1// SPDX-License-Identifier: GPL-2.0-or-later
2/*
3 * Handle incoming frames
4 * Linux ethernet bridge
5 *
6 * Authors:
7 * Lennert Buytenhek <buytenh@gnu.org>
8 */
9
10#include <linux/slab.h>
11#include <linux/kernel.h>
12#include <linux/netdevice.h>
13#include <linux/etherdevice.h>
14#include <linux/netfilter_bridge.h>
15#ifdef CONFIG_NETFILTER_FAMILY_BRIDGE
16#include <net/netfilter/nf_queue.h>
17#endif
18#include <linux/neighbour.h>
19#include <net/arp.h>
20#include <net/dsa.h>
21#include <linux/export.h>
22#include <linux/rculist.h>
23#include "br_private.h"
24#include "br_private_tunnel.h"
25
26static int
27br_netif_receive_skb(struct net *net, struct sock *sk, struct sk_buff *skb)
28{
29 br_drop_fake_rtable(skb);
30 return netif_receive_skb(skb);
31}
32
33static int br_pass_frame_up(struct sk_buff *skb, bool promisc)
34{
35 struct net_device *indev, *brdev = BR_INPUT_SKB_CB(skb)->brdev;
36 struct net_bridge *br = netdev_priv(dev: brdev);
37 struct net_bridge_vlan_group *vg;
38
39 dev_sw_netstats_rx_add(dev: brdev, len: skb->len);
40
41 vg = br_vlan_group_rcu(br);
42
43 /* Reset the offload_fwd_mark because there could be a stacked
44 * bridge above, and it should not think this bridge it doing
45 * that bridge's work forwarding out its ports.
46 */
47 br_switchdev_frame_unmark(skb);
48
49 /* Bridge is just like any other port. Make sure the
50 * packet is allowed except in promisc mode when someone
51 * may be running packet capture.
52 */
53 if (!(brdev->flags & IFF_PROMISC) &&
54 !br_allowed_egress(vg, skb)) {
55 kfree_skb(skb);
56 return NET_RX_DROP;
57 }
58
59 indev = skb->dev;
60 skb->dev = brdev;
61 skb = br_handle_vlan(br, NULL, vg, skb);
62 if (!skb)
63 return NET_RX_DROP;
64 /* update the multicast stats if the packet is IGMP/MLD */
65 br_multicast_count(br, NULL, skb, type: br_multicast_igmp_type(skb),
66 dir: BR_MCAST_DIR_TX);
67
68 BR_INPUT_SKB_CB(skb)->promisc = promisc;
69
70 return NF_HOOK(pf: NFPROTO_BRIDGE, NF_BR_LOCAL_IN,
71 net: dev_net(dev: indev), NULL, skb, in: indev, NULL,
72 okfn: br_netif_receive_skb);
73}
74
75/* note: already called with rcu_read_lock */
76int br_handle_frame_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
77{
78 enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED;
79 struct net_bridge_port *p = br_port_get_rcu(dev: skb->dev);
80 enum br_pkt_type pkt_type = BR_PKT_UNICAST;
81 struct net_bridge_fdb_entry *dst = NULL;
82 struct net_bridge_mcast_port *pmctx;
83 struct net_bridge_mdb_entry *mdst;
84 bool local_rcv, mcast_hit = false;
85 struct net_bridge_mcast *brmctx;
86 struct net_bridge_vlan *vlan;
87 struct net_bridge *br;
88 bool promisc;
89 u16 vid = 0;
90 u8 state;
91
92 if (!p)
93 goto drop;
94
95 br = p->br;
96
97 if (br_mst_is_enabled(br)) {
98 state = BR_STATE_FORWARDING;
99 } else {
100 if (p->state == BR_STATE_DISABLED) {
101 reason = SKB_DROP_REASON_BRIDGE_INGRESS_STP_STATE;
102 goto drop;
103 }
104
105 state = p->state;
106 }
107
108 brmctx = &p->br->multicast_ctx;
109 pmctx = &p->multicast_ctx;
110 if (!br_allowed_ingress(br: p->br, vg: nbp_vlan_group_rcu(p), skb, vid: &vid,
111 state: &state, vlan: &vlan))
112 goto out;
113
114 if (p->flags & BR_PORT_LOCKED) {
115 struct net_bridge_fdb_entry *fdb_src =
116 br_fdb_find_rcu(br, addr: eth_hdr(skb)->h_source, vid);
117
118 if (!fdb_src) {
119 /* FDB miss. Create locked FDB entry if MAB is enabled
120 * and drop the packet.
