]>
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 | ||
6700c270 DM |
114 | static void fake_update_pmtu(struct dst_entry *dst, struct sock *sk, |
115 | struct sk_buff *skb, u32 mtu) | |
631339f1 HX |
116 | { |
117 | } | |
118 | ||
6700c270 DM |
119 | static void fake_redirect(struct dst_entry *dst, struct sock *sk, |
120 | struct sk_buff *skb) | |
b587ee3b DM |
121 | { |
122 | } | |
123 | ||
6407d74c AH |
124 | static u32 *fake_cow_metrics(struct dst_entry *dst, unsigned long old) |
125 | { | |
126 | return NULL; | |
127 | } | |
128 | ||
f894cbf8 DM |
129 | static struct neighbour *fake_neigh_lookup(const struct dst_entry *dst, |
130 | struct sk_buff *skb, | |
131 | const void *daddr) | |
d3aaeb38 DM |
132 | { |
133 | return NULL; | |
134 | } | |
135 | ||
a13861a2 ED |
136 | static unsigned int fake_mtu(const struct dst_entry *dst) |
137 | { | |
138 | return dst->dev->mtu; | |
139 | } | |
140 | ||
631339f1 HX |
141 | static struct dst_ops fake_dst_ops = { |
142 | .family = AF_INET, | |
09640e63 | 143 | .protocol = cpu_to_be16(ETH_P_IP), |
631339f1 | 144 | .update_pmtu = fake_update_pmtu, |
b587ee3b | 145 | .redirect = fake_redirect, |
6407d74c | 146 | .cow_metrics = fake_cow_metrics, |
d3aaeb38 | 147 | .neigh_lookup = fake_neigh_lookup, |
a13861a2 | 148 | .mtu = fake_mtu, |
631339f1 HX |
149 | }; |
150 | ||
4adf0af6 SW |
151 | /* |
152 | * Initialize bogus route table used to keep netfilter happy. | |
1da177e4 LT |
153 | * Currently, we fill in the PMTU entry because netfilter |
154 | * refragmentation needs it, and the rt_flags entry because | |
155 | * ipt_REJECT needs it. Future netfilter modules might | |
4adf0af6 SW |
156 | * require us to fill additional fields. |
157 | */ | |
33eb9873 ED |
158 | static const u32 br_dst_default_metrics[RTAX_MAX] = { |
159 | [RTAX_MTU - 1] = 1500, | |
160 | }; | |
161 | ||
4adf0af6 SW |
162 | void br_netfilter_rtable_init(struct net_bridge *br) |
163 | { | |
164 | struct rtable *rt = &br->fake_rtable; | |
165 | ||
d8d1f30b CG |
166 | atomic_set(&rt->dst.__refcnt, 1); |
167 | rt->dst.dev = br->dev; | |
168 | rt->dst.path = &rt->dst; | |
33eb9873 | 169 | dst_init_metrics(&rt->dst, br_dst_default_metrics, true); |
a881e963 | 170 | rt->dst.flags = DST_NOXFRM | DST_NOPEER | DST_FAKE_RTABLE; |
d8d1f30b | 171 | rt->dst.ops = &fake_dst_ops; |
4adf0af6 SW |
172 | } |
173 | ||
174 | static inline struct rtable *bridge_parent_rtable(const struct net_device *dev) | |
175 | { | |
b5ed54e9 | 176 | struct net_bridge_port *port; |
177 | ||
178 | port = br_port_get_rcu(dev); | |
179 | return port ? &port->br->fake_rtable : NULL; | |
4adf0af6 | 180 | } |
1da177e4 | 181 | |
5dce971a SH |
182 | static inline struct net_device *bridge_parent(const struct net_device *dev) |
183 | { | |
b5ed54e9 | 184 | struct net_bridge_port *port; |
5dce971a | 185 | |
b5ed54e9 | 186 | port = br_port_get_rcu(dev); |
187 | return port ? port->br->dev : NULL; | |
5dce971a | 188 | } |
1da177e4 | 189 | |
fdeabdef SH |
190 | static inline struct nf_bridge_info *nf_bridge_alloc(struct sk_buff *skb) |
191 | { | |
192 | skb->nf_bridge = kzalloc(sizeof(struct nf_bridge_info), GFP_ATOMIC); | |
193 | if (likely(skb->nf_bridge)) | |
194 | atomic_set(&(skb->nf_bridge->use), 1); | |
195 | ||
196 | return skb->nf_bridge; | |
197 | } | |
198 | ||
2dc2f207 PM |
199 | static inline struct nf_bridge_info *nf_bridge_unshare(struct sk_buff *skb) |
200 | { | |
201 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
202 | ||
203 | if (atomic_read(&nf_bridge->use) > 1) { | |
204 | struct nf_bridge_info *tmp = nf_bridge_alloc(skb); | |
205 | ||
206 | if (tmp) { | |
207 | memcpy(tmp, nf_bridge, sizeof(struct nf_bridge_info)); | |
208 | atomic_set(&tmp->use, 1); | |
2dc2f207 | 209 | } |
4c3a76ab | 210 | nf_bridge_put(nf_bridge); |
2dc2f207 PM |
211 | nf_bridge = tmp; |
212 | } | |
213 | return nf_bridge; | |
214 | } | |
215 | ||
fc38582d PM |
216 | static inline void nf_bridge_push_encap_header(struct sk_buff *skb) |
217 | { | |
218 | unsigned int len = nf_bridge_encap_header_len(skb); | |
219 | ||
220 | skb_push(skb, len); | |
221 | skb->network_header -= len; | |
222 | } | |
223 | ||
224 | static inline void nf_bridge_pull_encap_header(struct sk_buff *skb) | |
fdeabdef | 225 | { |
fc38582d PM |
226 | unsigned int len = nf_bridge_encap_header_len(skb); |
227 | ||
228 | skb_pull(skb, len); | |
229 | skb->network_header += len; | |
230 | } | |
fdeabdef | 231 | |
fc38582d PM |
232 | static inline void nf_bridge_pull_encap_header_rcsum(struct sk_buff *skb) |
233 | { | |
234 | unsigned int len = nf_bridge_encap_header_len(skb); | |
235 | ||
236 | skb_pull_rcsum(skb, len); | |
237 | skb->network_header += len; | |
238 | } | |
239 | ||
240 | static inline void nf_bridge_save_header(struct sk_buff *skb) | |
241 | { | |
