Re: [PATCH 1/2] block: remove blk_zone_wp_offset()

[Christoph Hellwig <hch@lst.de>]

On Thu, Nov 06, 2025 at 04:06:26PM +0900, Damien Le Moal wrote:
> The helper function blk_zone_wp_offset() is called from
> disk_zone_wplug_sync_wp_offset(), and again called from
> blk_revalidate_seq_zone() right after the call to
> disk_zone_wplug_sync_wp_offset().
> 
> Change disk_zone_wplug_sync_wp_offset() to return the wp_offset it used
> for updating the target zone write plug to avoid this double call. With
> this change, blk_zone_wp_offset() can be open coded directly in
> disk_zone_wplug_sync_wp_offset(). This open-coding introduces 2 changes:
> handle the BLK_COND_ZONE_ACTIVE case, and return UINT_MAX as the
> wp_offset for a full zone, since the write pointer of full zones is
> invalid.
> 
> For the case where a zone does not have a zone write plug,
> disk_zone_wplug_sync_wp_offset() does nothing and returns 0. This in
> turn leads to blk_revalidate_seq_zone() to immediately return, which is
> exactly what we want (because there is no need to attempt removing a
> zone write plug that does not exist).

I like the use of disk_zone_wplug_sync_wp_offset in
blk_revalidate_seq_zone, but having the switch on the zone conditions
in a separate helper seems nicer than open coding it next to the
write plug handling.