |
1346 | */ | |
4feb88e5 | 1347 | if (net->ipv4.vif_table[vif].dev != skb->dev) { |
1da177e4 LT |
1348 | int true_vifi; |
1349 | ||
511c3f92 | 1350 | if (skb_rtable(skb)->fl.iif == 0) { |
1da177e4 LT |
1351 | /* It is our own packet, looped back. |
1352 | Very complicated situation... | |
1353 | ||
1354 | The best workaround until routing daemons will be | |
1355 | fixed is not to redistribute packet, if it was | |
1356 | send through wrong interface. It means, that | |
1357 | multicast applications WILL NOT work for | |
1358 | (S,G), which have default multicast route pointing | |
1359 | to wrong oif. In any case, it is not a good | |
1360 | idea to use multicasting applications on router. | |
1361 | */ | |
1362 | goto dont_forward; | |
1363 | } | |
1364 | ||
1365 | cache->mfc_un.res.wrong_if++; | |
1366 | true_vifi = ipmr_find_vif(skb->dev); | |
1367 | ||
4feb88e5 | 1368 | if (true_vifi >= 0 && net->ipv4.mroute_do_assert && |
1da177e4 LT |
1369 | /* pimsm uses asserts, when switching from RPT to SPT, |
1370 | so that we cannot check that packet arrived on an oif. | |
1371 | It is bad, but otherwise we would need to move pretty | |
1372 | large chunk of pimd to kernel. Ough... --ANK | |
1373 | */ | |
4feb88e5 | 1374 | (net->ipv4.mroute_do_pim || |
6f9374a9 | 1375 | cache->mfc_un.res.ttls[true_vifi] < 255) && |
e905a9ed | 1376 | time_after(jiffies, |
1da177e4 LT |
1377 | cache->mfc_un.res.last_assert + MFC_ASSERT_THRESH)) { |
1378 | cache->mfc_un.res.last_assert = jiffies; | |
4feb88e5 | 1379 | ipmr_cache_report(net, skb, true_vifi, IGMPMSG_WRONGVIF); |
1da177e4 LT |
1380 | } |
1381 | goto dont_forward; | |
1382 | } | |
1383 | ||
4feb88e5 BT |
1384 | net->ipv4.vif_table[vif].pkt_in++; |
1385 | net->ipv4.vif_table[vif].bytes_in += skb->len; | |
1da177e4 LT |
1386 | |
1387 | /* | |
1388 | * Forward the frame | |
1389 | */ | |
1390 | for (ct = cache->mfc_un.res.maxvif-1; ct >= cache->mfc_un.res.minvif; ct--) { | |
eddc9ec5 | 1391 | if (ip_hdr(skb)->ttl > cache->mfc_un.res.ttls[ct]) { |
1da177e4 LT |
1392 | if (psend != -1) { |
1393 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1394 | if (skb2) | |
1395 | ipmr_queue_xmit(skb2, cache, psend); | |
1396 | } | |
c354e124 | 1397 | psend = ct; |
1da177e4 LT |
1398 | } |
1399 | } | |
1400 | if (psend != -1) { | |
1401 | if (local) { | |
1402 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1403 | if (skb2) | |
1404 | ipmr_queue_xmit(skb2, cache, psend); | |
1405 | } else { | |
1406 | ipmr_queue_xmit(skb, cache, psend); | |
1407 | return 0; | |
1408 | } | |
1409 | } | |
1410 | ||
1411 | dont_forward: | |
1412 | if (!local) | |
1413 | kfree_skb(skb); | |
1414 | return 0; | |
1415 | } | |
1416 | ||
1417 | ||
1418 | /* | |
1419 | * Multicast packets for forwarding arrive here | |
1420 | */ | |
1421 | ||
1422 | int ip_mr_input(struct sk_buff *skb) | |
1423 | { | |
1424 | struct mfc_cache *cache; | |
4feb88e5 | 1425 | struct net *net = dev_net(skb->dev); |
511c3f92 | 1426 | int local = skb_rtable(skb)->rt_flags & RTCF_LOCAL; |
1da177e4 LT |
1427 | |
1428 | /* Packet is looped back after forward, it should not be | |
1429 | forwarded second time, but still can be delivered locally. | |
