Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * Handle firewalling | |
3 | * Linux ethernet bridge | |
4 | * | |
5 | * Authors: | |
8237908e BDS |
6 | * Lennert Buytenhek <buytenh@gnu.org> |
7 | * Bart De Schuymer <bdschuym@pandora.be> | |
1da177e4 LT |
8 | * |
9 | * This program is free software; you can redistribute it and/or | |
10 | * modify it under the terms of the GNU General Public License | |
11 | * as published by the Free Software Foundation; either version | |
12 | * 2 of the License, or (at your option) any later version. | |
13 | * | |
14 | * Lennert dedicates this file to Kerstin Wurdinger. | |
15 | */ | |
16 | ||
17 | #include <linux/module.h> | |
18 | #include <linux/kernel.h> | |
5a0e3ad6 | 19 | #include <linux/slab.h> |
1da177e4 LT |
20 | #include <linux/ip.h> |
21 | #include <linux/netdevice.h> | |
22 | #include <linux/skbuff.h> | |
14c85021 | 23 | #include <linux/if_arp.h> |
1da177e4 LT |
24 | #include <linux/if_ether.h> |
25 | #include <linux/if_vlan.h> | |
516299d2 MM |
26 | #include <linux/if_pppox.h> |
27 | #include <linux/ppp_defs.h> | |
1da177e4 LT |
28 | #include <linux/netfilter_bridge.h> |
29 | #include <linux/netfilter_ipv4.h> | |
30 | #include <linux/netfilter_ipv6.h> | |
31 | #include <linux/netfilter_arp.h> | |
32 | #include <linux/in_route.h> | |
f216f082 | 33 | #include <linux/inetdevice.h> |
14c85021 | 34 | |
1da177e4 LT |
35 | #include <net/ip.h> |
36 | #include <net/ipv6.h> | |
14c85021 ACM |
37 | #include <net/route.h> |
38 | ||
1da177e4 | 39 | #include <asm/uaccess.h> |
1da177e4 LT |
40 | #include "br_private.h" |
41 | #ifdef CONFIG_SYSCTL | |
42 | #include <linux/sysctl.h> | |
43 | #endif | |
44 | ||
45 | #define skb_origaddr(skb) (((struct bridge_skb_cb *) \ | |
46 | (skb->nf_bridge->data))->daddr.ipv4) | |
eddc9ec5 ACM |
47 | #define store_orig_dstaddr(skb) (skb_origaddr(skb) = ip_hdr(skb)->daddr) |
48 | #define dnat_took_place(skb) (skb_origaddr(skb) != ip_hdr(skb)->daddr) | |
1da177e4 | 49 | |
1da177e4 LT |
50 | #ifdef CONFIG_SYSCTL |
51 | static struct ctl_table_header *brnf_sysctl_header; | |
9c1ea148 BH |
52 | static int brnf_call_iptables __read_mostly = 1; |
53 | static int brnf_call_ip6tables __read_mostly = 1; | |
54 | static int brnf_call_arptables __read_mostly = 1; | |
47e0e1ca HX |
55 | static int brnf_filter_vlan_tagged __read_mostly = 0; |
56 | static int brnf_filter_pppoe_tagged __read_mostly = 0; | |
4981682c | 57 | static int brnf_pass_vlan_indev __read_mostly = 0; |
1da177e4 | 58 | #else |
4df53d8b PM |
59 | #define brnf_call_iptables 1 |
60 | #define brnf_call_ip6tables 1 | |
61 | #define brnf_call_arptables 1 | |
47e0e1ca HX |
62 | #define brnf_filter_vlan_tagged 0 |
63 | #define brnf_filter_pppoe_tagged 0 | |
4981682c | 64 | #define brnf_pass_vlan_indev 0 |
1da177e4 LT |
65 | #endif |
66 | ||
739e4505 FW |
67 | #define IS_IP(skb) \ |
68 | (!vlan_tx_tag_present(skb) && skb->protocol == htons(ETH_P_IP)) | |
69 | ||
70 | #define IS_IPV6(skb) \ | |
71 | (!vlan_tx_tag_present(skb) && skb->protocol == htons(ETH_P_IPV6)) | |
72 | ||
73 | #define IS_ARP(skb) \ | |
74 | (!vlan_tx_tag_present(skb) && skb->protocol == htons(ETH_P_ARP)) | |
75 | ||
b6f99a21 | 76 | static inline __be16 vlan_proto(const struct sk_buff *skb) |
8b42ec39 | 77 | { |
13937911 JG |
78 | if (vlan_tx_tag_present(skb)) |
79 | return skb->protocol; | |
80 | else if (skb->protocol == htons(ETH_P_8021Q)) | |
81 | return vlan_eth_hdr(skb)->h_vlan_encapsulated_proto; | |
82 | else | |
83 | return 0; | |
8b42ec39 SH |
84 | } |
85 | ||
86 | #define IS_VLAN_IP(skb) \ | |
13937911 | 87 | (vlan_proto(skb) == htons(ETH_P_IP) && \ |
8b42ec39 SH |
88 | brnf_filter_vlan_tagged) |
89 | ||
90 | #define IS_VLAN_IPV6(skb) \ | |
13937911 | 91 | (vlan_proto(skb) == htons(ETH_P_IPV6) && \ |
8b42ec39 SH |
92 | brnf_filter_vlan_tagged) |
93 | ||
94 | #define IS_VLAN_ARP(skb) \ | |
13937911 | 95 | (vlan_proto(skb) == htons(ETH_P_ARP) && \ |
8b42ec39 | 96 | brnf_filter_vlan_tagged) |
1da177e4 | 97 | |
516299d2 MM |
98 | static inline __be16 pppoe_proto(const struct sk_buff *skb) |
99 | { | |
100 | return *((__be16 *)(skb_mac_header(skb) + ETH_HLEN + | |
101 | sizeof(struct pppoe_hdr))); | |
102 | } | |
103 | ||
104 | #define IS_PPPOE_IP(skb) \ | |
105 | (skb->protocol == htons(ETH_P_PPP_SES) && \ | |
106 | pppoe_proto(skb) == htons(PPP_IP) && \ | |
107 | brnf_filter_pppoe_tagged) | |
108 | ||
109 | #define IS_PPPOE_IPV6(skb) \ | |
110 | (skb->protocol == htons(ETH_P_PPP_SES) && \ | |
111 | pppoe_proto(skb) == htons(PPP_IPV6) && \ | |
112 | brnf_filter_pppoe_tagged) | |
113 | ||
631339f1 HX |
114 | static void fake_update_pmtu(struct dst_entry *dst, u32 mtu) |
115 | { | |
116 | } | |
117 | ||
6407d74c AH |
118 | static u32 *fake_cow_metrics(struct dst_entry *dst, unsigned long old) |
119 | { | |
120 | return NULL; | |
121 | } | |
122 | ||
d3aaeb38 DM |
123 | static struct neighbour *fake_neigh_lookup(const struct dst_entry *dst, const void *daddr) |
124 | { | |
125 | return NULL; | |
126 | } | |
127 | ||
a13861a2 ED |
128 | static unsigned int fake_mtu(const struct dst_entry *dst) |
129 | { | |
130 | return dst->dev->mtu; | |
131 | } | |
132 | ||
631339f1 HX |
133 | static struct dst_ops fake_dst_ops = { |
134 | .family = AF_INET, | |
