Re: [PATCH v8 06/12] zloop: fix zone reporting consistency

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Medium] zloop: inconsistent empty zone report in zloop_report_zones()
--

commit 2a1f79e0caf4e50f76c7d14b20ef44343010c188
Author: Damien Le Moal <dlemoal@kernel.org>

zloop: fix zone reporting consistency

This commit moves the read of zone->cond inside zone->wp_lock in
zloop_report_zones() so that blkz.wp and blkz.cond are populated while
holding the same spinlock.

> diff --git a/drivers/block/zloop.c b/drivers/block/zloop.c
> index 65634a39b0bfc..e1ad7513fea9b 100644
> --- a/drivers/block/zloop.c
> +++ b/drivers/block/zloop.c
> @@ -992,8 +992,8 @@ static int zloop_report_zones(struct gendisk *disk, sector_t sector,
>  		blkz.len = zlo->zone_size;
>  		spin_lock(&zone->wp_lock);
>  		blkz.wp = zone->wp;
> -		spin_unlock(&zone->wp_lock);
>  		blkz.cond = zone->cond;
> +		spin_unlock(&zone->wp_lock);
>  		if (test_bit(ZLOOP_ZONE_CONV, &zone->flags)) {
>  			blkz.type = BLK_ZONE_TYPE_CONVENTIONAL;
>  			blkz.capacity = zlo->zone_size;

[Severity: Medium]
Can zloop_report_zones() still report blkz.cond == BLK_ZONE_COND_EMPTY
with blkz.wp > blkz.start when ordered_zone_append is enabled?

When a zone append to an empty zone does not fill the zone,
zloop_set_zone_append_sector() advances zone->wp under zone->wp_lock,
while transitioning zone->cond to BLK_ZONE_COND_IMP_OPEN via
zloop_do_open_zone() only happens later in the workqueue:

drivers/block/zloop.c:zloop_set_zone_append_sector() {
	...
	rq->__sector = zone->wp;
	zone->wp += blk_rq_sectors(rq);
	if (zone->wp >= zone_end)
		zloop_mark_full(zlo, zone);

	spin_unlock(&zone->wp_lock);
	...
}

CPU0                              CPU1 (kworker)
zloop_queue_rq()
  zloop_set_zone_append_sector()
    spin_lock(&zone->wp_lock)
    zone->wp += blk_rq_sectors(rq)
    spin_unlock(&zone->wp_lock)
  queue_work(..., &cmd->work)
zloop_report_zones()
  mutex_lock(&zone->lock)
  spin_lock(&zone->wp_lock)
  blkz.wp = zone->wp;
  blkz.cond = zone->cond;  <-- still BLK_ZONE_COND_EMPTY
  spin_unlock(&zone->wp_lock)
                                  zloop_rw()
                                    zloop_seq_write_prep()
                                      mutex_lock(&zone->lock)
                                      zloop_do_open_zone()

Reporting BLK_ZONE_COND_EMPTY with an advanced write pointer causes
disk_report_zone() -> disk_zone_wplug_sync_state() ->
blk_zone_wp_offset() to treat the write pointer offset as 0 instead of
blkz.wp - blkz.start.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261109095144.819709-1-dlemoal@kernel.org?part=6