242 | int header_size = ETH_HLEN + nf_bridge_encap_header_len(skb); | |
fdeabdef | 243 | |
d626f62b ACM |
244 | skb_copy_from_linear_data_offset(skb, -header_size, |
245 | skb->nf_bridge->data, header_size); | |
fdeabdef SH |
246 | } |
247 | ||
e179e632 BDS |
248 | static inline void nf_bridge_update_protocol(struct sk_buff *skb) |
249 | { | |
250 | if (skb->nf_bridge->mask & BRNF_8021Q) | |
251 | skb->protocol = htons(ETH_P_8021Q); | |
252 | else if (skb->nf_bridge->mask & BRNF_PPPoE) | |
253 | skb->protocol = htons(ETH_P_PPP_SES); | |
254 | } | |
255 | ||
462fb2af BD |
256 | /* When handing a packet over to the IP layer |
257 | * check whether we have a skb that is in the | |
258 | * expected format | |
259 | */ | |
260 | ||
d0280232 | 261 | static int br_parse_ip_options(struct sk_buff *skb) |
462fb2af BD |
262 | { |
263 | struct ip_options *opt; | |
b71d1d42 | 264 | const struct iphdr *iph; |
462fb2af BD |
265 | struct net_device *dev = skb->dev; |
266 | u32 len; | |
267 | ||
6caab7b0 SB |
268 | if (!pskb_may_pull(skb, sizeof(struct iphdr))) |
269 | goto inhdr_error; | |
270 | ||
462fb2af BD |
271 | iph = ip_hdr(skb); |
272 | opt = &(IPCB(skb)->opt); | |
273 | ||
274 | /* Basic sanity checks */ | |
275 | if (iph->ihl < 5 || iph->version != 4) | |
276 | goto inhdr_error; | |
277 | ||
278 | if (!pskb_may_pull(skb, iph->ihl*4)) | |
279 | goto inhdr_error; | |
280 | ||
281 | iph = ip_hdr(skb); | |
282 | if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl))) | |
283 | goto inhdr_error; | |
284 | ||
285 | len = ntohs(iph->tot_len); | |
286 | if (skb->len < len) { | |
287 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INTRUNCATEDPKTS); | |
288 | goto drop; | |
289 | } else if (len < (iph->ihl*4)) | |
290 | goto inhdr_error; | |
291 | ||
292 | if (pskb_trim_rcsum(skb, len)) { | |
293 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS); | |
294 | goto drop; | |
295 | } | |
296 | ||
f8e9881c ED |
297 | memset(IPCB(skb), 0, sizeof(struct inet_skb_parm)); |
298 | if (iph->ihl == 5) | |
462fb2af | 299 | return 0; |
462fb2af BD |
300 | |
301 | opt->optlen = iph->ihl*4 - sizeof(struct iphdr); | |
302 | if (ip_options_compile(dev_net(dev), opt, skb)) | |
303 | goto inhdr_error; | |
304 | ||
305 | /* Check correct handling of SRR option */ | |
306 | if (unlikely(opt->srr)) { | |
307 | struct in_device *in_dev = __in_dev_get_rcu(dev); | |
308 | if (in_dev && !IN_DEV_SOURCE_ROUTE(in_dev)) | |
309 | goto drop; | |
310 | ||
311 | if (ip_options_rcv_srr(skb)) | |
312 | goto drop; | |
313 | } | |
314 | ||
315 | return 0; | |
316 | ||
317 | inhdr_error: | |
318 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS); | |
319 | drop: | |
320 | return -1; | |
321 | } | |
322 | ||
e179e632 BDS |
323 | /* Fill in the header for fragmented IP packets handled by |
324 | * the IPv4 connection tracking code. | |
07317621 SH |
325 | */ |
326 | int nf_bridge_copy_header(struct sk_buff *skb) | |
327 | { | |
328 | int err; | |
e179e632 | 329 | unsigned int header_size; |
07317621 | 330 | |
e179e632 BDS |
331 | nf_bridge_update_protocol(skb); |
332 | header_size = ETH_HLEN + nf_bridge_encap_header_len(skb); | |
d9cc2048 | 333 | err = skb_cow_head(skb, header_size); |
07317621 SH |
334 | if (err) |
335 | return err; | |
336 | ||
27d7ff46 ACM |
337 | skb_copy_to_linear_data_offset(skb, -header_size, |
338 | skb->nf_bridge->data, header_size); | |
fc38582d | 339 | __skb_push(skb, nf_bridge_encap_header_len(skb)); |
07317621 SH |
340 | return 0; |
341 | } | |
342 | ||
1da177e4 LT |
343 | /* PF_BRIDGE/PRE_ROUTING *********************************************/ |
344 | /* Undo the changes made for ip6tables PREROUTING and continue the | |
345 | * bridge PRE_ROUTING hook. */ | |
346 | static int br_nf_pre_routing_finish_ipv6(struct sk_buff *skb) | |
347 | { | |
348 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
511c3f92 | 349 | struct rtable *rt; |
1da177e4 | 350 | |
1da177e4 LT |
351 | if (nf_bridge->mask & BRNF_PKT_TYPE) { |
352 | skb->pkt_type = PACKET_OTHERHOST; | |
353 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
354 | } | |
355 | nf_bridge->mask ^= BRNF_NF_BRIDGE_PREROUTING; | |
356 | ||
511c3f92 ED |
357 | rt = bridge_parent_rtable(nf_bridge->physindev); |
358 | if (!rt) { | |
4adf0af6 SW |
359 | kfree_skb(skb); |
360 | return 0; | |
361 | } | |
f9181f4f | 362 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
363 | |
364 | skb->dev = nf_bridge->physindev; | |
e179e632 | 365 | nf_bridge_update_protocol(skb); |
fc38582d | 366 | nf_bridge_push_encap_header(skb); |
713aefa3 | 367 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
368 | br_handle_frame_finish, 1); |
369 | ||
370 | return 0; | |
371 | } | |
372 | ||
e179e632 BDS |
373 | /* Obtain the correct destination MAC address, while preserving the original |
374 | * source MAC address. If we already know this address, we just copy it. If we | |
