Re: [PATCH 4/7] block: Avoid that request_fn is invoked on a dead queue

[Tejun Heo <tj@kernel.org>]

Hello,

On Fri, Oct 26, 2012 at 02:03:45PM +0200, Bart Van Assche wrote:
> A block driver may start cleaning up resources needed by its
> request_fn as soon as blk_cleanup_queue() finished, so request_fn
> must not be invoked after draining finished.

Please be a lot more detailed.  Please describe an example scenario
where things go wrong, explain why that's happening and how it's
fixed.

> +void inline __blk_run_queue_uncond(struct request_queue *q)
> +{
> +	if (unlikely(blk_queue_dead(q)))
> +		return;
> +
> +	q->request_fn(q);
> +}
...
> @@ -510,6 +529,7 @@ void blk_cleanup_queue(struct request_queue *q)
>  	/* drain all requests queued before DYING marking */
>  	spin_lock_irq(lock);
>  	__blk_drain_queue(q, true);
> +	queue_flag_set(QUEUE_FLAG_DEAD, q);
>  	spin_unlock_irq(lock);

And some comments explaining what the dead thing is doing would be
nice too; other than that, looks good to me.

Thanks!

-- 
tejun