| b.liu | e958203 | 2025-04-17 19:18:16 +0800 | [diff] [blame] | 1 | From: Nikolay Aleksandrov <nikolay@cumulusnetworks.com> |
| 2 | Date: Fri, 31 Jul 2020 19:26:16 +0300 |
| 3 | Subject: [PATCH] net: bridge: clear bridge's private skb space on xmit |
| 4 | |
| 5 | We need to clear all of the bridge private skb variables as they can be |
| 6 | stale due to the packet being recirculated through the stack and then |
| 7 | transmitted through the bridge device. Similar memset is already done on |
| 8 | bridge's input. We've seen cases where proxyarp_replied was 1 on routed |
| 9 | multicast packets transmitted through the bridge to ports with neigh |
| 10 | suppress which were getting dropped. Same thing can in theory happen with |
| 11 | the port isolation bit as well. |
| 12 | |
| 13 | Fixes: 821f1b21cabb ("bridge: add new BR_NEIGH_SUPPRESS port flag to suppress arp and nd flood") |
| 14 | Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com> |
| 15 | Signed-off-by: David S. Miller <davem@davemloft.net> |
| 16 | --- |
| 17 | |
| 18 | --- a/net/bridge/br_device.c |
| 19 | +++ b/net/bridge/br_device.c |
| 20 | @@ -35,6 +35,8 @@ netdev_tx_t br_dev_xmit(struct sk_buff * |
| 21 | const unsigned char *dest; |
| 22 | u16 vid = 0; |
| 23 | |
| 24 | + memset(skb->cb, 0, sizeof(struct br_input_skb_cb)); |
| 25 | + |
| 26 | rcu_read_lock(); |
| 27 | nf_ops = rcu_dereference(nf_br_ops); |
| 28 | if (nf_ops && nf_ops->br_dev_xmit_hook(skb)) { |