375 | * don't, we use the neighbour framework to find out. In both cases, we make | |
376 | * sure that br_handle_frame_finish() is called afterwards. | |
377 | */ | |
378 | static int br_nf_pre_routing_finish_bridge(struct sk_buff *skb) | |
379 | { | |
380 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
f6b72b62 | 381 | struct neighbour *neigh; |
e179e632 BDS |
382 | struct dst_entry *dst; |
383 | ||
384 | skb->dev = bridge_parent(skb->dev); | |
385 | if (!skb->dev) | |
386 | goto free_skb; | |
387 | dst = skb_dst(skb); | |
f9d75166 DM |
388 | neigh = dst_neigh_lookup_skb(dst, skb); |
389 | if (neigh) { | |
390 | int ret; | |
391 | ||
392 | if (neigh->hh.hh_len) { | |
393 | neigh_hh_bridge(&neigh->hh, skb); | |
394 | skb->dev = nf_bridge->physindev; | |
395 | ret = br_handle_frame_finish(skb); | |
396 | } else { | |
397 | /* the neighbour function below overwrites the complete | |
398 | * MAC header, so we save the Ethernet source address and | |
399 | * protocol number. | |
400 | */ | |
401 | skb_copy_from_linear_data_offset(skb, | |
402 | -(ETH_HLEN-ETH_ALEN), | |
403 | skb->nf_bridge->data, | |
404 | ETH_HLEN-ETH_ALEN); | |
405 | /* tell br_dev_xmit to continue with forwarding */ | |
406 | nf_bridge->mask |= BRNF_BRIDGED_DNAT; | |
407 | ret = neigh->output(neigh, skb); | |
408 | } | |
409 | neigh_release(neigh); | |
410 | return ret; | |
e179e632 BDS |
411 | } |
412 | free_skb: | |
413 | kfree_skb(skb); | |
414 | return 0; | |
415 | } | |
416 | ||
1da177e4 | 417 | /* This requires some explaining. If DNAT has taken place, |
ea2d9b41 | 418 | * we will need to fix up the destination Ethernet address. |
1da177e4 LT |
419 | * |
420 | * There are two cases to consider: | |
421 | * 1. The packet was DNAT'ed to a device in the same bridge | |
422 | * port group as it was received on. We can still bridge | |
423 | * the packet. | |
424 | * 2. The packet was DNAT'ed to a different device, either | |
425 | * a non-bridged device or another bridge port group. | |
426 | * The packet will need to be routed. | |
427 | * | |
428 | * The correct way of distinguishing between these two cases is to | |
429 | * call ip_route_input() and to look at skb->dst->dev, which is | |
430 | * changed to the destination device if ip_route_input() succeeds. | |
431 | * | |
ea2d9b41 | 432 | * Let's first consider the case that ip_route_input() succeeds: |
1da177e4 | 433 | * |
ea2d9b41 BDS |
434 | * If the output device equals the logical bridge device the packet |
435 | * came in on, we can consider this bridging. The corresponding MAC | |
436 | * address will be obtained in br_nf_pre_routing_finish_bridge. | |
1da177e4 LT |
437 | * Otherwise, the packet is considered to be routed and we just |
438 | * change the destination MAC address so that the packet will | |
f216f082 BDS |
439 | * later be passed up to the IP stack to be routed. For a redirected |
440 | * packet, ip_route_input() will give back the localhost as output device, | |
441 | * which differs from the bridge device. | |
1da177e4 | 442 | * |
ea2d9b41 | 443 | * Let's now consider the case that ip_route_input() fails: |
1da177e4 | 444 | * |
f216f082 BDS |
445 | * This can be because the destination address is martian, in which case |
446 | * the packet will be dropped. | |
ea2d9b41 BDS |
447 | * If IP forwarding is disabled, ip_route_input() will fail, while |
448 | * ip_route_output_key() can return success. The source | |
449 | * address for ip_route_output_key() is set to zero, so ip_route_output_key() | |
1da177e4 | 450 | * thinks we're handling a locally generated packet and won't care |
ea2d9b41 BDS |
451 | * if IP forwarding is enabled. If the output device equals the logical bridge |
452 | * device, we proceed as if ip_route_input() succeeded. If it differs from the | |
453 | * logical bridge port or if ip_route_output_key() fails we drop the packet. | |
454 | */ | |
1da177e4 LT |
455 | static int br_nf_pre_routing_finish(struct sk_buff *skb) |
456 | { | |
457 | struct net_device *dev = skb->dev; | |
eddc9ec5 | 458 | struct iphdr *iph = ip_hdr(skb); |
1da177e4 | 459 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; |
511c3f92 | 460 | struct rtable *rt; |
f216f082 | 461 | int err; |
1da177e4 | 462 | |
1da177e4 LT |
463 | if (nf_bridge->mask & BRNF_PKT_TYPE) { |
464 | skb->pkt_type = PACKET_OTHERHOST; | |
465 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
466 | } | |
467 | nf_bridge->mask ^= BRNF_NF_BRIDGE_PREROUTING; | |
1da177e4 | 468 | if (dnat_took_place(skb)) { |
f216f082 | 469 | if ((err = ip_route_input(skb, iph->daddr, iph->saddr, iph->tos, dev))) { |
f3abc9b9 | 470 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
f216f082 BDS |
471 | |
472 | /* If err equals -EHOSTUNREACH the error is due to a | |
