Merge tag 'net-6.10-rc0' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Current release - regressions:

   - virtio_net: fix missed error path rtnl_unlock after control queue
     locking rework

  Current release - new code bugs:

   - bpf: fix KASAN slab-out-of-bounds in percpu_array_map_gen_lookup,
     caused by missing nested map handling

   - drv: dsa: correct initialization order for KSZ88x3 ports

  Previous releases - regressions:

   - af_packet: do not call packet_read_pending() from
     tpacket_destruct_skb() fix performance regression

   - ipv6: fix route deleting failure when metric equals 0, don't assume
     0 means not set / default in this case

  Previous releases - always broken:

   - bridge: couple of syzbot-driven fixes"

* tag 'net-6.10-rc0' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (30 commits)
  selftests: net: local_termination: annotate the expected failures
  net: dsa: microchip: Correct initialization order for KSZ88x3 ports
  MAINTAINERS: net: Update reviewers for TI's Ethernet drivers
  dt-bindings: net: ti: Update maintainers list
  l2tp: fix ICMP error handling for UDP-encap sockets
  net: txgbe: fix to control VLAN strip
  net: wangxun: match VLAN CTAG and STAG features
  net: wangxun: fix to change Rx features
  af_packet: do not call packet_read_pending() from tpacket_destruct_skb()
  virtio_net: Fix missed rtnl_unlock
  netrom: fix possible dead-lock in nr_rt_ioctl()
  idpf: don't skip over ethtool tcp-data-split setting
  dt-bindings: net: qcom: ethernet: Allow dma-coherent
  bonding: fix oops during rmmod
  net/ipv6: Fix route deleting failure when metric equals 0
  selftests/net: reduce xfrm_policy test time
  selftests/bpf: Adjust btf_dump test to reflect recent change in file_operations
  selftests/bpf: Adjust test_access_variable_array after a kernel function name change
  selftests/net/lib: no need to record ns name if it already exist
  net: qrtr: ns: Fix module refcnt
  ...

7 files changed, 88 insertions, 32 deletions

diff --git a/net/bridge/br_device.c b/net/bridge/br_device.c
index 71b50001ca58..fb1115857e49 100644
--- a/net/bridge/br_device.c
+++ b/net/bridge/br_device.c
@@ -27,6 +27,7 @@ EXPORT_SYMBOL_GPL(nf_br_ops);
 /* net device transmit always called with BH disabled */
 netdev_tx_t br_dev_xmit(struct sk_buff *skb, struct net_device *dev)
 {
+	enum skb_drop_reason reason = pskb_may_pull_reason(skb, ETH_HLEN);
 	struct net_bridge_mcast_port *pmctx_null = NULL;
 	struct net_bridge *br = netdev_priv(dev);
 	struct net_bridge_mcast *brmctx = &br->multicast_ctx;
@@ -38,6 +39,11 @@ netdev_tx_t br_dev_xmit(struct sk_buff *skb, struct net_device *dev)
 	const unsigned char *dest;
 	u16 vid = 0;
 
+	if (unlikely(reason != SKB_NOT_DROPPED_YET)) {
+		kfree_skb_reason(skb, reason);
+		return NETDEV_TX_OK;
+	}
+
 	memset(skb->cb, 0, sizeof(struct br_input_skb_cb));
 	br_tc_skb_miss_set(skb, false);
 
diff --git a/net/bridge/br_mst.c b/net/bridge/br_mst.c
index ee680adcee17..3c66141d34d6 100644
--- a/net/bridge/br_mst.c
+++ b/net/bridge/br_mst.c
@@ -78,7 +78,7 @@ static void br_mst_vlan_set_state(struct net_bridge_port *p, struct net_bridge_v
 {
 	struct net_bridge_vlan_group *vg = nbp_vlan_group(p);
 
-	if (v->state == state)
+	if (br_vlan_get_state(v) == state)
 		return;
 
 	br_vlan_set_state(v, state);
@@ -100,11 +100,12 @@ int br_mst_set_state(struct net_bridge_port *p, u16 msti, u8 state,
 	};
 	struct net_bridge_vlan_group *vg;
 	struct net_bridge_vlan *v;
-	int err;
+	int err = 0;
 
+	rcu_read_lock();
 	vg = nbp_vlan_group(p);
 	if (!vg)
-		return 0;
+		goto out;
 
 	/* MSTI 0 (CST) state changes are notified via the regular
 	 * SWITCHDEV_ATTR_ID_PORT_STP_STATE.
@@ -112,17 +113,20 @@ int br_mst_set_state(struct net_bridge_port *p, u16 msti, u8 state,
 	if (msti) {
 		err = switchdev_port_attr_set(p->dev, &attr, extack);
 		if (err && err != -EOPNOTSUPP)
-			return err;
+			goto out;
 	}
 
