Re: [PATCH 2/4] scsi: remove an extra queue reference

[Ming Lei <ming.lei@redhat.com>]

On Tue, Oct 18, 2022 at 03:57:18PM +0200, Christoph Hellwig wrote:
> Now that blk_mq_destroy_queue does not release the queue reference, there
> is no need for a second queue reference to be held by the scsi_device.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
>  drivers/scsi/scsi_scan.c  | 1 -
>  drivers/scsi/scsi_sysfs.c | 1 -
>  2 files changed, 2 deletions(-)
> 
> diff --git a/drivers/scsi/scsi_scan.c b/drivers/scsi/scsi_scan.c
> index 5d27f5196de6f..0a95fa787fdf4 100644
> --- a/drivers/scsi/scsi_scan.c
> +++ b/drivers/scsi/scsi_scan.c
> @@ -344,7 +344,6 @@ static struct scsi_device *scsi_alloc_sdev(struct scsi_target *starget,
>  	sdev->request_queue = q;
>  	q->queuedata = sdev;
>  	__scsi_init_queue(sdev->host, q);
> -	WARN_ON_ONCE(!blk_get_queue(q));
>  
>  	depth = sdev->host->cmd_per_lun ?: 1;
>  
> diff --git a/drivers/scsi/scsi_sysfs.c b/drivers/scsi/scsi_sysfs.c
> index 1214c6f07bc64..c95177ca6ed26 100644
> --- a/drivers/scsi/scsi_sysfs.c
> +++ b/drivers/scsi/scsi_sysfs.c
> @@ -1478,7 +1478,6 @@ void __scsi_remove_device(struct scsi_device *sdev)
>  	mutex_unlock(&sdev->state_mutex);
>  
>  	blk_mq_destroy_queue(sdev->request_queue);
> -	blk_put_queue(sdev->request_queue);

The above put is counter-pair of blk_get_queue() in scsi_alloc_sdev, and
the original blk_put_queue() in blk_mq_destroy_queue() is counter-pair of
the initial get in blk_alloc_queue().

Now blk_put_queue() is moved out of blk_mq_destroy_queue(), I am wondering
how the scsi queue lifetime can work correctly with this patch? Or is there
bug in current scsi code?


Thanks,
Ming