473 | * martian destination or due to the fact that | |
474 | * forwarding is disabled. For most martian packets, | |
475 | * ip_route_output_key() will fail. It won't fail for 2 types of | |
476 | * martian destinations: loopback destinations and destination | |
477 | * 0.0.0.0. In both cases the packet will be dropped because the | |
478 | * destination is the loopback device and not the bridge. */ | |
479 | if (err != -EHOSTUNREACH || !in_dev || IN_DEV_FORWARD(in_dev)) | |
480 | goto free_skb; | |
1da177e4 | 481 | |
78fbfd8a DM |
482 | rt = ip_route_output(dev_net(dev), iph->daddr, 0, |
483 | RT_TOS(iph->tos), 0); | |
b23dd4fe | 484 | if (!IS_ERR(rt)) { |
1c011bed | 485 | /* - Bridged-and-DNAT'ed traffic doesn't |
f216f082 | 486 | * require ip_forwarding. */ |
b23dd4fe DM |
487 | if (rt->dst.dev == dev) { |
488 | skb_dst_set(skb, &rt->dst); | |
1da177e4 LT |
489 | goto bridged_dnat; |
490 | } | |
b23dd4fe | 491 | ip_rt_put(rt); |
1da177e4 | 492 | } |
f216f082 | 493 | free_skb: |
1da177e4 LT |
494 | kfree_skb(skb); |
495 | return 0; | |
496 | } else { | |
adf30907 | 497 | if (skb_dst(skb)->dev == dev) { |
1da177e4 | 498 | bridged_dnat: |
1da177e4 | 499 | skb->dev = nf_bridge->physindev; |
e179e632 | 500 | nf_bridge_update_protocol(skb); |
fc38582d | 501 | nf_bridge_push_encap_header(skb); |
713aefa3 JE |
502 | NF_HOOK_THRESH(NFPROTO_BRIDGE, |
503 | NF_BR_PRE_ROUTING, | |
1da177e4 LT |
504 | skb, skb->dev, NULL, |
505 | br_nf_pre_routing_finish_bridge, | |
506 | 1); | |
507 | return 0; | |
508 | } | |
789bc3e5 | 509 | memcpy(eth_hdr(skb)->h_dest, dev->dev_addr, ETH_ALEN); |
1da177e4 LT |
510 | skb->pkt_type = PACKET_HOST; |
511 | } | |
512 | } else { | |
511c3f92 ED |
513 | rt = bridge_parent_rtable(nf_bridge->physindev); |
514 | if (!rt) { | |
4adf0af6 SW |
515 | kfree_skb(skb); |
516 | return 0; | |
517 | } | |
f9181f4f | 518 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
519 | } |
520 | ||
521 | skb->dev = nf_bridge->physindev; | |
e179e632 | 522 | nf_bridge_update_protocol(skb); |
fc38582d | 523 | nf_bridge_push_encap_header(skb); |
713aefa3 | 524 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
525 | br_handle_frame_finish, 1); |
526 | ||
527 | return 0; | |
528 | } | |
529 | ||
4981682c PNA |
530 | static struct net_device *brnf_get_logical_dev(struct sk_buff *skb, const struct net_device *dev) |
531 | { | |
532 | struct net_device *vlan, *br; | |
533 | ||
534 | br = bridge_parent(dev); | |
535 | if (brnf_pass_vlan_indev == 0 || !vlan_tx_tag_present(skb)) | |
536 | return br; | |
537 | ||
86a9bad3 | 538 | vlan = __vlan_find_dev_deep(br, skb->vlan_proto, |
1fd9b1fc | 539 | vlan_tx_tag_get(skb) & VLAN_VID_MASK); |
4981682c PNA |
540 | |
541 | return vlan ? vlan : br; | |
542 | } | |
543 | ||
1da177e4 | 544 | /* Some common code for IPv4/IPv6 */ |
5dce971a | 545 | static struct net_device *setup_pre_routing(struct sk_buff *skb) |
1da177e4 LT |
546 | { |
547 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
548 | ||
549 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
550 | skb->pkt_type = PACKET_HOST; | |
551 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
552 | } | |
553 | ||
554 | nf_bridge->mask |= BRNF_NF_BRIDGE_PREROUTING; | |
555 | nf_bridge->physindev = skb->dev; | |
4981682c | 556 | skb->dev = brnf_get_logical_dev(skb, skb->dev); |
e179e632 BDS |
557 | if (skb->protocol == htons(ETH_P_8021Q)) |
558 | nf_bridge->mask |= BRNF_8021Q; | |
559 | else if (skb->protocol == htons(ETH_P_PPP_SES)) | |
560 | nf_bridge->mask |= BRNF_PPPoE; | |
5dce971a SH |
561 | |
562 | return skb->dev; | |
1da177e4 LT |
563 | } |
564 | ||
565 | /* We only check the length. A bridge shouldn't do any hop-by-hop stuff anyway */ | |
566 | static int check_hbh_len(struct sk_buff *skb) | |
567 | { | |
0660e03f | 568 | unsigned char *raw = (u8 *)(ipv6_hdr(skb) + 1); |
1da177e4 | 569 | u32 pkt_len; |
d56f90a7 ACM |
570 | const unsigned char *nh = skb_network_header(skb); |
571 | int off = raw - nh; | |
789bc3e5 | 572 | int len = (raw[1] + 1) << 3; |
1da177e4 LT |
573 | |
574 | if ((raw + len) - skb->data > skb_headlen(skb)) | |
575 | goto bad; | |
576 | ||
577 | off += 2; | |
578 | len -= 2; | |
579 | ||
580 | while (len > 0) { | |
d56f90a7 | 581 | int optlen = nh[off + 1] + 2; |
1da177e4 | 582 | |
d56f90a7 | 583 | switch (nh[off]) { |
1de5a71c | 584 | case IPV6_TLV_PAD1: |
1da177e4 LT |
585 | optlen = 1; |
586 | break; | |
587 | ||
588 | case IPV6_TLV_PADN: | |
589 | break; | |
590 | ||
591 | case IPV6_TLV_JUMBO: | |
d56f90a7 | 592 | if (nh[off + 1] != 4 || (off & 3) != 2) |
1da177e4 | 593 | goto bad; |
d56f90a7 | 594 | pkt_len = ntohl(*(__be32 *) (nh + off + 2)); |
b0366486 | 595 | if (pkt_len <= IPV6_MAXPLEN || |
0660e03f | 596 | ipv6_hdr(skb)->payload_len) |
b0366486 | 597 | goto bad; |
1da177e4 LT |
598 | if (pkt_len > skb->len - sizeof(struct ipv6hdr)) |
