]>
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> | |
c5136b15 | 33 | #include <linux/rculist.h> |
f216f082 | 34 | #include <linux/inetdevice.h> |
14c85021 | 35 | |
1da177e4 LT |
36 | #include <net/ip.h> |
37 | #include <net/ipv6.h> | |
efb6de9b | 38 | #include <net/addrconf.h> |
14c85021 | 39 | #include <net/route.h> |
56768644 | 40 | #include <net/netfilter/br_netfilter.h> |
5f6c253e | 41 | #include <net/netns/generic.h> |
14c85021 | 42 | |
7c0f6ba6 | 43 | #include <linux/uaccess.h> |
1da177e4 LT |
44 | #include "br_private.h" |
45 | #ifdef CONFIG_SYSCTL | |
46 | #include <linux/sysctl.h> | |
47 | #endif | |
48 | ||
c7d03a00 | 49 | static unsigned int brnf_net_id __read_mostly; |
5f6c253e | 50 | |
fb0b8628 CB |
51 | struct brnf_net { |
52 | bool enabled; | |
53 | ||
65de80f5 CB |
54 | #ifdef CONFIG_SYSCTL |
55 | struct ctl_table_header *ctl_hdr; | |
56 | #endif | |
57 | ||
fb0b8628 CB |
58 | /* default value is 1 */ |
59 | int call_iptables; | |
60 | int call_ip6tables; | |
61 | int call_arptables; | |
62 | ||
63 | /* default value is 0 */ | |
64 | int filter_vlan_tagged; | |
65 | int filter_pppoe_tagged; | |
66 | int pass_vlan_indev; | |
67 | }; | |
68 | ||
739e4505 | 69 | #define IS_IP(skb) \ |
df8a39de | 70 | (!skb_vlan_tag_present(skb) && skb->protocol == htons(ETH_P_IP)) |
739e4505 FW |
71 | |
72 | #define IS_IPV6(skb) \ | |
df8a39de | 73 | (!skb_vlan_tag_present(skb) && skb->protocol == htons(ETH_P_IPV6)) |
739e4505 FW |
74 | |
75 | #define IS_ARP(skb) \ | |
df8a39de | 76 | (!skb_vlan_tag_present(skb) && skb->protocol == htons(ETH_P_ARP)) |
739e4505 | 77 | |
b6f99a21 | 78 | static inline __be16 vlan_proto(const struct sk_buff *skb) |
8b42ec39 | 79 | { |
df8a39de | 80 | if (skb_vlan_tag_present(skb)) |
13937911 JG |
81 | return skb->protocol; |
82 | else if (skb->protocol == htons(ETH_P_8021Q)) | |
83 | return vlan_eth_hdr(skb)->h_vlan_encapsulated_proto; | |
84 | else | |
85 | return 0; | |
8b42ec39 SH |
86 | } |
87 | ||
fb0b8628 CB |
88 | static inline bool is_vlan_ip(const struct sk_buff *skb, const struct net *net) |
89 | { | |
90 | struct brnf_net *brnet = net_generic(net, brnf_net_id); | |
91 | ||
92 | return vlan_proto(skb) == htons(ETH_P_IP) && brnet->filter_vlan_tagged; | |
93 | } | |
94 | ||
95 | static inline bool is_vlan_ipv6(const struct sk_buff *skb, | |
96 | const struct net *net) | |
97 | { | |
98 | struct brnf_net *brnet = net_generic(net, brnf_net_id); | |
8b42ec39 | 99 | |
fb0b8628 CB |
100 | return vlan_proto(skb) == htons(ETH_P_IPV6) && |
101 | brnet->filter_vlan_tagged; | |
102 | } | |
8b42ec39 | 103 | |
fb0b8628 CB |
104 | static inline bool is_vlan_arp(const struct sk_buff *skb, const struct net *net) |
105 | { | |
106 | struct brnf_net *brnet = net_generic(net, brnf_net_id); | |
107 | ||
108 | return vlan_proto(skb) == htons(ETH_P_ARP) && brnet->filter_vlan_tagged; | |
109 | } | |
1da177e4 | 110 | |
516299d2 MM |
111 | static inline __be16 pppoe_proto(const struct sk_buff *skb) |
112 | { | |
113 | return *((__be16 *)(skb_mac_header(skb) + ETH_HLEN + | |
114 | sizeof(struct pppoe_hdr))); | |
115 | } | |
116 | ||
fb0b8628 CB |
117 | static inline bool is_pppoe_ip(const struct sk_buff *skb, const struct net *net) |
118 | { | |
119 | struct brnf_net *brnet = net_generic(net, brnf_net_id); | |
120 | ||
121 | return skb->protocol == htons(ETH_P_PPP_SES) && | |
122 | pppoe_proto(skb) == htons(PPP_IP) && brnet->filter_pppoe_tagged; | |
123 | } | |
124 | ||
125 | static inline bool is_pppoe_ipv6(const struct sk_buff *skb, | |
126 | const struct net *net) | |
127 | { | |
128 | struct brnf_net *brnet = net_generic(net, brnf_net_id); | |
516299d2 | 129 | |
fb0b8628 CB |
130 | return skb->protocol == htons(ETH_P_PPP_SES) && |
131 | pppoe_proto(skb) == htons(PPP_IPV6) && | |
132 | brnet->filter_pppoe_tagged; | |
133 | } | |
516299d2 | 134 | |
e70deecb FW |
135 | /* largest possible L2 header, see br_nf_dev_queue_xmit() */ |
136 | #define NF_BRIDGE_MAX_MAC_HEADER_LENGTH (PPPOE_SES_HLEN + ETH_HLEN) | |
137 | ||
e70deecb FW |
138 | struct brnf_frag_data { |
139 | char mac[NF_BRIDGE_MAX_MAC_HEADER_LENGTH]; | |
140 | u8 encap_size; | |
141 | u8 size; | |
d7b59742 FW |
142 | u16 vlan_tci; |
143 | __be16 vlan_proto; | |
e70deecb FW |
144 | }; |
145 | ||
146 | static DEFINE_PER_CPU(struct brnf_frag_data, brnf_frag_data_storage); | |
e70deecb | 147 | |
a9fcc6a4 FW |
148 | static void nf_bridge_info_free(struct sk_buff *skb) |
149 | { | |
150 | if (skb->nf_bridge) { | |
151 | nf_bridge_put(skb->nf_bridge); | |
152 | skb->nf_bridge = NULL; | |
153 | } | |
154 | } | |
155 | ||
5dce971a SH |
156 | static inline struct net_device *bridge_parent(const struct net_device *dev) |
157 | { | |
b5ed54e9 | 158 | struct net_bridge_port *port; |
5dce971a | 159 | |
b5ed54e9 | 160 | port = br_port_get_rcu(dev); |
161 | return port ? port->br->dev : NULL; | |
5dce971a | 162 | } |
1da177e4 | 163 | |
2dc2f207 PM |
164 | static inline struct nf_bridge_info *nf_bridge_unshare(struct sk_buff *skb) |
165 | { | |
166 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
167 | ||
53869ceb | 168 | if (refcount_read(&nf_bridge->use) > 1) { |
2dc2f207 PM |
169 | struct nf_bridge_info *tmp = nf_bridge_alloc(skb); |
170 | ||
171 | if (tmp) { | |
172 | memcpy(tmp, nf_bridge, sizeof(struct nf_bridge_info)); | |
53869ceb | 173 | refcount_set(&tmp->use, 1); |
2dc2f207 | 174 | } |
4c3a76ab | 175 | nf_bridge_put(nf_bridge); |
2dc2f207 PM |
176 | nf_bridge = tmp; |
177 | } | |
178 | return nf_bridge; | |
179 | } | |
180 | ||
230ac490 | 181 | unsigned int nf_bridge_encap_header_len(const struct sk_buff *skb) |
8d045163 FW |
182 | { |
183 | switch (skb->protocol) { | |
184 | case __cpu_to_be16(ETH_P_8021Q): | |
185 | return VLAN_HLEN; | |
186 | case __cpu_to_be16(ETH_P_PPP_SES): | |
