| 1 |
// SPDX-License-Identifier: GPL-2.0-or-later |
| 2 |
/* |
| 3 |
* Forwarding decision |
| 4 |
* Linux ethernet bridge |
| 5 |
* |
| 6 |
* Authors: |
| 7 |
* Lennert Buytenhek <[email protected]> |
| 8 |
*/ |
| 9 |
|
| 10 |
#include <linux/err.h> |
| 11 |
#include <linux/slab.h> |
| 12 |
#include <linux/kernel.h> |
| 13 |
#include <linux/netdevice.h> |
| 14 |
#include <linux/netpoll.h> |
| 15 |
#include <linux/skbuff.h> |
| 16 |
#include <linux/if_vlan.h> |
| 17 |
#include <linux/netfilter_bridge.h> |
| 18 |
#include "br_private.h" |
| 19 |
|
| 20 |
/* Don't forward packets to originating port or forwarding disabled */ |
| 21 |
static inline int should_deliver(const struct net_bridge_port *p, |
| 22 |
const struct sk_buff *skb) |
| 23 |
{ |
| 24 |
struct net_bridge_vlan_group *vg; |
| 25 |
|
| 26 |
vg = nbp_vlan_group_rcu(p); |
| 27 |
return ((p->flags & BR_HAIRPIN_MODE) || skb->dev != p->dev) && |
| 28 |
(br_mst_is_enabled(p) || p->state == BR_STATE_FORWARDING) && |
| 29 |
br_allowed_egress(vg, skb) && nbp_switchdev_allowed_egress(p, skb) && |
| 30 |
!br_skb_isolated(p, skb); |
| 31 |
} |
| 32 |
|
| 33 |
int br_dev_queue_push_xmit(struct net *net, struct sock *sk, struct sk_buff *skb) |
| 34 |
{ |
| 35 |
skb_push(skb, ETH_HLEN); |
| 36 |
if (!is_skb_forwardable(skb->dev, skb)) |
| 37 |
goto drop; |
| 38 |
|
| 39 |
br_drop_fake_rtable(skb); |
| 40 |
|
| 41 |
if (skb->ip_summed == CHECKSUM_PARTIAL && |
| 42 |
eth_type_vlan(skb->protocol)) { |
| 43 |
int depth; |
| 44 |
|
| 45 |
if (!vlan_get_protocol_and_depth(skb, skb->protocol, &depth)) |
| 46 |
goto drop; |
| 47 |
|
| 48 |
skb_set_network_header(skb, depth); |
| 49 |
} |
| 50 |
|
| 51 |
br_switchdev_frame_set_offload_fwd_mark(skb); |
| 52 |
|
| 53 |
dev_queue_xmit(skb); |
| 54 |
|
| 55 |
return 0; |
| 56 |
|
| 57 |
drop: |
| 58 |
kfree_skb(skb); |
| 59 |
return 0; |
| 60 |
} |
| 61 |
EXPORT_SYMBOL_GPL(br_dev_queue_push_xmit); |
| 62 |
|
| 63 |
int br_forward_finish(struct net *net, struct sock *sk, struct sk_buff *skb) |
| 64 |
{ |
| 65 |
skb_clear_tstamp(skb); |
| 66 |
return NF_HOOK(NFPROTO_BRIDGE, NF_BR_POST_ROUTING, |
| 67 |
net, sk, skb, NULL, skb->dev, |
| 68 |
br_dev_queue_push_xmit); |
| 69 |
|
| 70 |
} |
| 71 |
EXPORT_SYMBOL_GPL(br_forward_finish); |
| 72 |
|
| 73 |
static void __br_forward(const struct net_bridge_port *to, |
| 74 |
struct sk_buff *skb, bool local_orig) |
| 75 |
{ |
| 76 |
struct net_bridge_vlan_group *vg; |
| 77 |
struct net_device *indev; |
| 78 |
struct net *net; |
| 79 |
int br_hook; |
| 80 |
|
| 81 |
/* Mark the skb for forwarding offload early so that br_handle_vlan() |
| 82 |
* can know whether to pop the VLAN header on egress or keep it. |
| 83 |
*/ |
| 84 |
nbp_switchdev_frame_mark_tx_fwd_offload(to, skb); |
| 85 |
|
| 86 |
vg = nbp_vlan_group_rcu(to); |
| 87 |
skb = br_handle_vlan(to->br, to, vg, skb); |
| 88 |
if (!skb) |
| 89 |
return; |
| 90 |
|
| 91 |
indev = skb->dev; |