599 | goto bad; | |
b0366486 | 600 | if (pskb_trim_rcsum(skb, |
789bc3e5 | 601 | pkt_len + sizeof(struct ipv6hdr))) |
b0366486 | 602 | goto bad; |
d56f90a7 | 603 | nh = skb_network_header(skb); |
1da177e4 LT |
604 | break; |
605 | default: | |
606 | if (optlen > len) | |
607 | goto bad; | |
608 | break; | |
609 | } | |
610 | off += optlen; | |
611 | len -= optlen; | |
612 | } | |
613 | if (len == 0) | |
614 | return 0; | |
615 | bad: | |
616 | return -1; | |
617 | ||
618 | } | |
619 | ||
620 | /* Replicate the checks that IPv6 does on packet reception and pass the packet | |
621 | * to ip6tables, which doesn't support NAT, so things are fairly simple. */ | |
622 | static unsigned int br_nf_pre_routing_ipv6(unsigned int hook, | |
789bc3e5 SH |
623 | struct sk_buff *skb, |
624 | const struct net_device *in, | |
625 | const struct net_device *out, | |
626 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 627 | { |
b71d1d42 | 628 | const struct ipv6hdr *hdr; |
1da177e4 | 629 | u32 pkt_len; |
1da177e4 LT |
630 | |
631 | if (skb->len < sizeof(struct ipv6hdr)) | |
deef4b52 | 632 | return NF_DROP; |
1da177e4 LT |
633 | |
634 | if (!pskb_may_pull(skb, sizeof(struct ipv6hdr))) | |
deef4b52 | 635 | return NF_DROP; |
1da177e4 | 636 | |
0660e03f | 637 | hdr = ipv6_hdr(skb); |
1da177e4 LT |
638 | |
639 | if (hdr->version != 6) | |
deef4b52 | 640 | return NF_DROP; |
1da177e4 LT |
641 | |
642 | pkt_len = ntohs(hdr->payload_len); | |
643 | ||
644 | if (pkt_len || hdr->nexthdr != NEXTHDR_HOP) { | |
645 | if (pkt_len + sizeof(struct ipv6hdr) > skb->len) | |
deef4b52 | 646 | return NF_DROP; |
b38dfee3 | 647 | if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr))) |
deef4b52 | 648 | return NF_DROP; |
1da177e4 LT |
649 | } |
650 | if (hdr->nexthdr == NEXTHDR_HOP && check_hbh_len(skb)) | |
deef4b52 | 651 | return NF_DROP; |
1da177e4 | 652 | |
789bc3e5 | 653 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 654 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 655 | return NF_DROP; |
5dce971a SH |
656 | if (!setup_pre_routing(skb)) |
657 | return NF_DROP; | |
1da177e4 | 658 | |
e179e632 | 659 | skb->protocol = htons(ETH_P_IPV6); |
713aefa3 | 660 | NF_HOOK(NFPROTO_IPV6, NF_INET_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
661 | br_nf_pre_routing_finish_ipv6); |
662 | ||
663 | return NF_STOLEN; | |
1da177e4 LT |
664 | } |
665 | ||
666 | /* Direct IPv6 traffic to br_nf_pre_routing_ipv6. | |
667 | * Replicate the checks that IPv4 does on packet reception. | |
668 | * Set skb->dev to the bridge device (i.e. parent of the | |
669 | * receiving device) to make netfilter happy, the REDIRECT | |
670 | * target in particular. Save the original destination IP | |
671 | * address to be able to detect DNAT afterwards. */ | |
3db05fea | 672 | static unsigned int br_nf_pre_routing(unsigned int hook, struct sk_buff *skb, |
ee02b3a6 SH |
673 | const struct net_device *in, |
674 | const struct net_device *out, | |
675 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 676 | { |
4df53d8b PM |
677 | struct net_bridge_port *p; |
678 | struct net_bridge *br; | |
e7c243c9 EP |
679 | __u32 len = nf_bridge_encap_header_len(skb); |
680 | ||
e7c243c9 | 681 | if (unlikely(!pskb_may_pull(skb, len))) |
deef4b52 | 682 | return NF_DROP; |
1da177e4 | 683 | |
e490c1de | 684 | p = br_port_get_rcu(in); |
4df53d8b | 685 | if (p == NULL) |
deef4b52 | 686 | return NF_DROP; |
4df53d8b PM |
687 | br = p->br; |
688 | ||
739e4505 | 689 | if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) { |
4df53d8b | 690 | if (!brnf_call_ip6tables && !br->nf_call_ip6tables) |
1da177e4 | 691 | return NF_ACCEPT; |
4df53d8b | 692 | |
fc38582d | 693 | nf_bridge_pull_encap_header_rcsum(skb); |
1da177e4 LT |
694 | return br_nf_pre_routing_ipv6(hook, skb, in, out, okfn); |
695 | } | |
4df53d8b PM |
696 | |
697 | if (!brnf_call_iptables && !br->nf_call_iptables) | |
1da177e4 | 698 | return NF_ACCEPT; |
1da177e4 | 699 | |
739e4505 | 700 | if (!IS_IP(skb) && !IS_VLAN_IP(skb) && !IS_PPPOE_IP(skb)) |
1da177e4 LT |
701 | return NF_ACCEPT; |
702 | ||
fc38582d | 703 | nf_bridge_pull_encap_header_rcsum(skb); |
1da177e4 | 704 | |
462fb2af | 705 | if (br_parse_ip_options(skb)) |
deef4b52 | 706 | return NF_DROP; |
17762060 | 707 | |
789bc3e5 | 708 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 709 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 710 | return NF_DROP; |
5dce971a SH |
711 | if (!setup_pre_routing(skb)) |
712 | return NF_DROP; | |
1da177e4 | 713 | store_orig_dstaddr(skb); |
e179e632 | 714 | skb->protocol = htons(ETH_P_IP); |
1da177e4 | 715 | |
713aefa3 | 716 | NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
717 | br_nf_pre_routing_finish); |
718 | ||
719 | return NF_STOLEN; | |
1da177e4 LT |
720 | } |
721 | ||
722 | ||