187 | return PPPOE_SES_HLEN; | |
188 | default: | |
189 | return 0; | |
190 | } | |
191 | } | |
192 | ||
fc38582d | 193 | static inline void nf_bridge_pull_encap_header(struct sk_buff *skb) |
fdeabdef | 194 | { |
fc38582d PM |
195 | unsigned int len = nf_bridge_encap_header_len(skb); |
196 | ||
197 | skb_pull(skb, len); | |
198 | skb->network_header += len; | |
199 | } | |
fdeabdef | 200 | |
fc38582d PM |
201 | static inline void nf_bridge_pull_encap_header_rcsum(struct sk_buff *skb) |
202 | { | |
203 | unsigned int len = nf_bridge_encap_header_len(skb); | |
204 | ||
205 | skb_pull_rcsum(skb, len); | |
206 | skb->network_header += len; | |
207 | } | |
208 | ||
462fb2af BD |
209 | /* When handing a packet over to the IP layer |
210 | * check whether we have a skb that is in the | |
211 | * expected format | |
212 | */ | |
213 | ||
c1444c63 | 214 | static int br_validate_ipv4(struct net *net, struct sk_buff *skb) |
462fb2af | 215 | { |
b71d1d42 | 216 | const struct iphdr *iph; |
462fb2af BD |
217 | u32 len; |
218 | ||
6caab7b0 SB |
219 | if (!pskb_may_pull(skb, sizeof(struct iphdr))) |
220 | goto inhdr_error; | |
221 | ||
462fb2af | 222 | iph = ip_hdr(skb); |
462fb2af BD |
223 | |
224 | /* Basic sanity checks */ | |
225 | if (iph->ihl < 5 || iph->version != 4) | |
226 | goto inhdr_error; | |
227 | ||
228 | if (!pskb_may_pull(skb, iph->ihl*4)) | |
229 | goto inhdr_error; | |
230 | ||
231 | iph = ip_hdr(skb); | |
232 | if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl))) | |
233 | goto inhdr_error; | |
234 | ||
235 | len = ntohs(iph->tot_len); | |
236 | if (skb->len < len) { | |
b45386ef | 237 | __IP_INC_STATS(net, IPSTATS_MIB_INTRUNCATEDPKTS); |
462fb2af BD |
238 | goto drop; |
239 | } else if (len < (iph->ihl*4)) | |
240 | goto inhdr_error; | |
241 | ||
242 | if (pskb_trim_rcsum(skb, len)) { | |
b45386ef | 243 | __IP_INC_STATS(net, IPSTATS_MIB_INDISCARDS); |
462fb2af BD |
244 | goto drop; |
245 | } | |
246 | ||
f8e9881c | 247 | memset(IPCB(skb), 0, sizeof(struct inet_skb_parm)); |
7677e868 HX |
248 | /* We should really parse IP options here but until |
249 | * somebody who actually uses IP options complains to | |
250 | * us we'll just silently ignore the options because | |
251 | * we're lazy! | |
252 | */ | |
462fb2af BD |
253 | return 0; |
254 | ||
255 | inhdr_error: | |
b45386ef | 256 | __IP_INC_STATS(net, IPSTATS_MIB_INHDRERRORS); |
462fb2af BD |
257 | drop: |
258 | return -1; | |
259 | } | |
260 | ||
230ac490 | 261 | void nf_bridge_update_protocol(struct sk_buff *skb) |
4a9d2f20 | 262 | { |
3eaf4025 FW |
263 | switch (skb->nf_bridge->orig_proto) { |
264 | case BRNF_PROTO_8021Q: | |
4a9d2f20 | 265 | skb->protocol = htons(ETH_P_8021Q); |
3eaf4025 FW |
266 | break; |
267 | case BRNF_PROTO_PPPOE: | |
4a9d2f20 | 268 | skb->protocol = htons(ETH_P_PPP_SES); |
3eaf4025 FW |
269 | break; |
270 | case BRNF_PROTO_UNCHANGED: | |
271 | break; | |
272 | } | |
4a9d2f20 FW |
273 | } |
274 | ||
e179e632 BDS |
275 | /* Obtain the correct destination MAC address, while preserving the original |
276 | * source MAC address. If we already know this address, we just copy it. If we | |
277 | * don't, we use the neighbour framework to find out. In both cases, we make | |
278 | * sure that br_handle_frame_finish() is called afterwards. | |
279 | */ | |
0c4b51f0 | 280 | int br_nf_pre_routing_finish_bridge(struct net *net, struct sock *sk, struct sk_buff *skb) |
e179e632 | 281 | { |
f6b72b62 | 282 | struct neighbour *neigh; |
e179e632 BDS |
283 | struct dst_entry *dst; |
284 | ||
285 | skb->dev = bridge_parent(skb->dev); | |
286 | if (!skb->dev) | |
287 | goto free_skb; | |
288 | dst = skb_dst(skb); | |
f9d75166 DM |
289 | neigh = dst_neigh_lookup_skb(dst, skb); |
290 | if (neigh) { | |
38330783 | 291 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
f9d75166 DM |
292 | int ret; |
293 | ||
1cca7fd0 | 294 | if ((neigh->nud_state & NUD_CONNECTED) && neigh->hh.hh_len) { |
f9d75166 DM |
295 | neigh_hh_bridge(&neigh->hh, skb); |
296 | skb->dev = nf_bridge->physindev; | |
0c4b51f0 | 297 | ret = br_handle_frame_finish(net, sk, skb); |
f9d75166 DM |
298 | } else { |
299 | /* the neighbour function below overwrites the complete | |
300 | * MAC header, so we save the Ethernet source address and | |
301 | * protocol number. | |
302 | */ | |
303 | skb_copy_from_linear_data_offset(skb, | |
304 | -(ETH_HLEN-ETH_ALEN), | |
e70deecb | 305 | nf_bridge->neigh_header, |
f9d75166 DM |
306 | ETH_HLEN-ETH_ALEN); |
307 | /* tell br_dev_xmit to continue with forwarding */ | |
72b1e5e4 | 308 | nf_bridge->bridged_dnat = 1; |
93fdd47e | 309 | /* FIXME Need to refragment */ |
f9d75166 DM |
310 | ret = neigh->output(neigh, skb); |
311 | } | |
312 | neigh_release(neigh); | |
313 | return ret; | |
e179e632 BDS |
314 | } |
315 | free_skb: | |
316 | kfree_skb(skb); | |
317 | return 0; | |
318 | } | |
319 | ||
230ac490 PNA |
320 | static inline bool |
321 | br_nf_ipv4_daddr_was_changed(const struct sk_buff *skb, | |
322 | const struct nf_bridge_info *nf_bridge) | |
8cae308d | 323 | { |
230ac490 | 324 | return ip_hdr(skb)->daddr != nf_bridge->ipv4_daddr; |
8cae308d BT |
325 | } |
326 | ||
1da177e4 | 327 | /* This requires some explaining. If DNAT has taken place, |
ea2d9b41 | 328 | * we will need to fix up the destination Ethernet address. |
faecbb45 | 329 | * This is also true when SNAT takes place (for the reply direction). |
1da177e4 LT |
330 | * |
331 | * There are two cases to consider: | |
332 | * 1. The packet was DNAT'ed to a device in the same bridge | |
333 | * port group as it was received on. We can still bridge | |
334 | * the packet. | |
335 | * 2. The packet was DNAT'ed to a different device, either | |
336 | * a non-bridged device or another bridge port group. | |
