]>
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 | 37 | #include <net/route.h> |
56768644 | 38 | #include <net/netfilter/br_netfilter.h> |
14c85021 | 39 | |
c055d5b0 FW |
40 | #if IS_ENABLED(CONFIG_NF_CONNTRACK) |
41 | #include <net/netfilter/nf_conntrack.h> | |
42 | #endif | |
43 | ||
1da177e4 | 44 | #include <asm/uaccess.h> |
1da177e4 LT |
45 | #include "br_private.h" |
46 | #ifdef CONFIG_SYSCTL | |
47 | #include <linux/sysctl.h> | |
48 | #endif | |
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 | 67 | #define IS_IP(skb) \ |
df8a39de | 68 | (!skb_vlan_tag_present(skb) && skb->protocol == htons(ETH_P_IP)) |
739e4505 FW |
69 | |
70 | #define IS_IPV6(skb) \ | |
df8a39de | 71 | (!skb_vlan_tag_present(skb) && skb->protocol == htons(ETH_P_IPV6)) |
739e4505 FW |
72 | |
73 | #define IS_ARP(skb) \ | |
df8a39de | 74 | (!skb_vlan_tag_present(skb) && skb->protocol == htons(ETH_P_ARP)) |
739e4505 | 75 | |
b6f99a21 | 76 | static inline __be16 vlan_proto(const struct sk_buff *skb) |
8b42ec39 | 77 | { |
df8a39de | 78 | if (skb_vlan_tag_present(skb)) |
13937911 JG |
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 | ||
e70deecb FW |
114 | /* largest possible L2 header, see br_nf_dev_queue_xmit() */ |
115 | #define NF_BRIDGE_MAX_MAC_HEADER_LENGTH (PPPOE_SES_HLEN + ETH_HLEN) | |
116 | ||
117 | #if IS_ENABLED(CONFIG_NF_DEFRAG_IPV4) | |
118 | struct brnf_frag_data { | |
119 | char mac[NF_BRIDGE_MAX_MAC_HEADER_LENGTH]; | |
120 | u8 encap_size; | |
121 | u8 size; | |
122 | }; | |
123 | ||
124 | static DEFINE_PER_CPU(struct brnf_frag_data, brnf_frag_data_storage); | |
125 | #endif | |
126 | ||
4adf0af6 SW |
127 | static inline struct rtable *bridge_parent_rtable(const struct net_device *dev) |
128 | { | |
b5ed54e9 | 129 | struct net_bridge_port *port; |
130 | ||
131 | port = br_port_get_rcu(dev); | |
132 | return port ? &port->br->fake_rtable : NULL; | |
4adf0af6 | 133 | } |
1da177e4 | 134 | |
5dce971a SH |
135 | static inline struct net_device *bridge_parent(const struct net_device *dev) |
136 | { | |
b5ed54e9 | 137 | struct net_bridge_port *port; |
5dce971a | 138 | |
b5ed54e9 | 139 | port = br_port_get_rcu(dev); |
140 | return port ? port->br->dev : NULL; | |
5dce971a | 141 | } |
1da177e4 | 142 | |
fdeabdef SH |
143 | static inline struct nf_bridge_info *nf_bridge_alloc(struct sk_buff *skb) |
144 | { | |
145 | skb->nf_bridge = kzalloc(sizeof(struct nf_bridge_info), GFP_ATOMIC); | |
146 | if (likely(skb->nf_bridge)) | |
147 | atomic_set(&(skb->nf_bridge->use), 1); | |
148 | ||
149 | return skb->nf_bridge; | |
150 | } | |
151 | ||
2dc2f207 PM |
152 | static inline struct nf_bridge_info *nf_bridge_unshare(struct sk_buff *skb) |
153 | { | |
154 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
155 | ||
156 | if (atomic_read(&nf_bridge->use) > 1) { | |
157 | struct nf_bridge_info *tmp = nf_bridge_alloc(skb); | |
158 | ||
159 | if (tmp) { | |
160 | memcpy(tmp, nf_bridge, sizeof(struct nf_bridge_info)); | |
161 | atomic_set(&tmp->use, 1); | |
2dc2f207 | 162 | } |
4c3a76ab | 163 | nf_bridge_put(nf_bridge); |
2dc2f207 PM |
164 | nf_bridge = tmp; |
165 | } | |
166 | return nf_bridge; | |
167 | } | |
168 | ||
8d045163 FW |
169 | static unsigned int nf_bridge_encap_header_len(const struct sk_buff *skb) |
170 | { | |
171 | switch (skb->protocol) { | |
172 | case __cpu_to_be16(ETH_P_8021Q): | |
173 | return VLAN_HLEN; | |
174 | case __cpu_to_be16(ETH_P_PPP_SES): | |
175 | return PPPOE_SES_HLEN; | |
176 | default: | |
177 | return 0; | |
178 | } | |
179 | } | |
180 | ||
fc38582d PM |
181 | static inline void nf_bridge_push_encap_header(struct sk_buff *skb) |
182 | { | |
183 | unsigned int len = nf_bridge_encap_header_len(skb); | |
184 | ||
185 | skb_push(skb, len); | |
186 | skb->network_header -= len; | |
187 | } | |
188 | ||
189 | static inline void nf_bridge_pull_encap_header(struct sk_buff *skb) | |
fdeabdef | 190 | { |
fc38582d PM |
191 | unsigned int len = nf_bridge_encap_header_len(skb); |
192 | ||
193 | skb_pull(skb, len); | |
194 | skb->network_header += len; | |
195 | } | |
fdeabdef | 196 | |
fc38582d PM |
197 | static inline void nf_bridge_pull_encap_header_rcsum(struct sk_buff *skb) |
198 | { | |
199 | unsigned int len = nf_bridge_encap_header_len(skb); | |
200 | ||
201 | skb_pull_rcsum(skb, len); | |
202 | skb->network_header += len; | |
203 | } | |
204 | ||
462fb2af BD |
205 | /* When handing a packet over to the IP layer |
206 | * check whether we have a skb that is in the | |
207 | * expected format | |
208 | */ | |
209 | ||
d0280232 | 210 | static int br_parse_ip_options(struct sk_buff *skb) |
462fb2af | 211 | { |
b71d1d42 | 212 | const struct iphdr *iph; |
462fb2af BD |
213 | struct net_device *dev = skb->dev; |
214 | u32 len; | |
215 | ||
6caab7b0 SB |
216 | if (!pskb_may_pull(skb, sizeof(struct iphdr))) |
217 | goto inhdr_error; | |
218 | ||
462fb2af | 219 | iph = ip_hdr(skb); |
462fb2af BD |
220 | |
221 | /* Basic sanity checks */ | |
222 | if (iph->ihl < 5 || iph->version != 4) | |
223 | goto inhdr_error; | |
224 | ||
225 | if (!pskb_may_pull(skb, iph->ihl*4)) | |
226 | goto inhdr_error; | |
227 | ||
228 | iph = ip_hdr(skb); | |
229 | if (unlikely(ip_fast_csum((u8 *)iph, iph->ihl))) | |
