[ofa-general] ***SPAM*** Re: [PATCH v2] ipoib: fix hang while bringing down uninitialized interface

[Yossi Etigin]

Looks OK to me.

Roland Dreier wrote:
>  > Fix bug #1172: If a pkey for an interface is not found during
>  > initialization, then poll_timer is left uninitialized. When the
>  > device is brought down, ipoib tries to del_timer_sync() it. This
>  > call hangs in an infinite loop in lock_timer_base(), because
>  > timer_base is NULL. We should check whether the timer was really
>  > initialized.
> 
> Sorry for being so slow to get to this.
> 
> But does it work to just make sure the timer is always initialized?
> Seems cleaner that way, and it makes the code an insignificant bit
> smaller as a bonus.  ie does the patch below fix things too?
> 
>  drivers/infiniband/ulp/ipoib/ipoib_ib.c |    7 +++----
>  1 files changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
> index 0e748ae..28eb6f0 100644
> --- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
> +++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
> @@ -685,10 +685,6 @@ int ipoib_ib_dev_open(struct net_device *dev)
>  	queue_delayed_work(ipoib_workqueue, &priv->ah_reap_task,
>  			   round_jiffies_relative(HZ));
>  
> -	init_timer(&priv->poll_timer);
> -	priv->poll_timer.function = ipoib_ib_tx_timer_func;
> -	priv->poll_timer.data = (unsigned long)dev;
> -
>  	set_bit(IPOIB_FLAG_INITIALIZED, &priv->flags);
>  
>  	return 0;
> @@ -906,6 +902,9 @@ int ipoib_ib_dev_init(struct net_device *dev, struct ib_device *ca, int port)
>  		return -ENODEV;
>  	}
>  
> +	setup_timer(&priv->poll_timer, ipoib_ib_tx_timer_func,
> +		    (unsigned long) dev);
> +
>  	if (dev->flags & IFF_UP) {
>  		if (ipoib_ib_dev_open(dev)) {
>  			ipoib_transport_dev_cleanup(dev);
>