337 | * The packet will need to be routed. | |
338 | * | |
339 | * The correct way of distinguishing between these two cases is to | |
340 | * call ip_route_input() and to look at skb->dst->dev, which is | |
341 | * changed to the destination device if ip_route_input() succeeds. | |
342 | * | |
ea2d9b41 | 343 | * Let's first consider the case that ip_route_input() succeeds: |
1da177e4 | 344 | * |
ea2d9b41 BDS |
345 | * If the output device equals the logical bridge device the packet |
346 | * came in on, we can consider this bridging. The corresponding MAC | |
347 | * address will be obtained in br_nf_pre_routing_finish_bridge. | |
1da177e4 LT |
348 | * Otherwise, the packet is considered to be routed and we just |
349 | * change the destination MAC address so that the packet will | |
f216f082 BDS |
350 | * later be passed up to the IP stack to be routed. For a redirected |
351 | * packet, ip_route_input() will give back the localhost as output device, | |
352 | * which differs from the bridge device. | |
1da177e4 | 353 | * |
ea2d9b41 | 354 | * Let's now consider the case that ip_route_input() fails: |
1da177e4 | 355 | * |
f216f082 BDS |
356 | * This can be because the destination address is martian, in which case |
357 | * the packet will be dropped. | |
ea2d9b41 BDS |
358 | * If IP forwarding is disabled, ip_route_input() will fail, while |
359 | * ip_route_output_key() can return success. The source | |
360 | * address for ip_route_output_key() is set to zero, so ip_route_output_key() | |
1da177e4 | 361 | * thinks we're handling a locally generated packet and won't care |
ea2d9b41 BDS |
362 | * if IP forwarding is enabled. If the output device equals the logical bridge |
363 | * device, we proceed as if ip_route_input() succeeded. If it differs from the | |
364 | * logical bridge port or if ip_route_output_key() fails we drop the packet. | |
365 | */ | |
0c4b51f0 | 366 | static int br_nf_pre_routing_finish(struct net *net, struct sock *sk, struct sk_buff *skb) |
1da177e4 LT |
367 | { |
368 | struct net_device *dev = skb->dev; | |
eddc9ec5 | 369 | struct iphdr *iph = ip_hdr(skb); |
38330783 | 370 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
511c3f92 | 371 | struct rtable *rt; |
f216f082 | 372 | int err; |
93fdd47e | 373 | |
411ffb4f | 374 | nf_bridge->frag_max_size = IPCB(skb)->frag_max_size; |
1da177e4 | 375 | |
a1e67951 | 376 | if (nf_bridge->pkt_otherhost) { |
1da177e4 | 377 | skb->pkt_type = PACKET_OTHERHOST; |
a1e67951 | 378 | nf_bridge->pkt_otherhost = false; |
1da177e4 | 379 | } |
72b1e5e4 | 380 | nf_bridge->in_prerouting = 0; |
230ac490 | 381 | if (br_nf_ipv4_daddr_was_changed(skb, nf_bridge)) { |
f216f082 | 382 | if ((err = ip_route_input(skb, iph->daddr, iph->saddr, iph->tos, dev))) { |
f3abc9b9 | 383 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
f216f082 BDS |
384 | |
385 | /* If err equals -EHOSTUNREACH the error is due to a | |
386 | * martian destination or due to the fact that | |
387 | * forwarding is disabled. For most martian packets, | |
388 | * ip_route_output_key() will fail. It won't fail for 2 types of | |
389 | * martian destinations: loopback destinations and destination | |
390 | * 0.0.0.0. In both cases the packet will be dropped because the | |
391 | * destination is the loopback device and not the bridge. */ | |
392 | if (err != -EHOSTUNREACH || !in_dev || IN_DEV_FORWARD(in_dev)) | |
393 | goto free_skb; | |
1da177e4 | 394 | |
f2d74cf8 | 395 | rt = ip_route_output(net, iph->daddr, 0, |
78fbfd8a | 396 | RT_TOS(iph->tos), 0); |
b23dd4fe | 397 | if (!IS_ERR(rt)) { |
1c011bed | 398 | /* - Bridged-and-DNAT'ed traffic doesn't |
f216f082 | 399 | * require ip_forwarding. */ |
b23dd4fe DM |
400 | if (rt->dst.dev == dev) { |
401 | skb_dst_set(skb, &rt->dst); | |
1da177e4 LT |
402 | goto bridged_dnat; |
403 | } | |
b23dd4fe | 404 | ip_rt_put(rt); |
1da177e4 | 405 | } |
f216f082 | 406 | free_skb: |
1da177e4 LT |
407 | kfree_skb(skb); |
408 | return 0; | |
409 | } else { | |
adf30907 | 410 | if (skb_dst(skb)->dev == dev) { |
1da177e4 | 411 | bridged_dnat: |
1da177e4 | 412 | skb->dev = nf_bridge->physindev; |
e179e632 | 413 | nf_bridge_update_protocol(skb); |
fc38582d | 414 | nf_bridge_push_encap_header(skb); |
c5136b15 FW |
415 | br_nf_hook_thresh(NF_BR_PRE_ROUTING, |
416 | net, sk, skb, skb->dev, | |
417 | NULL, | |
14221cc4 | 418 | br_nf_pre_routing_finish_bridge); |
1da177e4 LT |
419 | return 0; |
420 | } | |
04091142 | 421 | ether_addr_copy(eth_hdr(skb)->h_dest, dev->dev_addr); |
1da177e4 LT |
422 | skb->pkt_type = PACKET_HOST; |
423 | } | |
424 | } else { | |
511c3f92 ED |
425 | rt = bridge_parent_rtable(nf_bridge->physindev); |
426 | if (!rt) { | |
4adf0af6 SW |
427 | kfree_skb(skb); |
428 | return 0; | |
429 | } | |
f9181f4f | 430 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
431 | } |
432 | ||
433 | skb->dev = nf_bridge->physindev; | |
e179e632 | 434 | nf_bridge_update_protocol(skb); |
fc38582d | 435 | nf_bridge_push_encap_header(skb); |
c5136b15 FW |
436 | br_nf_hook_thresh(NF_BR_PRE_ROUTING, net, sk, skb, skb->dev, NULL, |
437 | br_handle_frame_finish); | |
1da177e4 LT |
438 | return 0; |
439 | } | |
440 | ||
fb0b8628 CB |
441 | static struct net_device *brnf_get_logical_dev(struct sk_buff *skb, |
442 | const struct net_device *dev, | |
443 | const struct net *net) | |
4981682c PNA |
444 | { |
445 | struct net_device *vlan, *br; | |
fb0b8628 | 446 | struct brnf_net *brnet = net_generic(net, brnf_net_id); |
4981682c PNA |
447 | |
448 | br = bridge_parent(dev); | |
fb0b8628 CB |
449 | |
450 | if (brnet->pass_vlan_indev == 0 || !skb_vlan_tag_present(skb)) | |
4981682c PNA |
451 | return br; |
452 | ||
f06c7f9f | 453 | vlan = __vlan_find_dev_deep_rcu(br, skb->vlan_proto, |
df8a39de | 454 | skb_vlan_tag_get(skb) & VLAN_VID_MASK); |