-	list_for_each_entry(v, &vg->vlan_list, vlist) {
+	err = 0;
+	list_for_each_entry_rcu(v, &vg->vlan_list, vlist) {
 		if (v->brvlan->msti != msti)
 			continue;
 
 		br_mst_vlan_set_state(p, v, state);
 	}
 
-	return 0;
+out:
+	rcu_read_unlock();
+	return err;
 }
 
 static void br_mst_vlan_sync_state(struct net_bridge_vlan *pv, u16 msti)
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index c43b0616742e..bbc2a0dd9314 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -4445,7 +4445,7 @@ static void rtmsg_to_fib6_config(struct net *net,
 		.fc_table = l3mdev_fib_table_by_index(net, rtmsg->rtmsg_ifindex) ?
 			 : RT6_TABLE_MAIN,
 		.fc_ifindex = rtmsg->rtmsg_ifindex,
-		.fc_metric = rtmsg->rtmsg_metric ? : IP6_RT_PRIO_USER,
+		.fc_metric = rtmsg->rtmsg_metric,
 		.fc_expires = rtmsg->rtmsg_info,
 		.fc_dst_len = rtmsg->rtmsg_dst_len,
 		.fc_src_len = rtmsg->rtmsg_src_len,
@@ -4475,6 +4475,9 @@ int ipv6_route_ioctl(struct net *net, unsigned int cmd, struct in6_rtmsg *rtmsg)
 	rtnl_lock();
 	switch (cmd) {
 	case SIOCADDRT:
+		/* Only do the default setting of fc_metric in route adding */
+		if (cfg.fc_metric == 0)
+			cfg.fc_metric = IP6_RT_PRIO_USER;
 		err = ip6_route_add(&cfg, GFP_KERNEL, NULL);
 		break;
 	case SIOCDELRT:
diff --git a/net/l2tp/l2tp_core.c b/net/l2tp/l2tp_core.c
index 7d519a46a844..88a34db265d8 100644
--- a/net/l2tp/l2tp_core.c
+++ b/net/l2tp/l2tp_core.c
@@ -910,22 +910,20 @@ pass:
 	return 1;
 }
 
-/* UDP encapsulation receive handler. See net/ipv4/udp.c.
- * Return codes:
- * 0 : success.
- * <0: error
- * >0: skb should be passed up to userspace as UDP.
+/* UDP encapsulation receive and error receive handlers.
+ * See net/ipv4/udp.c for details.
+ *
+ * Note that these functions are called from inside an
+ * RCU-protected region, but without the socket being locked.
+ *
+ * Hence we use rcu_dereference_sk_user_data to access the
+ * tunnel data structure rather the usual l2tp_sk_to_tunnel
+ * accessor function.
  */
 int l2tp_udp_encap_recv(struct sock *sk, struct sk_buff *skb)
 {
 	struct l2tp_tunnel *tunnel;
 
-	/* Note that this is called from the encap_rcv hook inside an
-	 * RCU-protected region, but without the socket being locked.
-	 * Hence we use rcu_dereference_sk_user_data to access the
-	 * tunnel data structure rather the usual l2tp_sk_to_tunnel
-	 * accessor function.
-	 */
 	tunnel = rcu_dereference_sk_user_data(sk);
 	if (!tunnel)
 		goto pass_up;
@@ -942,6 +940,29 @@ pass_up:
 }
 EXPORT_SYMBOL_GPL(l2tp_udp_encap_recv);
 
+static void l2tp_udp_encap_err_recv(struct sock *sk, struct sk_buff *skb, int err,
+				    __be16 port, u32 info, u8 *payload)
+{
+	struct l2tp_tunnel *tunnel;
+
+	tunnel = rcu_dereference_sk_user_data(sk);
+	if (!tunnel || tunnel->fd < 0)
+		return;
+
+	sk->sk_err = err;
+	sk_error_report(sk);
+
+	if (ip_hdr(skb)->version == IPVERSION) {
+		if (inet_test_bit(RECVERR, sk))
+			return ip_icmp_error(sk, skb, err, port, info, payload);
+#if IS_ENABLED(CONFIG_IPV6)
+	} else {
+		if (inet6_test_bit(RECVERR6, sk))
+			return ipv6_icmp_error(sk, skb, err, port, info, payload);
+#endif
+	}
+}
+
 /************************************************************************
  * Transmit handling
  ***********************************************************************/
@@ -1516,6 +1537,7 @@ int l2tp_tunnel_register(struct l2tp_tunnel *tunnel, struct net *net,
 			.sk_user_data = tunnel,
 			.encap_type = UDP_ENCAP_L2TPINUDP,
 			.encap_rcv = l2tp_udp_encap_recv,
+			.encap_err_rcv = l2tp_udp_encap_err_recv,
 			.encap_destroy = l2tp_udp_encap_destroy,
 		};
 
diff --git a/net/netrom/nr_route.c b/net/netrom/nr_route.c
index 70480869ad1c..bd2b17b219ae 100644
--- a/net/netrom/nr_route.c
+++ b/net/netrom/nr_route.c
@@ -285,22 +285,14 @@ static int __must_check nr_add_node(ax25_address *nr, const char *mnemonic,
 	return 0;
 }
 