09640e63 | 135 | .protocol = cpu_to_be16(ETH_P_IP), |
631339f1 | 136 | .update_pmtu = fake_update_pmtu, |
6407d74c | 137 | .cow_metrics = fake_cow_metrics, |
d3aaeb38 | 138 | .neigh_lookup = fake_neigh_lookup, |
a13861a2 | 139 | .mtu = fake_mtu, |
631339f1 HX |
140 | }; |
141 | ||
4adf0af6 SW |
142 | /* |
143 | * Initialize bogus route table used to keep netfilter happy. | |
1da177e4 LT |
144 | * Currently, we fill in the PMTU entry because netfilter |
145 | * refragmentation needs it, and the rt_flags entry because | |
146 | * ipt_REJECT needs it. Future netfilter modules might | |
4adf0af6 SW |
147 | * require us to fill additional fields. |
148 | */ | |
33eb9873 ED |
149 | static const u32 br_dst_default_metrics[RTAX_MAX] = { |
150 | [RTAX_MTU - 1] = 1500, | |
151 | }; | |
152 | ||
4adf0af6 SW |
153 | void br_netfilter_rtable_init(struct net_bridge *br) |
154 | { | |
155 | struct rtable *rt = &br->fake_rtable; | |
156 | ||
d8d1f30b CG |
157 | atomic_set(&rt->dst.__refcnt, 1); |
158 | rt->dst.dev = br->dev; | |
159 | rt->dst.path = &rt->dst; | |
33eb9873 | 160 | dst_init_metrics(&rt->dst, br_dst_default_metrics, true); |
a881e963 | 161 | rt->dst.flags = DST_NOXFRM | DST_NOPEER | DST_FAKE_RTABLE; |
d8d1f30b | 162 | rt->dst.ops = &fake_dst_ops; |
4adf0af6 SW |
163 | } |
164 | ||
165 | static inline struct rtable *bridge_parent_rtable(const struct net_device *dev) | |
166 | { | |
b5ed54e9 | 167 | struct net_bridge_port *port; |
168 | ||
169 | port = br_port_get_rcu(dev); | |
170 | return port ? &port->br->fake_rtable : NULL; | |
4adf0af6 | 171 | } |
1da177e4 | 172 | |
5dce971a SH |
173 | static inline struct net_device *bridge_parent(const struct net_device *dev) |
174 | { | |
b5ed54e9 | 175 | struct net_bridge_port *port; |
5dce971a | 176 | |
b5ed54e9 | 177 | port = br_port_get_rcu(dev); |
178 | return port ? port->br->dev : NULL; | |
5dce971a | 179 | } |
1da177e4 | 180 | |
fdeabdef SH |
181 | static inline struct nf_bridge_info *nf_bridge_alloc(struct sk_buff *skb) |
182 | { | |
183 | skb->nf_bridge = kzalloc(sizeof(struct nf_bridge_info), GFP_ATOMIC); | |
184 | if (likely(skb->nf_bridge)) | |
185 | atomic_set(&(skb->nf_bridge->use), 1); | |
186 | ||
187 | return skb->nf_bridge; | |
188 | } | |
189 | ||
2dc2f207 PM |
190 | static inline struct nf_bridge_info *nf_bridge_unshare(struct sk_buff *skb) |
191 | { | |
192 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
193 | ||
194 | if (atomic_read(&nf_bridge->use) > 1) { | |
195 | struct nf_bridge_info *tmp = nf_bridge_alloc(skb); | |
196 | ||
197 | if (tmp) { | |
198 | memcpy(tmp, nf_bridge, sizeof(struct nf_bridge_info)); | |
199 | atomic_set(&tmp->use, 1); | |
2dc2f207 | 200 | } |
4c3a76ab | 201 | nf_bridge_put(nf_bridge); |
2dc2f207 PM |
202 | nf_bridge = tmp; |
203 | } | |
204 | return nf_bridge; | |
205 | } | |
206 | ||
fc38582d PM |
207 | static inline void nf_bridge_push_encap_header(struct sk_buff *skb) |
208 | { | |
209 | unsigned int len = nf_bridge_encap_header_len(skb); | |
210 | ||
211 | skb_push(skb, len); | |
212 | skb->network_header -= len; | |
213 | } | |
214 | ||
215 | static inline void nf_bridge_pull_encap_header(struct sk_buff *skb) | |
fdeabdef | 216 | { |
fc38582d PM |
217 | unsigned int len = nf_bridge_encap_header_len(skb); |
218 | ||
219 | skb_pull(skb, len); | |
220 | skb->network_header += len; | |
221 | } | |
fdeabdef | 222 | |
fc38582d PM |
223 | static inline void nf_bridge_pull_encap_header_rcsum(struct sk_buff *skb) |
224 | { | |
225 | unsigned int len = nf_bridge_encap_header_len(skb); | |
226 | ||
227 | skb_pull_rcsum(skb, len); | |
228 | skb->network_header += len; | |
229 | } | |
230 | ||
231 | static inline void nf_bridge_save_header(struct sk_buff *skb) | |
232 | { | |
233 | int header_size = ETH_HLEN + nf_bridge_encap_header_len(skb); | |
fdeabdef | 234 | |
d626f62b ACM |
235 | skb_copy_from_linear_data_offset(skb, -header_size, |
236 | skb->nf_bridge->data, header_size); | |
fdeabdef SH |
237 | } |
238 | ||
e179e632 BDS |
239 | static inline void nf_bridge_update_protocol(struct sk_buff *skb) |
240 | { | |
241 | if (skb->nf_bridge->mask & BRNF_8021Q) | |
242 | skb->protocol = htons(ETH_P_8021Q); | |
243 | else if (skb->nf_bridge->mask & BRNF_PPPoE) | |
244 | skb->protocol = htons(ETH_P_PPP_SES); | |
245 | } | |
246 | ||
462fb2af BD |
247 | /* When handing a packet over to the IP layer |
248 | * check whether we have a skb that is in the | |
249 | * expected format | |
250 | */ | |
251 | ||
d0280232 | 252 | static int br_parse_ip_options(struct sk_buff *skb) |
462fb2af BD |
253 | { |
254 | struct ip_options *opt; | |
b71d1d42 | 255 | const struct iphdr *iph; |
462fb2af BD |
256 | struct net_device *dev = skb->dev; |
257 | u32 len; | |
258 | ||
259 | iph = ip_hdr(skb); | |
260 | opt = &(IPCB(skb)->opt); | |
261 | ||
262 | /* Basic sanity checks */ | |
263 | if (iph->ihl < 5 || iph->version != 4) | |
264 | goto inhdr_error; | |
265 | ||
266 | if (!pskb_may_pull(skb, iph->ihl*4)) | |
267 | goto inhdr_error; | |
268 | ||
269 | iph = ip_hdr(skb); | |
270 | if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl))) | |
271 | goto inhdr_error; | |
272 | ||
273 | len = ntohs(iph->tot_len); | |
274 | if (skb->len < len) { | |
275 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INTRUNCATEDPKTS); | |