4981682c PNA |
455 | |
456 | return vlan ? vlan : br; | |
457 | } | |
458 | ||
1da177e4 | 459 | /* Some common code for IPv4/IPv6 */ |
fb0b8628 | 460 | struct net_device *setup_pre_routing(struct sk_buff *skb, const struct net *net) |
1da177e4 | 461 | { |
38330783 | 462 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
1da177e4 LT |
463 | |
464 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
465 | skb->pkt_type = PACKET_HOST; | |
a1e67951 | 466 | nf_bridge->pkt_otherhost = true; |
1da177e4 LT |
467 | } |
468 | ||
72b1e5e4 | 469 | nf_bridge->in_prerouting = 1; |
1da177e4 | 470 | nf_bridge->physindev = skb->dev; |
fb0b8628 | 471 | skb->dev = brnf_get_logical_dev(skb, skb->dev, net); |
3eaf4025 | 472 | |
e179e632 | 473 | if (skb->protocol == htons(ETH_P_8021Q)) |
3eaf4025 | 474 | nf_bridge->orig_proto = BRNF_PROTO_8021Q; |
e179e632 | 475 | else if (skb->protocol == htons(ETH_P_PPP_SES)) |
3eaf4025 | 476 | nf_bridge->orig_proto = BRNF_PROTO_PPPOE; |
5dce971a | 477 | |
6b8dbcf2 FW |
478 | /* Must drop socket now because of tproxy. */ |
479 | skb_orphan(skb); | |
5dce971a | 480 | return skb->dev; |
1da177e4 LT |
481 | } |
482 | ||
1da177e4 LT |
483 | /* Direct IPv6 traffic to br_nf_pre_routing_ipv6. |
484 | * Replicate the checks that IPv4 does on packet reception. | |
485 | * Set skb->dev to the bridge device (i.e. parent of the | |
486 | * receiving device) to make netfilter happy, the REDIRECT | |
487 | * target in particular. Save the original destination IP | |
488 | * address to be able to detect DNAT afterwards. */ | |
06198b34 | 489 | static unsigned int br_nf_pre_routing(void *priv, |
795aa6ef | 490 | struct sk_buff *skb, |
238e54c9 | 491 | const struct nf_hook_state *state) |
1da177e4 | 492 | { |
faecbb45 | 493 | struct nf_bridge_info *nf_bridge; |
4df53d8b PM |
494 | struct net_bridge_port *p; |
495 | struct net_bridge *br; | |
e7c243c9 | 496 | __u32 len = nf_bridge_encap_header_len(skb); |
fb0b8628 | 497 | struct brnf_net *brnet; |
e7c243c9 | 498 | |
e7c243c9 | 499 | if (unlikely(!pskb_may_pull(skb, len))) |
deef4b52 | 500 | return NF_DROP; |
1da177e4 | 501 | |
238e54c9 | 502 | p = br_port_get_rcu(state->in); |
4df53d8b | 503 | if (p == NULL) |
deef4b52 | 504 | return NF_DROP; |
4df53d8b PM |
505 | br = p->br; |
506 | ||
fb0b8628 CB |
507 | brnet = net_generic(state->net, brnf_net_id); |
508 | if (IS_IPV6(skb) || is_vlan_ipv6(skb, state->net) || | |
509 | is_pppoe_ipv6(skb, state->net)) { | |
510 | if (!brnet->call_ip6tables && | |
fd701d89 | 511 | !br_opt_get(br, BROPT_NF_CALL_IP6TABLES)) |
1da177e4 | 512 | return NF_ACCEPT; |
4df53d8b | 513 | |
fc38582d | 514 | nf_bridge_pull_encap_header_rcsum(skb); |
06198b34 | 515 | return br_nf_pre_routing_ipv6(priv, skb, state); |
1da177e4 | 516 | } |
4df53d8b | 517 | |
fb0b8628 | 518 | if (!brnet->call_iptables && !br_opt_get(br, BROPT_NF_CALL_IPTABLES)) |
1da177e4 | 519 | return NF_ACCEPT; |
1da177e4 | 520 | |
fb0b8628 CB |
521 | if (!IS_IP(skb) && !is_vlan_ip(skb, state->net) && |
522 | !is_pppoe_ip(skb, state->net)) | |
1da177e4 LT |
523 | return NF_ACCEPT; |
524 | ||
fc38582d | 525 | nf_bridge_pull_encap_header_rcsum(skb); |
1da177e4 | 526 | |
c1444c63 | 527 | if (br_validate_ipv4(state->net, skb)) |
deef4b52 | 528 | return NF_DROP; |
17762060 | 529 | |
789bc3e5 | 530 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 531 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 532 | return NF_DROP; |
fb0b8628 | 533 | if (!setup_pre_routing(skb, state->net)) |
5dce971a | 534 | return NF_DROP; |
c055d5b0 | 535 | |
faecbb45 FW |
536 | nf_bridge = nf_bridge_info_get(skb); |
537 | nf_bridge->ipv4_daddr = ip_hdr(skb)->daddr; | |
538 | ||
e179e632 | 539 | skb->protocol = htons(ETH_P_IP); |
3442e060 | 540 | skb->transport_header = skb->network_header + ip_hdr(skb)->ihl * 4; |
1da177e4 | 541 | |
29a26a56 | 542 | NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, state->net, state->sk, skb, |
7026b1dd | 543 | skb->dev, NULL, |
1da177e4 LT |
544 | br_nf_pre_routing_finish); |
545 | ||
546 | return NF_STOLEN; | |
1da177e4 LT |
547 | } |
548 | ||
549 | ||
1da177e4 | 550 | /* PF_BRIDGE/FORWARD *************************************************/ |
0c4b51f0 | 551 | static int br_nf_forward_finish(struct net *net, struct sock *sk, struct sk_buff *skb) |
1da177e4 | 552 | { |
38330783 | 553 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
1da177e4 | 554 | struct net_device *in; |
1da177e4 | 555 | |
fb0b8628 | 556 | if (!IS_ARP(skb) && !is_vlan_arp(skb, net)) { |
0b67c43c | 557 | |
efb6de9b | 558 | if (skb->protocol == htons(ETH_P_IP)) |
411ffb4f | 559 | nf_bridge->frag_max_size = IPCB(skb)->frag_max_size; |
efb6de9b BT |
560 | |
561 | if (skb->protocol == htons(ETH_P_IPV6)) | |
562 | nf_bridge->frag_max_size = IP6CB(skb)->frag_max_size; | |
0b67c43c | 563 | |
1da177e4 | 564 | in = nf_bridge->physindev; |
a1e67951 | 565 | if (nf_bridge->pkt_otherhost) { |
1da177e4 | 566 | skb->pkt_type = PACKET_OTHERHOST; |
a1e67951 | 567 | nf_bridge->pkt_otherhost = false; |
1da177e4 | 568 | } |
e94c6743 | 569 | nf_bridge_update_protocol(skb); |
1da177e4 LT |
570 | } else { |
571 | in = *((struct net_device **)(skb->cb)); | |
572 | } | |
fc38582d | 573 | nf_bridge_push_encap_header(skb); |
e179e632 | 574 | |
1610a73c PNA |
575 | br_nf_hook_thresh(NF_BR_FORWARD, net, sk, skb, in, skb->dev, |
576 | br_forward_finish); | |
1da177e4 LT |
577 | return 0; |
578 | } | |
579 | ||
739e4505 | 580 | |
1da177e4 LT |
581 | /* This is the 'purely bridged' case. For IP, we pass the packet to |
582 | * netfilter with indev and outdev set to the bridge device, | |
583 | * but we are still able to filter on the 'real' indev/outdev | |
584 | * because of the physdev module. For ARP, indev and outdev are the | |
