net/mlx5: Avoid reloading already removed devices

This change “net/mlx5: Avoid reloading already removed devices” (commit 4642a65) in Linux kernel is authored by Alaa Hleihel <alaa [at] mellanox.com> on Sun May 19 11:11:49 2019 +0300.

Description of "net/mlx5: Avoid reloading already removed devices"

The change “net/mlx5: Avoid reloading already removed devices” introduces changes as follows.

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>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Linux kernel releases containing commit 4642a65

The Linux kernel releases containing this commit are as follows.

Linux kernel code changes from "net/mlx5: Avoid reloading already removed devices"

There are 25 lines of Linux source code added/deleted in this change. Code changes to Linux kernel are as follows.

 drivers/net/ethernet/mellanox/mlx5/core/dev.c | 25 +++++++++++++++++--
 1 file 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 37ba7c78859d..1c225be9c7db 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/dev.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c
@@ -342,11 +342,32 @@ 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);
 }
 

The commit for this change in Linux stable tree is 4642a65 (patch).

Last modified: 2020/02/08