230 | goto inhdr_error; | |
231 | ||
232 | len = ntohs(iph->tot_len); | |
233 | if (skb->len < len) { | |
234 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INTRUNCATEDPKTS); | |
235 | goto drop; | |
236 | } else if (len < (iph->ihl*4)) | |
237 | goto inhdr_error; | |
238 | ||
239 | if (pskb_trim_rcsum(skb, len)) { | |
240 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INDISCARDS); | |
241 | goto drop; | |
242 | } | |
243 | ||
f8e9881c | 244 | memset(IPCB(skb), 0, sizeof(struct inet_skb_parm)); |
7677e868 HX |
245 | /* We should really parse IP options here but until |
246 | * somebody who actually uses IP options complains to | |
247 | * us we'll just silently ignore the options because | |
248 | * we're lazy! | |
249 | */ | |
462fb2af BD |
250 | return 0; |
251 | ||
252 | inhdr_error: | |
253 | IP_INC_STATS_BH(dev_net(dev), IPSTATS_MIB_INHDRERRORS); | |
254 | drop: | |
255 | return -1; | |
256 | } | |
257 | ||
4a9d2f20 FW |
258 | static void nf_bridge_update_protocol(struct sk_buff *skb) |
259 | { | |
260 | if (skb->nf_bridge->mask & BRNF_8021Q) | |
261 | skb->protocol = htons(ETH_P_8021Q); | |
262 | else if (skb->nf_bridge->mask & BRNF_PPPoE) | |
263 | skb->protocol = htons(ETH_P_PPP_SES); | |
264 | } | |
265 | ||
1da177e4 LT |
266 | /* PF_BRIDGE/PRE_ROUTING *********************************************/ |
267 | /* Undo the changes made for ip6tables PREROUTING and continue the | |
268 | * bridge PRE_ROUTING hook. */ | |
269 | static int br_nf_pre_routing_finish_ipv6(struct sk_buff *skb) | |
270 | { | |
271 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
511c3f92 | 272 | struct rtable *rt; |
1da177e4 | 273 | |
1da177e4 LT |
274 | if (nf_bridge->mask & BRNF_PKT_TYPE) { |
275 | skb->pkt_type = PACKET_OTHERHOST; | |
276 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
277 | } | |
278 | nf_bridge->mask ^= BRNF_NF_BRIDGE_PREROUTING; | |
279 | ||
511c3f92 ED |
280 | rt = bridge_parent_rtable(nf_bridge->physindev); |
281 | if (!rt) { | |
4adf0af6 SW |
282 | kfree_skb(skb); |
283 | return 0; | |
284 | } | |
f9181f4f | 285 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
286 | |
287 | skb->dev = nf_bridge->physindev; | |
e179e632 | 288 | nf_bridge_update_protocol(skb); |
fc38582d | 289 | nf_bridge_push_encap_header(skb); |
713aefa3 | 290 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
291 | br_handle_frame_finish, 1); |
292 | ||
293 | return 0; | |
294 | } | |
295 | ||
e179e632 BDS |
296 | /* Obtain the correct destination MAC address, while preserving the original |
297 | * source MAC address. If we already know this address, we just copy it. If we | |
298 | * don't, we use the neighbour framework to find out. In both cases, we make | |
299 | * sure that br_handle_frame_finish() is called afterwards. | |
300 | */ | |
301 | static int br_nf_pre_routing_finish_bridge(struct sk_buff *skb) | |
302 | { | |
303 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
f6b72b62 | 304 | struct neighbour *neigh; |
e179e632 BDS |
305 | struct dst_entry *dst; |
306 | ||
307 | skb->dev = bridge_parent(skb->dev); | |
308 | if (!skb->dev) | |
309 | goto free_skb; | |
310 | dst = skb_dst(skb); | |
f9d75166 DM |
311 | neigh = dst_neigh_lookup_skb(dst, skb); |
312 | if (neigh) { | |
313 | int ret; | |
314 | ||
315 | if (neigh->hh.hh_len) { | |
316 | neigh_hh_bridge(&neigh->hh, skb); | |
317 | skb->dev = nf_bridge->physindev; | |
318 | ret = br_handle_frame_finish(skb); | |
319 | } else { | |
320 | /* the neighbour function below overwrites the complete | |
321 | * MAC header, so we save the Ethernet source address and | |
322 | * protocol number. | |
323 | */ | |
324 | skb_copy_from_linear_data_offset(skb, | |
325 | -(ETH_HLEN-ETH_ALEN), | |
e70deecb | 326 | nf_bridge->neigh_header, |
f9d75166 DM |
327 | ETH_HLEN-ETH_ALEN); |
328 | /* tell br_dev_xmit to continue with forwarding */ | |
329 | nf_bridge->mask |= BRNF_BRIDGED_DNAT; | |
93fdd47e | 330 | /* FIXME Need to refragment */ |
f9d75166 DM |
331 | ret = neigh->output(neigh, skb); |
332 | } | |
333 | neigh_release(neigh); | |
334 | return ret; | |
e179e632 BDS |
335 | } |
336 | free_skb: | |
337 | kfree_skb(skb); | |
338 | return 0; | |
339 | } | |
340 | ||
c055d5b0 FW |
341 | static bool dnat_took_place(const struct sk_buff *skb) |
342 | { | |
343 | #if IS_ENABLED(CONFIG_NF_CONNTRACK) | |
344 | enum ip_conntrack_info ctinfo; | |
345 | struct nf_conn *ct; | |
346 | ||
347 | ct = nf_ct_get(skb, &ctinfo); | |
348 | if (!ct || nf_ct_is_untracked(ct)) | |
349 | return false; | |
350 | ||
351 | return test_bit(IPS_DST_NAT_BIT, &ct->status); | |
352 | #else | |
353 | return false; | |
354 | #endif | |
355 | } | |
356 | ||
1da177e4 | 357 | /* This requires some explaining. If DNAT has taken place, |
ea2d9b41 | 358 | * we will need to fix up the destination Ethernet address. |
1da177e4 LT |
359 | * |
360 | * There are two cases to consider: | |
361 | * 1. The packet was DNAT'ed to a device in the same bridge | |
362 | * port group as it was received on. We can still bridge | |
363 | * the packet. | |
364 | * 2. The packet was DNAT'ed to a different device, either | |
365 | * a non-bridged device or another bridge port group. | |