1430 | */ | |
1431 | if (IPCB(skb)->flags&IPSKB_FORWARDED) | |
1432 | goto dont_forward; | |
1433 | ||
1434 | if (!local) { | |
1435 | if (IPCB(skb)->opt.router_alert) { | |
1436 | if (ip_call_ra_chain(skb)) | |
1437 | return 0; | |
eddc9ec5 | 1438 | } else if (ip_hdr(skb)->protocol == IPPROTO_IGMP){ |
1da177e4 LT |
1439 | /* IGMPv1 (and broken IGMPv2 implementations sort of |
1440 | Cisco IOS <= 11.2(8)) do not put router alert | |
1441 | option to IGMP packets destined to routable | |
1442 | groups. It is very bad, because it means | |
1443 | that we can forward NO IGMP messages. | |
1444 | */ | |
1445 | read_lock(&mrt_lock); | |
4feb88e5 | 1446 | if (net->ipv4.mroute_sk) { |
2715bcf9 | 1447 | nf_reset(skb); |
4feb88e5 | 1448 | raw_rcv(net->ipv4.mroute_sk, skb); |
1da177e4 LT |
1449 | read_unlock(&mrt_lock); |
1450 | return 0; | |
1451 | } | |
1452 | read_unlock(&mrt_lock); | |
1453 | } | |
1454 | } | |
1455 | ||
1456 | read_lock(&mrt_lock); | |
4feb88e5 | 1457 | cache = ipmr_cache_find(net, ip_hdr(skb)->saddr, ip_hdr(skb)->daddr); |
1da177e4 LT |
1458 | |
1459 | /* | |
1460 | * No usable cache entry | |
1461 | */ | |
c354e124 | 1462 | if (cache == NULL) { |
1da177e4 LT |
1463 | int vif; |
1464 | ||
1465 | if (local) { | |
1466 | struct sk_buff *skb2 = skb_clone(skb, GFP_ATOMIC); | |
1467 | ip_local_deliver(skb); | |
1468 | if (skb2 == NULL) { | |
1469 | read_unlock(&mrt_lock); | |
1470 | return -ENOBUFS; | |
1471 | } | |
1472 | skb = skb2; | |
1473 | } | |
1474 | ||
1475 | vif = ipmr_find_vif(skb->dev); | |
1476 | if (vif >= 0) { | |
4feb88e5 | 1477 | int err = ipmr_cache_unresolved(net, vif, skb); |
1da177e4 LT |
1478 | read_unlock(&mrt_lock); |
1479 | ||
1480 | return err; | |
1481 | } | |
1482 | read_unlock(&mrt_lock); | |
1483 | kfree_skb(skb); | |
1484 | return -ENODEV; | |
1485 | } | |
1486 | ||
1487 | ip_mr_forward(skb, cache, local); | |
1488 | ||
1489 | read_unlock(&mrt_lock); | |
1490 | ||
1491 | if (local) | |
1492 | return ip_local_deliver(skb); | |
1493 | ||
1494 | return 0; | |
1495 | ||
1496 | dont_forward: | |
1497 | if (local) | |
1498 | return ip_local_deliver(skb); | |
1499 | kfree_skb(skb); | |
1500 | return 0; | |
1501 | } | |
1502 | ||
b1879204 IJ |
1503 | #ifdef CONFIG_IP_PIMSM |
1504 | static int __pim_rcv(struct sk_buff *skb, unsigned int pimlen) | |
1da177e4 | 1505 | { |
b1879204 IJ |
1506 | struct net_device *reg_dev = NULL; |
1507 | struct iphdr *encap; | |
4feb88e5 | 1508 | struct net *net = dev_net(skb->dev); |
1da177e4 | 1509 | |
b1879204 | 1510 | encap = (struct iphdr *)(skb_transport_header(skb) + pimlen); |
1da177e4 LT |
1511 | /* |
1512 | Check that: | |
1513 | a. packet is really destinted to a multicast group | |
1514 | b. packet is not a NULL-REGISTER | |
1515 | c. packet is not truncated | |
1516 | */ | |
f97c1e0c | 1517 | if (!ipv4_is_multicast(encap->daddr) || |
1da177e4 | 1518 | encap->tot_len == 0 || |
b1879204 IJ |
1519 | ntohs(encap->tot_len) + pimlen > skb->len) |
1520 | return 1; | |
1da177e4 LT |
1521 | |
1522 | read_lock(&mrt_lock); | |
4feb88e5 BT |
1523 | if (net->ipv4.mroute_reg_vif_num >= 0) |
1524 | reg_dev = net->ipv4.vif_table[net->ipv4.mroute_reg_vif_num].dev; | |