585 | * bridge ports. */ | |
06198b34 | 586 | static unsigned int br_nf_forward_ip(void *priv, |
795aa6ef | 587 | struct sk_buff *skb, |
238e54c9 | 588 | const struct nf_hook_state *state) |
1da177e4 | 589 | { |
1da177e4 | 590 | struct nf_bridge_info *nf_bridge; |
5dce971a | 591 | struct net_device *parent; |
76108cea | 592 | u_int8_t pf; |
1da177e4 LT |
593 | |
594 | if (!skb->nf_bridge) | |
595 | return NF_ACCEPT; | |
596 | ||
2dc2f207 PM |
597 | /* Need exclusive nf_bridge_info since we might have multiple |
598 | * different physoutdevs. */ | |
599 | if (!nf_bridge_unshare(skb)) | |
600 | return NF_DROP; | |
601 | ||
38330783 FW |
602 | nf_bridge = nf_bridge_info_get(skb); |
603 | if (!nf_bridge) | |
604 | return NF_DROP; | |
605 | ||
238e54c9 | 606 | parent = bridge_parent(state->out); |
5dce971a SH |
607 | if (!parent) |
608 | return NF_DROP; | |
609 | ||
fb0b8628 CB |
610 | if (IS_IP(skb) || is_vlan_ip(skb, state->net) || |
611 | is_pppoe_ip(skb, state->net)) | |
aa740f46 | 612 | pf = NFPROTO_IPV4; |
fb0b8628 CB |
613 | else if (IS_IPV6(skb) || is_vlan_ipv6(skb, state->net) || |
614 | is_pppoe_ipv6(skb, state->net)) | |
aa740f46 | 615 | pf = NFPROTO_IPV6; |
a2bd40ad HX |
616 | else |
617 | return NF_ACCEPT; | |
1da177e4 | 618 | |
3db05fea | 619 | nf_bridge_pull_encap_header(skb); |
1da177e4 | 620 | |
1da177e4 LT |
621 | if (skb->pkt_type == PACKET_OTHERHOST) { |
622 | skb->pkt_type = PACKET_HOST; | |
a1e67951 | 623 | nf_bridge->pkt_otherhost = true; |
1da177e4 LT |
624 | } |
625 | ||
0b67c43c | 626 | if (pf == NFPROTO_IPV4) { |
c1444c63 | 627 | if (br_validate_ipv4(state->net, skb)) |
0b67c43c | 628 | return NF_DROP; |
411ffb4f | 629 | IPCB(skb)->frag_max_size = nf_bridge->frag_max_size; |
0b67c43c | 630 | } |
6b1e960f | 631 | |
efb6de9b | 632 | if (pf == NFPROTO_IPV6) { |
c1444c63 | 633 | if (br_validate_ipv6(state->net, skb)) |
efb6de9b BT |
634 | return NF_DROP; |
635 | IP6CB(skb)->frag_max_size = nf_bridge->frag_max_size; | |
636 | } | |
637 | ||
1da177e4 | 638 | nf_bridge->physoutdev = skb->dev; |
aa740f46 | 639 | if (pf == NFPROTO_IPV4) |
e179e632 BDS |
640 | skb->protocol = htons(ETH_P_IP); |
641 | else | |
642 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 643 | |
29a26a56 | 644 | NF_HOOK(pf, NF_INET_FORWARD, state->net, NULL, skb, |
fb0b8628 | 645 | brnf_get_logical_dev(skb, state->in, state->net), |
238e54c9 | 646 | parent, br_nf_forward_finish); |
1da177e4 LT |
647 | |
648 | return NF_STOLEN; | |
649 | } | |
650 | ||
06198b34 | 651 | static unsigned int br_nf_forward_arp(void *priv, |
795aa6ef | 652 | struct sk_buff *skb, |
238e54c9 | 653 | const struct nf_hook_state *state) |
1da177e4 | 654 | { |
4df53d8b PM |
655 | struct net_bridge_port *p; |
656 | struct net_bridge *br; | |
1da177e4 | 657 | struct net_device **d = (struct net_device **)(skb->cb); |
fb0b8628 | 658 | struct brnf_net *brnet; |
1da177e4 | 659 | |
238e54c9 | 660 | p = br_port_get_rcu(state->out); |
4df53d8b PM |
661 | if (p == NULL) |
662 | return NF_ACCEPT; | |
663 | br = p->br; | |
664 | ||
fb0b8628 CB |
665 | brnet = net_generic(state->net, brnf_net_id); |
666 | if (!brnet->call_arptables && !br_opt_get(br, BROPT_NF_CALL_ARPTABLES)) | |
1da177e4 | 667 | return NF_ACCEPT; |
1da177e4 | 668 | |
739e4505 | 669 | if (!IS_ARP(skb)) { |
fb0b8628 | 670 | if (!is_vlan_arp(skb, state->net)) |
1da177e4 | 671 | return NF_ACCEPT; |
3db05fea | 672 | nf_bridge_pull_encap_header(skb); |
1da177e4 LT |
673 | } |
674 | ||
d0a92be0 | 675 | if (arp_hdr(skb)->ar_pln != 4) { |
fb0b8628 | 676 | if (is_vlan_arp(skb, state->net)) |
3db05fea | 677 | nf_bridge_push_encap_header(skb); |
1da177e4 LT |
678 | return NF_ACCEPT; |
679 | } | |
238e54c9 | 680 | *d = state->in; |
29a26a56 | 681 | NF_HOOK(NFPROTO_ARP, NF_ARP_FORWARD, state->net, state->sk, skb, |
7026b1dd | 682 | state->in, state->out, br_nf_forward_finish); |
1da177e4 LT |
683 | |
684 | return NF_STOLEN; | |
685 | } | |
686 | ||
6532948b | 687 | static int br_nf_push_frag_xmit(struct net *net, struct sock *sk, struct sk_buff *skb) |
8bd63cf1 | 688 | { |
e70deecb | 689 | struct brnf_frag_data *data; |
8bd63cf1 | 690 | int err; |
8bd63cf1 | 691 | |
e70deecb FW |
692 | data = this_cpu_ptr(&brnf_frag_data_storage); |
693 | err = skb_cow_head(skb, data->size); | |
8bd63cf1 | 694 | |
e70deecb | 695 | if (err) { |
8bd63cf1 FW |
696 | kfree_skb(skb); |
697 | return 0; | |
698 | } | |
699 | ||
d7b59742 FW |
700 | if (data->vlan_tci) { |
701 | skb->vlan_tci = data->vlan_tci; | |
702 | skb->vlan_proto = data->vlan_proto; | |
703 | } | |
704 | ||
e70deecb FW |
705 | skb_copy_to_linear_data_offset(skb, -data->size, data->mac, data->size); |
706 | __skb_push(skb, data->encap_size); | |
707 | ||
a9fcc6a4 | 708 | nf_bridge_info_free(skb); |
0c4b51f0 | 709 | return br_dev_queue_push_xmit(net, sk, skb); |
8bd63cf1 FW |
710 | } |
711 | ||
8d4df0b9 EB |
712 | static int |
713 | br_nf_ip_fragment(struct net *net, struct sock *sk, struct sk_buff *skb, | |
694869b3 | 714 | int (*output)(struct net *, struct sock *, struct sk_buff *)) |
49d16b23 | 715 | { |
fedbb6b4 | 716 | unsigned int mtu = ip_skb_dst_mtu(sk, skb); |
49d16b23 | 717 | struct iphdr *iph = ip_hdr(skb); |
49d16b23 AZ |
718 | |
719 | if (unlikely(((iph->frag_off & htons(IP_DF)) && !skb->ignore_df) || | |
720 | (IPCB(skb)->frag_max_size && | |
721 | IPCB(skb)->frag_max_size > mtu))) { | |
8d4df0b9 | 722 | IP_INC_STATS(net, IPSTATS_MIB_FRAGFAILS); |
49d16b23 AZ |
723 | kfree_skb(skb); |
724 | return -EMSGSIZE; | |
725 | } | |
726 | ||
694869b3 | 727 | return ip_do_fragment(net, sk, skb, output); |
49d16b23 AZ |
728 | } |
729 | ||
33b1f313 FW |
730 | static unsigned int nf_bridge_mtu_reduction(const struct sk_buff *skb) |
731 | { | |
732 | if (skb->nf_bridge->orig_proto == BRNF_PROTO_PPPOE) | |