121 */
122 if (p->flags & BR_PORT_MAB)
123 br_fdb_update(br, source: p, addr: eth_hdr(skb)->h_source,
124 vid, BIT(BR_FDB_LOCKED));
125 goto drop;
126 } else if (READ_ONCE(fdb_src->dst) != p ||
127 test_bit(BR_FDB_LOCAL, &fdb_src->flags)) {
128 /* FDB mismatch. Drop the packet without roaming. */
129 goto drop;
130 } else if (test_bit(BR_FDB_LOCKED, &fdb_src->flags)) {
131 /* FDB match, but entry is locked. Refresh it and drop
132 * the packet.
133 */
134 br_fdb_update(br, source: p, addr: eth_hdr(skb)->h_source, vid,
135 BIT(BR_FDB_LOCKED));
136 goto drop;
137 }
138 }
139
140 nbp_switchdev_frame_mark(p, skb);
141
142 /* insert into forwarding database after filtering to avoid spoofing */
143 if (p->flags & BR_LEARNING)
144 br_fdb_update(br, source: p, addr: eth_hdr(skb)->h_source, vid, flags: 0);
145
146 promisc = !!(br->dev->flags & IFF_PROMISC);
147 local_rcv = promisc;
148
149 if (is_multicast_ether_addr(addr: eth_hdr(skb)->h_dest)) {
150 /* by definition the broadcast is also a multicast address */
151 if (is_broadcast_ether_addr(addr: eth_hdr(skb)->h_dest)) {
152 pkt_type = BR_PKT_BROADCAST;
153 local_rcv = true;
154 } else {
155 pkt_type = BR_PKT_MULTICAST;
156 if (br_multicast_rcv(brmctx: &brmctx, pmctx: &pmctx, vlan, skb, vid))
157 goto drop;
158 }
159 }
160
161 if (state == BR_STATE_LEARNING) {
162 reason = SKB_DROP_REASON_BRIDGE_INGRESS_STP_STATE;
163 goto drop;
164 }
165
166 BR_INPUT_SKB_CB(skb)->brdev = br->dev;
167 BR_INPUT_SKB_CB(skb)->src_port_isolated = !!(p->flags & BR_ISOLATED);
168
169 if (IS_ENABLED(CONFIG_INET) &&
170 (skb->protocol == htons(ETH_P_ARP) ||
171 skb->protocol == htons(ETH_P_RARP))) {
172 br_do_proxy_suppress_arp(skb, br, vid, p);
173 } else if (IS_ENABLED(CONFIG_IPV6) &&
174 skb->protocol == htons(ETH_P_IPV6) &&
175 br_opt_get(br, opt: BROPT_NEIGH_SUPPRESS_ENABLED) &&
176 pskb_may_pull(skb, len: sizeof(struct ipv6hdr) +
177 sizeof(struct nd_msg)) &&
178 ipv6_hdr(skb)->nexthdr == IPPROTO_ICMPV6) {
179 struct nd_msg *msg, _msg;
180
181 msg = br_is_nd_neigh_msg(skb, m: &_msg);
182 if (msg)
183 br_do_suppress_nd(skb, br, vid, p, msg);
184 }
185
186 switch (pkt_type) {
187 case BR_PKT_MULTICAST:
188 mdst = br_mdb_entry_skb_get(brmctx, skb, vid);
189 if ((mdst || BR_INPUT_SKB_CB_MROUTERS_ONLY(skb)) &&
190 br_multicast_querier_exists(brmctx, eth: eth_hdr(skb), mdb: mdst)) {
191 if ((mdst && mdst->host_joined) ||
192 br_multicast_is_router(brmctx, skb) ||
193 br->dev->flags & IFF_ALLMULTI) {
194 local_rcv = true;
195 DEV_STATS_INC(br->dev, multicast);
196 }
197 mcast_hit = true;
198 } else {
199 local_rcv = true;
200 DEV_STATS_INC(br->dev, multicast);
201 }
202 break;
203 case BR_PKT_UNICAST:
204 dst = br_fdb_find_rcu(br, addr: eth_hdr(skb)->h_dest, vid);
205 break;
206 default:
207 break;
208 }
209
210 if (dst) {
211 unsigned long now = jiffies;
212
213 if (test_bit(BR_FDB_LOCAL, &dst->flags))
214 return br_pass_frame_up(skb, promisc: false);
215
216 if (now != dst->used)
217 dst->used = now;
218 br_forward(to: dst->dst, skb, local_rcv, local_orig: false);
219 } else {
220 if (!mcast_hit)
221 br_flood(br, skb, pkt_type, local_rcv, local_orig: false, vid);
222 else
223 br_multicast_flood(mdst, skb, brmctx, local_rcv, local_orig: false);
224 }
225
226 if (local_rcv)
227 return br_pass_frame_up(skb, promisc);
228
229out:
230 return 0;
231drop:
232 kfree_skb_reason(skb, reason);
233 goto out;
234}
235EXPORT_SYMBOL_GPL(br_handle_frame_finish);
236
237static void __br_handle_local_finish(struct sk_buff *skb)
238{
239 struct net_bridge_port *p = br_port_get_rcu(dev: skb->dev);