1da177e4 LT |
1525 | if (reg_dev) |
1526 | dev_hold(reg_dev); | |
1527 | read_unlock(&mrt_lock); | |
1528 | ||
e905a9ed | 1529 | if (reg_dev == NULL) |
b1879204 | 1530 | return 1; |
1da177e4 | 1531 | |
b0e380b1 | 1532 | skb->mac_header = skb->network_header; |
1da177e4 | 1533 | skb_pull(skb, (u8*)encap - skb->data); |
31c7711b | 1534 | skb_reset_network_header(skb); |
1da177e4 | 1535 | skb->dev = reg_dev; |
1da177e4 LT |
1536 | skb->protocol = htons(ETH_P_IP); |
1537 | skb->ip_summed = 0; | |
1538 | skb->pkt_type = PACKET_HOST; | |
adf30907 | 1539 | skb_dst_drop(skb); |
cf3677ae PE |
1540 | reg_dev->stats.rx_bytes += skb->len; |
1541 | reg_dev->stats.rx_packets++; | |
1da177e4 LT |
1542 | nf_reset(skb); |
1543 | netif_rx(skb); | |
1544 | dev_put(reg_dev); | |
b1879204 | 1545 | |
1da177e4 | 1546 | return 0; |
b1879204 IJ |
1547 | } |
1548 | #endif | |
1549 | ||
1550 | #ifdef CONFIG_IP_PIMSM_V1 | |
1551 | /* | |
1552 | * Handle IGMP messages of PIMv1 | |
1553 | */ | |
1554 | ||
1555 | int pim_rcv_v1(struct sk_buff * skb) | |
1556 | { | |
1557 | struct igmphdr *pim; | |
4feb88e5 | 1558 | struct net *net = dev_net(skb->dev); |
b1879204 IJ |
1559 | |
1560 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) | |
1561 | goto drop; | |
1562 | ||
1563 | pim = igmp_hdr(skb); | |
1564 | ||
4feb88e5 | 1565 | if (!net->ipv4.mroute_do_pim || |
b1879204 IJ |
1566 | pim->group != PIM_V1_VERSION || pim->code != PIM_V1_REGISTER) |
1567 | goto drop; | |
1568 | ||
1569 | if (__pim_rcv(skb, sizeof(*pim))) { | |
1570 | drop: | |
1571 | kfree_skb(skb); | |
1572 | } | |
1da177e4 LT |
1573 | return 0; |
1574 | } | |
1575 | #endif | |
1576 | ||
1577 | #ifdef CONFIG_IP_PIMSM_V2 | |
1578 | static int pim_rcv(struct sk_buff * skb) | |
1579 | { | |
1580 | struct pimreghdr *pim; | |
1da177e4 | 1581 | |
b1879204 | 1582 | if (!pskb_may_pull(skb, sizeof(*pim) + sizeof(struct iphdr))) |
1da177e4 LT |
1583 | goto drop; |
1584 | ||
9c70220b | 1585 | pim = (struct pimreghdr *)skb_transport_header(skb); |
e905a9ed | 1586 | if (pim->type != ((PIM_VERSION<<4)|(PIM_REGISTER)) || |
1da177e4 | 1587 | (pim->flags&PIM_NULL_REGISTER) || |
e905a9ed | 1588 | (ip_compute_csum((void *)pim, sizeof(*pim)) != 0 && |
d3bc23e7 | 1589 | csum_fold(skb_checksum(skb, 0, skb->len, 0)))) |
1da177e4 LT |
1590 | goto drop; |
1591 | ||
b1879204 IJ |
1592 | if (__pim_rcv(skb, sizeof(*pim))) { |
1593 | drop: | |
1594 | kfree_skb(skb); | |
1595 | } | |
1da177e4 LT |
1596 | return 0; |
1597 | } | |
1598 | #endif | |
1599 | ||
1600 | static int | |
1601 | ipmr_fill_mroute(struct sk_buff *skb, struct mfc_cache *c, struct rtmsg *rtm) | |
1602 | { | |
1603 | int ct; | |
1604 | struct rtnexthop *nhp; | |
4feb88e5 BT |
1605 | struct net *net = mfc_net(c); |
1606 | struct net_device *dev = net->ipv4.vif_table[c->mfc_parent].dev; | |
27a884dc | 1607 | u8 *b = skb_tail_pointer(skb); |
1da177e4 LT |
1608 | struct rtattr *mp_head; |
1609 | ||
1610 | if (dev) | |
1611 | RTA_PUT(skb, RTA_IIF, 4, &dev->ifindex); | |
1612 | ||
c354e124 | 1613 | mp_head = (struct rtattr *)skb_put(skb, RTA_LENGTH(0)); |
1da177e4 LT |
1614 | |
1615 | for (ct = c->mfc_un.res.minvif; ct < c->mfc_un.res.maxvif; ct++) { | |