733 | return PPPOE_SES_HLEN; | |
734 | return 0; | |
735 | } | |
736 | ||
0c4b51f0 | 737 | static int br_nf_dev_queue_xmit(struct net *net, struct sock *sk, struct sk_buff *skb) |
2e2f7aef | 738 | { |
4ca60d08 FW |
739 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
740 | unsigned int mtu, mtu_reserved; | |
462fb2af | 741 | |
efb6de9b | 742 | mtu_reserved = nf_bridge_mtu_reduction(skb); |
4ca60d08 FW |
743 | mtu = skb->dev->mtu; |
744 | ||
745 | if (nf_bridge->frag_max_size && nf_bridge->frag_max_size < mtu) | |
746 | mtu = nf_bridge->frag_max_size; | |
efb6de9b | 747 | |
4ca60d08 | 748 | if (skb_is_gso(skb) || skb->len + mtu_reserved <= mtu) { |
a9fcc6a4 | 749 | nf_bridge_info_free(skb); |
0c4b51f0 | 750 | return br_dev_queue_push_xmit(net, sk, skb); |
a9fcc6a4 | 751 | } |
7a8d831d | 752 | |
93fdd47e HX |
753 | /* This is wrong! We should preserve the original fragment |
754 | * boundaries by preserving frag_list rather than refragmenting. | |
755 | */ | |
c9322458 AB |
756 | if (IS_ENABLED(CONFIG_NF_DEFRAG_IPV4) && |
757 | skb->protocol == htons(ETH_P_IP)) { | |
e70deecb FW |
758 | struct brnf_frag_data *data; |
759 | ||
c1444c63 | 760 | if (br_validate_ipv4(net, skb)) |
dd302b59 | 761 | goto drop; |
411ffb4f BT |
762 | |
763 | IPCB(skb)->frag_max_size = nf_bridge->frag_max_size; | |
e70deecb FW |
764 | |
765 | nf_bridge_update_protocol(skb); | |
766 | ||
767 | data = this_cpu_ptr(&brnf_frag_data_storage); | |
d7b59742 FW |
768 | |
769 | data->vlan_tci = skb->vlan_tci; | |
770 | data->vlan_proto = skb->vlan_proto; | |
e70deecb FW |
771 | data->encap_size = nf_bridge_encap_header_len(skb); |
772 | data->size = ETH_HLEN + data->encap_size; | |
773 | ||
774 | skb_copy_from_linear_data_offset(skb, -data->size, data->mac, | |
775 | data->size); | |
776 | ||
694869b3 | 777 | return br_nf_ip_fragment(net, sk, skb, br_nf_push_frag_xmit); |
e70deecb | 778 | } |
c9322458 AB |
779 | if (IS_ENABLED(CONFIG_NF_DEFRAG_IPV6) && |
780 | skb->protocol == htons(ETH_P_IPV6)) { | |
efb6de9b BT |
781 | const struct nf_ipv6_ops *v6ops = nf_get_ipv6_ops(); |
782 | struct brnf_frag_data *data; | |
462fb2af | 783 | |
c1444c63 | 784 | if (br_validate_ipv6(net, skb)) |
dd302b59 | 785 | goto drop; |
efb6de9b BT |
786 | |
787 | IP6CB(skb)->frag_max_size = nf_bridge->frag_max_size; | |
788 | ||
789 | nf_bridge_update_protocol(skb); | |
790 | ||
791 | data = this_cpu_ptr(&brnf_frag_data_storage); | |
792 | data->encap_size = nf_bridge_encap_header_len(skb); | |
793 | data->size = ETH_HLEN + data->encap_size; | |
794 | ||
795 | skb_copy_from_linear_data_offset(skb, -data->size, data->mac, | |
796 | data->size); | |
797 | ||
798 | if (v6ops) | |
7d8c6e39 | 799 | return v6ops->fragment(net, sk, skb, br_nf_push_frag_xmit); |
dd302b59 FW |
800 | |
801 | kfree_skb(skb); | |
802 | return -EMSGSIZE; | |
efb6de9b | 803 | } |
a9fcc6a4 | 804 | nf_bridge_info_free(skb); |
0c4b51f0 | 805 | return br_dev_queue_push_xmit(net, sk, skb); |
dd302b59 FW |
806 | drop: |
807 | kfree_skb(skb); | |
808 | return 0; | |
c197facc | 809 | } |
1da177e4 LT |
810 | |
811 | /* PF_BRIDGE/POST_ROUTING ********************************************/ | |
06198b34 | 812 | static unsigned int br_nf_post_routing(void *priv, |
795aa6ef | 813 | struct sk_buff *skb, |
238e54c9 | 814 | const struct nf_hook_state *state) |
1da177e4 | 815 | { |
38330783 | 816 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
1da177e4 | 817 | struct net_device *realoutdev = bridge_parent(skb->dev); |
76108cea | 818 | u_int8_t pf; |
1da177e4 | 819 | |
e4bb9bcb FW |
820 | /* if nf_bridge is set, but ->physoutdev is NULL, this packet came in |
821 | * on a bridge, but was delivered locally and is now being routed: | |
822 | * | |
823 | * POST_ROUTING was already invoked from the ip stack. | |
824 | */ | |
825 | if (!nf_bridge || !nf_bridge->physoutdev) | |
81d9ddae PM |
826 | return NF_ACCEPT; |
827 | ||
5dce971a SH |
828 | if (!realoutdev) |
829 | return NF_DROP; | |
830 | ||
fb0b8628 CB |
831 | if (IS_IP(skb) || is_vlan_ip(skb, state->net) || |
832 | is_pppoe_ip(skb, state->net)) | |
aa740f46 | 833 | pf = NFPROTO_IPV4; |
fb0b8628 CB |
834 | else if (IS_IPV6(skb) || is_vlan_ipv6(skb, state->net) || |
835 | is_pppoe_ipv6(skb, state->net)) | |
aa740f46 | 836 | pf = NFPROTO_IPV6; |
a2bd40ad HX |
837 | else |
838 | return NF_ACCEPT; | |
1da177e4 | 839 | |
1da177e4 LT |
840 | /* We assume any code from br_dev_queue_push_xmit onwards doesn't care |
841 | * about the value of skb->pkt_type. */ | |
842 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
843 | skb->pkt_type = PACKET_HOST; | |
a1e67951 | 844 | nf_bridge->pkt_otherhost = true; |
1da177e4 LT |
845 | } |
846 | ||
fc38582d | 847 | nf_bridge_pull_encap_header(skb); |
aa740f46 | 848 | if (pf == NFPROTO_IPV4) |
e179e632 BDS |
849 | skb->protocol = htons(ETH_P_IP); |
850 | else | |
851 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 852 | |
29a26a56 | 853 | NF_HOOK(pf, NF_INET_POST_ROUTING, state->net, state->sk, skb, |
7026b1dd | 854 | NULL, realoutdev, |
2e2f7aef | 855 | br_nf_dev_queue_xmit); |
1da177e4 LT |
856 | |
857 | return NF_STOLEN; | |
1da177e4 LT |
858 | } |
859 | ||
1da177e4 LT |
860 | /* IP/SABOTAGE *****************************************************/ |
861 | /* Don't hand locally destined packets to PF_INET(6)/PRE_ROUTING | |
862 | * for the second time. */ | |
06198b34 | 863 | static unsigned int ip_sabotage_in(void *priv, |
795aa6ef | 864 | struct sk_buff *skb, |
238e54c9 | 865 | const struct nf_hook_state *state) |
1da177e4 | 866 | { |
ec9a51be DA |
867 | if (skb->nf_bridge && !skb->nf_bridge->in_prerouting && |
868 | !netif_is_l3_master(skb->dev)) { | |
06fd3a39 PNA |
869 | state->okfn(state->net, state->sk, skb); |
870 | return NF_STOLEN; | |
871 | } | |