276 | goto drop; | |
277 | } else if (len < (iph->ihl*4)) | |
278 | goto inhdr_error; | |
279 | ||
280 | if (pskb_trim_rcsum(skb, len)) { | |
281 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS); | |
282 | goto drop; | |
283 | } | |
284 | ||
f8e9881c ED |
285 | memset(IPCB(skb), 0, sizeof(struct inet_skb_parm)); |
286 | if (iph->ihl == 5) | |
462fb2af | 287 | return 0; |
462fb2af BD |
288 | |
289 | opt->optlen = iph->ihl*4 - sizeof(struct iphdr); | |
290 | if (ip_options_compile(dev_net(dev), opt, skb)) | |
291 | goto inhdr_error; | |
292 | ||
293 | /* Check correct handling of SRR option */ | |
294 | if (unlikely(opt->srr)) { | |
295 | struct in_device *in_dev = __in_dev_get_rcu(dev); | |
296 | if (in_dev && !IN_DEV_SOURCE_ROUTE(in_dev)) | |
297 | goto drop; | |
298 | ||
299 | if (ip_options_rcv_srr(skb)) | |
300 | goto drop; | |
301 | } | |
302 | ||
303 | return 0; | |
304 | ||
305 | inhdr_error: | |
306 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS); | |
307 | drop: | |
308 | return -1; | |
309 | } | |
310 | ||
e179e632 BDS |
311 | /* Fill in the header for fragmented IP packets handled by |
312 | * the IPv4 connection tracking code. | |
07317621 SH |
313 | */ |
314 | int nf_bridge_copy_header(struct sk_buff *skb) | |
315 | { | |
316 | int err; | |
e179e632 | 317 | unsigned int header_size; |
07317621 | 318 | |
e179e632 BDS |
319 | nf_bridge_update_protocol(skb); |
320 | header_size = ETH_HLEN + nf_bridge_encap_header_len(skb); | |
d9cc2048 | 321 | err = skb_cow_head(skb, header_size); |
07317621 SH |
322 | if (err) |
323 | return err; | |
324 | ||
27d7ff46 ACM |
325 | skb_copy_to_linear_data_offset(skb, -header_size, |
326 | skb->nf_bridge->data, header_size); | |
fc38582d | 327 | __skb_push(skb, nf_bridge_encap_header_len(skb)); |
07317621 SH |
328 | return 0; |
329 | } | |
330 | ||
1da177e4 LT |
331 | /* PF_BRIDGE/PRE_ROUTING *********************************************/ |
332 | /* Undo the changes made for ip6tables PREROUTING and continue the | |
333 | * bridge PRE_ROUTING hook. */ | |
334 | static int br_nf_pre_routing_finish_ipv6(struct sk_buff *skb) | |
335 | { | |
336 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
511c3f92 | 337 | struct rtable *rt; |
1da177e4 | 338 | |
1da177e4 LT |
339 | if (nf_bridge->mask & BRNF_PKT_TYPE) { |
340 | skb->pkt_type = PACKET_OTHERHOST; | |
341 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
342 | } | |
343 | nf_bridge->mask ^= BRNF_NF_BRIDGE_PREROUTING; | |
344 | ||
511c3f92 ED |
345 | rt = bridge_parent_rtable(nf_bridge->physindev); |
346 | if (!rt) { | |
4adf0af6 SW |
347 | kfree_skb(skb); |
348 | return 0; | |
349 | } | |
f9181f4f | 350 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
351 | |
352 | skb->dev = nf_bridge->physindev; | |
e179e632 | 353 | nf_bridge_update_protocol(skb); |
fc38582d | 354 | nf_bridge_push_encap_header(skb); |
713aefa3 | 355 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
356 | br_handle_frame_finish, 1); |
357 | ||
358 | return 0; | |
359 | } | |
360 | ||
e179e632 BDS |
361 | /* Obtain the correct destination MAC address, while preserving the original |
362 | * source MAC address. If we already know this address, we just copy it. If we | |
363 | * don't, we use the neighbour framework to find out. In both cases, we make | |
364 | * sure that br_handle_frame_finish() is called afterwards. | |
365 | */ | |
366 | static int br_nf_pre_routing_finish_bridge(struct sk_buff *skb) | |
367 | { | |
368 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
f6b72b62 | 369 | struct neighbour *neigh; |
e179e632 BDS |
370 | struct dst_entry *dst; |
371 | ||
372 | skb->dev = bridge_parent(skb->dev); | |
373 | if (!skb->dev) | |
374 | goto free_skb; | |
375 | dst = skb_dst(skb); | |
27217455 | 376 | neigh = dst_get_neighbour_noref(dst); |
f6b72b62 DM |
377 | if (neigh->hh.hh_len) { |
378 | neigh_hh_bridge(&neigh->hh, skb); | |
e179e632 BDS |
379 | skb->dev = nf_bridge->physindev; |
380 | return br_handle_frame_finish(skb); | |
8f40b161 | 381 | } else { |
e179e632 BDS |
382 | /* the neighbour function below overwrites the complete |
383 | * MAC header, so we save the Ethernet source address and | |
384 | * protocol number. */ | |
385 | skb_copy_from_linear_data_offset(skb, -(ETH_HLEN-ETH_ALEN), skb->nf_bridge->data, ETH_HLEN-ETH_ALEN); | |
386 | /* tell br_dev_xmit to continue with forwarding */ | |
387 | nf_bridge->mask |= BRNF_BRIDGED_DNAT; | |
8f40b161 | 388 | return neigh->output(neigh, skb); |
e179e632 BDS |
389 | } |
390 | free_skb: | |
391 | kfree_skb(skb); | |
392 | return 0; | |
393 | } | |
394 | ||
1da177e4 | 395 | /* This requires some explaining. If DNAT has taken place, |
ea2d9b41 | 396 | * we will need to fix up the destination Ethernet address. |
1da177e4 LT |
397 | * |
398 | * There are two cases to consider: | |
399 | * 1. The packet was DNAT'ed to a device in the same bridge | |
400 | * port group as it was received on. We can still bridge | |
401 | * the packet. | |
402 | * 2. The packet was DNAT'ed to a different device, either | |
403 | * a non-bridged device or another bridge port group. | |
404 | * The packet will need to be routed. | |
405 | * | |
406 | * The correct way of distinguishing between these two cases is to | |