1616 | if (c->mfc_un.res.ttls[ct] < 255) { | |
1617 | if (skb_tailroom(skb) < RTA_ALIGN(RTA_ALIGN(sizeof(*nhp)) + 4)) | |
1618 | goto rtattr_failure; | |
c354e124 | 1619 | nhp = (struct rtnexthop *)skb_put(skb, RTA_ALIGN(sizeof(*nhp))); |
1da177e4 LT |
1620 | nhp->rtnh_flags = 0; |
1621 | nhp->rtnh_hops = c->mfc_un.res.ttls[ct]; | |
4feb88e5 | 1622 | nhp->rtnh_ifindex = net->ipv4.vif_table[ct].dev->ifindex; |
1da177e4 LT |
1623 | nhp->rtnh_len = sizeof(*nhp); |
1624 | } | |
1625 | } | |
1626 | mp_head->rta_type = RTA_MULTIPATH; | |
27a884dc | 1627 | mp_head->rta_len = skb_tail_pointer(skb) - (u8 *)mp_head; |
1da177e4 LT |
1628 | rtm->rtm_type = RTN_MULTICAST; |
1629 | return 1; | |
1630 | ||
1631 | rtattr_failure: | |
dc5fc579 | 1632 | nlmsg_trim(skb, b); |
1da177e4 LT |
1633 | return -EMSGSIZE; |
1634 | } | |
1635 | ||
4feb88e5 BT |
1636 | int ipmr_get_route(struct net *net, |
1637 | struct sk_buff *skb, struct rtmsg *rtm, int nowait) | |
1da177e4 LT |
1638 | { |
1639 | int err; | |
1640 | struct mfc_cache *cache; | |
511c3f92 | 1641 | struct rtable *rt = skb_rtable(skb); |
1da177e4 LT |
1642 | |
1643 | read_lock(&mrt_lock); | |
4feb88e5 | 1644 | cache = ipmr_cache_find(net, rt->rt_src, rt->rt_dst); |
1da177e4 | 1645 | |
c354e124 | 1646 | if (cache == NULL) { |
72287490 | 1647 | struct sk_buff *skb2; |
eddc9ec5 | 1648 | struct iphdr *iph; |
1da177e4 LT |
1649 | struct net_device *dev; |
1650 | int vif; | |
1651 | ||
1652 | if (nowait) { | |
1653 | read_unlock(&mrt_lock); | |
1654 | return -EAGAIN; | |
1655 | } | |
1656 | ||
1657 | dev = skb->dev; | |
1658 | if (dev == NULL || (vif = ipmr_find_vif(dev)) < 0) { | |
1659 | read_unlock(&mrt_lock); | |
1660 | return -ENODEV; | |
1661 | } | |
72287490 AK |
1662 | skb2 = skb_clone(skb, GFP_ATOMIC); |
1663 | if (!skb2) { | |
1664 | read_unlock(&mrt_lock); | |
1665 | return -ENOMEM; | |
1666 | } | |
1667 | ||
e2d1bca7 ACM |
1668 | skb_push(skb2, sizeof(struct iphdr)); |
1669 | skb_reset_network_header(skb2); | |
eddc9ec5 ACM |
1670 | iph = ip_hdr(skb2); |
1671 | iph->ihl = sizeof(struct iphdr) >> 2; | |
1672 | iph->saddr = rt->rt_src; | |
1673 | iph->daddr = rt->rt_dst; | |
1674 | iph->version = 0; | |
4feb88e5 | 1675 | err = ipmr_cache_unresolved(net, vif, skb2); |
1da177e4 LT |
1676 | read_unlock(&mrt_lock); |
1677 | return err; | |
1678 | } | |
1679 | ||
1680 | if (!nowait && (rtm->rtm_flags&RTM_F_NOTIFY)) | |
1681 | cache->mfc_flags |= MFC_NOTIFY; | |
1682 | err = ipmr_fill_mroute(skb, cache, rtm); | |
1683 | read_unlock(&mrt_lock); | |
1684 | return err; | |
1685 | } | |
1686 | ||
e905a9ed | 1687 | #ifdef CONFIG_PROC_FS |
1da177e4 LT |
1688 | /* |
1689 | * The /proc interfaces to multicast routing /proc/ip_mr_cache /proc/ip_mr_vif | |
1690 | */ | |
1691 | struct ipmr_vif_iter { | |
f6bb4514 | 1692 | struct seq_net_private p; |
1da177e4 LT |
1693 | int ct; |
1694 | }; | |
1695 | ||
f6bb4514 BT |
1696 | static struct vif_device *ipmr_vif_seq_idx(struct net *net, |
1697 | struct ipmr_vif_iter *iter, | |
1da177e4 LT |
1698 | loff_t pos) |
1699 | { | |
f6bb4514 BT |
1700 | for (iter->ct = 0; iter->ct < net->ipv4.maxvif; ++iter->ct) { |
