On 9/22/26 1:25 AM, Muchun Song wrote:
> After the DAX configuration locking was converted to rwsems, successful
> lock acquisition leaves rc set to zero in mapping_store(). Two error
> paths can return this stale value: an unbound region driver and a
> misaligned range size. The sysfs write therefore reports success without
> allocating the requested range.
>
> Return -ENXIO when the region driver is not bound and -EINVAL when the
> range size is misaligned.
>
> Fixes: c05ae9d85b47 ("dax/bus.c: replace driver-core lock usage by a local
> rwsem")
> Assisted-by: LLM
> Signed-off-by: Muchun Song <[email protected]>
Reviewed-by: Dave Jiang <[email protected]>
> ---
> v2:
> - Return explicit errors before Device DAX range allocation
> (suggested by Dave Jiang)
> - Move the patch before range alignment validation for independent
> backports (suggested by Dave Jiang)
> ---
> drivers/dax/bus.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/dax/bus.c b/drivers/dax/bus.c
> index b809e1a264af..e40c25401cf0 100644
> --- a/drivers/dax/bus.c
> +++ b/drivers/dax/bus.c
> @@ -1192,7 +1192,7 @@ static ssize_t mapping_store(struct device *dev, struct
> device_attribute *attr,
> return rc;
> if (!dax_region->dev->driver) {
> up_write(&dax_region_rwsem);
> - return rc;
> + return -ENXIO;
> }
> rc = down_write_killable(&dax_dev_rwsem);
> if (rc) {
> @@ -1201,8 +1201,12 @@ static ssize_t mapping_store(struct device *dev,
> struct device_attribute *attr,
> }
>
> to_alloc = range_len(&r);
> - if (alloc_is_aligned(dev_dax, to_alloc))
> + if (!alloc_is_aligned(dev_dax, to_alloc)) {
> + dev_dbg(dev, "%s: size: %zu misaligned\n", __func__, to_alloc);
> + rc = -EINVAL;
> + } else {
> rc = alloc_dev_dax_range(dev_dax, r.start, to_alloc);
> + }
> up_write(&dax_dev_rwsem);
> up_write(&dax_region_rwsem);
>
