net/mlx5e: Fix nullptr in mlx5e_hairpin_get_mdev()

[ Upstream commit b1c2f6312c ]

The result of __dev_get_by_index() is not checked for NULL and then gets
dereferenced immediately.

Also, __dev_get_by_index() must be called while holding either RTNL lock
or @dev_base_lock, which isn't satisfied by mlx5e_hairpin_get_mdev() or
its callers. This makes the underlying hlist_for_each_entry() loop not
safe, and can have adverse effects in itself.

Fix by using dev_get_by_index() and handling nullptr return value when
ifindex device is not found. Update mlx5e_hairpin_get_mdev() callers to
check for possible PTR_ERR() result.

Fixes: 77ab67b7f0 ("net/mlx5e: Basic setup of hairpin object")
Addresses-Coverity: ("Dereference null return value")
Signed-off-by: Dima Chumak <dchumak@nvidia.com>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/net/ethernet/mellanox/mlx5/core/en_tc.c

@@ -481,12 +481,32 @@ static void mlx5e_detach_mod_hdr(struct mlx5e_priv *priv,
 static
 struct mlx5_core_dev *mlx5e_hairpin_get_mdev(struct net *net, int ifindex)
 {
+	struct mlx5_core_dev *mdev;
 	struct net_device *netdev;
 	struct mlx5e_priv *priv;
 
-	netdev = __dev_get_by_index(net, ifindex);
+	netdev = dev_get_by_index(net, ifindex);
+	if (!netdev)
+		return ERR_PTR(-ENODEV);
+
 	priv = netdev_priv(netdev);
-	return priv->mdev;
+	mdev = priv->mdev;
+	dev_put(netdev);
+
+	/* Mirred tc action holds a refcount on the ifindex net_device (see
+	 * net/sched/act_mirred.c:tcf_mirred_get_dev). So, it's okay to continue using mdev
+	 * after dev_put(netdev), while we're in the context of adding a tc flow.
+	 *
+	 * The mdev pointer corresponds to the peer/out net_device of a hairpin. It is then
+	 * stored in a hairpin object, which exists until all flows, that refer to it, get
+	 * removed.
+	 *
+	 * On the other hand, after a hairpin object has been created, the peer net_device may
+	 * be removed/unbound while there are still some hairpin flows that are using it. This
+	 * case is handled by mlx5e_tc_hairpin_update_dead_peer, which is hooked to
+	 * NETDEV_UNREGISTER event of the peer net_device.
+	 */
+	return mdev;
 }
 
 static int mlx5e_hairpin_create_transport(struct mlx5e_hairpin *hp)
@@ -685,6 +705,10 @@ mlx5e_hairpin_create(struct mlx5e_priv *priv, struct mlx5_hairpin_params *params
 
 	func_mdev = priv->mdev;
 	peer_mdev = mlx5e_hairpin_get_mdev(dev_net(priv->netdev), peer_ifindex);
+	if (IS_ERR(peer_mdev)) {
+		err = PTR_ERR(peer_mdev);
+		goto create_pair_err;
+	}
 
 	pair = mlx5_core_hairpin_create(func_mdev, peer_mdev, params);
 	if (IS_ERR(pair)) {
@@ -823,6 +847,11 @@ static int mlx5e_hairpin_flow_add(struct mlx5e_priv *priv,
 	int err;
 
 	peer_mdev = mlx5e_hairpin_get_mdev(dev_net(priv->netdev), peer_ifindex);
+	if (IS_ERR(peer_mdev)) {
+		NL_SET_ERR_MSG_MOD(extack, "invalid ifindex of mirred device");
+		return PTR_ERR(peer_mdev);
+	}
+
 	if (!MLX5_CAP_GEN(priv->mdev, hairpin) || !MLX5_CAP_GEN(peer_mdev, hairpin)) {
 		NL_SET_ERR_MSG_MOD(extack, "hairpin is not supported");
 		return -EOPNOTSUPP;