amd64/180791: Kernel crash on ifdown and kldunload mlxen

[Shahar Klein]

>Number:         180791
>Category:       amd64
>Synopsis:       Kernel crash on ifdown and kldunload mlxen
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-amd64
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Wed Jul 24 07:30:00 UTC 2013
>Closed-Date:
>Last-Modified:
>Originator:     Shahar Klein
>Release:        9.1
>Organization:
Mellanox
>Environment:
FreeBSD 9.1-RELEASE #0 r+95f2241: Mon Jul 15 16:48:47 IDT 2013   .../sys/GENERIC  amd64
>Description:
Two related problems:

1. ifdown on an eth interface(using mlxen) panics the kernel.
2. kldunload mlxen when the interface is up - crashes the kernel
>How-To-Repeat:
1. bring a Mellanox HCA up and try to ifdown it.
2. kldunload mlxen when an eth interface is up(using mlxen)
>Fix:
### first fix ####

<pre>
diff --git a/sys/ofed/drivers/net/mlx4/en_netdev.c b/sys/ofed/drivers/net/mlx4/en_netdev.c
index acb2c12..157f05d 100755
--- a/sys/ofed/drivers/net/mlx4/en_netdev.c
+++ b/sys/ofed/drivers/net/mlx4/en_netdev.c
@@ -495,11 +495,6 @@ static void mlx4_en_do_get_stats(struct work_struct *work)

                queue_delayed_work(mdev->workqueue, &priv->stats_task, STATS_DELAY);
        }
-   if (mdev->mac_removed[MLX4_MAX_PORTS + 1 - priv->port]) {
-           panic("mlx4_en_do_get_stats: Unexpected mac removed for %d\n",
-               priv->port);
-           mdev->mac_removed[MLX4_MAX_PORTS + 1 - priv->port] = 0;
-   }
        mutex_unlock(&mdev->state_lock);
 }
</pre>



### second fix ###

<pre>
diff --git a/sys/ofed/drivers/net/mlx4/en_netdev.c b/sys/ofed/drivers/net/mlx4/en_netdev.c
index 260952b..807858a 100755
--- a/sys/ofed/drivers/net/mlx4/en_netdev.c
+++ b/sys/ofed/drivers/net/mlx4/en_netdev.c
@@ -921,6 +921,8 @@ void mlx4_en_destroy_netdev(struct net_device *dev)
        if (priv->sysctl)
                sysctl_ctx_free(&priv->conf_ctx);

+        mlx4_en_stop_port(dev);
+
        cancel_delayed_work(&priv->stats_task);
        /* flush any pending task for this netdev */
        flush_workqueue(mdev->workqueue);
</pre>

>Release-Note:
>Audit-Trail:
>Unformatted: