Re: [PATCH 5/6] NLM: Convert lockd to use kthreads

[Neil Brown <neilb@suse.de>]

On Saturday January 5, jlayton@redhat.com wrote:
> Have lockd_up start lockd using kthread_run. With this change,
> lockd_down now blocks until lockd actually exits, so there's no longer
> need for the waitqueue code at the end of lockd_down. This also means
> that only one lockd can be running at a time which simplifies the code
> within lockd's main loop.
> 
> Signed-off-by: Jeff Layton <jlayton@redhat.com>
> ---


> -	module_put_and_exit(0);
> +	module_put(THIS_MODULE);
> +	return 0;

This changes bothers me.  Putting the last ref to a module in code
inside that module is not safe, which is why module_put_and_exit
exists. 

So this module_put is either unsafe or not needed.  I think the
latter.

As you say in the comment, lockd_down now blocks until lockd actually
exits.  As every caller for lockd_down will own a reference to the
lockd module, the lockd thread no longer needs to own a reference too.
So I think it is safe to remove the module_put, and also remove the
__module_get at the top of the lockd function.

Also, I suspect that the "no users" and "no lockd running" messages
should probably be changed to BUGs as they really should be
impossible, not just unlikely.

Also:

> 	 * Check whether there's a new lockd process before
> 	 * shutting down the hosts and clearing the slot.
> 	 */

This comment should go as the actual check has gone.

The rest looks fine. It is a substantial improvement, thanks.

NeilBrown