1701 | if (!VIF_EXISTS(net, iter->ct)) | |
1da177e4 | 1702 | continue; |
e905a9ed | 1703 | if (pos-- == 0) |
f6bb4514 | 1704 | return &net->ipv4.vif_table[iter->ct]; |
1da177e4 LT |
1705 | } |
1706 | return NULL; | |
1707 | } | |
1708 | ||
1709 | static void *ipmr_vif_seq_start(struct seq_file *seq, loff_t *pos) | |
ba93ef74 | 1710 | __acquires(mrt_lock) |
1da177e4 | 1711 | { |
f6bb4514 BT |
1712 | struct net *net = seq_file_net(seq); |
1713 | ||
1da177e4 | 1714 | read_lock(&mrt_lock); |
f6bb4514 | 1715 | return *pos ? ipmr_vif_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
1716 | : SEQ_START_TOKEN; |
1717 | } | |
1718 | ||
1719 | static void *ipmr_vif_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
1720 | { | |
1721 | struct ipmr_vif_iter *iter = seq->private; | |
f6bb4514 | 1722 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1723 | |
1724 | ++*pos; | |
1725 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 1726 | return ipmr_vif_seq_idx(net, iter, 0); |
e905a9ed | 1727 | |
f6bb4514 BT |
1728 | while (++iter->ct < net->ipv4.maxvif) { |
1729 | if (!VIF_EXISTS(net, iter->ct)) | |
1da177e4 | 1730 | continue; |
f6bb4514 | 1731 | return &net->ipv4.vif_table[iter->ct]; |
1da177e4 LT |
1732 | } |
1733 | return NULL; | |
1734 | } | |
1735 | ||
1736 | static void ipmr_vif_seq_stop(struct seq_file *seq, void *v) | |
ba93ef74 | 1737 | __releases(mrt_lock) |
1da177e4 LT |
1738 | { |
1739 | read_unlock(&mrt_lock); | |
1740 | } | |
1741 | ||
1742 | static int ipmr_vif_seq_show(struct seq_file *seq, void *v) | |
1743 | { | |
f6bb4514 BT |
1744 | struct net *net = seq_file_net(seq); |
1745 | ||
1da177e4 | 1746 | if (v == SEQ_START_TOKEN) { |
e905a9ed | 1747 | seq_puts(seq, |
1da177e4 LT |
1748 | "Interface BytesIn PktsIn BytesOut PktsOut Flags Local Remote\n"); |
1749 | } else { | |
1750 | const struct vif_device *vif = v; | |
1751 | const char *name = vif->dev ? vif->dev->name : "none"; | |
1752 | ||
1753 | seq_printf(seq, | |
1754 | "%2Zd %-10s %8ld %7ld %8ld %7ld %05X %08X %08X\n", | |
f6bb4514 | 1755 | vif - net->ipv4.vif_table, |
e905a9ed | 1756 | name, vif->bytes_in, vif->pkt_in, |
1da177e4 LT |
1757 | vif->bytes_out, vif->pkt_out, |
1758 | vif->flags, vif->local, vif->remote); | |
1759 | } | |
1760 | return 0; | |
1761 | } | |
1762 | ||
f690808e | 1763 | static const struct seq_operations ipmr_vif_seq_ops = { |
1da177e4 LT |
1764 | .start = ipmr_vif_seq_start, |
1765 | .next = ipmr_vif_seq_next, | |
1766 | .stop = ipmr_vif_seq_stop, | |
1767 | .show = ipmr_vif_seq_show, | |
1768 | }; | |
1769 | ||
1770 | static int ipmr_vif_open(struct inode *inode, struct file *file) | |
1771 | { | |
f6bb4514 BT |
1772 | return seq_open_net(inode, file, &ipmr_vif_seq_ops, |
1773 | sizeof(struct ipmr_vif_iter)); | |
1da177e4 LT |
1774 | } |
1775 | ||
9a32144e | 1776 | static const struct file_operations ipmr_vif_fops = { |
1da177e4 LT |
1777 | .owner = THIS_MODULE, |
1778 | .open = ipmr_vif_open, | |
1779 | .read = seq_read, | |
1780 | .llseek = seq_lseek, | |
f6bb4514 | 1781 | .release = seq_release_net, |
1da177e4 LT |
1782 | }; |
1783 | ||
1784 | struct ipmr_mfc_iter { | |
f6bb4514 | 1785 | struct seq_net_private p; |
1da177e4 LT |
