Re: [PATCH V2] blk-mq: fix race between complete and BLK_EH_RESET_TIMER

["jianchao.wang" <jianchao.w.wang@xxxxxxxxxx>]

Hi Ming

On 04/12/2018 07:38 AM, Ming Lei wrote:
> +	 *
> +	 * Cover complete vs BLK_EH_RESET_TIMER race in slow path with
> +	 * helding queue lock.
>  	 */
>  	hctx_lock(hctx, &srcu_idx);
>  	if (blk_mq_rq_aborted_gstate(rq) != rq->gstate)
>  		__blk_mq_complete_request(rq);
> +	else {
> +		unsigned long flags;
> +		bool need_complete = false;
> +
> +		spin_lock_irqsave(q->queue_lock, flags);
> +		if (!blk_mq_rq_aborted_gstate(rq))
> +			need_complete = true;
> +		else
> +			blk_mq_rq_update_state(rq, MQ_RQ_COMPLETE_IN_TIMEOUT);
> +		spin_unlock_irqrestore(q->queue_lock, flags);

What if the .timeout return BLK_EH_HANDLED during this ?
timeout context                     irq context
  .timeout()
                                     blk_mq_complete_request
                                       set state to MQ_RQ_COMPLETE_IN_TIMEOUT
  __blk_mq_complete_request
    WARN_ON_ONCE(blk_mq_rq_state(rq) 
     != MQ_RQ_IN_FLIGHT);

If further upon blk_mq_free_request, the final freed request maybe changed to MQ_RQ_COMPLETE_IN_TIMEOUT
instead of MQ_RQ_IDLE.

Thanks
Jianchao