366 | * The packet will need to be routed. | |
367 | * | |
368 | * The correct way of distinguishing between these two cases is to | |
369 | * call ip_route_input() and to look at skb->dst->dev, which is | |
370 | * changed to the destination device if ip_route_input() succeeds. | |
371 | * | |
ea2d9b41 | 372 | * Let's first consider the case that ip_route_input() succeeds: |
1da177e4 | 373 | * |
ea2d9b41 BDS |
374 | * If the output device equals the logical bridge device the packet |
375 | * came in on, we can consider this bridging. The corresponding MAC | |
376 | * address will be obtained in br_nf_pre_routing_finish_bridge. | |
1da177e4 LT |
377 | * Otherwise, the packet is considered to be routed and we just |
378 | * change the destination MAC address so that the packet will | |
f216f082 BDS |
379 | * later be passed up to the IP stack to be routed. For a redirected |
380 | * packet, ip_route_input() will give back the localhost as output device, | |
381 | * which differs from the bridge device. | |
1da177e4 | 382 | * |
ea2d9b41 | 383 | * Let's now consider the case that ip_route_input() fails: |
1da177e4 | 384 | * |
f216f082 BDS |
385 | * This can be because the destination address is martian, in which case |
386 | * the packet will be dropped. | |
ea2d9b41 BDS |
387 | * If IP forwarding is disabled, ip_route_input() will fail, while |
388 | * ip_route_output_key() can return success. The source | |
389 | * address for ip_route_output_key() is set to zero, so ip_route_output_key() | |
1da177e4 | 390 | * thinks we're handling a locally generated packet and won't care |
ea2d9b41 BDS |
391 | * if IP forwarding is enabled. If the output device equals the logical bridge |
392 | * device, we proceed as if ip_route_input() succeeded. If it differs from the | |
393 | * logical bridge port or if ip_route_output_key() fails we drop the packet. | |
394 | */ | |
1da177e4 LT |
395 | static int br_nf_pre_routing_finish(struct sk_buff *skb) |
396 | { | |
397 | struct net_device *dev = skb->dev; | |
eddc9ec5 | 398 | struct iphdr *iph = ip_hdr(skb); |
1da177e4 | 399 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; |
511c3f92 | 400 | struct rtable *rt; |
f216f082 | 401 | int err; |
93fdd47e HX |
402 | int frag_max_size; |
403 | ||
404 | frag_max_size = IPCB(skb)->frag_max_size; | |
405 | BR_INPUT_SKB_CB(skb)->frag_max_size = frag_max_size; | |
1da177e4 | 406 | |
1da177e4 LT |
407 | if (nf_bridge->mask & BRNF_PKT_TYPE) { |
408 | skb->pkt_type = PACKET_OTHERHOST; | |
409 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
410 | } | |
411 | nf_bridge->mask ^= BRNF_NF_BRIDGE_PREROUTING; | |
1da177e4 | 412 | if (dnat_took_place(skb)) { |
f216f082 | 413 | if ((err = ip_route_input(skb, iph->daddr, iph->saddr, iph->tos, dev))) { |
f3abc9b9 | 414 | struct in_device *in_dev = __in_dev_get_rcu(dev); |
f216f082 BDS |
415 | |
416 | /* If err equals -EHOSTUNREACH the error is due to a | |
417 | * martian destination or due to the fact that | |
418 | * forwarding is disabled. For most martian packets, | |
419 | * ip_route_output_key() will fail. It won't fail for 2 types of | |
420 | * martian destinations: loopback destinations and destination | |
421 | * 0.0.0.0. In both cases the packet will be dropped because the | |
422 | * destination is the loopback device and not the bridge. */ | |
423 | if (err != -EHOSTUNREACH || !in_dev || IN_DEV_FORWARD(in_dev)) | |
424 | goto free_skb; | |
1da177e4 | 425 | |
78fbfd8a DM |
426 | rt = ip_route_output(dev_net(dev), iph->daddr, 0, |
427 | RT_TOS(iph->tos), 0); | |
b23dd4fe | 428 | if (!IS_ERR(rt)) { |
1c011bed | 429 | /* - Bridged-and-DNAT'ed traffic doesn't |
f216f082 | 430 | * require ip_forwarding. */ |
b23dd4fe DM |
431 | if (rt->dst.dev == dev) { |
432 | skb_dst_set(skb, &rt->dst); | |
1da177e4 LT |
433 | goto bridged_dnat; |
434 | } | |
b23dd4fe | 435 | ip_rt_put(rt); |
1da177e4 | 436 | } |
f216f082 | 437 | free_skb: |
1da177e4 LT |
438 | kfree_skb(skb); |
439 | return 0; | |
440 | } else { | |
adf30907 | 441 | if (skb_dst(skb)->dev == dev) { |
1da177e4 | 442 | bridged_dnat: |
1da177e4 | 443 | skb->dev = nf_bridge->physindev; |
e179e632 | 444 | nf_bridge_update_protocol(skb); |
fc38582d | 445 | nf_bridge_push_encap_header(skb); |
713aefa3 JE |
446 | NF_HOOK_THRESH(NFPROTO_BRIDGE, |
447 | NF_BR_PRE_ROUTING, | |
1da177e4 LT |
448 | skb, skb->dev, NULL, |
449 | br_nf_pre_routing_finish_bridge, | |
450 | 1); | |
451 | return 0; | |
452 | } | |
04091142 | 453 | ether_addr_copy(eth_hdr(skb)->h_dest, dev->dev_addr); |
1da177e4 LT |
454 | skb->pkt_type = PACKET_HOST; |
455 | } | |
456 | } else { | |
511c3f92 ED |
457 | rt = bridge_parent_rtable(nf_bridge->physindev); |
458 | if (!rt) { | |
4adf0af6 SW |
459 | kfree_skb(skb); |
460 | return 0; | |
461 | } | |
f9181f4f | 462 | skb_dst_set_noref(skb, &rt->dst); |
1da177e4 LT |
463 | } |
464 | ||
465 | skb->dev = nf_bridge->physindev; | |
e179e632 | 466 | nf_bridge_update_protocol(skb); |
fc38582d | 467 | nf_bridge_push_encap_header(skb); |
713aefa3 | 468 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
469 | br_handle_frame_finish, 1); |
470 | ||
471 | return 0; | |
472 | } | |
473 | ||
4981682c PNA |