1da177e4 LT |
872 | |
873 | return NF_ACCEPT; | |
874 | } | |
875 | ||
e5de75bf PNA |
876 | /* This is called when br_netfilter has called into iptables/netfilter, |
877 | * and DNAT has taken place on a bridge-forwarded packet. | |
878 | * | |
879 | * neigh->output has created a new MAC header, with local br0 MAC | |
880 | * as saddr. | |
881 | * | |
882 | * This restores the original MAC saddr of the bridged packet | |
883 | * before invoking bridge forward logic to transmit the packet. | |
884 | */ | |
885 | static void br_nf_pre_routing_finish_bridge_slow(struct sk_buff *skb) | |
886 | { | |
38330783 | 887 | struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); |
e5de75bf PNA |
888 | |
889 | skb_pull(skb, ETH_HLEN); | |
72b1e5e4 | 890 | nf_bridge->bridged_dnat = 0; |
e5de75bf | 891 | |
e70deecb FW |
892 | BUILD_BUG_ON(sizeof(nf_bridge->neigh_header) != (ETH_HLEN - ETH_ALEN)); |
893 | ||
894 | skb_copy_to_linear_data_offset(skb, -(ETH_HLEN - ETH_ALEN), | |
895 | nf_bridge->neigh_header, | |
896 | ETH_HLEN - ETH_ALEN); | |
e5de75bf | 897 | skb->dev = nf_bridge->physindev; |
7fb48c5b FW |
898 | |
899 | nf_bridge->physoutdev = NULL; | |
0c4b51f0 | 900 | br_handle_frame_finish(dev_net(skb->dev), NULL, skb); |
e5de75bf PNA |
901 | } |
902 | ||
1a4ba64d | 903 | static int br_nf_dev_xmit(struct sk_buff *skb) |
e5de75bf | 904 | { |
72b1e5e4 | 905 | if (skb->nf_bridge && skb->nf_bridge->bridged_dnat) { |
e5de75bf PNA |
906 | br_nf_pre_routing_finish_bridge_slow(skb); |
907 | return 1; | |
908 | } | |
909 | return 0; | |
910 | } | |
1a4ba64d PNA |
911 | |
912 | static const struct nf_br_ops br_ops = { | |
913 | .br_dev_xmit_hook = br_nf_dev_xmit, | |
914 | }; | |
e5de75bf | 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 */ | |
591bb278 | 918 | static const struct nf_hook_ops br_nf_ops[] = { |
1490fd89 CG |
919 | { |
920 | .hook = br_nf_pre_routing, | |
aa740f46 | 921 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
922 | .hooknum = NF_BR_PRE_ROUTING, |
923 | .priority = NF_BR_PRI_BRNF, | |
924 | }, | |
1490fd89 CG |
925 | { |
926 | .hook = br_nf_forward_ip, | |
aa740f46 | 927 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
928 | .hooknum = NF_BR_FORWARD, |
929 | .priority = NF_BR_PRI_BRNF - 1, | |
930 | }, | |
931 | { | |
932 | .hook = br_nf_forward_arp, | |
aa740f46 | 933 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
934 | .hooknum = NF_BR_FORWARD, |
935 | .priority = NF_BR_PRI_BRNF, | |
936 | }, | |
1490fd89 CG |
937 | { |
938 | .hook = br_nf_post_routing, | |
aa740f46 | 939 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
940 | .hooknum = NF_BR_POST_ROUTING, |
941 | .priority = NF_BR_PRI_LAST, | |
942 | }, | |
943 | { | |
944 | .hook = ip_sabotage_in, | |
aa740f46 | 945 | .pf = NFPROTO_IPV4, |
1490fd89 CG |
946 | .hooknum = NF_INET_PRE_ROUTING, |
947 | .priority = NF_IP_PRI_FIRST, | |
948 | }, | |
949 | { | |
950 | .hook = ip_sabotage_in, | |
aa740f46 | 951 | .pf = NFPROTO_IPV6, |
1490fd89 CG |
952 | .hooknum = NF_INET_PRE_ROUTING, |
953 | .priority = NF_IP6_PRI_FIRST, | |
954 | }, | |
1da177e4 LT |
955 | }; |
956 | ||
5f6c253e FW |
957 | static int brnf_device_event(struct notifier_block *unused, unsigned long event, |
958 | void *ptr) | |
959 | { | |
960 | struct net_device *dev = netdev_notifier_info_to_dev(ptr); | |
961 | struct brnf_net *brnet; | |
962 | struct net *net; | |
963 | int ret; | |
964 | ||
965 | if (event != NETDEV_REGISTER || !(dev->priv_flags & IFF_EBRIDGE)) | |
966 | return NOTIFY_DONE; | |
967 | ||
968 | ASSERT_RTNL(); | |
969 | ||
970 | net = dev_net(dev); | |
971 | brnet = net_generic(net, brnf_net_id); | |
972 | if (brnet->enabled) | |
973 | return NOTIFY_OK; | |
974 | ||
975 | ret = nf_register_net_hooks(net, br_nf_ops, ARRAY_SIZE(br_nf_ops)); | |
976 | if (ret) | |
977 | return NOTIFY_BAD; | |
978 | ||
979 | brnet->enabled = true; | |
980 | return NOTIFY_OK; | |
981 | } | |
982 | ||
5f6c253e FW |
983 | static struct notifier_block brnf_notifier __read_mostly = { |
984 | .notifier_call = brnf_device_event, | |
985 | }; | |
986 | ||
c5136b15 FW |
987 | /* recursively invokes nf_hook_slow (again), skipping already-called |
988 | * hooks (< NF_BR_PRI_BRNF). | |
989 | * | |
990 | * Called with rcu read lock held. | |
991 | */ | |
992 | int br_nf_hook_thresh(unsigned int hook, struct net *net, | |
993 | struct sock *sk, struct sk_buff *skb, | |
994 | struct net_device *indev, | |
995 | struct net_device *outdev, | |
996 | int (*okfn)(struct net *, struct sock *, | |
997 | struct sk_buff *)) | |
998 | { | |
960632ec | 999 | const struct nf_hook_entries *e; |
c5136b15 | 1000 | struct nf_hook_state state; |
960632ec AC |
1001 | struct nf_hook_ops **ops; |
1002 | unsigned int i; | |
c5136b15 FW |
1003 | int ret; |
1004 | ||
960632ec AC |
1005 | e = rcu_dereference(net->nf.hooks[NFPROTO_BRIDGE][hook]); |
1006 | if (!e) | |
c5136b15 FW |
1007 | return okfn(net, sk, skb); |
1008 | ||
960632ec AC |
1009 | ops = nf_hook_entries_get_hook_ops(e); |
1010 | for (i = 0; i < e->num_hook_entries && | |
1011 | ops[i]->priority <= NF_BR_PRI_BRNF; i++) | |
1012 | ; | |
1013 | ||
01886bd9 | 1014 | nf_hook_state_init(&state, hook, NFPROTO_BRIDGE, indev, outdev, |
1610a73c | 1015 | sk, net, okfn); |
c5136b15 | 1016 | |
960632ec | 1017 | ret = nf_hook_slow(skb, &state, e, i); |
c5136b15 FW |
1018 | if (ret == 1) |
1019 | ret = okfn(net, sk, skb); | |
1020 | ||
1021 | return ret; | |
1022 | } | |
1023 | ||
1da177e4 LT |
1024 | #ifdef CONFIG_SYSCTL |
1025 | static | |
fe2c6338 | 1026 | int brnf_sysctl_call_tables(struct ctl_table *ctl, int write, |
56b148eb | 1027 | void __user *buffer, size_t *lenp, loff_t *ppos) |
1da177e4 LT |
1028 | { |
1029 | int ret; | |
1030 | ||
8d65af78 | 1031 | ret = proc_dointvec(ctl, write, buffer, lenp, ppos); |