| 92 |
skb->dev = to->dev; |
| 93 |
if (!local_orig) { |
| 94 |
if (skb_warn_if_lro(skb)) { |
| 95 |
kfree_skb(skb); |
| 96 |
return; |
| 97 |
} |
| 98 |
br_hook = NF_BR_FORWARD; |
| 99 |
skb_forward_csum(skb); |
| 100 |
net = dev_net(indev); |
| 101 |
} else { |
| 102 |
if (unlikely(netpoll_tx_running(to->br->dev))) { |
| 103 |
skb_push(skb, ETH_HLEN); |
| 104 |
if (!is_skb_forwardable(skb->dev, skb)) |
| 105 |
kfree_skb(skb); |
| 106 |
else |
| 107 |
br_netpoll_send_skb(to, skb); |
| 108 |
return; |
| 109 |
} |
| 110 |
br_hook = NF_BR_LOCAL_OUT; |
| 111 |
net = dev_net(skb->dev); |
| 112 |
indev = NULL; |
| 113 |
} |
| 114 |
|
| 115 |
NF_HOOK(NFPROTO_BRIDGE, br_hook, |
| 116 |
net, NULL, skb, indev, skb->dev, |
| 117 |
br_forward_finish); |
| 118 |
} |
| 119 |
|
| 120 |
static int deliver_clone(const struct net_bridge_port *prev, |
| 121 |
struct sk_buff *skb, bool local_orig) |
| 122 |
{ |
| 123 |
struct net_device *dev = BR_INPUT_SKB_CB(skb)->brdev; |
| 124 |
|
| 125 |
skb = skb_clone(skb, GFP_ATOMIC); |
| 126 |
if (!skb) { |
| 127 |
DEV_STATS_INC(dev, tx_dropped); |
| 128 |
return -ENOMEM; |
| 129 |
} |
| 130 |
|
| 131 |
__br_forward(prev, skb, local_orig); |
| 132 |
return 0; |
| 133 |
} |
| 134 |
|
| 135 |
/** |
| 136 |
* br_forward - forward a packet to a specific port |
| 137 |
* @to: destination port |
| 138 |
* @skb: packet being forwarded |
| 139 |
* @local_rcv: packet will be received locally after forwarding |
| 140 |
* @local_orig: packet is locally originated |
| 141 |
* |
| 142 |
* Should be called with rcu_read_lock. |
| 143 |
*/ |
| 144 |
void br_forward(const struct net_bridge_port *to, |
| 145 |
struct sk_buff *skb, bool local_rcv, bool local_orig) |
| 146 |
{ |
| 147 |
if (unlikely(!to)) |
| 148 |
goto out; |
| 149 |
|
| 150 |
/* redirect to backup link if the destination port is down */ |
| 151 |
if (rcu_access_pointer(to->backup_port) && |
| 152 |
(!netif_carrier_ok(to->dev) || !netif_running(to->dev))) { |
| 153 |
struct net_bridge_port *backup_port; |
| 154 |
|
| 155 |
backup_port = rcu_dereference(to->backup_port); |
| 156 |
if (unlikely(!backup_port)) |
| 157 |
goto out; |
| 158 |
BR_INPUT_SKB_CB(skb)->backup_nhid = READ_ONCE(to->backup_nhid); |
| 159 |
to = backup_port; |
| 160 |
} |
| 161 |
|
| 162 |
if (should_deliver(to, skb)) { |
| 163 |
if (local_rcv) |
| 164 |
deliver_clone(to, skb, local_orig); |
| 165 |
else |
| 166 |
__br_forward(to, skb, local_orig); |
| 167 |
return; |
| 168 |
} |
| 169 |
|
| 170 |
out: |
| 171 |
if (!local_rcv) |
| 172 |
kfree_skb(skb); |
| 173 |
} |
| 174 |
EXPORT_SYMBOL_GPL(br_forward); |
| 175 |
|
| 176 |
static struct net_bridge_port *maybe_deliver( |
| 177 |
struct net_bridge_port *prev, struct net_bridge_port *p, |
| 178 |
struct sk_buff *skb, bool local_orig) |
| 179 |
{ |
| 180 |
u8 igmp_type = br_multicast_igmp_type(skb); |
| 181 |
int err; |
| 182 |
|
| 183 |
if (!should_deliver(p, skb)) |