474 | static struct net_device *brnf_get_logical_dev(struct sk_buff *skb, const struct net_device *dev) |
475 | { | |
476 | struct net_device *vlan, *br; | |
477 | ||
478 | br = bridge_parent(dev); | |
df8a39de | 479 | if (brnf_pass_vlan_indev == 0 || !skb_vlan_tag_present(skb)) |
4981682c PNA |
480 | return br; |
481 | ||
f06c7f9f | 482 | vlan = __vlan_find_dev_deep_rcu(br, skb->vlan_proto, |
df8a39de | 483 | skb_vlan_tag_get(skb) & VLAN_VID_MASK); |
4981682c PNA |
484 | |
485 | return vlan ? vlan : br; | |
486 | } | |
487 | ||
1da177e4 | 488 | /* Some common code for IPv4/IPv6 */ |
5dce971a | 489 | static struct net_device *setup_pre_routing(struct sk_buff *skb) |
1da177e4 LT |
490 | { |
491 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
492 | ||
493 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
494 | skb->pkt_type = PACKET_HOST; | |
495 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
496 | } | |
497 | ||
498 | nf_bridge->mask |= BRNF_NF_BRIDGE_PREROUTING; | |
499 | nf_bridge->physindev = skb->dev; | |
4981682c | 500 | skb->dev = brnf_get_logical_dev(skb, skb->dev); |
e179e632 BDS |
501 | if (skb->protocol == htons(ETH_P_8021Q)) |
502 | nf_bridge->mask |= BRNF_8021Q; | |
503 | else if (skb->protocol == htons(ETH_P_PPP_SES)) | |
504 | nf_bridge->mask |= BRNF_PPPoE; | |
5dce971a | 505 | |
6b8dbcf2 FW |
506 | /* Must drop socket now because of tproxy. */ |
507 | skb_orphan(skb); | |
5dce971a | 508 | return skb->dev; |
1da177e4 LT |
509 | } |
510 | ||
511 | /* We only check the length. A bridge shouldn't do any hop-by-hop stuff anyway */ | |
512 | static int check_hbh_len(struct sk_buff *skb) | |
513 | { | |
0660e03f | 514 | unsigned char *raw = (u8 *)(ipv6_hdr(skb) + 1); |
1da177e4 | 515 | u32 pkt_len; |
d56f90a7 ACM |
516 | const unsigned char *nh = skb_network_header(skb); |
517 | int off = raw - nh; | |
789bc3e5 | 518 | int len = (raw[1] + 1) << 3; |
1da177e4 LT |
519 | |
520 | if ((raw + len) - skb->data > skb_headlen(skb)) | |
521 | goto bad; | |
522 | ||
523 | off += 2; | |
524 | len -= 2; | |
525 | ||
526 | while (len > 0) { | |
d56f90a7 | 527 | int optlen = nh[off + 1] + 2; |
1da177e4 | 528 | |
d56f90a7 | 529 | switch (nh[off]) { |
1de5a71c | 530 | case IPV6_TLV_PAD1: |
1da177e4 LT |
531 | optlen = 1; |
532 | break; | |
533 | ||
534 | case IPV6_TLV_PADN: | |
535 | break; | |
536 | ||
537 | case IPV6_TLV_JUMBO: | |
d56f90a7 | 538 | if (nh[off + 1] != 4 || (off & 3) != 2) |
1da177e4 | 539 | goto bad; |
d56f90a7 | 540 | pkt_len = ntohl(*(__be32 *) (nh + off + 2)); |
b0366486 | 541 | if (pkt_len <= IPV6_MAXPLEN || |
0660e03f | 542 | ipv6_hdr(skb)->payload_len) |
b0366486 | 543 | goto bad; |
1da177e4 LT |
544 | if (pkt_len > skb->len - sizeof(struct ipv6hdr)) |
545 | goto bad; | |
b0366486 | 546 | if (pskb_trim_rcsum(skb, |
789bc3e5 | 547 | pkt_len + sizeof(struct ipv6hdr))) |
b0366486 | 548 | goto bad; |
d56f90a7 | 549 | nh = skb_network_header(skb); |
1da177e4 LT |
550 | break; |
551 | default: | |
552 | if (optlen > len) | |
553 | goto bad; | |
554 | break; | |
555 | } | |
556 | off += optlen; | |
557 | len -= optlen; | |
558 | } | |
559 | if (len == 0) | |
560 | return 0; | |
561 | bad: | |
562 | return -1; | |
563 | ||
564 | } | |
565 | ||
566 | /* Replicate the checks that IPv6 does on packet reception and pass the packet | |
567 | * to ip6tables, which doesn't support NAT, so things are fairly simple. */ | |
795aa6ef | 568 | static unsigned int br_nf_pre_routing_ipv6(const struct nf_hook_ops *ops, |
789bc3e5 SH |
569 | struct sk_buff *skb, |
570 | const struct net_device *in, | |
571 | const struct net_device *out, | |
572 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 573 | { |
b71d1d42 | 574 | const struct ipv6hdr *hdr; |
1da177e4 | 575 | u32 pkt_len; |
1da177e4 LT |
576 | |
577 | if (skb->len < sizeof(struct ipv6hdr)) | |
deef4b52 | 578 | return NF_DROP; |
1da177e4 LT |
579 | |
580 | if (!pskb_may_pull(skb, sizeof(struct ipv6hdr))) | |
deef4b52 | 581 | return NF_DROP; |
1da177e4 | 582 | |
0660e03f | 583 | hdr = ipv6_hdr(skb); |
1da177e4 LT |
584 | |
585 | if (hdr->version != 6) | |
deef4b52 | 586 | return NF_DROP; |
1da177e4 LT |
587 | |
588 | pkt_len = ntohs(hdr->payload_len); | |
589 | ||
590 | if (pkt_len || hdr->nexthdr != NEXTHDR_HOP) { | |
591 | if (pkt_len + sizeof(struct ipv6hdr) > skb->len) | |
deef4b52 | 592 | return NF_DROP; |
b38dfee3 | 593 | if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr))) |
deef4b52 | 594 | return NF_DROP; |
1da177e4 LT |
595 | } |
596 | if (hdr->nexthdr == NEXTHDR_HOP && check_hbh_len(skb)) | |
deef4b52 | 597 | return NF_DROP; |
1da177e4 | 598 | |
789bc3e5 | 599 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 600 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 601 | return NF_DROP; |
5dce971a SH |
602 | if (!setup_pre_routing(skb)) |
603 | return NF_DROP; | |
1da177e4 | 604 | |
e179e632 | 605 | skb->protocol = htons(ETH_P_IPV6); |
713aefa3 | 606 | NF_HOOK(NFPROTO_IPV6, NF_INET_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
607 | br_nf_pre_routing_finish_ipv6); |
608 | ||
609 | return NF_STOLEN; | |
1da177e4 LT |
610 | } |
611 | ||
612 | /* Direct IPv6 traffic to br_nf_pre_routing_ipv6. | |
613 | * Replicate the checks that IPv4 does on packet reception. | |
