Re: [PATCH RFC 2/7] blk-mq: delay tag fair sharing until fail to get driver tag

[Bart Van Assche]

On 6/18/23 09:07, Yu Kuai wrote:
> Start tag fair sharing when a device start to issue io will waste
> resources, same number of tags will be assigned to each disk/hctx,
> and such tags can't be used for other disk/hctx, which means a disk/hctx
> can't use more than assinged tags even if there are still lots of tags
                      ^^^^^^^^
                      assigned
> that is assinged to other disks are unused.
          ^^^^^^^^
          assigned
> Add a new api blk_mq_driver_tag_busy(), it will be called when get
> driver tag failed, and move tag sharing from blk_mq_tag_busy() to
> blk_mq_driver_tag_busy().


> +	spin_lock_irq(&tags->lock);
> +	WRITE_ONCE(tags->ctl.share_queues, tags->ctl.active_queues);
> +	blk_mq_update_wake_batch(tags, tags->ctl.share_queues);
> +	spin_unlock_irq(&tags->lock);
> +}

Are all tags->ctl.share_queues changes protected by tags->lock? If so, please
use a regular assignment to update that member variable instead of using
WRITE_ONCE().

> @@ -735,6 +736,7 @@ struct request *blk_mq_alloc_request_hctx(struct request_queue *q,
>   
>   struct tag_sharing_ctl {
>   	unsigned int active_queues;
> +	unsigned int share_queues;
>   };

Please rename "share_queues" into "shared_queues" such that the names of
both struct members start with an adjective.

Thanks,

Bart.