vxlan: fix hlist corruption

It's not a good idea to add the same hlist_node to two different hash lists.
This leads to various hard to debug memory corruptions.

Fixes: b1be00a6c39f ("vxlan: support both IPv4 and IPv6 sockets in a single vxlan device")
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

2 files changed, 30 insertions, 12 deletions

diff --git a/drivers/net/vxlan.c b/drivers/net/vxlan.c
index 47d6e65851aa16..b04e103350fb7b 100644
--- a/drivers/net/vxlan.c
+++ b/drivers/net/vxlan.c
@@ -229,25 +229,25 @@ static struct vxlan_sock *vxlan_find_sock(struct net *net, sa_family_t family,
 static struct vxlan_dev *vxlan_vs_find_vni(struct vxlan_sock *vs, int ifindex,
 					   __be32 vni)
 {
-	struct vxlan_dev *vxlan;
+	struct vxlan_dev_node *node;
 
 	/* For flow based devices, map all packets to VNI 0 */
 	if (vs->flags & VXLAN_F_COLLECT_METADATA)
 		vni = 0;
 
-	hlist_for_each_entry_rcu(vxlan, vni_head(vs, vni), hlist) {
-		if (vxlan->default_dst.remote_vni != vni)
+	hlist_for_each_entry_rcu(node, vni_head(vs, vni), hlist) {
+		if (node->vxlan->default_dst.remote_vni != vni)
 			continue;
 
 		if (IS_ENABLED(CONFIG_IPV6)) {
-			const struct vxlan_config *cfg = &vxlan->cfg;
+			const struct vxlan_config *cfg = &node->vxlan->cfg;
 
 			if ((cfg->flags & VXLAN_F_IPV6_LINKLOCAL) &&
 			    cfg->remote_ifindex != ifindex)
 				continue;
 		}
 
-		return vxlan;
+		return node->vxlan;
 	}
 
 	return NULL;
@@ -2387,17 +2387,22 @@ static void vxlan_vs_del_dev(struct vxlan_dev *vxlan)
 	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
 
 	spin_lock(&vn->sock_lock);
-	hlist_del_init_rcu(&vxlan->hlist);
+	hlist_del_init_rcu(&vxlan->hlist4.hlist);
+#if IS_ENABLED(CONFIG_IPV6)
+	hlist_del_init_rcu(&vxlan->hlist6.hlist);
+#endif
 	spin_unlock(&vn->sock_lock);
 }
 
-static void vxlan_vs_add_dev(struct vxlan_sock *vs, struct vxlan_dev *vxlan)
+static void vxlan_vs_add_dev(struct vxlan_sock *vs, struct vxlan_dev *vxlan,
+			     struct vxlan_dev_node *node)
 {
 	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
 	__be32 vni = vxlan->default_dst.remote_vni;
 
+	node->vxlan = vxlan;
 	spin_lock(&vn->sock_lock);
-	hlist_add_head_rcu(&vxlan->hlist, vni_head(vs, vni));
+	hlist_add_head_rcu(&node->hlist, vni_head(vs, vni));
 	spin_unlock(&vn->sock_lock);
 }
 
@@ -2849,6 +2854,7 @@ static int __vxlan_sock_add(struct vxlan_dev *vxlan, bool ipv6)
 {
 	struct vxlan_net *vn = net_generic(vxlan->net, vxlan_net_id);
 	struct vxlan_sock *vs = NULL;
+	struct vxlan_dev_node *node;
 
 	if (!vxlan->cfg.no_share) {
 		spin_lock(&vn->sock_lock);
@@ -2866,12 +2872,16 @@ static int __vxlan_sock_add(struct vxlan_dev *vxlan, bool ipv6)
 	if (IS_ERR(vs))
 		return PTR_ERR(vs);
 #if IS_ENABLED(CONFIG_IPV6)
-	if (ipv6)
+	if (ipv6) {
 		rcu_assign_pointer(vxlan->vn6_sock, vs);
-	else
+		node = &vxlan->hlist6;
+	} else
 #endif
+	{
 		rcu_assign_pointer(vxlan->vn4_sock, vs);
-	vxlan_vs_add_dev(vs, vxlan);
+		node = &vxlan->hlist4;
+	}
+	vxlan_vs_add_dev(vs, vxlan, node);
 	return 0;
 }
 
diff --git a/include/net/vxlan.h b/include/net/vxlan.h
index b816a0a6686e2b..326e8498b10e0e 100644
--- a/include/net/vxlan.h
+++ b/include/net/vxlan.h
@@ -221,9 +221,17 @@ struct vxlan_config {
 	bool			no_share;
 };
 
+struct vxlan_dev_node {
+	struct hlist_node hlist;
+	struct vxlan_dev *vxlan;
+};
+
 /* Pseudo network device */
 struct vxlan_dev {
-	struct hlist_node hlist;	/* vni hash table */
+	struct vxlan_dev_node hlist4;	/* vni hash table for IPv4 socket */
+#if IS_ENABLED(CONFIG_IPV6)
+	struct vxlan_dev_node hlist6;	/* vni hash table for IPv6 socket */
+#endif
 	struct list_head  next;		/* vxlan's per namespace list */
 	struct vxlan_sock __rcu *vn4_sock;	/* listening socket for IPv4 */
 #if IS_ENABLED(CONFIG_IPV6)