723 | /* PF_BRIDGE/LOCAL_IN ************************************************/ | |
724 | /* The packet is locally destined, which requires a real | |
725 | * dst_entry, so detach the fake one. On the way up, the | |
726 | * packet would pass through PRE_ROUTING again (which already | |
727 | * took place when the packet entered the bridge), but we | |
728 | * register an IPv4 PRE_ROUTING 'sabotage' hook that will | |
729 | * prevent this from happening. */ | |
3db05fea | 730 | static unsigned int br_nf_local_in(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
731 | const struct net_device *in, |
732 | const struct net_device *out, | |
733 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 734 | { |
a881e963 | 735 | br_drop_fake_rtable(skb); |
1da177e4 LT |
736 | return NF_ACCEPT; |
737 | } | |
738 | ||
1da177e4 LT |
739 | /* PF_BRIDGE/FORWARD *************************************************/ |
740 | static int br_nf_forward_finish(struct sk_buff *skb) | |
741 | { | |
742 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
743 | struct net_device *in; | |
1da177e4 | 744 | |
739e4505 | 745 | if (!IS_ARP(skb) && !IS_VLAN_ARP(skb)) { |
1da177e4 LT |
746 | in = nf_bridge->physindev; |
747 | if (nf_bridge->mask & BRNF_PKT_TYPE) { | |
748 | skb->pkt_type = PACKET_OTHERHOST; | |
749 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
750 | } | |
e94c6743 | 751 | nf_bridge_update_protocol(skb); |
1da177e4 LT |
752 | } else { |
753 | in = *((struct net_device **)(skb->cb)); | |
754 | } | |
fc38582d | 755 | nf_bridge_push_encap_header(skb); |
e179e632 | 756 | |
713aefa3 | 757 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_FORWARD, skb, in, |
789bc3e5 | 758 | skb->dev, br_forward_finish, 1); |
1da177e4 LT |
759 | return 0; |
760 | } | |
761 | ||
739e4505 | 762 | |
1da177e4 LT |
763 | /* This is the 'purely bridged' case. For IP, we pass the packet to |
764 | * netfilter with indev and outdev set to the bridge device, | |
765 | * but we are still able to filter on the 'real' indev/outdev | |
766 | * because of the physdev module. For ARP, indev and outdev are the | |
767 | * bridge ports. */ | |
3db05fea | 768 | static unsigned int br_nf_forward_ip(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
769 | const struct net_device *in, |
770 | const struct net_device *out, | |
771 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 772 | { |
1da177e4 | 773 | struct nf_bridge_info *nf_bridge; |
5dce971a | 774 | struct net_device *parent; |
76108cea | 775 | u_int8_t pf; |
1da177e4 LT |
776 | |
777 | if (!skb->nf_bridge) | |
778 | return NF_ACCEPT; | |
779 | ||
2dc2f207 PM |
780 | /* Need exclusive nf_bridge_info since we might have multiple |
781 | * different physoutdevs. */ | |
782 | if (!nf_bridge_unshare(skb)) | |
783 | return NF_DROP; | |
784 | ||
5dce971a SH |
785 | parent = bridge_parent(out); |
786 | if (!parent) | |
787 | return NF_DROP; | |
788 | ||
739e4505 | 789 | if (IS_IP(skb) || IS_VLAN_IP(skb) || IS_PPPOE_IP(skb)) |
aa740f46 | 790 | pf = NFPROTO_IPV4; |
739e4505 | 791 | else if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) |
aa740f46 | 792 | pf = NFPROTO_IPV6; |
a2bd40ad HX |
793 | else |
794 | return NF_ACCEPT; | |
1da177e4 | 795 | |
3db05fea | 796 | nf_bridge_pull_encap_header(skb); |
1da177e4 | 797 | |
1da177e4 LT |
798 | nf_bridge = skb->nf_bridge; |
799 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
800 | skb->pkt_type = PACKET_HOST; | |
801 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
802 | } | |
803 | ||
aa740f46 | 804 | if (pf == NFPROTO_IPV4 && br_parse_ip_options(skb)) |
6b1e960f HX |
805 | return NF_DROP; |
806 | ||
1da177e4 LT |
807 | /* The physdev module checks on this */ |
808 | nf_bridge->mask |= BRNF_BRIDGED; | |
809 | nf_bridge->physoutdev = skb->dev; | |
aa740f46 | 810 | if (pf == NFPROTO_IPV4) |
e179e632 BDS |
811 | skb->protocol = htons(ETH_P_IP); |
812 | else | |
813 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 814 | |
4981682c | 815 | NF_HOOK(pf, NF_INET_FORWARD, skb, brnf_get_logical_dev(skb, in), parent, |
5dce971a | 816 | br_nf_forward_finish); |
1da177e4 LT |
817 | |
818 | return NF_STOLEN; | |
819 | } | |
820 | ||
3db05fea | 821 | static unsigned int br_nf_forward_arp(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
822 | const struct net_device *in, |
823 | const struct net_device *out, | |
824 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 825 | { |
4df53d8b PM |
826 | struct net_bridge_port *p; |
827 | struct net_bridge *br; | |
1da177e4 LT |
828 | struct net_device **d = (struct net_device **)(skb->cb); |
829 | ||
e490c1de | 830 | p = br_port_get_rcu(out); |
4df53d8b PM |
831 | if (p == NULL) |
832 | return NF_ACCEPT; | |
833 | br = p->br; | |
834 | ||
835 | if (!brnf_call_arptables && !br->nf_call_arptables) | |
1da177e4 | 836 | return NF_ACCEPT; |
