net/mlx5: Skip mlx5_unload_one if mlx5_load_one fails
authorHuy Nguyen <huyn@mellanox.com>
Tue, 8 Aug 2017 18:17:00 +0000 (13:17 -0500)
committerSaeed Mahameed <saeedm@mellanox.com>
Wed, 30 Aug 2017 18:20:43 +0000 (21:20 +0300)
commitb3cb5388499c5e219324bfe7da2e46cbad82bfcf
treea955b5290d59ec1e87ba1e7cebae113d69da139b
parent672d0880b7798a917bcc622308f25a0fbb991dab
net/mlx5: Skip mlx5_unload_one if mlx5_load_one fails

There is an issue where the firmware fails during mlx5_load_one,
the health_care timer detects the issue and schedules a health_care call.
Then the mlx5_load_one detects the issue, cleans up and quits. Then
the health_care starts and calls mlx5_unload_one to clean up the resources
that no longer exist and causes kernel panic.

The root cause is that the bit MLX5_INTERFACE_STATE_DOWN is not set
after mlx5_load_one fails. The solution is removing the bit
MLX5_INTERFACE_STATE_DOWN and quit mlx5_unload_one if the
bit MLX5_INTERFACE_STATE_UP is not set. The bit MLX5_INTERFACE_STATE_DOWN
is redundant and we can use MLX5_INTERFACE_STATE_UP instead.

Fixes: 5fc7197d3a25 ("net/mlx5: Add pci shutdown callback")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Reviewed-by: Daniel Jurgens <danielj@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
drivers/net/ethernet/mellanox/mlx5/core/main.c
include/linux/mlx5/driver.h