1786 | struct mfc_cache **cache; |
1787 | int ct; | |
1788 | }; | |
1789 | ||
1790 | ||
f6bb4514 BT |
1791 | static struct mfc_cache *ipmr_mfc_seq_idx(struct net *net, |
1792 | struct ipmr_mfc_iter *it, loff_t pos) | |
1da177e4 LT |
1793 | { |
1794 | struct mfc_cache *mfc; | |
1795 | ||
f6bb4514 | 1796 | it->cache = net->ipv4.mfc_cache_array; |
1da177e4 | 1797 | read_lock(&mrt_lock); |
e905a9ed | 1798 | for (it->ct = 0; it->ct < MFC_LINES; it->ct++) |
f6bb4514 | 1799 | for (mfc = net->ipv4.mfc_cache_array[it->ct]; |
2bb8b26c | 1800 | mfc; mfc = mfc->next) |
e905a9ed | 1801 | if (pos-- == 0) |
1da177e4 LT |
1802 | return mfc; |
1803 | read_unlock(&mrt_lock); | |
1804 | ||
1805 | it->cache = &mfc_unres_queue; | |
1806 | spin_lock_bh(&mfc_unres_lock); | |
132adf54 | 1807 | for (mfc = mfc_unres_queue; mfc; mfc = mfc->next) |
f6bb4514 BT |
1808 | if (net_eq(mfc_net(mfc), net) && |
1809 | pos-- == 0) | |
1da177e4 LT |
1810 | return mfc; |
1811 | spin_unlock_bh(&mfc_unres_lock); | |
1812 | ||
1813 | it->cache = NULL; | |
1814 | return NULL; | |
1815 | } | |
1816 | ||
1817 | ||
1818 | static void *ipmr_mfc_seq_start(struct seq_file *seq, loff_t *pos) | |
1819 | { | |
1820 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 BT |
1821 | struct net *net = seq_file_net(seq); |
1822 | ||
1da177e4 LT |
1823 | it->cache = NULL; |
1824 | it->ct = 0; | |
f6bb4514 | 1825 | return *pos ? ipmr_mfc_seq_idx(net, seq->private, *pos - 1) |
1da177e4 LT |
1826 | : SEQ_START_TOKEN; |
1827 | } | |
1828 | ||
1829 | static void *ipmr_mfc_seq_next(struct seq_file *seq, void *v, loff_t *pos) | |
1830 | { | |
1831 | struct mfc_cache *mfc = v; | |
1832 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 1833 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1834 | |
1835 | ++*pos; | |
1836 | ||
1837 | if (v == SEQ_START_TOKEN) | |
f6bb4514 | 1838 | return ipmr_mfc_seq_idx(net, seq->private, 0); |
1da177e4 LT |
1839 | |
1840 | if (mfc->next) | |
1841 | return mfc->next; | |
e905a9ed YH |
1842 | |
1843 | if (it->cache == &mfc_unres_queue) | |
1da177e4 LT |
1844 | goto end_of_list; |
1845 | ||
f6bb4514 | 1846 | BUG_ON(it->cache != net->ipv4.mfc_cache_array); |
1da177e4 LT |
1847 | |
1848 | while (++it->ct < MFC_LINES) { | |
f6bb4514 | 1849 | mfc = net->ipv4.mfc_cache_array[it->ct]; |
1da177e4 LT |
1850 | if (mfc) |
1851 | return mfc; | |
1852 | } | |
1853 | ||
1854 | /* exhausted cache_array, show unresolved */ | |
1855 | read_unlock(&mrt_lock); | |
1856 | it->cache = &mfc_unres_queue; | |
1857 | it->ct = 0; | |
e905a9ed | 1858 | |
1da177e4 LT |
1859 | spin_lock_bh(&mfc_unres_lock); |
1860 | mfc = mfc_unres_queue; | |
f6bb4514 BT |
1861 | while (mfc && !net_eq(mfc_net(mfc), net)) |
1862 | mfc = mfc->next; | |
e905a9ed | 1863 | if (mfc) |
1da177e4 LT |
1864 | return mfc; |
1865 | ||
1866 | end_of_list: | |
1867 | spin_unlock_bh(&mfc_unres_lock); | |
1868 | it->cache = NULL; | |
1869 | ||
1870 | return NULL; | |
1871 | } | |
1872 | ||
1873 | static void ipmr_mfc_seq_stop(struct seq_file *seq, void *v) | |
1874 | { | |
1875 | struct ipmr_mfc_iter *it = seq->private; | |
f6bb4514 | 1876 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1877 | |
