net/mlx5: Avoid reloading already removed devices

Prior to reloading a device we must first verify that it was not already
removed. Otherwise, the attempt to remove the device will do nothing, and
in that case we will end up proceeding with adding an new device that no
one was expecting to remove, leaving behind used resources such as EQs that
causes a failure to destroy comp EQs and syndrome (0x30f433).

Fix that by making sure that we try to remove and add a device (based on a
protocol) only if the device is already added.

Fixes: c5447c70594b ("net/mlx5: E-Switch, Reload IB interface when switching devlink modes")
Signed-off-by: Alaa Hleihel <alaa@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>

1 files changed, 23 insertions, 2 deletions

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
index ebc046fa97d3..f6b1da99e6c2 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/dev.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
@@ -248,11 +248,32 @@ void mlx5_unregister_interface(struct mlx5_interface *intf)
 }
 EXPORT_SYMBOL(mlx5_unregister_interface);
 
+/* Must be called with intf_mutex held */
+static bool mlx5_has_added_dev_by_protocol(struct mlx5_core_dev *mdev, int protocol)
+{
+	struct mlx5_device_context *dev_ctx;
+	struct mlx5_interface *intf;
+	bool found = false;
+
+	list_for_each_entry(intf, &intf_list, list) {
+		if (intf->protocol == protocol) {
+			dev_ctx = mlx5_get_device(intf, &mdev->priv);
+			if (dev_ctx && test_bit(MLX5_INTERFACE_ADDED, &dev_ctx->state))
+				found = true;
+			break;
+		}
+	}
+
+	return found;
+}
+
 void mlx5_reload_interface(struct mlx5_core_dev *mdev, int protocol)
 {
 	mutex_lock(&mlx5_intf_mutex);
-	mlx5_remove_dev_by_protocol(mdev, protocol);
-	mlx5_add_dev_by_protocol(mdev, protocol);
+	if (mlx5_has_added_dev_by_protocol(mdev, protocol)) {
+		mlx5_remove_dev_by_protocol(mdev, protocol);
+		mlx5_add_dev_by_protocol(mdev, protocol);
+	}
 	mutex_unlock(&mlx5_intf_mutex);
 }