240 u16 vid = 0;
241
242 /* check if vlan is allowed, to avoid spoofing */
243 if ((p->flags & BR_LEARNING) &&
244 nbp_state_should_learn(p) &&
245 !br_opt_get(br: p->br, opt: BROPT_NO_LL_LEARN) &&
246 br_should_learn(p, skb, vid: &vid))
247 br_fdb_update(br: p->br, source: p, addr: eth_hdr(skb)->h_source, vid, flags: 0);
248}
249
250/* note: already called with rcu_read_lock */
251static int br_handle_local_finish(struct net *net, struct sock *sk, struct sk_buff *skb)
252{
253 __br_handle_local_finish(skb);
254
255 /* return 1 to signal the okfn() was called so it's ok to use the skb */
256 return 1;
257}
258
259static int nf_hook_bridge_pre(struct sk_buff *skb, struct sk_buff **pskb)
260{
261#ifdef CONFIG_NETFILTER_FAMILY_BRIDGE
262 struct nf_hook_entries *e = NULL;
263 struct nf_hook_state state;
264 unsigned int verdict, i;
265 struct net *net;
266 int ret;
267
268 net = dev_net(dev: skb->dev);
269#ifdef HAVE_JUMP_LABEL
270 if (!static_key_false(&nf_hooks_needed[NFPROTO_BRIDGE][NF_BR_PRE_ROUTING]))
271 goto frame_finish;
272#endif
273
274 e = rcu_dereference(net->nf.hooks_bridge[NF_BR_PRE_ROUTING]);
275 if (!e)
276 goto frame_finish;
277
278 nf_hook_state_init(p: &state, NF_BR_PRE_ROUTING,
279 pf: NFPROTO_BRIDGE, indev: skb->dev, NULL, NULL,
280 net, okfn: br_handle_frame_finish);
281
282 for (i = 0; i < e->num_hook_entries; i++) {
283 verdict = nf_hook_entry_hookfn(entry: &e->hooks[i], skb, state: &state);
284 switch (verdict & NF_VERDICT_MASK) {
285 case NF_ACCEPT:
286 if (BR_INPUT_SKB_CB(skb)->br_netfilter_broute) {
287 *pskb = skb;
288 return RX_HANDLER_PASS;
289 }
290 break;
291 case NF_DROP:
292 kfree_skb(skb);
293 return RX_HANDLER_CONSUMED;
294 case NF_QUEUE:
295 ret = nf_queue(skb, state: &state, index: i, verdict);
296 if (ret == 1)
297 continue;
298 return RX_HANDLER_CONSUMED;
299 default: /* STOLEN */
300 return RX_HANDLER_CONSUMED;
301 }
302 }
303frame_finish:
304 net = dev_net(dev: skb->dev);
305 br_handle_frame_finish(net, NULL, skb);
306#else
307 br_handle_frame_finish(dev_net(skb->dev), NULL, skb);
308#endif
309 return RX_HANDLER_CONSUMED;
310}
311
312/* Return 0 if the frame was not processed otherwise 1
313 * note: already called with rcu_read_lock
314 */
315static int br_process_frame_type(struct net_bridge_port *p,
316 struct sk_buff *skb)
317{
318 struct br_frame_type *tmp;
319
320 hlist_for_each_entry_rcu(tmp, &p->br->frame_type_list, list)
321 if (unlikely(tmp->type == skb->protocol))
322 return tmp->frame_handler(p, skb);
323
324 return 0;
325}
326
327/*
328 * Return NULL if skb is handled
329 * note: already called with rcu_read_lock
330 */
331static rx_handler_result_t br_handle_frame(struct sk_buff **pskb)
332{
333 enum skb_drop_reason reason = SKB_DROP_REASON_NOT_SPECIFIED;
334 struct net_bridge_port *p;
335 struct sk_buff *skb = *pskb;
336 const unsigned char *dest = eth_hdr(skb)->h_dest;
337
338 if (unlikely(skb->pkt_type == PACKET_LOOPBACK))
339 return RX_HANDLER_PASS;
340
341 if (!is_valid_ether_addr(addr: eth_hdr(skb)->h_source)) {
342 reason = SKB_DROP_REASON_MAC_INVALID_SOURCE;
343 goto drop;
344 }
345
346 skb = skb_share_check(skb, GFP_ATOMIC);
347 if (!skb)
348 return RX_HANDLER_CONSUMED;
349
350 memset(skb->cb, 0, sizeof(struct br_input_skb_cb));
351 br_tc_skb_miss_set(skb, miss: false);
352
353 p = br_port_get_rcu(dev: skb->dev);
354 if (p->flags & BR_VLAN_TUNNEL)
355 br_handle_ingress_vlan_tunnel(skb, p, vg: nbp_vlan_group_rcu(p));
356
357 if (unlikely(is_link_local_ether_addr(dest))) {
