On Mon, Mar 13, 2017 at 09:01:08PM +0000, Bart Van Assche wrote: > > + /* > > + * @request_count may become stale because of schedule > > + * out, so check the list again. > > + */ > > The above comment was relevant as long as there was a request_count assignment > above blk_mq_sched_get_request(). This patch moves that assignment inside if > (plug && q->nr_hw_queues == 1). Does that mean that the above comment should be > removed entirely? I don't think so - for the !blk_queue_nomerges cases we still rely on blk_attempt_plug_merge calculatіng request_count, so the comment still applies.