Currently, remove and reload flows can run in parallel to module cleanup.
This design is error prone. For example: aux_drivers callbacks are called
from both cleanup and remove flows with different lockings, which can
cause a deadlock[1].
Hence, serialize module cleanup with reload and remove.
[1]
cleanup remove
------- ------
auxiliary_driver_unregister();
devl_lock()
auxiliary_device_delete(mlx5e_aux)
device_lock(mlx5e_aux)
devl_lock()
device_lock(mlx5e_aux)
Fixes: 912cebf420c2 ("net/mlx5e: Connect ethernet part to auxiliary bus")
Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
mlx5_core_verify_params();
mlx5_register_debugfs();
- err = pci_register_driver(&mlx5_core_driver);
+ err = mlx5e_init();
if (err)
goto err_debug;
if (err)
goto err_sf;
- err = mlx5e_init();
+ err = pci_register_driver(&mlx5_core_driver);
if (err)
- goto err_en;
+ goto err_pci;
return 0;
-err_en:
+err_pci:
mlx5_sf_driver_unregister();
err_sf:
- pci_unregister_driver(&mlx5_core_driver);
+ mlx5e_cleanup();
err_debug:
mlx5_unregister_debugfs();
return err;
static void __exit mlx5_cleanup(void)
{
- mlx5e_cleanup();
- mlx5_sf_driver_unregister();
pci_unregister_driver(&mlx5_core_driver);
+ mlx5_sf_driver_unregister();
+ mlx5e_cleanup();
mlx5_unregister_debugfs();
}