Re: [PATCH 3/3] dm: make dm_set_zones_restrictions work on the queue limits

[Mike Snitzer <snitzer@kernel.org>]

On Fri, May 24, 2024 at 04:29:11PM +0200, Christoph Hellwig wrote:
> Don't stuff the values directly into the queue without any
> synchronization, but instead delay applying the queue limits in
> the caller and let dm_set_zones_restrictions work on the limit
> structure.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
>  drivers/md/dm-table.c | 16 +++++++++-------
>  drivers/md/dm-zone.c  | 11 ++++++-----
>  drivers/md/dm.h       |  3 ++-
>  3 files changed, 17 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c
> index e291b78b307b13..6e4ee829005e92 100644
> --- a/drivers/md/dm-table.c
> +++ b/drivers/md/dm-table.c
> @@ -1981,10 +1981,6 @@ int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
>  	if (!dm_table_supports_secure_erase(t))
>  		limits->max_secure_erase_sectors = 0;
>  
> -	r = queue_limits_set(q, limits);
> -	if (r)
> -		return r;
> -
>  	if (dm_table_supports_flush(t, (1UL << QUEUE_FLAG_WC))) {
>  		wc = true;
>  		if (dm_table_supports_flush(t, (1UL << QUEUE_FLAG_FUA)))
> @@ -2036,12 +2032,18 @@ int dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
>  	 * For a zoned target, setup the zones related queue attributes
>  	 * and resources necessary for zone append emulation if necessary.
>  	 */
> -	if (blk_queue_is_zoned(q)) {
> -		r = dm_set_zones_restrictions(t, q);
> -		if (r)
> +	if (limits->zoned) {
> +		r = dm_set_zones_restrictions(t, q, limits);
> +		if (r) {
> +			queue_limits_cancel_update(q);

Not aware of a preceding queue_limits_start_update().

Otherwise, rest of patch looks fine.