614 | * Set skb->dev to the bridge device (i.e. parent of the | |
615 | * receiving device) to make netfilter happy, the REDIRECT | |
616 | * target in particular. Save the original destination IP | |
617 | * address to be able to detect DNAT afterwards. */ | |
795aa6ef PM |
618 | static unsigned int br_nf_pre_routing(const struct nf_hook_ops *ops, |
619 | struct sk_buff *skb, | |
ee02b3a6 SH |
620 | const struct net_device *in, |
621 | const struct net_device *out, | |
622 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 623 | { |
4df53d8b PM |
624 | struct net_bridge_port *p; |
625 | struct net_bridge *br; | |
e7c243c9 EP |
626 | __u32 len = nf_bridge_encap_header_len(skb); |
627 | ||
e7c243c9 | 628 | if (unlikely(!pskb_may_pull(skb, len))) |
deef4b52 | 629 | return NF_DROP; |
1da177e4 | 630 | |
e490c1de | 631 | p = br_port_get_rcu(in); |
4df53d8b | 632 | if (p == NULL) |
deef4b52 | 633 | return NF_DROP; |
4df53d8b PM |
634 | br = p->br; |
635 | ||
739e4505 | 636 | if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) { |
4df53d8b | 637 | if (!brnf_call_ip6tables && !br->nf_call_ip6tables) |
1da177e4 | 638 | return NF_ACCEPT; |
4df53d8b | 639 | |
fc38582d | 640 | nf_bridge_pull_encap_header_rcsum(skb); |
795aa6ef | 641 | return br_nf_pre_routing_ipv6(ops, skb, in, out, okfn); |
1da177e4 | 642 | } |
4df53d8b PM |
643 | |
644 | if (!brnf_call_iptables && !br->nf_call_iptables) | |
1da177e4 | 645 | return NF_ACCEPT; |
1da177e4 | 646 | |
739e4505 | 647 | if (!IS_IP(skb) && !IS_VLAN_IP(skb) && !IS_PPPOE_IP(skb)) |
1da177e4 LT |
648 | return NF_ACCEPT; |
649 | ||
fc38582d | 650 | nf_bridge_pull_encap_header_rcsum(skb); |
1da177e4 | 651 | |
462fb2af | 652 | if (br_parse_ip_options(skb)) |
deef4b52 | 653 | return NF_DROP; |
17762060 | 654 | |
789bc3e5 | 655 | nf_bridge_put(skb->nf_bridge); |
fdeabdef | 656 | if (!nf_bridge_alloc(skb)) |
1da177e4 | 657 | return NF_DROP; |
5dce971a SH |
658 | if (!setup_pre_routing(skb)) |
659 | return NF_DROP; | |
c055d5b0 | 660 | |
e179e632 | 661 | skb->protocol = htons(ETH_P_IP); |
1da177e4 | 662 | |
713aefa3 | 663 | NF_HOOK(NFPROTO_IPV4, NF_INET_PRE_ROUTING, skb, skb->dev, NULL, |
1da177e4 LT |
664 | br_nf_pre_routing_finish); |
665 | ||
666 | return NF_STOLEN; | |
1da177e4 LT |
667 | } |
668 | ||
669 | ||
670 | /* PF_BRIDGE/LOCAL_IN ************************************************/ | |
671 | /* The packet is locally destined, which requires a real | |
672 | * dst_entry, so detach the fake one. On the way up, the | |
673 | * packet would pass through PRE_ROUTING again (which already | |
674 | * took place when the packet entered the bridge), but we | |
675 | * register an IPv4 PRE_ROUTING 'sabotage' hook that will | |
676 | * prevent this from happening. */ | |
795aa6ef PM |
677 | static unsigned int br_nf_local_in(const struct nf_hook_ops *ops, |
678 | struct sk_buff *skb, | |
789bc3e5 SH |
679 | const struct net_device *in, |
680 | const struct net_device *out, | |
681 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 682 | { |
a881e963 | 683 | br_drop_fake_rtable(skb); |
1da177e4 LT |
684 | return NF_ACCEPT; |
685 | } | |
686 | ||
1da177e4 LT |
687 | /* PF_BRIDGE/FORWARD *************************************************/ |
688 | static int br_nf_forward_finish(struct sk_buff *skb) | |
689 | { | |
690 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
691 | struct net_device *in; | |
1da177e4 | 692 | |
739e4505 | 693 | if (!IS_ARP(skb) && !IS_VLAN_ARP(skb)) { |
0b67c43c FW |
694 | int frag_max_size; |
695 | ||
696 | if (skb->protocol == htons(ETH_P_IP)) { | |
697 | frag_max_size = IPCB(skb)->frag_max_size; | |
698 | BR_INPUT_SKB_CB(skb)->frag_max_size = frag_max_size; | |
699 | } | |
700 | ||
1da177e4 LT |
701 | in = nf_bridge->physindev; |
702 | if (nf_bridge->mask & BRNF_PKT_TYPE) { | |
703 | skb->pkt_type = PACKET_OTHERHOST; | |
704 | nf_bridge->mask ^= BRNF_PKT_TYPE; | |
705 | } | |
e94c6743 | 706 | nf_bridge_update_protocol(skb); |
1da177e4 LT |
707 | } else { |
708 | in = *((struct net_device **)(skb->cb)); | |
709 | } | |
fc38582d | 710 | nf_bridge_push_encap_header(skb); |
e179e632 | 711 | |
713aefa3 | 712 | NF_HOOK_THRESH(NFPROTO_BRIDGE, NF_BR_FORWARD, skb, in, |
789bc3e5 | 713 | skb->dev, br_forward_finish, 1); |
1da177e4 LT |
714 | return 0; |
715 | } | |
716 | ||
739e4505 | 717 | |
1da177e4 LT |
718 | /* This is the 'purely bridged' case. For IP, we pass the packet to |
719 | * netfilter with indev and outdev set to the bridge device, | |
720 | * but we are still able to filter on the 'real' indev/outdev | |
721 | * because of the physdev module. For ARP, indev and outdev are the | |
722 | * bridge ports. */ | |
795aa6ef PM |
723 | static unsigned int br_nf_forward_ip(const struct nf_hook_ops *ops, |
724 | struct sk_buff *skb, | |
789bc3e5 SH |
725 | const struct net_device *in, |
726 | const struct net_device *out, | |
727 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 728 | { |
1da177e4 | 729 | struct nf_bridge_info *nf_bridge; |
5dce971a | 730 | struct net_device *parent; |
76108cea | 731 | u_int8_t pf; |
1da177e4 LT |
732 | |
733 | if (!skb->nf_bridge) | |
734 | return NF_ACCEPT; | |
735 | ||
2dc2f207 PM |