| 184 |
return prev; |
| 185 |
|
| 186 |
nbp_switchdev_frame_mark_tx_fwd_to_hwdom(p, skb); |
| 187 |
|
| 188 |
if (!prev) |
| 189 |
goto out; |
| 190 |
|
| 191 |
err = deliver_clone(prev, skb, local_orig); |
| 192 |
if (err) |
| 193 |
return ERR_PTR(err); |
| 194 |
out: |
| 195 |
br_multicast_count(p->br, p, skb, igmp_type, BR_MCAST_DIR_TX); |
| 196 |
|
| 197 |
return p; |
| 198 |
} |
| 199 |
|
| 200 |
/* called under rcu_read_lock */ |
| 201 |
void br_flood(struct net_bridge *br, struct sk_buff *skb, |
| 202 |
enum br_pkt_type pkt_type, bool local_rcv, bool local_orig, |
| 203 |
u16 vid) |
| 204 |
{ |
| 205 |
enum skb_drop_reason reason = SKB_DROP_REASON_NO_TX_TARGET; |
| 206 |
struct net_bridge_port *prev = NULL; |
| 207 |
struct net_bridge_port *p; |
| 208 |
|
| 209 |
br_tc_skb_miss_set(skb, pkt_type != BR_PKT_BROADCAST); |
| 210 |
|
| 211 |
list_for_each_entry_rcu(p, &br->port_list, list) { |
| 212 |
/* Do not flood unicast traffic to ports that turn it off, nor |
| 213 |
* other traffic if flood off, except for traffic we originate |
| 214 |
*/ |
| 215 |
switch (pkt_type) { |
| 216 |
case BR_PKT_UNICAST: |
| 217 |
if (!(p->flags & BR_FLOOD)) |
| 218 |
continue; |
| 219 |
break; |
| 220 |
case BR_PKT_MULTICAST: |
| 221 |
if (!(p->flags & BR_MCAST_FLOOD) && skb->dev != br->dev) |
| 222 |
continue; |
| 223 |
break; |
| 224 |
case BR_PKT_BROADCAST: |
| 225 |
if (!(p->flags & BR_BCAST_FLOOD) && skb->dev != br->dev) |
| 226 |
continue; |
| 227 |
break; |
| 228 |
} |
| 229 |
|
| 230 |
/* Do not flood to ports that enable proxy ARP */ |
| 231 |
if (p->flags & BR_PROXYARP) |
| 232 |
continue; |
| 233 |
if (BR_INPUT_SKB_CB(skb)->proxyarp_replied && |
| 234 |
((p->flags & BR_PROXYARP_WIFI) || |
| 235 |
br_is_neigh_suppress_enabled(p, vid))) |
| 236 |
continue; |
| 237 |
|
| 238 |
prev = maybe_deliver(prev, p, skb, local_orig); |
| 239 |
if (IS_ERR(prev)) { |
| 240 |
reason = PTR_ERR(prev) == -ENOMEM ? SKB_DROP_REASON_NOMEM : |
| 241 |
SKB_DROP_REASON_NOT_SPECIFIED; |
| 242 |
goto out; |
| 243 |
} |
| 244 |
} |
| 245 |
|
| 246 |
if (!prev) |
| 247 |
goto out; |
| 248 |
|
| 249 |
if (local_rcv) |
| 250 |
deliver_clone(prev, skb, local_orig); |
| 251 |
else |
| 252 |
__br_forward(prev, skb, local_orig); |
| 253 |
return; |
| 254 |
|
| 255 |
out: |
| 256 |
if (!local_rcv) |
| 257 |
kfree_skb_reason(skb, reason); |
| 258 |
} |
| 259 |
|
| 260 |
#ifdef CONFIG_BRIDGE_IGMP_SNOOPING |
| 261 |
static void maybe_deliver_addr(struct net_bridge_port *p, struct sk_buff *skb, |
| 262 |
const unsigned char *addr, bool local_orig) |
| 263 |
{ |
| 264 |
struct net_device *dev = BR_INPUT_SKB_CB(skb)->brdev; |
| 265 |
const unsigned char *src = eth_hdr(skb)->h_source; |
| 266 |
struct sk_buff *nskb; |
| 267 |
|
| 268 |
if (!should_deliver(p, skb)) |
| 269 |
return; |
| 270 |
|
| 271 |
/* Even with hairpin, no soliloquies - prevent breaking IPv6 DAD */ |
| 272 |