1da177e4 | 837 | |
739e4505 | 838 | if (!IS_ARP(skb)) { |
8b42ec39 | 839 | if (!IS_VLAN_ARP(skb)) |
1da177e4 | 840 | return NF_ACCEPT; |
3db05fea | 841 | nf_bridge_pull_encap_header(skb); |
1da177e4 LT |
842 | } |
843 | ||
d0a92be0 | 844 | if (arp_hdr(skb)->ar_pln != 4) { |
fc38582d | 845 | if (IS_VLAN_ARP(skb)) |
3db05fea | 846 | nf_bridge_push_encap_header(skb); |
1da177e4 LT |
847 | return NF_ACCEPT; |
848 | } | |
849 | *d = (struct net_device *)in; | |
fdc9314c | 850 | NF_HOOK(NFPROTO_ARP, NF_ARP_FORWARD, skb, (struct net_device *)in, |
1da177e4 LT |
851 | (struct net_device *)out, br_nf_forward_finish); |
852 | ||
853 | return NF_STOLEN; | |
854 | } | |
855 | ||
e6373c4c | 856 | #if IS_ENABLED(CONFIG_NF_CONNTRACK_IPV4) |
2e2f7aef PM |
857 | static int br_nf_dev_queue_xmit(struct sk_buff *skb) |
858 | { | |
462fb2af BD |
859 | int ret; |
860 | ||
e179e632 | 861 | if (skb->nfct != NULL && skb->protocol == htons(ETH_P_IP) && |
6c79bf0f | 862 | skb->len + nf_bridge_mtu_reduction(skb) > skb->dev->mtu && |
87f94b4e | 863 | !skb_is_gso(skb)) { |
462fb2af BD |
864 | if (br_parse_ip_options(skb)) |
865 | /* Drop invalid packet */ | |
866 | return NF_DROP; | |
867 | ret = ip_fragment(skb, br_dev_queue_push_xmit); | |
87f94b4e | 868 | } else |
462fb2af BD |
869 | ret = br_dev_queue_push_xmit(skb); |
870 | ||
871 | return ret; | |
2e2f7aef | 872 | } |
c197facc | 873 | #else |
874 | static int br_nf_dev_queue_xmit(struct sk_buff *skb) | |
875 | { | |
876 | return br_dev_queue_push_xmit(skb); | |
877 | } | |
878 | #endif | |
1da177e4 LT |
879 | |
880 | /* PF_BRIDGE/POST_ROUTING ********************************************/ | |
3db05fea | 881 | static unsigned int br_nf_post_routing(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
882 | const struct net_device *in, |
883 | const struct net_device *out, | |
884 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 885 | { |
3db05fea | 886 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; |
1da177e4 | 887 | struct net_device *realoutdev = bridge_parent(skb->dev); |
76108cea | 888 | u_int8_t pf; |
1da177e4 | 889 | |
ea2d9b41 | 890 | if (!nf_bridge || !(nf_bridge->mask & BRNF_BRIDGED)) |
81d9ddae PM |
891 | return NF_ACCEPT; |
892 | ||
5dce971a SH |
893 | if (!realoutdev) |
894 | return NF_DROP; | |
895 | ||
739e4505 | 896 | if (IS_IP(skb) || IS_VLAN_IP(skb) || IS_PPPOE_IP(skb)) |
aa740f46 | 897 | pf = NFPROTO_IPV4; |
739e4505 | 898 | else if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) |
aa740f46 | 899 | pf = NFPROTO_IPV6; |
a2bd40ad HX |
900 | else |
901 | return NF_ACCEPT; | |
1da177e4 | 902 | |
1da177e4 LT |
903 | /* We assume any code from br_dev_queue_push_xmit onwards doesn't care |
904 | * about the value of skb->pkt_type. */ | |
905 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
906 | skb->pkt_type = PACKET_HOST; | |
907 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
908 | } | |
909 | ||
fc38582d | 910 | nf_bridge_pull_encap_header(skb); |
1da177e4 | 911 | nf_bridge_save_header(skb); |
aa740f46 | 912 | if (pf == NFPROTO_IPV4) |
e179e632 BDS |
913 | skb->protocol = htons(ETH_P_IP); |
914 | else | |
915 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 916 | |
6e23ae2a | 917 | NF_HOOK(pf, NF_INET_POST_ROUTING, skb, NULL, realoutdev, |
2e2f7aef | 918 | br_nf_dev_queue_xmit); |
1da177e4 LT |
919 | |
920 | return NF_STOLEN; | |
1da177e4 LT |
921 | } |
922 | ||
1da177e4 LT |
923 | /* IP/SABOTAGE *****************************************************/ |
924 | /* Don't hand locally destined packets to PF_INET(6)/PRE_ROUTING | |
925 | * for the second time. */ | |
3db05fea | 926 | static unsigned int ip_sabotage_in(unsigned int hook, struct sk_buff *skb, |
789bc3e5 SH |
927 | const struct net_device *in, |
928 | const struct net_device *out, | |
929 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 930 | { |
3db05fea HX |
931 | if (skb->nf_bridge && |
932 | !(skb->nf_bridge->mask & BRNF_NF_BRIDGE_PREROUTING)) { | |
1da177e4 LT |
933 | return NF_STOP; |
934 | } | |
935 | ||
936 | return NF_ACCEPT; | |
937 | } | |
938 | ||
ea2d9b41 BDS |
939 | /* For br_nf_post_routing, we need (prio = NF_BR_PRI_LAST), because |
940 | * br_dev_queue_push_xmit is called afterwards */ | |
1999414a | 941 | static struct nf_hook_ops br_nf_ops[] __read_mostly = { |
1490fd89 CG |
942 | { |
943 | .hook = br_nf_pre_routing, | |
944 | .owner = THIS_MODULE, | |
aa740f46 | 945 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
946 | .hooknum = NF_BR_PRE_ROUTING, |
947 | .priority = NF_BR_PRI_BRNF, | |
948 | }, | |
949 | { | |
950 | .hook = br_nf_local_in, | |
951 | .owner = THIS_MODULE, | |
aa740f46 | 952 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
953 | .hooknum = NF_BR_LOCAL_IN, |
954 | .priority = NF_BR_PRI_BRNF, | |
955 | }, | |
956 | { | |
957 | .hook = br_nf_forward_ip, | |