736 | /* Need exclusive nf_bridge_info since we might have multiple |
737 | * different physoutdevs. */ | |
738 | if (!nf_bridge_unshare(skb)) | |
739 | return NF_DROP; | |
740 | ||
5dce971a SH |
741 | parent = bridge_parent(out); |
742 | if (!parent) | |
743 | return NF_DROP; | |
744 | ||
739e4505 | 745 | if (IS_IP(skb) || IS_VLAN_IP(skb) || IS_PPPOE_IP(skb)) |
aa740f46 | 746 | pf = NFPROTO_IPV4; |
739e4505 | 747 | else if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) |
aa740f46 | 748 | pf = NFPROTO_IPV6; |
a2bd40ad HX |
749 | else |
750 | return NF_ACCEPT; | |
1da177e4 | 751 | |
3db05fea | 752 | nf_bridge_pull_encap_header(skb); |
1da177e4 | 753 | |
1da177e4 LT |
754 | nf_bridge = skb->nf_bridge; |
755 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
756 | skb->pkt_type = PACKET_HOST; | |
757 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
758 | } | |
759 | ||
0b67c43c FW |
760 | if (pf == NFPROTO_IPV4) { |
761 | int frag_max = BR_INPUT_SKB_CB(skb)->frag_max_size; | |
762 | ||
763 | if (br_parse_ip_options(skb)) | |
764 | return NF_DROP; | |
765 | ||
766 | IPCB(skb)->frag_max_size = frag_max; | |
767 | } | |
6b1e960f | 768 | |
1da177e4 | 769 | nf_bridge->physoutdev = skb->dev; |
aa740f46 | 770 | if (pf == NFPROTO_IPV4) |
e179e632 BDS |
771 | skb->protocol = htons(ETH_P_IP); |
772 | else | |
773 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 774 | |
4981682c | 775 | NF_HOOK(pf, NF_INET_FORWARD, skb, brnf_get_logical_dev(skb, in), parent, |
5dce971a | 776 | br_nf_forward_finish); |
1da177e4 LT |
777 | |
778 | return NF_STOLEN; | |
779 | } | |
780 | ||
795aa6ef PM |
781 | static unsigned int br_nf_forward_arp(const struct nf_hook_ops *ops, |
782 | struct sk_buff *skb, | |
789bc3e5 SH |
783 | const struct net_device *in, |
784 | const struct net_device *out, | |
785 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 786 | { |
4df53d8b PM |
787 | struct net_bridge_port *p; |
788 | struct net_bridge *br; | |
1da177e4 LT |
789 | struct net_device **d = (struct net_device **)(skb->cb); |
790 | ||
e490c1de | 791 | p = br_port_get_rcu(out); |
4df53d8b PM |
792 | if (p == NULL) |
793 | return NF_ACCEPT; | |
794 | br = p->br; | |
795 | ||
796 | if (!brnf_call_arptables && !br->nf_call_arptables) | |
1da177e4 | 797 | return NF_ACCEPT; |
1da177e4 | 798 | |
739e4505 | 799 | if (!IS_ARP(skb)) { |
8b42ec39 | 800 | if (!IS_VLAN_ARP(skb)) |
1da177e4 | 801 | return NF_ACCEPT; |
3db05fea | 802 | nf_bridge_pull_encap_header(skb); |
1da177e4 LT |
803 | } |
804 | ||
d0a92be0 | 805 | if (arp_hdr(skb)->ar_pln != 4) { |
fc38582d | 806 | if (IS_VLAN_ARP(skb)) |
3db05fea | 807 | nf_bridge_push_encap_header(skb); |
1da177e4 LT |
808 | return NF_ACCEPT; |
809 | } | |
810 | *d = (struct net_device *)in; | |
fdc9314c | 811 | NF_HOOK(NFPROTO_ARP, NF_ARP_FORWARD, skb, (struct net_device *)in, |
1da177e4 LT |
812 | (struct net_device *)out, br_nf_forward_finish); |
813 | ||
814 | return NF_STOLEN; | |
815 | } | |
816 | ||
aff09ce3 | 817 | #if IS_ENABLED(CONFIG_NF_DEFRAG_IPV4) |
e70deecb | 818 | static int br_nf_push_frag_xmit(struct sk_buff *skb) |
8bd63cf1 | 819 | { |
e70deecb | 820 | struct brnf_frag_data *data; |
8bd63cf1 | 821 | int err; |
8bd63cf1 | 822 | |
e70deecb FW |
823 | data = this_cpu_ptr(&brnf_frag_data_storage); |
824 | err = skb_cow_head(skb, data->size); | |
8bd63cf1 | 825 | |
e70deecb | 826 | if (err) { |
8bd63cf1 FW |
827 | kfree_skb(skb); |
828 | return 0; | |
829 | } | |
830 | ||
e70deecb FW |
831 | skb_copy_to_linear_data_offset(skb, -data->size, data->mac, data->size); |
832 | __skb_push(skb, data->encap_size); | |
833 | ||
8bd63cf1 FW |
834 | return br_dev_queue_push_xmit(skb); |
835 | } | |
836 | ||
2e2f7aef PM |
837 | static int br_nf_dev_queue_xmit(struct sk_buff *skb) |
838 | { | |
462fb2af | 839 | int ret; |
93fdd47e | 840 | int frag_max_size; |
7a8d831d | 841 | unsigned int mtu_reserved; |
462fb2af | 842 | |
7a8d831d FW |
843 | if (skb_is_gso(skb) || skb->protocol != htons(ETH_P_IP)) |
844 | return br_dev_queue_push_xmit(skb); | |
845 | ||
846 | mtu_reserved = nf_bridge_mtu_reduction(skb); | |
93fdd47e HX |
847 | /* This is wrong! We should preserve the original fragment |
848 | * boundaries by preserving frag_list rather than refragmenting. | |
849 | */ | |
7a8d831d | 850 | if (skb->len + mtu_reserved > skb->dev->mtu) { |
e70deecb FW |
851 | struct brnf_frag_data *data; |
852 | ||
93fdd47e | 853 | frag_max_size = BR_INPUT_SKB_CB(skb)->frag_max_size; |
462fb2af BD |
854 | if (br_parse_ip_options(skb)) |
855 | /* Drop invalid packet */ | |
856 | return NF_DROP; | |
93fdd47e | 857 | IPCB(skb)->frag_max_size = frag_max_size; |
e70deecb FW |
858 | |
859 | nf_bridge_update_protocol(skb); | |
860 | ||
861 | data = this_cpu_ptr(&brnf_frag_data_storage); | |
862 | data->encap_size = nf_bridge_encap_header_len(skb); | |
863 | data->size = ETH_HLEN + data->encap_size; | |
864 | ||
865 | skb_copy_from_linear_data_offset(skb, -data->size, data->mac, | |
866 | data->size); | |
867 | ||
8bd63cf1 | 868 | ret = ip_fragment(skb, br_nf_push_frag_xmit); |
e70deecb | 869 | } else { |
462fb2af | 870 | ret = br_dev_queue_push_xmit(skb); |
e70deecb | 871 | } |
462fb2af BD |
872 | |
873 | return ret; | |