1878 | if (it->cache == &mfc_unres_queue) | |
1879 | spin_unlock_bh(&mfc_unres_lock); | |
f6bb4514 | 1880 | else if (it->cache == net->ipv4.mfc_cache_array) |
1da177e4 LT |
1881 | read_unlock(&mrt_lock); |
1882 | } | |
1883 | ||
1884 | static int ipmr_mfc_seq_show(struct seq_file *seq, void *v) | |
1885 | { | |
1886 | int n; | |
f6bb4514 | 1887 | struct net *net = seq_file_net(seq); |
1da177e4 LT |
1888 | |
1889 | if (v == SEQ_START_TOKEN) { | |
e905a9ed | 1890 | seq_puts(seq, |
1da177e4 LT |
1891 | "Group Origin Iif Pkts Bytes Wrong Oifs\n"); |
1892 | } else { | |
1893 | const struct mfc_cache *mfc = v; | |
1894 | const struct ipmr_mfc_iter *it = seq->private; | |
e905a9ed | 1895 | |
999890b2 | 1896 | seq_printf(seq, "%08lX %08lX %-3hd", |
1da177e4 LT |
1897 | (unsigned long) mfc->mfc_mcastgrp, |
1898 | (unsigned long) mfc->mfc_origin, | |
1ea472e2 | 1899 | mfc->mfc_parent); |
1da177e4 LT |
1900 | |
1901 | if (it->cache != &mfc_unres_queue) { | |
1ea472e2 BT |
1902 | seq_printf(seq, " %8lu %8lu %8lu", |
1903 | mfc->mfc_un.res.pkt, | |
1904 | mfc->mfc_un.res.bytes, | |
1905 | mfc->mfc_un.res.wrong_if); | |
132adf54 SH |
1906 | for (n = mfc->mfc_un.res.minvif; |
1907 | n < mfc->mfc_un.res.maxvif; n++ ) { | |
f6bb4514 | 1908 | if (VIF_EXISTS(net, n) && |
cf958ae3 BT |
1909 | mfc->mfc_un.res.ttls[n] < 255) |
1910 | seq_printf(seq, | |
e905a9ed | 1911 | " %2d:%-3d", |
1da177e4 LT |
1912 | n, mfc->mfc_un.res.ttls[n]); |
1913 | } | |
1ea472e2 BT |
1914 | } else { |
1915 | /* unresolved mfc_caches don't contain | |
1916 | * pkt, bytes and wrong_if values | |
1917 | */ | |
1918 | seq_printf(seq, " %8lu %8lu %8lu", 0ul, 0ul, 0ul); | |
1da177e4 LT |
1919 | } |
1920 | seq_putc(seq, '\n'); | |
1921 | } | |
1922 | return 0; | |
1923 | } | |
1924 | ||
f690808e | 1925 | static const struct seq_operations ipmr_mfc_seq_ops = { |
1da177e4 LT |
1926 | .start = ipmr_mfc_seq_start, |
1927 | .next = ipmr_mfc_seq_next, | |
1928 | .stop = ipmr_mfc_seq_stop, | |
1929 | .show = ipmr_mfc_seq_show, | |
1930 | }; | |
1931 | ||
1932 | static int ipmr_mfc_open(struct inode *inode, struct file *file) | |
1933 | { | |
f6bb4514 BT |
1934 | return seq_open_net(inode, file, &ipmr_mfc_seq_ops, |
1935 | sizeof(struct ipmr_mfc_iter)); | |
1da177e4 LT |
1936 | } |
1937 | ||
9a32144e | 1938 | static const struct file_operations ipmr_mfc_fops = { |
1da177e4 LT |
1939 | .owner = THIS_MODULE, |
1940 | .open = ipmr_mfc_open, | |
1941 | .read = seq_read, | |
1942 | .llseek = seq_lseek, | |
f6bb4514 | 1943 | .release = seq_release_net, |
1da177e4 | 1944 | }; |
e905a9ed | 1945 | #endif |
1da177e4 LT |
1946 | |
1947 | #ifdef CONFIG_IP_PIMSM_V2 | |
1948 | static struct net_protocol pim_protocol = { | |
1949 | .handler = pim_rcv, | |
403dbb97 | 1950 | .netns_ok = 1, |
1da177e4 LT |
1951 | }; |
1952 | #endif | |
1953 | ||
1954 | ||
1955 | /* | |
1956 | * Setup for IP multicast routing | |
1957 | */ | |
cf958ae3 BT |
1958 | static int __net_init ipmr_net_init(struct net *net) |
1959 | { | |
1960 | int err = 0; | |
1961 | ||
1962 | net->ipv4.vif_table = kcalloc(MAXVIFS, sizeof(struct vif_device), | |
1963 | GFP_KERNEL); | |
1964 | if (!net->ipv4.vif_table) { | |
1965 | err = -ENOMEM; | |