407 | * call ip_route_input() and to look at skb->dst->dev, which is | |
408 | * changed to the destination device if ip_route_input() succeeds. | |
409 | * | |
ea2d9b41 | 410 | * Let's first consider the case that ip_route_input() succeeds: |
1da177e4 | 411 | * |
ea2d9b41 BDS |
412 | * If the output device equals the logical bridge device the packet |
413 | * came in on, we can consider this bridging. The corresponding MAC | |
414 | * address will be obtained in br_nf_pre_routing_finish_bridge. | |
1da177e4 LT |
415 | * Otherwise, the packet is considered to be routed and we just |
416 | * change the destination MAC address so that the packet will | |
f216f082 BDS |
417 | * later be passed up to the IP stack to be routed. For a redirected |
418 | * packet, ip_route_input() will give back the localhost as output device, | |
419 | * which differs from the bridge device. | |
1da177e4 | 420 | * |
ea2d9b41 | 421 | * Let's now consider the case that ip_route_input() fails: |
1da177e4 | 422 | * |
f216f082 BDS |
423 | * This can be because the destination address is martian, in which case |
424 | * the packet will be dropped. | |
ea2d9b41 BDS |
425 | * If IP forwarding is disabled, ip_route_input() will fail, while |
426 | * ip_route_output_key() can return success. The source | |
427 | * address for ip_route_output_key() is set to zero, so ip_route_output_key() | |
1da177e4 | 428 | * thinks we're handling a locally generated packet and won't care |
ea2d9b41 BDS |
429 | * if IP forwarding is enabled. If the output device equals the logical bridge |
430 | * device, we proceed as if ip_route_input() succeeded. If it differs from the | |
431 | * logical bridge port or if ip_route_output_key() fails we drop the packet. | |
432 | */ | |
1da177e4 LT |
433 | static int br_nf_pre_routing_finish(struct sk_buff *skb) |
434 | { | |
435 | struct net_device *dev = skb->dev; | |
eddc9ec5 | 436 | struct iphdr *iph = ip_hdr(skb); |
1da177e4 | 437 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; |
511c3f92 | 438 | struct rtable *rt; |
f216f082 | 439 | int err; |
1da177e4 | 440 | |
1da177e4 LT |
441 | if (nf_bridge->mask & BRNF_PKT_TYPE) { |
442 | skb->pkt_type = PACKET_OTHERHOST; | |
443 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
444 | } | |
445 | nf_bridge->mask ^= BRNF_NF_BRIDGE_PREROUTING; | |
1da177e4 | 446 | if (dnat_took_place(skb)) { |
f216f082 | 447 | if ((err = ip_route_input(skb, iph->daddr, iph->saddr, iph->tos, dev))) { |
f3abc9b9 | 448 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
f216f082 BDS |
449 | |
450 | /* If err equals -EHOSTUNREACH the error is due to a | |
451 | * martian destination or due to the fact that | |
452 | * forwarding is disabled. For most martian packets, | |
453 | * ip_route_output_key() will fail. It won't fail for 2 types of | |
454 | * martian destinations: loopback destinations and destination | |
455 | * 0.0.0.0. In both cases the packet will be dropped because the | |
456 | * destination is the loopback device and not the bridge. */ | |
457 | if (err != -EHOSTUNREACH || !in_dev || IN_DEV_FORWARD(in_dev)) | |
458 | goto free_skb; | |
1da177e4 | 459 | |
78fbfd8a DM |
460 | rt = ip_route_output(dev_net(dev), iph->daddr, 0, |
461 | RT_TOS(iph->tos), 0); | |
b23dd4fe | 462 | if (!IS_ERR(rt)) { |
1c011bed | 463 | /* - Bridged-and-DNAT'ed traffic doesn't |
f216f082 | 464 | * require ip_forwarding. */ |
b23dd4fe DM |
465 | if (rt->dst.dev == dev) { |
466 | skb_dst_set(skb, &rt->dst); | |
1da177e4 LT |
467 | goto bridged_dnat; |
468 | } | |
b23dd4fe | 469 | ip_rt_put(rt); |
1da177e4 | 470 | } |
f216f082 | 471 | free_skb: |
1da177e4 LT |
472 | kfree_skb(skb); |
473 | return 0; | |
474 | } else { | |
adf30907 | 475 | if (skb_dst(skb)->dev == dev) { |
1da177e4 | 476 | bridged_dnat: |
1da177e4 | 477 | skb->dev = nf_bridge->physindev; |
e179e632 | 478 | nf_bridge_update_protocol(skb); |
fc38582d | 479 | nf_bridge_push_encap_header(skb); |
713aefa3 JE |
480 | NF_HOOK_THRESH(NFPROTO_BRIDGE, |
481 | NF_BR_PRE_ROUTING, | |
1da177e4 LT |
482 | skb, skb->dev, NULL, |
483 | br_nf_pre_routing_finish_bridge, | |
484 | 1); | |
485 | return 0; | |
486 | } | |
789bc3e5 | 487 | memcpy(eth_hdr(skb)->h_dest, dev->dev_addr, ETH_ALEN); |
1da177e4 LT |
488 | skb->pkt_type = PACKET_HOST; |
489 | } | |
490 | } else { | |
511c3f92 ED |
491 | rt = bridge_parent_rtable(nf_bridge->physindev); |
492 | if (!rt) { | |
4adf0af6 SW |
493 | kfree_skb(skb); |
494 | return 0; | |
495 | } | |
f9181f4f | 496 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
497 | } |
498 | ||
499 | skb->dev = nf_bridge->physindev; | |
e179e632 | 500 | nf_bridge_update_protocol(skb); |
fc38582d | 501 | nf_bridge_push_encap_header(skb); |
713aefa3 | 502 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
503 | br_handle_frame_finish, 1); |
504 | ||
505 | return 0; | |
506 | } | |
507 | ||
4981682c PNA |
508 | static struct net_device *brnf_get_logical_dev(struct sk_buff *skb, const struct net_device *dev) |
509 | { | |
510 | struct net_device *vlan, *br; | |
511 | ||
512 | br = bridge_parent(dev); | |
513 | if (brnf_pass_vlan_indev == 0 || !vlan_tx_tag_present(skb)) | |
514 | return br; | |
515 | ||
516 | vlan = __vlan_find_dev_deep(br, vlan_tx_tag_get(skb) & VLAN_VID_MASK); | |
517 | ||