-static inline void __nr_remove_node(struct nr_node *nr_node)
+static void nr_remove_node_locked(struct nr_node *nr_node)
 {
+	lockdep_assert_held(&nr_node_list_lock);
+
 	hlist_del_init(&nr_node->node_node);
 	nr_node_put(nr_node);
 }
 
-#define nr_remove_node_locked(__node) \
-	__nr_remove_node(__node)
-
-static void nr_remove_node(struct nr_node *nr_node)
-{
-	spin_lock_bh(&nr_node_list_lock);
-	__nr_remove_node(nr_node);
-	spin_unlock_bh(&nr_node_list_lock);
-}
-
 static inline void __nr_remove_neigh(struct nr_neigh *nr_neigh)
 {
 	hlist_del_init(&nr_neigh->neigh_node);
@@ -339,6 +331,7 @@ static int nr_del_node(ax25_address *callsign, ax25_address *neighbour, struct n
 		return -EINVAL;
 	}
 
+	spin_lock_bh(&nr_node_list_lock);
 	nr_node_lock(nr_node);
 	for (i = 0; i < nr_node->count; i++) {
 		if (nr_node->routes[i].neighbour == nr_neigh) {
@@ -352,7 +345,7 @@ static int nr_del_node(ax25_address *callsign, ax25_address *neighbour, struct n
 			nr_node->count--;
 
 			if (nr_node->count == 0) {
-				nr_remove_node(nr_node);
+				nr_remove_node_locked(nr_node);
 			} else {
 				switch (i) {
 				case 0:
@@ -367,12 +360,14 @@ static int nr_del_node(ax25_address *callsign, ax25_address *neighbour, struct n
 				nr_node_put(nr_node);
 			}
 			nr_node_unlock(nr_node);
+			spin_unlock_bh(&nr_node_list_lock);
 
 			return 0;
 		}
 	}
 	nr_neigh_put(nr_neigh);
 	nr_node_unlock(nr_node);
+	spin_unlock_bh(&nr_node_list_lock);
 	nr_node_put(nr_node);
 
 	return -EINVAL;
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index 8c6d3fbb4ed8..ea3ebc160e25 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -2522,8 +2522,7 @@ static void tpacket_destruct_skb(struct sk_buff *skb)
 		ts = __packet_set_timestamp(po, ph, skb);
 		__packet_set_status(po, ph, TP_STATUS_AVAILABLE | ts);
 
-		if (!packet_read_pending(&po->tx_ring))
-			complete(&po->skb_completion);
+		complete(&po->skb_completion);
 	}
 
 	sock_wfree(skb);
diff --git a/net/qrtr/ns.c b/net/qrtr/ns.c
index abb0c70ffc8b..654a3cc0d347 100644
--- a/net/qrtr/ns.c
+++ b/net/qrtr/ns.c
@@ -725,6 +725,24 @@ int qrtr_ns_init(void)
 	if (ret < 0)
 		goto err_wq;
 
+	/* As the qrtr ns socket owner and creator is the same module, we have
+	 * to decrease the qrtr module reference count to guarantee that it
+	 * remains zero after the ns socket is created, otherwise, executing
+	 * "rmmod" command is unable to make the qrtr module deleted after the
+	 *  qrtr module is inserted successfully.
+	 *
+	 * However, the reference count is increased twice in
+	 * sock_create_kern(): one is to increase the reference count of owner
+	 * of qrtr socket's proto_ops struct; another is to increment the
+	 * reference count of owner of qrtr proto struct. Therefore, we must
+	 * decrement the module reference count twice to ensure that it keeps
+	 * zero after server's listening socket is created. Of course, we
+	 * must bump the module reference count twice as well before the socket
+	 * is closed.
+	 */
+	module_put(qrtr_ns.sock->ops->owner);
+	module_put(qrtr_ns.sock->sk->sk_prot_creator->owner);
+
 	return 0;
 
 err_wq:
@@ -739,6 +757,15 @@ void qrtr_ns_remove(void)
 {
 	cancel_work_sync(&qrtr_ns.work);
 	destroy_workqueue(qrtr_ns.workqueue);
+
+	/* sock_release() expects the two references that were put during
+	 * qrtr_ns_init(). This function is only called during module remove,
+	 * so try_stop_module() has already set the refcnt to 0. Use
+	 * __module_get() instead of try_module_get() to successfully take two
+	 * references.
+	 */
+	__module_get(qrtr_ns.sock->ops->owner);
+	__module_get(qrtr_ns.sock->sk->sk_prot_creator->owner);
 	sock_release(qrtr_ns.sock);
 }
 EXPORT_SYMBOL_GPL(qrtr_ns_remove);