958 | .owner = THIS_MODULE, | |
aa740f46 | 959 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
960 | .hooknum = NF_BR_FORWARD, |
961 | .priority = NF_BR_PRI_BRNF - 1, | |
962 | }, | |
963 | { | |
964 | .hook = br_nf_forward_arp, | |
965 | .owner = THIS_MODULE, | |
aa740f46 | 966 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
967 | .hooknum = NF_BR_FORWARD, |
968 | .priority = NF_BR_PRI_BRNF, | |
969 | }, | |
1490fd89 CG |
970 | { |
971 | .hook = br_nf_post_routing, | |
972 | .owner = THIS_MODULE, | |
aa740f46 | 973 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
974 | .hooknum = NF_BR_POST_ROUTING, |
975 | .priority = NF_BR_PRI_LAST, | |
976 | }, | |
977 | { | |
978 | .hook = ip_sabotage_in, | |
979 | .owner = THIS_MODULE, | |
aa740f46 | 980 | .pf = NFPROTO_IPV4, |
1490fd89 CG |
981 | .hooknum = NF_INET_PRE_ROUTING, |
982 | .priority = NF_IP_PRI_FIRST, | |
983 | }, | |
984 | { | |
985 | .hook = ip_sabotage_in, | |
986 | .owner = THIS_MODULE, | |
aa740f46 | 987 | .pf = NFPROTO_IPV6, |
1490fd89 CG |
988 | .hooknum = NF_INET_PRE_ROUTING, |
989 | .priority = NF_IP6_PRI_FIRST, | |
990 | }, | |
1da177e4 LT |
991 | }; |
992 | ||
993 | #ifdef CONFIG_SYSCTL | |
994 | static | |
fe2c6338 | 995 | int brnf_sysctl_call_tables(struct ctl_table *ctl, int write, |
789bc3e5 | 996 | void __user * buffer, size_t * lenp, loff_t * ppos) |
1da177e4 LT |
997 | { |
998 | int ret; | |
999 | ||
8d65af78 | 1000 | ret = proc_dointvec(ctl, write, buffer, lenp, ppos); |
1da177e4 LT |
1001 | |
1002 | if (write && *(int *)(ctl->data)) | |
1003 | *(int *)(ctl->data) = 1; | |
1004 | return ret; | |
1005 | } | |
1006 | ||
fe2c6338 | 1007 | static struct ctl_table brnf_table[] = { |
1da177e4 | 1008 | { |
1da177e4 LT |
1009 | .procname = "bridge-nf-call-arptables", |
1010 | .data = &brnf_call_arptables, | |
1011 | .maxlen = sizeof(int), | |
1012 | .mode = 0644, | |
6d9f239a | 1013 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1014 | }, |
1015 | { | |
1da177e4 LT |
1016 | .procname = "bridge-nf-call-iptables", |
1017 | .data = &brnf_call_iptables, | |
1018 | .maxlen = sizeof(int), | |
1019 | .mode = 0644, | |
6d9f239a | 1020 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1021 | }, |
1022 | { | |
1da177e4 LT |
1023 | .procname = "bridge-nf-call-ip6tables", |
1024 | .data = &brnf_call_ip6tables, | |
1025 | .maxlen = sizeof(int), | |
1026 | .mode = 0644, | |
6d9f239a | 1027 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1028 | }, |
1029 | { | |
1da177e4 LT |
1030 | .procname = "bridge-nf-filter-vlan-tagged", |
1031 | .data = &brnf_filter_vlan_tagged, | |
1032 | .maxlen = sizeof(int), | |
1033 | .mode = 0644, | |
6d9f239a | 1034 | .proc_handler = brnf_sysctl_call_tables, |
516299d2 MM |
1035 | }, |
1036 | { | |
516299d2 MM |
1037 | .procname = "bridge-nf-filter-pppoe-tagged", |
1038 | .data = &brnf_filter_pppoe_tagged, | |
1039 | .maxlen = sizeof(int), | |
1040 | .mode = 0644, | |
6d9f239a | 1041 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 | 1042 | }, |
4981682c PNA |
1043 | { |
1044 | .procname = "bridge-nf-pass-vlan-input-dev", | |
1045 | .data = &brnf_pass_vlan_indev, | |
1046 | .maxlen = sizeof(int), | |
1047 | .mode = 0644, | |
1048 | .proc_handler = brnf_sysctl_call_tables, | |
1049 | }, | |
f8572d8f | 1050 | { } |
1da177e4 | 1051 | }; |
1da177e4 LT |
1052 | #endif |
1053 | ||
5eb87f45 | 1054 | int __init br_netfilter_init(void) |
1da177e4 | 1055 | { |
5eb87f45 | 1056 | int ret; |
1da177e4 | 1057 | |
fc66f95c | 1058 | ret = dst_entries_init(&fake_dst_ops); |
5eb87f45 | 1059 | if (ret < 0) |
1da177e4 | 1060 | return ret; |
fc66f95c ED |
1061 | |
1062 | ret = nf_register_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); | |
1063 | if (ret < 0) { | |
1064 | dst_entries_destroy(&fake_dst_ops); | |
1065 | return ret; | |
1066 | } | |
1da177e4 | 1067 | #ifdef CONFIG_SYSCTL |
ec8f23ce | 1068 | brnf_sysctl_header = register_net_sysctl(&init_net, "net/bridge", brnf_table); |
1da177e4 | 1069 | if (brnf_sysctl_header == NULL) { |
789bc3e5 SH |
1070 | printk(KERN_WARNING |
1071 | "br_netfilter: can't register to sysctl.\n"); | |
5eb87f45 | 1072 | nf_unregister_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
fc66f95c | 1073 | dst_entries_destroy(&fake_dst_ops); |
5eb87f45 | 1074 | return -ENOMEM; |
1da177e4 LT |
1075 | } |
1076 | #endif | |
1da177e4 | 1077 | printk(KERN_NOTICE "Bridge firewalling registered\n"); |
1da177e4 LT |
1078 | return 0; |
1079 | } | |
1080 | ||
1081 | void br_netfilter_fini(void) | |
1082 | { | |
5eb87f45 | 1083 | nf_unregister_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
1da177e4 | 1084 | #ifdef CONFIG_SYSCTL |
5dd3df10 | 1085 | unregister_net_sysctl_table(brnf_sysctl_header); |
1da177e4 | 1086 | #endif |
fc66f95c | 1087 | dst_entries_destroy(&fake_dst_ops); |
1da177e4 | 1088 | } |