2e2f7aef | 874 | } |
c197facc | 875 | #else |
876 | static int br_nf_dev_queue_xmit(struct sk_buff *skb) | |
877 | { | |
878 | return br_dev_queue_push_xmit(skb); | |
879 | } | |
880 | #endif | |
1da177e4 LT |
881 | |
882 | /* PF_BRIDGE/POST_ROUTING ********************************************/ | |
795aa6ef PM |
883 | static unsigned int br_nf_post_routing(const struct nf_hook_ops *ops, |
884 | struct sk_buff *skb, | |
789bc3e5 SH |
885 | const struct net_device *in, |
886 | const struct net_device *out, | |
887 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 888 | { |
3db05fea | 889 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; |
1da177e4 | 890 | struct net_device *realoutdev = bridge_parent(skb->dev); |
76108cea | 891 | u_int8_t pf; |
1da177e4 | 892 | |
e4bb9bcb FW |
893 | /* if nf_bridge is set, but ->physoutdev is NULL, this packet came in |
894 | * on a bridge, but was delivered locally and is now being routed: | |
895 | * | |
896 | * POST_ROUTING was already invoked from the ip stack. | |
897 | */ | |
898 | if (!nf_bridge || !nf_bridge->physoutdev) | |
81d9ddae PM |
899 | return NF_ACCEPT; |
900 | ||
5dce971a SH |
901 | if (!realoutdev) |
902 | return NF_DROP; | |
903 | ||
739e4505 | 904 | if (IS_IP(skb) || IS_VLAN_IP(skb) || IS_PPPOE_IP(skb)) |
aa740f46 | 905 | pf = NFPROTO_IPV4; |
739e4505 | 906 | else if (IS_IPV6(skb) || IS_VLAN_IPV6(skb) || IS_PPPOE_IPV6(skb)) |
aa740f46 | 907 | pf = NFPROTO_IPV6; |
a2bd40ad HX |
908 | else |
909 | return NF_ACCEPT; | |
1da177e4 | 910 | |
1da177e4 LT |
911 | /* We assume any code from br_dev_queue_push_xmit onwards doesn't care |
912 | * about the value of skb->pkt_type. */ | |
913 | if (skb->pkt_type == PACKET_OTHERHOST) { | |
914 | skb->pkt_type = PACKET_HOST; | |
915 | nf_bridge->mask |= BRNF_PKT_TYPE; | |
916 | } | |
917 | ||
fc38582d | 918 | nf_bridge_pull_encap_header(skb); |
aa740f46 | 919 | if (pf == NFPROTO_IPV4) |
e179e632 BDS |
920 | skb->protocol = htons(ETH_P_IP); |
921 | else | |
922 | skb->protocol = htons(ETH_P_IPV6); | |
1da177e4 | 923 | |
6e23ae2a | 924 | NF_HOOK(pf, NF_INET_POST_ROUTING, skb, NULL, realoutdev, |
2e2f7aef | 925 | br_nf_dev_queue_xmit); |
1da177e4 LT |
926 | |
927 | return NF_STOLEN; | |
1da177e4 LT |
928 | } |
929 | ||
1da177e4 LT |
930 | /* IP/SABOTAGE *****************************************************/ |
931 | /* Don't hand locally destined packets to PF_INET(6)/PRE_ROUTING | |
932 | * for the second time. */ | |
795aa6ef PM |
933 | static unsigned int ip_sabotage_in(const struct nf_hook_ops *ops, |
934 | struct sk_buff *skb, | |
789bc3e5 SH |
935 | const struct net_device *in, |
936 | const struct net_device *out, | |
937 | int (*okfn)(struct sk_buff *)) | |
1da177e4 | 938 | { |
3db05fea HX |
939 | if (skb->nf_bridge && |
940 | !(skb->nf_bridge->mask & BRNF_NF_BRIDGE_PREROUTING)) { | |
1da177e4 LT |
941 | return NF_STOP; |
942 | } | |
943 | ||
944 | return NF_ACCEPT; | |
945 | } | |
946 | ||
e5de75bf PNA |
947 | /* This is called when br_netfilter has called into iptables/netfilter, |
948 | * and DNAT has taken place on a bridge-forwarded packet. | |
949 | * | |
950 | * neigh->output has created a new MAC header, with local br0 MAC | |
951 | * as saddr. | |
952 | * | |
953 | * This restores the original MAC saddr of the bridged packet | |
954 | * before invoking bridge forward logic to transmit the packet. | |
955 | */ | |
956 | static void br_nf_pre_routing_finish_bridge_slow(struct sk_buff *skb) | |
957 | { | |
958 | struct nf_bridge_info *nf_bridge = skb->nf_bridge; | |
959 | ||
960 | skb_pull(skb, ETH_HLEN); | |
961 | nf_bridge->mask &= ~BRNF_BRIDGED_DNAT; | |
962 | ||
e70deecb FW |
963 | BUILD_BUG_ON(sizeof(nf_bridge->neigh_header) != (ETH_HLEN - ETH_ALEN)); |
964 | ||
965 | skb_copy_to_linear_data_offset(skb, -(ETH_HLEN - ETH_ALEN), | |
966 | nf_bridge->neigh_header, | |
967 | ETH_HLEN - ETH_ALEN); | |
e5de75bf PNA |
968 | skb->dev = nf_bridge->physindev; |
969 | br_handle_frame_finish(skb); | |
970 | } | |
971 | ||
1a4ba64d | 972 | static int br_nf_dev_xmit(struct sk_buff *skb) |
e5de75bf PNA |
973 | { |
974 | if (skb->nf_bridge && (skb->nf_bridge->mask & BRNF_BRIDGED_DNAT)) { | |
975 | br_nf_pre_routing_finish_bridge_slow(skb); | |
976 | return 1; | |
977 | } | |
978 | return 0; | |
979 | } | |
1a4ba64d PNA |
980 | |
981 | static const struct nf_br_ops br_ops = { | |
982 | .br_dev_xmit_hook = br_nf_dev_xmit, | |
983 | }; | |
e5de75bf | 984 | |
4b7fd5d9 PNA |
985 | void br_netfilter_enable(void) |
986 | { | |
987 | } | |
988 | EXPORT_SYMBOL_GPL(br_netfilter_enable); | |
989 | ||
ea2d9b41 BDS |
990 | /* For br_nf_post_routing, we need (prio = NF_BR_PRI_LAST), because |
991 | * br_dev_queue_push_xmit is called afterwards */ | |
1999414a | 992 | static struct nf_hook_ops br_nf_ops[] __read_mostly = { |
1490fd89 CG |
993 | { |
994 | .hook = br_nf_pre_routing, | |
995 | .owner = THIS_MODULE, | |
aa740f46 | 996 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
997 | .hooknum = NF_BR_PRE_ROUTING, |
998 | .priority = NF_BR_PRI_BRNF, | |
999 | }, | |
1000 | { | |
1001 | .hook = br_nf_local_in, | |
1002 | .owner = THIS_MODULE, | |
aa740f46 | 1003 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
