Re: [PATCH v3] IB/IPoIB: ibX: failed to create mcg debug file

[Shamir Rabinovitch <shamir.rabinovitch-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>]

On Mon, Mar 27, 2017 at 11:17:14PM +0300, Shamir Rabinovitch wrote:
> 
> > >
> > > +#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
> > > +static int ipoib_netdev_event(struct notifier_block *this,
> > > +			      unsigned long event, void *ptr)
> > > +{
> > > +	struct netdev_notifier_info *ni = ptr;
> > > +	struct net_device *dev = ni->dev;
> > > +
> > > +	if (dev->netdev_ops->ndo_open != ipoib_open)
> > > +		return NOTIFY_DONE;
> > > +
> > > +	switch (event) {
> > > +	case NETDEV_REGISTER:
> > > +		ipoib_create_debug_files(dev);
> > > +		break;
> > > +	case NETDEV_CHANGENAME:
> > > +		ipoib_debugfs_rename(dev);
> > 
> > Why do we need explicit ipoib_debugfs_rename function?
> > Will it work by simply calling ipoib_delete_debug_files
> > and immediately after that ipoib_create_debug_files?
> > 
> > 
> > > +		break;
> > > +	case NETDEV_UNREGISTER:
> > > +		ipoib_delete_debug_files(dev);
> > > +		break;
> > > +	}
> > > +
> > > +	return NOTIFY_DONE;
> > > +}
> > > +#endif
> > > +
> 
> Hi Leon, 
> 
> Good point. 
> I will have look on this idea and get back to you.
> 
> BR, Shamir

Hi Leon,

It seems that the difference between using debugfs_rename and a
combination of debugfs_create_file and debugfs_remove is mainly the
atomicy of the rename operation with regard to the file system.
debugfs_rename is atomic operation while the above combination is not.
As result I see kernel panic when the rename operation interleave with
the delete due to module unload. So after carefully considering what you
suggest I think it might introduce unexpected issues.

BR, Shamir  
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html