518 | return vlan ? vlan : br; | |
519 | } | |
520 | ||
1da177e4 | 521 | /* Some common code for IPv4/IPv6 */ |
5dce971a | 522 | static struct net_device *setup_pre_routing(struct sk_buff *skb) |
1da177e4 LT |
523 | { |
524 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
525 | ||
526 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
527 | skb->pkt_type = PACKET_HOST; | |
528 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
529 | } | |
530 | ||
531 | nf_bridge->mask |= BRNF_NF_BRIDGE_PREROUTING; | |
532 | nf_bridge->physindev = skb->dev; | |
4981682c | 533 | skb->dev = brnf_get_logical_dev(skb, skb->dev); |
e179e632 BDS |
534 | if (skb->protocol == htons(ETH_P_8021Q)) |
535 | nf_bridge->mask |= BRNF_8021Q; | |
536 | else if (skb->protocol == htons(ETH_P_PPP_SES)) | |
537 | nf_bridge->mask |= BRNF_PPPoE; | |
5dce971a SH |
538 | |
539 | return skb->dev; | |
1da177e4 LT |
540 | } |
541 | ||
542 | /* We only check the length. A bridge shouldn't do any hop-by-hop stuff anyway */ | |
543 | static int check_hbh_len(struct sk_buff *skb) | |
544 | { | |
0660e03f | 545 | unsigned char *raw = (u8 *)(ipv6_hdr(skb) + 1); |
1da177e4 | 546 | u32 pkt_len; |
d56f90a7 ACM |
547 | const unsigned char *nh = skb_network_header(skb); |
548 | int off = raw - nh; | |
789bc3e5 | 549 | int len = (raw[1] + 1) << 3; |
1da177e4 LT |
550 | |
551 | if ((raw + len) - skb->data > skb_headlen(skb)) | |
552 | goto bad; | |
553 | ||
554 | off += 2; | |
555 | len -= 2; | |
556 | ||
557 | while (len > 0) { | |
d56f90a7 | 558 | int optlen = nh[off + 1] + 2; |
1da177e4 | 559 | |
d56f90a7 | 560 | switch (nh[off]) { |
1de5a71c | 561 | case IPV6_TLV_PAD1: |
1da177e4 LT |
562 | optlen = 1; |
563 | break; | |
564 | ||
565 | case IPV6_TLV_PADN: | |
566 | break; | |
567 | ||
568 | case IPV6_TLV_JUMBO: | |
d56f90a7 | 569 | if (nh[off + 1] != 4 || (off & 3) != 2) |
1da177e4 | 570 | goto bad; |
d56f90a7 | 571 | pkt_len = ntohl(*(__be32 *) (nh + off + 2)); |
b0366486 | 572 | if (pkt_len <= IPV6_MAXPLEN || |
0660e03f | 573 | ipv6_hdr(skb)->payload_len) |
b0366486 | 574 | goto bad; |
1da177e4 LT |
575 | if (pkt_len > skb->len - sizeof(struct ipv6hdr)) |
576 | goto bad; | |
b0366486 | 577 | if (pskb_trim_rcsum(skb, |
789bc3e5 | 578 | pkt_len + sizeof(struct ipv6hdr))) |
b0366486 | 579 | goto bad; |
d56f90a7 | 580 | nh = skb_network_header(skb); |
1da177e4 LT |
581 | break; |
582 | default: | |
583 | if (optlen > len) | |
584 | goto bad; | |
585 | break; | |
586 | } | |
587 | off += optlen; | |
588 | len -= optlen; | |
589 | } | |
590 | if (len == 0) | |
591 | return 0; | |
592 | bad: | |
593 | return -1; | |
594 | ||
595 | } | |
596 | ||
597 | /* Replicate the checks that IPv6 does on packet reception and pass the packet | |
598 | * to ip6tables, which doesn't support NAT, so things are fairly simple. */ | |
599 | static unsigned int br_nf_pre_routing_ipv6(unsigned int hook, | |
789bc3e5 SH |
600 | struct sk_buff *skb, |
601 | const struct net_device *in, | |
602 | const struct net_device *out, | |
603 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 604 | { |
b71d1d42 | 605 | const struct ipv6hdr *hdr; |
1da177e4 | 606 | u32 pkt_len; |
1da177e4 LT |
607 | |
608 | if (skb->len < sizeof(struct ipv6hdr)) | |
deef4b52 | 609 | return NF_DROP; |
1da177e4 LT |
610 | |
611 | if (!pskb_may_pull(skb, sizeof(struct ipv6hdr))) | |
deef4b52 | 612 | return NF_DROP; |
1da177e4 | 613 | |
0660e03f | 614 | hdr = ipv6_hdr(skb); |
1da177e4 LT |
615 | |
616 | if (hdr->version != 6) | |
deef4b52 | 617 | return NF_DROP; |
1da177e4 LT |
618 | |
619 | pkt_len = ntohs(hdr->payload_len); | |
620 | ||
621 | if (pkt_len || hdr->nexthdr != NEXTHDR_HOP) { | |
622 | if (pkt_len + sizeof(struct ipv6hdr) > skb->len) | |
deef4b52 | 623 | return NF_DROP; |
b38dfee3 | 624 | if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr))) |
deef4b52 | 625 | return NF_DROP; |
1da177e4 LT |
626 | } |
627 | if (hdr->nexthdr == NEXTHDR_HOP && check_hbh_len(skb)) | |
deef4b52 | 628 | return NF_DROP; |
1da177e4 | 629 | |
789bc3e5 | 630 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 631 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 632 | return NF_DROP; |
5dce971a SH |
633 | if (!setup_pre_routing(skb)) |
634 | return NF_DROP; | |
1da177e4 | 635 | |
e179e632 | 636 | skb->protocol = htons(ETH_P_IPV6); |
713aefa3 | 637 | NF_HOOK(NFPROTO_IPV6, NF_INET_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
638 | br_nf_pre_routing_finish_ipv6); |
639 | ||
640 | return NF_STOLEN; | |
1da177e4 LT |
641 | } |
642 | ||
643 | /* Direct IPv6 traffic to br_nf_pre_routing_ipv6. | |
644 | * Replicate the checks that IPv4 does on packet reception. | |
645 | * Set skb->dev to the bridge device (i.e. parent of the | |
646 | * receiving device) to make netfilter happy, the REDIRECT | |
647 | * target in particular. Save the original destination IP | |
648 | * address to be able to detect DNAT afterwards. */ | |
3db05fea | 649 | static unsigned int br_nf_pre_routing(unsigned int hook, struct sk_buff *skb, |
ee02b3a6 SH |
650 | const struct net_device *in, |
651 | const struct net_device *out, | |
652 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 653 | { |
4df53d8b PM |
654 | struct net_bridge_port *p; |
655 | struct net_bridge *br; | |
e7c243c9 EP |
