On Tue, Sep 19, 2017 at 04:49:15PM +0000, Bart Van Assche wrote: > On Wed, 2017-09-20 at 00:04 +0800, Ming Lei wrote: > > Run queue at end_io is definitely wrong, because blk-mq has SCHED_RESTART > > to do that already. > > Sorry but I disagree. If SCHED_RESTART is set that causes the blk-mq core to > reexamine the software queues and the hctx dispatch list but not the requeue > list. If a block driver returns BLK_STS_RESOURCE then requests end up on the > requeue list. Hence the following code in scsi_end_request(): > > if (scsi_target(sdev)->single_lun || !list_empty(&sdev->host->starved_list)) > kblockd_schedule_work(&sdev->requeue_work); > else > blk_mq_run_hw_queues(q, true); Let's focus on dm-rq. I have explained before, dm-rq is different with SCSI's, we don't need to requeue request any more in dm-rq if blk_get_request() returns NULL in multipath_clone_and_map(), since SCHED_RESTART can cover that definitely. Not mention dm_mq_delay_requeue_request() will run the queue explicitly if it has to be done. That needn't SCHED_RESTART to cover, right? -- Ming