1004 | .hooknum = NF_BR_LOCAL_IN, |
1005 | .priority = NF_BR_PRI_BRNF, | |
1006 | }, | |
1007 | { | |
1008 | .hook = br_nf_forward_ip, | |
1009 | .owner = THIS_MODULE, | |
aa740f46 | 1010 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
1011 | .hooknum = NF_BR_FORWARD, |
1012 | .priority = NF_BR_PRI_BRNF - 1, | |
1013 | }, | |
1014 | { | |
1015 | .hook = br_nf_forward_arp, | |
1016 | .owner = THIS_MODULE, | |
aa740f46 | 1017 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
1018 | .hooknum = NF_BR_FORWARD, |
1019 | .priority = NF_BR_PRI_BRNF, | |
1020 | }, | |
1490fd89 CG |
1021 | { |
1022 | .hook = br_nf_post_routing, | |
1023 | .owner = THIS_MODULE, | |
aa740f46 | 1024 | .pf = NFPROTO_BRIDGE, |
1490fd89 CG |
1025 | .hooknum = NF_BR_POST_ROUTING, |
1026 | .priority = NF_BR_PRI_LAST, | |
1027 | }, | |
1028 | { | |
1029 | .hook = ip_sabotage_in, | |
1030 | .owner = THIS_MODULE, | |
aa740f46 | 1031 | .pf = NFPROTO_IPV4, |
1490fd89 CG |
1032 | .hooknum = NF_INET_PRE_ROUTING, |
1033 | .priority = NF_IP_PRI_FIRST, | |
1034 | }, | |
1035 | { | |
1036 | .hook = ip_sabotage_in, | |
1037 | .owner = THIS_MODULE, | |
aa740f46 | 1038 | .pf = NFPROTO_IPV6, |
1490fd89 CG |
1039 | .hooknum = NF_INET_PRE_ROUTING, |
1040 | .priority = NF_IP6_PRI_FIRST, | |
1041 | }, | |
1da177e4 LT |
1042 | }; |
1043 | ||
1044 | #ifdef CONFIG_SYSCTL | |
1045 | static | |
fe2c6338 | 1046 | int brnf_sysctl_call_tables(struct ctl_table *ctl, int write, |
56b148eb | 1047 | void __user *buffer, size_t *lenp, loff_t *ppos) |
1da177e4 LT |
1048 | { |
1049 | int ret; | |
1050 | ||
8d65af78 | 1051 | ret = proc_dointvec(ctl, write, buffer, lenp, ppos); |
1da177e4 LT |
1052 | |
1053 | if (write && *(int *)(ctl->data)) | |
1054 | *(int *)(ctl->data) = 1; | |
1055 | return ret; | |
1056 | } | |
1057 | ||
fe2c6338 | 1058 | static struct ctl_table brnf_table[] = { |
1da177e4 | 1059 | { |
1da177e4 LT |
1060 | .procname = "bridge-nf-call-arptables", |
1061 | .data = &brnf_call_arptables, | |
1062 | .maxlen = sizeof(int), | |
1063 | .mode = 0644, | |
6d9f239a | 1064 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1065 | }, |
1066 | { | |
1da177e4 LT |
1067 | .procname = "bridge-nf-call-iptables", |
1068 | .data = &brnf_call_iptables, | |
1069 | .maxlen = sizeof(int), | |
1070 | .mode = 0644, | |
6d9f239a | 1071 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1072 | }, |
1073 | { | |
1da177e4 LT |
1074 | .procname = "bridge-nf-call-ip6tables", |
1075 | .data = &brnf_call_ip6tables, | |
1076 | .maxlen = sizeof(int), | |
1077 | .mode = 0644, | |
6d9f239a | 1078 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 LT |
1079 | }, |
1080 | { | |
1da177e4 LT |
1081 | .procname = "bridge-nf-filter-vlan-tagged", |
1082 | .data = &brnf_filter_vlan_tagged, | |
1083 | .maxlen = sizeof(int), | |
1084 | .mode = 0644, | |
6d9f239a | 1085 | .proc_handler = brnf_sysctl_call_tables, |
516299d2 MM |
1086 | }, |
1087 | { | |
516299d2 MM |
1088 | .procname = "bridge-nf-filter-pppoe-tagged", |
1089 | .data = &brnf_filter_pppoe_tagged, | |
1090 | .maxlen = sizeof(int), | |
1091 | .mode = 0644, | |
6d9f239a | 1092 | .proc_handler = brnf_sysctl_call_tables, |
1da177e4 | 1093 | }, |
4981682c PNA |
1094 | { |
1095 | .procname = "bridge-nf-pass-vlan-input-dev", | |
1096 | .data = &brnf_pass_vlan_indev, | |
1097 | .maxlen = sizeof(int), | |
1098 | .mode = 0644, | |
1099 | .proc_handler = brnf_sysctl_call_tables, | |
1100 | }, | |
f8572d8f | 1101 | { } |
1da177e4 | 1102 | }; |
1da177e4 LT |
1103 | #endif |
1104 | ||
34666d46 | 1105 | static int __init br_netfilter_init(void) |
1da177e4 | 1106 | { |
5eb87f45 | 1107 | int ret; |
1da177e4 | 1108 | |
34666d46 | 1109 | ret = nf_register_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
5eb87f45 | 1110 | if (ret < 0) |
1da177e4 | 1111 | return ret; |
fc66f95c | 1112 | |
1da177e4 | 1113 | #ifdef CONFIG_SYSCTL |
ec8f23ce | 1114 | brnf_sysctl_header = register_net_sysctl(&init_net, "net/bridge", brnf_table); |
1da177e4 | 1115 | if (brnf_sysctl_header == NULL) { |
789bc3e5 SH |
1116 | printk(KERN_WARNING |
1117 | "br_netfilter: can't register to sysctl.\n"); | |
96727239 GU |
1118 | nf_unregister_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
1119 | return -ENOMEM; | |
1da177e4 LT |
1120 | } |
1121 | #endif | |
1a4ba64d | 1122 | RCU_INIT_POINTER(nf_br_ops, &br_ops); |
1da177e4 | 1123 | printk(KERN_NOTICE "Bridge firewalling registered\n"); |
1da177e4 LT |
1124 | return 0; |
1125 | } | |
1126 | ||
34666d46 | 1127 | static void __exit br_netfilter_fini(void) |
1da177e4 | 1128 | { |
1a4ba64d | 1129 | RCU_INIT_POINTER(nf_br_ops, NULL); |
5eb87f45 | 1130 | nf_unregister_hooks(br_nf_ops, ARRAY_SIZE(br_nf_ops)); |
1da177e4 | 1131 | #ifdef CONFIG_SYSCTL |
5dd3df10 | 1132 | unregister_net_sysctl_table(brnf_sysctl_header); |
1da177e4 LT |
1133 | #endif |
1134 | } | |
34666d46 PNA |
1135 | |
1136 | module_init(br_netfilter_init); | |
1137 | module_exit(br_netfilter_fini); | |
1138 | ||
1139 | MODULE_LICENSE("GPL"); | |
1140 | MODULE_AUTHOR("Lennert Buytenhek <buytenh@gnu.org>"); | |
1141 | MODULE_AUTHOR("Bart De Schuymer <bdschuym@pandora.be>"); | |
1142 | MODULE_DESCRIPTION("Linux ethernet netfilter firewall bridge"); |