656 | __u32 len = nf_bridge_encap_header_len(skb); |
657 | ||
e7c243c9 | 658 | if (unlikely(!pskb_may_pull(skb, len))) |
deef4b52 | 659 | return NF_DROP; |
1da177e4 | 660 | |
e490c1de | 661 | p = br_port_get_rcu(in); |
4df53d8b | 662 | if (p == NULL) |
deef4b52 | 663 | return NF_DROP; |
4df53d8b PM |
664 | br = p->br; |
665 | ||
739e4505 | 666 | if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) { |
4df53d8b | 667 | if (!brnf_call_ip6tables && !br->nf_call_ip6tables) |
1da177e4 | 668 | return NF_ACCEPT; |
4df53d8b | 669 | |
fc38582d | 670 | nf_bridge_pull_encap_header_rcsum(skb); |
1da177e4 LT |
671 | return br_nf_pre_routing_ipv6(hook, skb, in, out, okfn); |
672 | } | |
4df53d8b PM |
673 | |
674 | if (!brnf_call_iptables && !br->nf_call_iptables) | |
1da177e4 | 675 | return NF_ACCEPT; |
1da177e4 | 676 | |
739e4505 | 677 | if (!IS_IP(skb) && !IS_VLAN_IP(skb) && !IS_PPPOE_IP(skb)) |
1da177e4 LT |
678 | return NF_ACCEPT; |
679 | ||
fc38582d | 680 | nf_bridge_pull_encap_header_rcsum(skb); |
1da177e4 | 681 | |
462fb2af | 682 | if (br_parse_ip_options(skb)) |
deef4b52 | 683 | return NF_DROP; |
17762060 | 684 | |
789bc3e5 | 685 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 686 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 687 | return NF_DROP; |
5dce971a SH |
688 | if (!setup_pre_routing(skb)) |
689 | return NF_DROP; | |
1da177e4 | 690 | store_orig_dstaddr(skb); |
e179e632 | 691 | skb->protocol = htons(ETH_P_IP); |
1da177e4 | 692 | |
713aefa3 | 693 | NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
694 | br_nf_pre_routing_finish); |
695 | ||
696 | return NF_STOLEN; | |
1da177e4 LT |
697 | } |
698 | ||
699 | ||
700 | /* PF_BRIDGE/LOCAL_IN ************************************************/ | |
701 | /* The packet is locally destined, which requires a real | |
702 | * dst_entry, so detach the fake one. On the way up, the | |
703 | * packet would pass through PRE_ROUTING again (which already | |
704 | * took place when the packet entered the bridge), but we | |
705 | * register an IPv4 PRE_ROUTING 'sabotage' hook that will | |
706 | * prevent this from happening. */ | |
3db05fea | 707 | static unsigned int br_nf_local_in(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
708 | const struct net_device *in, |
709 | const struct net_device *out, | |
710 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 711 | { |
a881e963 | 712 | br_drop_fake_rtable(skb); |
1da177e4 LT |
713 | return NF_ACCEPT; |
714 | } | |
715 | ||
1da177e4 LT |
716 | /* PF_BRIDGE/FORWARD *************************************************/ |
717 | static int br_nf_forward_finish(struct sk_buff *skb) | |
718 | { | |
719 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
720 | struct net_device *in; | |
1da177e4 | 721 | |
739e4505 | 722 | if (!IS_ARP(skb) && !IS_VLAN_ARP(skb)) { |
1da177e4 LT |
723 | in = nf_bridge->physindev; |
724 | if (nf_bridge->mask & BRNF_PKT_TYPE) { | |
725 | skb->pkt_type = PACKET_OTHERHOST; | |
726 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
727 | } | |
e94c6743 | 728 | nf_bridge_update_protocol(skb); |
1da177e4 LT |
729 | } else { |
730 | in = *((struct net_device **)(skb->cb)); | |
731 | } | |
fc38582d | 732 | nf_bridge_push_encap_header(skb); |
e179e632 | 733 | |
713aefa3 | 734 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_FORWARD, skb, in, |
789bc3e5 | 735 | skb->dev, br_forward_finish, 1); |
1da177e4 LT |
736 | return 0; |
737 | } | |
738 | ||
739e4505 | 739 | |
1da177e4 LT |
740 | /* This is the 'purely bridged' case. For IP, we pass the packet to |
741 | * netfilter with indev and outdev set to the bridge device, | |
742 | * but we are still able to filter on the 'real' indev/outdev | |
743 | * because of the physdev module. For ARP, indev and outdev are the | |
744 | * bridge ports. */ | |
3db05fea | 745 | static unsigned int br_nf_forward_ip(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
746 | const struct net_device *in, |
747 | const struct net_device *out, | |
748 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 749 | { |
1da177e4 | 750 | struct nf_bridge_info *nf_bridge; |
5dce971a | 751 | struct net_device *parent; |
76108cea | 752 | u_int8_t pf; |
1da177e4 LT |
753 | |
754 | if (!skb->nf_bridge) | |
755 | return NF_ACCEPT; | |
756 | ||
2dc2f207 PM |
757 | /* Need exclusive nf_bridge_info since we might have multiple |
758 | * different physoutdevs. */ | |
759 | if (!nf_bridge_unshare(skb)) | |
760 | return NF_DROP; | |
761 | ||
5dce971a SH |
762 | parent = bridge_parent(out); |
763 | if (!parent) | |
764 | return NF_DROP; | |
765 | ||
739e4505 | 766 | if (IS_IP(skb) || IS_VLAN_IP(skb) || IS_PPPOE_IP(skb)) |
1da177e4 | 767 | pf = PF_INET; |
739e4505 | 768 | else if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) |
1da177e4 | 769 | pf = PF_INET6; |
a2bd40ad HX |
770 | else |
771 | return NF_ACCEPT; | |
1da177e4 | 772 | |
3db05fea | 773 | nf_bridge_pull_encap_header(skb); |
1da177e4 | 774 | |
1da177e4 LT |
775 | nf_bridge = skb->nf_bridge; |
776 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
777 | skb->pkt_type = PACKET_HOST; | |
778 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
779 | } | |
780 | ||
cb685528 | 781 | if (pf == PF_INET && br_parse_ip_options(skb)) |
6b1e960f HX |
782 | return NF_DROP; |