if (skb->dev == p->dev && ether_addr_equal(src, addr)) |
| 273 |
return; |
| 274 |
|
| 275 |
__skb_push(skb, ETH_HLEN); |
| 276 |
nskb = pskb_copy(skb, GFP_ATOMIC); |
| 277 |
__skb_pull(skb, ETH_HLEN); |
| 278 |
if (!nskb) { |
| 279 |
DEV_STATS_INC(dev, tx_dropped); |
| 280 |
return; |
| 281 |
} |
| 282 |
|
| 283 |
skb = nskb; |
| 284 |
__skb_pull(skb, ETH_HLEN); |
| 285 |
if (!is_broadcast_ether_addr(addr)) |
| 286 |
memcpy(eth_hdr(skb)->h_dest, addr, ETH_ALEN); |
| 287 |
|
| 288 |
__br_forward(p, skb, local_orig); |
| 289 |
} |
| 290 |
|
| 291 |
/* called with rcu_read_lock */ |
| 292 |
void br_multicast_flood(struct net_bridge_mdb_entry *mdst, |
| 293 |
struct sk_buff *skb, |
| 294 |
struct net_bridge_mcast *brmctx, |
| 295 |
bool local_rcv, bool local_orig) |
| 296 |
{ |
| 297 |
enum skb_drop_reason reason = SKB_DROP_REASON_NO_TX_TARGET; |
| 298 |
struct net_bridge_port *prev = NULL; |
| 299 |
struct net_bridge_port_group *p; |
| 300 |
bool allow_mode_include = true; |
| 301 |
struct hlist_node *rp; |
| 302 |
|
| 303 |
rp = br_multicast_get_first_rport_node(brmctx, skb); |
| 304 |
|
| 305 |
if (mdst) { |
| 306 |
p = rcu_dereference(mdst->ports); |
| 307 |
if (br_multicast_should_handle_mode(brmctx, mdst->addr.proto) && |
| 308 |
br_multicast_is_star_g(&mdst->addr)) |
| 309 |
allow_mode_include = false; |
| 310 |
} else { |
| 311 |
p = NULL; |
| 312 |
br_tc_skb_miss_set(skb, true); |
| 313 |
} |
| 314 |
|
| 315 |
while (p || rp) { |
| 316 |
struct net_bridge_port *port, *lport, *rport; |
| 317 |
|
| 318 |
lport = p ? p->key.port : NULL; |
| 319 |
rport = br_multicast_rport_from_node_skb(rp, skb); |
| 320 |
|
| 321 |
if ((unsigned long)lport > (unsigned long)rport) { |
| 322 |
port = lport; |
| 323 |
|
| 324 |
if (port->flags & BR_MULTICAST_TO_UNICAST) { |
| 325 |
maybe_deliver_addr(lport, skb, p->eth_addr, |
| 326 |
local_orig); |
| 327 |
goto delivered; |
| 328 |
} |
| 329 |
if ((!allow_mode_include && |
| 330 |
p->filter_mode == MCAST_INCLUDE) || |
| 331 |
(p->flags & MDB_PG_FLAGS_BLOCKED)) |
| 332 |
goto delivered; |
| 333 |
} else { |
| 334 |
port = rport; |
| 335 |
} |
| 336 |
|
| 337 |
prev = maybe_deliver(prev, port, skb, local_orig); |
| 338 |
if (IS_ERR(prev)) { |
| 339 |
reason = PTR_ERR(prev) == -ENOMEM ? SKB_DROP_REASON_NOMEM : |
| 340 |
SKB_DROP_REASON_NOT_SPECIFIED; |
| 341 |
goto out; |
| 342 |
} |
| 343 |
delivered: |
| 344 |
if ((unsigned long)lport >= (unsigned long)port) |
| 345 |
p = rcu_dereference(p->next); |
| 346 |
if ((unsigned long)rport >= (unsigned long)port) |
| 347 |
rp = rcu_dereference(hlist_next_rcu(rp)); |
| 348 |
} |
| 349 |
|
| 350 |
if (!prev) |
| 351 |
goto out; |
| 352 |
|
| 353 |
if (local_rcv) |
| 354 |
deliver_clone(prev, skb, local_orig); |
| 355 |
else |
| 356 |
__br_forward(prev, skb, local_orig); |
| 357 |
return; |
| 358 |
|
| 359 |
out: |
| 360 |
if (!local_rcv) |
| 361 |
kfree_skb_reason(skb, reason); |
| 362 |
} |
| 363 |
#endif |