1da177e4 LT |
1032 | |
1033 | if (write && *(int *)(ctl->data)) | |
1034 | *(int *)(ctl->data) = 1; | |
1035 | return ret; | |
1036 | } | |
1037 | ||
fe2c6338 | 1038 | static struct ctl_table brnf_table[] = { |
1da177e4 | 1039 | { |
1da177e4 | 1040 | .procname = "bridge-nf-call-arptables", |
1da177e4 LT |
1041 | .maxlen = sizeof(int), |
1042 | .mode = 0644, | |
6d9f239a | 1043 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1044 | }, |
1045 | { | |
1da177e4 | 1046 | .procname = "bridge-nf-call-iptables", |
1da177e4 LT |
1047 | .maxlen = sizeof(int), |
1048 | .mode = 0644, | |
6d9f239a | 1049 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1050 | }, |
1051 | { | |
1da177e4 | 1052 | .procname = "bridge-nf-call-ip6tables", |
1da177e4 LT |
1053 | .maxlen = sizeof(int), |
1054 | .mode = 0644, | |
6d9f239a | 1055 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1056 | }, |
1057 | { | |
1da177e4 | 1058 | .procname = "bridge-nf-filter-vlan-tagged", |
1da177e4 LT |
1059 | .maxlen = sizeof(int), |
1060 | .mode = 0644, | |
6d9f239a | 1061 | .proc_handler = brnf_sysctl_call_tables, |
516299d2 MM |
1062 | }, |
1063 | { | |
516299d2 | 1064 | .procname = "bridge-nf-filter-pppoe-tagged", |
516299d2 MM |
1065 | .maxlen = sizeof(int), |
1066 | .mode = 0644, | |
6d9f239a | 1067 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 | 1068 | }, |
4981682c PNA |
1069 | { |
1070 | .procname = "bridge-nf-pass-vlan-input-dev", | |
4981682c PNA |
1071 | .maxlen = sizeof(int), |
1072 | .mode = 0644, | |
1073 | .proc_handler = brnf_sysctl_call_tables, | |
1074 | }, | |
f8572d8f | 1075 | { } |
1da177e4 | 1076 | }; |
fb0b8628 CB |
1077 | |
1078 | static inline void br_netfilter_sysctl_default(struct brnf_net *brnf) | |
1079 | { | |
1080 | brnf->call_iptables = 1; | |
1081 | brnf->call_ip6tables = 1; | |
1082 | brnf->call_arptables = 1; | |
1083 | brnf->filter_vlan_tagged = 0; | |
1084 | brnf->filter_pppoe_tagged = 0; | |
1085 | brnf->pass_vlan_indev = 0; | |
1086 | } | |
1087 | ||
65de80f5 CB |
1088 | static int br_netfilter_sysctl_init_net(struct net *net) |
1089 | { | |
1090 | struct ctl_table *table = brnf_table; | |
1091 | struct brnf_net *brnet; | |
1092 | ||
1093 | if (!net_eq(net, &init_net)) { | |
1094 | table = kmemdup(table, sizeof(brnf_table), GFP_KERNEL); | |
1095 | if (!table) | |
1096 | return -ENOMEM; | |
1097 | } | |
1098 | ||
1099 | brnet = net_generic(net, brnf_net_id); | |
1100 | table[0].data = &brnet->call_arptables; | |
1101 | table[1].data = &brnet->call_iptables; | |
1102 | table[2].data = &brnet->call_ip6tables; | |
1103 | table[3].data = &brnet->filter_vlan_tagged; | |
1104 | table[4].data = &brnet->filter_pppoe_tagged; | |
1105 | table[5].data = &brnet->pass_vlan_indev; | |
1106 | ||
1107 | br_netfilter_sysctl_default(brnet); | |
1108 | ||
1109 | brnet->ctl_hdr = register_net_sysctl(net, "net/bridge", table); | |
1110 | if (!brnet->ctl_hdr) { | |
1111 | if (!net_eq(net, &init_net)) | |
1112 | kfree(table); | |
1113 | ||
1114 | return -ENOMEM; | |
1115 | } | |
1116 | ||
1117 | return 0; | |
1118 | } | |
1119 | ||
1120 | static void br_netfilter_sysctl_exit_net(struct net *net, | |
1121 | struct brnf_net *brnet) | |
1122 | { | |
386e630c CB |
1123 | struct ctl_table *table = brnet->ctl_hdr->ctl_table_arg; |
1124 | ||
65de80f5 CB |
1125 | unregister_net_sysctl_table(brnet->ctl_hdr); |
1126 | if (!net_eq(net, &init_net)) | |
386e630c | 1127 | kfree(table); |
65de80f5 CB |
1128 | } |
1129 | ||
1130 | static int __net_init brnf_init_net(struct net *net) | |
1131 | { | |
1132 | return br_netfilter_sysctl_init_net(net); | |
1133 | } | |
1134 | #endif | |
1135 | ||
1136 | static void __net_exit brnf_exit_net(struct net *net) | |
1137 | { | |
1138 | struct brnf_net *brnet; | |
1139 | ||
1140 | brnet = net_generic(net, brnf_net_id); | |
1141 | if (brnet->enabled) { | |
1142 | nf_unregister_net_hooks(net, br_nf_ops, ARRAY_SIZE(br_nf_ops)); | |
1143 | brnet->enabled = false; | |
1144 | } | |
1145 | ||
1146 | #ifdef CONFIG_SYSCTL | |
1147 | br_netfilter_sysctl_exit_net(net, brnet); | |
1da177e4 | 1148 | #endif |
65de80f5 CB |
1149 | } |
1150 | ||
1151 | static struct pernet_operations brnf_net_ops __read_mostly = { | |
1152 | #ifdef CONFIG_SYSCTL | |
1153 | .init = brnf_init_net, | |
1154 | #endif | |
1155 | .exit = brnf_exit_net, | |
1156 | .id = &brnf_net_id, | |
1157 | .size = sizeof(struct brnf_net), | |
1158 | }; | |
1da177e4 | 1159 | |
34666d46 | 1160 | static int __init br_netfilter_init(void) |
1da177e4 | 1161 | { |
5eb87f45 | 1162 | int ret; |
1da177e4 | 1163 | |
5f6c253e | 1164 | ret = register_pernet_subsys(&brnf_net_ops); |
5eb87f45 | 1165 | if (ret < 0) |
1da177e4 | 1166 | return ret; |
fc66f95c | 1167 | |
5f6c253e FW |
1168 | ret = register_netdevice_notifier(&brnf_notifier); |
1169 | if (ret < 0) { | |
1170 | unregister_pernet_subsys(&brnf_net_ops); | |
1171 | return ret; | |
1172 | } | |
1173 | ||
1a4ba64d | 1174 | RCU_INIT_POINTER(nf_br_ops, &br_ops); |
1da177e4 | 1175 | printk(KERN_NOTICE "Bridge firewalling registered\n"); |
1da177e4 LT |
1176 | return 0; |
1177 | } | |
1178 | ||
34666d46 | 1179 | static void __exit br_netfilter_fini(void) |
1da177e4 | 1180 | { |
1a4ba64d | 1181 | RCU_INIT_POINTER(nf_br_ops, NULL); |
5f6c253e FW |
1182 | unregister_netdevice_notifier(&brnf_notifier); |
1183 | unregister_pernet_subsys(&brnf_net_ops); | |
1da177e4 | 1184 | } |
34666d46 PNA |
1185 | |
1186 | module_init(br_netfilter_init); | |
1187 | module_exit(br_netfilter_fini); | |
1188 | ||
1189 | MODULE_LICENSE("GPL"); | |
1190 | MODULE_AUTHOR("Lennert Buytenhek <buytenh@gnu.org>"); | |
1191 | MODULE_AUTHOR("Bart De Schuymer <bdschuym@pandora.be>"); | |
1192 | MODULE_DESCRIPTION("Linux ethernet netfilter firewall bridge"); |