783 | ||
1da177e4 LT |
784 | /* The physdev module checks on this */ |
785 | nf_bridge->mask |= BRNF_BRIDGED; | |
786 | nf_bridge->physoutdev = skb->dev; | |
e179e632 BDS |
787 | if (pf == PF_INET) |
788 | skb->protocol = htons(ETH_P_IP); | |
789 | else | |
790 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 791 | |
4981682c | 792 | NF_HOOK(pf, NF_INET_FORWARD, skb, brnf_get_logical_dev(skb, in), parent, |
5dce971a | 793 | br_nf_forward_finish); |
1da177e4 LT |
794 | |
795 | return NF_STOLEN; | |
796 | } | |
797 | ||
3db05fea | 798 | static unsigned int br_nf_forward_arp(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
799 | const struct net_device *in, |
800 | const struct net_device *out, | |
801 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 802 | { |
4df53d8b PM |
803 | struct net_bridge_port *p; |
804 | struct net_bridge *br; | |
1da177e4 LT |
805 | struct net_device **d = (struct net_device **)(skb->cb); |
806 | ||
e490c1de | 807 | p = br_port_get_rcu(out); |
4df53d8b PM |
808 | if (p == NULL) |
809 | return NF_ACCEPT; | |
810 | br = p->br; | |
811 | ||
812 | if (!brnf_call_arptables && !br->nf_call_arptables) | |
1da177e4 | 813 | return NF_ACCEPT; |
1da177e4 | 814 | |
739e4505 | 815 | if (!IS_ARP(skb)) { |
8b42ec39 | 816 | if (!IS_VLAN_ARP(skb)) |
1da177e4 | 817 | return NF_ACCEPT; |
3db05fea | 818 | nf_bridge_pull_encap_header(skb); |
1da177e4 LT |
819 | } |
820 | ||
d0a92be0 | 821 | if (arp_hdr(skb)->ar_pln != 4) { |
fc38582d | 822 | if (IS_VLAN_ARP(skb)) |
3db05fea | 823 | nf_bridge_push_encap_header(skb); |
1da177e4 LT |
824 | return NF_ACCEPT; |
825 | } | |
826 | *d = (struct net_device *)in; | |
fdc9314c | 827 | NF_HOOK(NFPROTO_ARP, NF_ARP_FORWARD, skb, (struct net_device *)in, |
1da177e4 LT |
828 | (struct net_device *)out, br_nf_forward_finish); |
829 | ||
830 | return NF_STOLEN; | |
831 | } | |
832 | ||
e6373c4c | 833 | #if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV4) |
2e2f7aef PM |
834 | static int br_nf_dev_queue_xmit(struct sk_buff *skb) |
835 | { | |
462fb2af BD |
836 | int ret; |
837 | ||
e179e632 | 838 | if (skb->nfct != NULL && skb->protocol == htons(ETH_P_IP) && |
6c79bf0f | 839 | skb->len + nf_bridge_mtu_reduction(skb) > skb->dev->mtu && |
87f94b4e | 840 | !skb_is_gso(skb)) { |
462fb2af BD |
841 | if (br_parse_ip_options(skb)) |
842 | /* Drop invalid packet */ | |
843 | return NF_DROP; | |
844 | ret = ip_fragment(skb, br_dev_queue_push_xmit); | |
87f94b4e | 845 | } else |
462fb2af BD |
846 | ret = br_dev_queue_push_xmit(skb); |
847 | ||
848 | return ret; | |
2e2f7aef | 849 | } |
c197facc | 850 | #else |
851 | static int br_nf_dev_queue_xmit(struct sk_buff *skb) | |
852 | { | |
853 | return br_dev_queue_push_xmit(skb); | |
854 | } | |
855 | #endif | |
1da177e4 LT |
856 | |
857 | /* PF_BRIDGE/POST_ROUTING ********************************************/ | |
3db05fea | 858 | static unsigned int br_nf_post_routing(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
859 | const struct net_device *in, |
860 | const struct net_device *out, | |
861 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 862 | { |
3db05fea | 863 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; |
1da177e4 | 864 | struct net_device *realoutdev = bridge_parent(skb->dev); |
76108cea | 865 | u_int8_t pf; |
1da177e4 | 866 | |
ea2d9b41 | 867 | if (!nf_bridge || !(nf_bridge->mask & BRNF_BRIDGED)) |
81d9ddae PM |
868 | return NF_ACCEPT; |
869 | ||
5dce971a SH |
870 | if (!realoutdev) |
871 | return NF_DROP; | |
872 | ||
739e4505 | 873 | if (IS_IP(skb) || IS_VLAN_IP(skb) || IS_PPPOE_IP(skb)) |
1da177e4 | 874 | pf = PF_INET; |
739e4505 | 875 | else if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) |
1da177e4 | 876 | pf = PF_INET6; |
a2bd40ad HX |
877 | else |
878 | return NF_ACCEPT; | |
1da177e4 | 879 | |
1da177e4 LT |
880 | /* We assume any code from br_dev_queue_push_xmit onwards doesn't care |
881 | * about the value of skb->pkt_type. */ | |
882 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
883 | skb->pkt_type = PACKET_HOST; | |
884 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
885 | } | |
886 | ||
fc38582d | 887 | nf_bridge_pull_encap_header(skb); |
1da177e4 | 888 | nf_bridge_save_header(skb); |
e179e632 BDS |
889 | if (pf == PF_INET) |
890 | skb->protocol = htons(ETH_P_IP); | |
891 | else | |
892 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 893 | |
6e23ae2a | 894 | NF_HOOK(pf, NF_INET_POST_ROUTING, skb, NULL, realoutdev, |
2e2f7aef | 895 | br_nf_dev_queue_xmit); |
1da177e4 LT |
896 | |
897 | return NF_STOLEN; | |
1da177e4 LT |
898 | } |
899 | ||
1da177e4 LT |
900 | /* IP/SABOTAGE *****************************************************/ |
901 | /* Don't hand locally destined packets to PF_INET(6)/PRE_ROUTING | |
902 | * for the second time. */ | |
3db05fea | 903 | static unsigned int ip_sabotage_in(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
904 | const struct net_device *in, |
905 | const struct net_device *out, | |
906 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 907 | { |
3db05fea HX |
908 | if (skb->nf_bridge && |
909 | !(skb->nf_bridge->mask & BRNF_NF_BRIDGE_PREROUTING)) { | |
1da177e4 LT |
910 | return NF_STOP; |
911 | } | |
912 | ||
913 | return NF_ACCEPT; | |
914 | } | |
915 | ||
