Re: [PATCH v5 8/9] scsi: sd_zbc: always revalidate zones for disks supporting head depopulation

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] scsi: data race on bitfield sdkp->modify_zones_supported vs sysfs updates
--

commit 008f16ea9959160e815da59ad94c230e5abd0803
Author: Damien Le Moal <dlemoal@kernel.org>

scsi: sd_zbc: always revalidate zones for disks supporting head depopulation

This commit prevents the block layer from skipping blk_revalidate_disk_zones()
for ZBC devices that support head depopulation. It introduces a new bitfield,
modify_zones_supported, to remember if the device supports the REMOVE ELEMENT
AND MODIFY ZONES command.

> diff --git a/drivers/scsi/sd.h b/drivers/scsi/sd.h
> index 574af82430169..6a72371fee786 100644
> --- a/drivers/scsi/sd.h
> +++ b/drivers/scsi/sd.h
> @@ -156,6 +156,7 @@ struct scsi_disk {
>  	unsigned	ignore_medium_access_errors : 1;
>  	unsigned	rscs : 1; /* reduced stream control support */
>  	unsigned	use_atomic_write_boundary : 1;
> +	unsigned	modify_zones_supported : 1;
>  };
>  #define to_scsi_disk(obj) container_of(obj, struct scsi_disk, disk_dev)

[Severity: Medium]
Is it safe to place this bitfield here? It shares the same machine word as
the sysfs-writable (WCE, cache_override) and EH-writable
(ignore_medium_access_errors) bitfields.

> diff --git a/drivers/scsi/sd_zbc.c b/drivers/scsi/sd_zbc.c
> index 8cbf171e6fabf..1bf5f7ea0d9c5 100644
> --- a/drivers/scsi/sd_zbc.c
> +++ b/drivers/scsi/sd_zbc.c
[ ... ]
> @@ -620,6 +641,9 @@ int sd_zbc_read_zones(struct scsi_disk *sdkp, struct queue_limits *lim,
>  	if (ret != 0)
>  		goto err;
>  
> +	/* Check if REMOVE ELEMENT AND MODIFY ZONES is supported. */
> +	sdkp->modify_zones_supported = sd_zbc_check_modify_zones(sdkp, buf);
> +
>  	nr_zones = round_up(sdkp->capacity, zone_blocks) >> ilog2(zone_blocks);
>  	if (nr_zones > INT_MAX) {
>  		sd_printk(KERN_ERR, sdkp, "Too many zones (%llu)\n",

[Severity: Medium]
Can this lockless update race with other bitfield updates, leading to a lost
write?

Because bitfield updates compile down to non-atomic read-modify-write
operations, and no locks synchronize sysfs stores like cache_type_store()
with disk revalidation, concurrent modifications to these bitfields could
result in a data race and lost updates.

If the modify_zones_supported update is lost, sd_zbc_revalidate_zones()
might skip blk_revalidate_disk_zones(), leading to a stale block-layer view
of offline/read-only zones.

Alternatively, if a sysfs update to cache_type (WCE or cache_override) is
lost, it could lead to an unsafe or unexpected write cache configuration.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261012130121.1902634-1-dlemoal@kernel.org?part=8