6.12-stable review patch. If anyone has any objections, please let me know.
------------------
From: Ido Schimmel idosch@nvidia.com
[ Upstream commit 804b09be09f8af4eda5346a72361459ba21fcf1b ]
The Tx path does not run from an RCU read-side critical section which makes the current lockless accesses to FDB entries invalid. As far as I am aware, this has not been a problem in practice, but traces will be generated once we transition the FDB lookup to rhashtable_lookup().
Add rcu_read_{lock,unlock}() around the handling of FDB entries in the Tx path. Remove the RCU read-side critical section from vxlan_xmit_nh() as now the function is always called from an RCU read-side critical section.
Reviewed-by: Petr Machata petrm@nvidia.com Signed-off-by: Ido Schimmel idosch@nvidia.com Link: https://patch.msgid.link/20250415121143.345227-2-idosch@nvidia.com Reviewed-by: Nikolay Aleksandrov razor@blackwall.org Signed-off-by: Paolo Abeni pabeni@redhat.com Stable-dep-of: 1f5d2fd1ca04 ("vxlan: Fix NPD in {arp,neigh}_reduce() when using nexthop objects") Signed-off-by: Sasha Levin sashal@kernel.org --- drivers/net/vxlan/vxlan_core.c | 14 ++++++++------ 1 file changed, 8 insertions(+), 6 deletions(-)
diff --git a/drivers/net/vxlan/vxlan_core.c b/drivers/net/vxlan/vxlan_core.c index 8853fcb7eb7f2..1d431e3fc71ea 100644 --- a/drivers/net/vxlan/vxlan_core.c +++ b/drivers/net/vxlan/vxlan_core.c @@ -1909,12 +1909,15 @@ static int arp_reduce(struct net_device *dev, struct sk_buff *skb, __be32 vni) goto out; }
+ rcu_read_lock(); f = vxlan_find_mac(vxlan, n->ha, vni); if (f && vxlan_addr_any(&(first_remote_rcu(f)->remote_ip))) { /* bridge-local neighbor */ neigh_release(n); + rcu_read_unlock(); goto out; } + rcu_read_unlock();
reply = arp_create(ARPOP_REPLY, ETH_P_ARP, sip, dev, tip, sha, n->ha, sha); @@ -2632,14 +2635,10 @@ static void vxlan_xmit_nh(struct sk_buff *skb, struct net_device *dev, memset(&nh_rdst, 0, sizeof(struct vxlan_rdst)); hash = skb_get_hash(skb);
- rcu_read_lock(); nh = rcu_dereference(f->nh); - if (!nh) { - rcu_read_unlock(); + if (!nh) goto drop; - } do_xmit = vxlan_fdb_nh_path_select(nh, hash, &nh_rdst); - rcu_read_unlock();
if (likely(do_xmit)) vxlan_xmit_one(skb, dev, vni, &nh_rdst, did_rsc); @@ -2766,6 +2765,7 @@ static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev) }
eth = eth_hdr(skb); + rcu_read_lock(); f = vxlan_find_mac(vxlan, eth->h_dest, vni); did_rsc = false;
@@ -2788,7 +2788,7 @@ static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev) vxlan_vnifilter_count(vxlan, vni, NULL, VXLAN_VNI_STATS_TX_DROPS, 0); kfree_skb_reason(skb, SKB_DROP_REASON_NO_TX_TARGET); - return NETDEV_TX_OK; + goto out; } }
@@ -2813,6 +2813,8 @@ static netdev_tx_t vxlan_xmit(struct sk_buff *skb, struct net_device *dev) kfree_skb_reason(skb, SKB_DROP_REASON_NO_TX_TARGET); }
+out: + rcu_read_unlock(); return NETDEV_TX_OK; }