1966 | goto fail; | |
1967 | } | |
2bb8b26c BT |
1968 | |
1969 | /* Forwarding cache */ | |
1970 | net->ipv4.mfc_cache_array = kcalloc(MFC_LINES, | |
1971 | sizeof(struct mfc_cache *), | |
1972 | GFP_KERNEL); | |
1973 | if (!net->ipv4.mfc_cache_array) { | |
1974 | err = -ENOMEM; | |
1975 | goto fail_mfc_cache; | |
1976 | } | |
6c5143db BT |
1977 | |
1978 | #ifdef CONFIG_IP_PIMSM | |
1979 | net->ipv4.mroute_reg_vif_num = -1; | |
1980 | #endif | |
f6bb4514 BT |
1981 | |
1982 | #ifdef CONFIG_PROC_FS | |
1983 | err = -ENOMEM; | |
1984 | if (!proc_net_fops_create(net, "ip_mr_vif", 0, &ipmr_vif_fops)) | |
1985 | goto proc_vif_fail; | |
1986 | if (!proc_net_fops_create(net, "ip_mr_cache", 0, &ipmr_mfc_fops)) | |
1987 | goto proc_cache_fail; | |
1988 | #endif | |
2bb8b26c BT |
1989 | return 0; |
1990 | ||
f6bb4514 BT |
1991 | #ifdef CONFIG_PROC_FS |
1992 | proc_cache_fail: | |
1993 | proc_net_remove(net, "ip_mr_vif"); | |
1994 | proc_vif_fail: | |
1995 | kfree(net->ipv4.mfc_cache_array); | |
1996 | #endif | |
2bb8b26c BT |
1997 | fail_mfc_cache: |
1998 | kfree(net->ipv4.vif_table); | |
cf958ae3 BT |
1999 | fail: |
2000 | return err; | |
2001 | } | |
2002 | ||
2003 | static void __net_exit ipmr_net_exit(struct net *net) | |
2004 | { | |
f6bb4514 BT |
2005 | #ifdef CONFIG_PROC_FS |
2006 | proc_net_remove(net, "ip_mr_cache"); | |
2007 | proc_net_remove(net, "ip_mr_vif"); | |
2008 | #endif | |
2bb8b26c | 2009 | kfree(net->ipv4.mfc_cache_array); |
cf958ae3 BT |
2010 | kfree(net->ipv4.vif_table); |
2011 | } | |
2012 | ||
2013 | static struct pernet_operations ipmr_net_ops = { | |
2014 | .init = ipmr_net_init, | |
2015 | .exit = ipmr_net_exit, | |
2016 | }; | |
e905a9ed | 2017 | |
03d2f897 | 2018 | int __init ip_mr_init(void) |
1da177e4 | 2019 | { |
03d2f897 WC |
2020 | int err; |
2021 | ||
1da177e4 LT |
2022 | mrt_cachep = kmem_cache_create("ip_mrt_cache", |
2023 | sizeof(struct mfc_cache), | |
e5d679f3 | 2024 | 0, SLAB_HWCACHE_ALIGN|SLAB_PANIC, |
20c2df83 | 2025 | NULL); |
03d2f897 WC |
2026 | if (!mrt_cachep) |
2027 | return -ENOMEM; | |
2028 | ||
cf958ae3 BT |
2029 | err = register_pernet_subsys(&ipmr_net_ops); |
2030 | if (err) | |
2031 | goto reg_pernet_fail; | |
2032 | ||
b24b8a24 | 2033 | setup_timer(&ipmr_expire_timer, ipmr_expire_process, 0); |
03d2f897 WC |
2034 | err = register_netdevice_notifier(&ip_mr_notifier); |
2035 | if (err) | |
2036 | goto reg_notif_fail; | |
403dbb97 TG |
2037 | #ifdef CONFIG_IP_PIMSM_V2 |
2038 | if (inet_add_protocol(&pim_protocol, IPPROTO_PIM) < 0) { | |
2039 | printk(KERN_ERR "ip_mr_init: can't add PIM protocol\n"); | |
2040 | err = -EAGAIN; | |
2041 | goto add_proto_fail; | |
2042 | } | |
2043 | #endif | |
03d2f897 | 2044 | return 0; |
f6bb4514 | 2045 | |
403dbb97 TG |
2046 | #ifdef CONFIG_IP_PIMSM_V2 |
2047 | add_proto_fail: | |
2048 | unregister_netdevice_notifier(&ip_mr_notifier); | |
2049 | #endif | |
c3e38896 BT |
2050 | reg_notif_fail: |
2051 | del_timer(&ipmr_expire_timer); | |
cf958ae3 BT |
2052 | unregister_pernet_subsys(&ipmr_net_ops); |
2053 | reg_pernet_fail: | |
c3e38896 | 2054 | kmem_cache_destroy(mrt_cachep); |
03d2f897 | 2055 | return err; |
1da177e4 | 2056 | } |