ea2d9b41 BDS |
916 | /* For br_nf_post_routing, we need (prio = NF_BR_PRI_LAST), because |
917 | * br_dev_queue_push_xmit is called afterwards */ | |
1999414a | 918 | static struct nf_hook_ops br_nf_ops[] __read_mostly = { |
1490fd89 CG |
919 | { |
920 | .hook = br_nf_pre_routing, | |
921 | .owner = THIS_MODULE, | |
922 | .pf = PF_BRIDGE, | |
923 | .hooknum = NF_BR_PRE_ROUTING, | |
924 | .priority = NF_BR_PRI_BRNF, | |
925 | }, | |
926 | { | |
927 | .hook = br_nf_local_in, | |
928 | .owner = THIS_MODULE, | |
929 | .pf = PF_BRIDGE, | |
930 | .hooknum = NF_BR_LOCAL_IN, | |
931 | .priority = NF_BR_PRI_BRNF, | |
932 | }, | |
933 | { | |
934 | .hook = br_nf_forward_ip, | |
935 | .owner = THIS_MODULE, | |
936 | .pf = PF_BRIDGE, | |
937 | .hooknum = NF_BR_FORWARD, | |
938 | .priority = NF_BR_PRI_BRNF - 1, | |
939 | }, | |
940 | { | |
941 | .hook = br_nf_forward_arp, | |
942 | .owner = THIS_MODULE, | |
943 | .pf = PF_BRIDGE, | |
944 | .hooknum = NF_BR_FORWARD, | |
945 | .priority = NF_BR_PRI_BRNF, | |
946 | }, | |
1490fd89 CG |
947 | { |
948 | .hook = br_nf_post_routing, | |
949 | .owner = THIS_MODULE, | |
950 | .pf = PF_BRIDGE, | |
951 | .hooknum = NF_BR_POST_ROUTING, | |
952 | .priority = NF_BR_PRI_LAST, | |
953 | }, | |
954 | { | |
955 | .hook = ip_sabotage_in, | |
956 | .owner = THIS_MODULE, | |
957 | .pf = PF_INET, | |
958 | .hooknum = NF_INET_PRE_ROUTING, | |
959 | .priority = NF_IP_PRI_FIRST, | |
960 | }, | |
961 | { | |
962 | .hook = ip_sabotage_in, | |
963 | .owner = THIS_MODULE, | |
964 | .pf = PF_INET6, | |
965 | .hooknum = NF_INET_PRE_ROUTING, | |
966 | .priority = NF_IP6_PRI_FIRST, | |
967 | }, | |
1da177e4 LT |
968 | }; |
969 | ||
970 | #ifdef CONFIG_SYSCTL | |
971 | static | |
8d65af78 | 972 | int brnf_sysctl_call_tables(ctl_table * ctl, int write, |
789bc3e5 | 973 | void __user * buffer, size_t * lenp, loff_t * ppos) |
1da177e4 LT |
974 | { |
975 | int ret; | |
976 | ||
8d65af78 | 977 | ret = proc_dointvec(ctl, write, buffer, lenp, ppos); |
1da177e4 LT |
978 | |
979 | if (write && *(int *)(ctl->data)) | |
980 | *(int *)(ctl->data) = 1; | |
981 | return ret; | |
982 | } | |
983 | ||
984 | static ctl_table brnf_table[] = { | |
985 | { | |
1da177e4 LT |
986 | .procname = "bridge-nf-call-arptables", |
987 | .data = &brnf_call_arptables, | |
988 | .maxlen = sizeof(int), | |
989 | .mode = 0644, | |
6d9f239a | 990 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
991 | }, |
992 | { | |
1da177e4 LT |
993 | .procname = "bridge-nf-call-iptables", |
994 | .data = &brnf_call_iptables, | |
995 | .maxlen = sizeof(int), | |
996 | .mode = 0644, | |
6d9f239a | 997 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
998 | }, |
999 | { | |
1da177e4 LT |
1000 | .procname = "bridge-nf-call-ip6tables", |
1001 | .data = &brnf_call_ip6tables, | |
1002 | .maxlen = sizeof(int), | |
1003 | .mode = 0644, | |
6d9f239a | 1004 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1005 | }, |
1006 | { | |
1da177e4 LT |
1007 | .procname = "bridge-nf-filter-vlan-tagged", |
1008 | .data = &brnf_filter_vlan_tagged, | |
1009 | .maxlen = sizeof(int), | |
1010 | .mode = 0644, | |
6d9f239a | 1011 | .proc_handler = brnf_sysctl_call_tables, |
516299d2 MM |
1012 | }, |
1013 | { | |
516299d2 MM |
1014 | .procname = "bridge-nf-filter-pppoe-tagged", |
1015 | .data = &brnf_filter_pppoe_tagged, | |
1016 | .maxlen = sizeof(int), | |
1017 | .mode = 0644, | |
6d9f239a | 1018 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 | 1019 | }, |
4981682c PNA |
1020 | { |
1021 | .procname = "bridge-nf-pass-vlan-input-dev", | |
1022 | .data = &brnf_pass_vlan_indev, | |
1023 | .maxlen = sizeof(int), | |
1024 | .mode = 0644, | |
1025 | .proc_handler = brnf_sysctl_call_tables, | |
1026 | }, | |
f8572d8f | 1027 | { } |
1da177e4 | 1028 | }; |
1da177e4 LT |
1029 | #endif |
1030 | ||
5eb87f45 | 1031 | int __init br_netfilter_init(void) |
1da177e4 | 1032 | { |
5eb87f45 | 1033 | int ret; |
1da177e4 | 1034 | |
fc66f95c | 1035 | ret = dst_entries_init(&fake_dst_ops); |
5eb87f45 | 1036 | if (ret < 0) |
1da177e4 | 1037 | return ret; |
fc66f95c ED |
1038 | |
1039 | ret = nf_register_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); | |
1040 | if (ret < 0) { | |
1041 | dst_entries_destroy(&fake_dst_ops); | |
1042 | return ret; | |
1043 | } | |
1da177e4 | 1044 | #ifdef CONFIG_SYSCTL |
ec8f23ce | 1045 | brnf_sysctl_header = register_net_sysctl(&init_net, "net/bridge", brnf_table); |
1da177e4 | 1046 | if (brnf_sysctl_header == NULL) { |
789bc3e5 SH |
1047 | printk(KERN_WARNING |
1048 | "br_netfilter: can't register to sysctl.\n"); | |
5eb87f45 | 1049 | nf_unregister_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
fc66f95c | 1050 | dst_entries_destroy(&fake_dst_ops); |
5eb87f45 | 1051 | return -ENOMEM; |
1da177e4 LT |
1052 | } |
1053 | #endif | |
1da177e4 | 1054 | printk(KERN_NOTICE "Bridge firewalling registered\n"); |
1da177e4 LT |
1055 | return 0; |
1056 | } | |
1057 | ||
1058 | void br_netfilter_fini(void) | |
1059 | { | |
5eb87f45 | 1060 | nf_unregister_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
1da177e4 | 1061 | #ifdef CONFIG_SYSCTL |
5dd3df10 | 1062 | unregister_net_sysctl_table(brnf_sysctl_header); |
1da177e4 | 1063 | #endif |
fc66f95c | 1064 | dst_entries_destroy(&fake_dst_ops); |
1da177e4 | 1065 | } |