358 u16 fwd_mask = p->br->group_fwd_mask_required;
359
360 /*
361 * See IEEE 802.1D Table 7-10 Reserved addresses
362 *
363 * Assignment Value
364 * Bridge Group Address 01-80-C2-00-00-00
365 * (MAC Control) 802.3 01-80-C2-00-00-01
366 * (Link Aggregation) 802.3 01-80-C2-00-00-02
367 * 802.1X PAE address 01-80-C2-00-00-03
368 *
369 * 802.1AB LLDP 01-80-C2-00-00-0E
370 *
371 * Others reserved for future standardization
372 */
373 fwd_mask |= p->group_fwd_mask;
374 switch (dest[5]) {
375 case 0x00: /* Bridge Group Address */
376 /* If STP is turned off,
377 then must forward to keep loop detection */
378 if (p->br->stp_enabled == BR_NO_STP ||
379 fwd_mask & (1u << dest[5]))
380 goto forward;
381 *pskb = skb;
382 __br_handle_local_finish(skb);
383 return RX_HANDLER_PASS;
384
385 case 0x01: /* IEEE MAC (Pause) */
386 reason = SKB_DROP_REASON_MAC_IEEE_MAC_CONTROL;
387 goto drop;
388
389 case 0x0E: /* 802.1AB LLDP */
390 fwd_mask |= p->br->group_fwd_mask;
391 if (fwd_mask & (1u << dest[5]))
392 goto forward;
393 *pskb = skb;
394 __br_handle_local_finish(skb);
395 return RX_HANDLER_PASS;
396
397 default:
398 /* Allow selective forwarding for most other protocols */
399 fwd_mask |= p->br->group_fwd_mask;
400 if (fwd_mask & (1u << dest[5]))
401 goto forward;
402 }
403
404 BR_INPUT_SKB_CB(skb)->promisc = false;
405
406 /* The else clause should be hit when nf_hook():
407 * - returns < 0 (drop/error)
408 * - returns = 0 (stolen/nf_queue)
409 * Thus return 1 from the okfn() to signal the skb is ok to pass
410 */
411 if (NF_HOOK(pf: NFPROTO_BRIDGE, NF_BR_LOCAL_IN,
412 net: dev_net(dev: skb->dev), NULL, skb, in: skb->dev, NULL,
413 okfn: br_handle_local_finish) == 1) {
414 return RX_HANDLER_PASS;
415 } else {
416 return RX_HANDLER_CONSUMED;
417 }
418 }
419
420 if (unlikely(br_process_frame_type(p, skb)))
421 return RX_HANDLER_PASS;
422
423forward:
424 if (br_mst_is_enabled(br: p->br))
425 goto defer_stp_filtering;
426
427 switch (p->state) {
428 case BR_STATE_FORWARDING:
429 case BR_STATE_LEARNING:
430defer_stp_filtering:
431 if (ether_addr_equal(addr1: p->br->dev->dev_addr, addr2: dest))
432 skb->pkt_type = PACKET_HOST;
433
434 return nf_hook_bridge_pre(skb, pskb);
435 default:
436 reason = SKB_DROP_REASON_BRIDGE_INGRESS_STP_STATE;
437drop:
438 kfree_skb_reason(skb, reason);
439 }
440 return RX_HANDLER_CONSUMED;
441}
442
443/* This function has no purpose other than to appease the br_port_get_rcu/rtnl
444 * helpers which identify bridged ports according to the rx_handler installed
445 * on them (so there _needs_ to be a bridge rx_handler even if we don't need it
446 * to do anything useful). This bridge won't support traffic to/from the stack,
447 * but only hardware bridging. So return RX_HANDLER_PASS so we don't steal
448 * frames from the ETH_P_XDSA packet_type handler.
449 */
450static rx_handler_result_t br_handle_frame_dummy(struct sk_buff **pskb)
451{
452 return RX_HANDLER_PASS;
453}
454
455rx_handler_func_t *br_get_rx_handler(const struct net_device *dev)
456{
457 if (netdev_uses_dsa(dev))
458 return br_handle_frame_dummy;
459
460 return br_handle_frame;
461}
462
463void br_add_frame(struct net_bridge *br, struct br_frame_type *ft)
464{
465 hlist_add_head_rcu(n: &ft->list, h: &br->frame_type_list);
466}
467
468void br_del_frame(struct net_bridge *br, struct br_frame_type *ft)
469{
470 struct br_frame_type *tmp;
471
472 hlist_for_each_entry(tmp, &br->frame_type_list, list)
473 if (ft == tmp) {
474 hlist_del_rcu(n: &ft->list);
475 return;
476 }
477}
478

source code of linux/net/bridge/br_input.c