On Wed, Sep 27, 2023 at 08:11:11AM -0700, Darrick J. Wong wrote:
> > @@ -56,14 +56,11 @@ struct block_device {
> > void * bd_holder;
>
> Hmmm. get_bdev_super from patch 3 now requires that bd_holder is a
> pointer to a struct super_block. AFAICT it's only called in conjunction
> with fs_holder_ops, so I suggest that the declaration for that should
> grow a comment to that effect:
>
> /*
> * For filesystems, the @holder argument passed to blkdev_get_* and
> * bd_prepare_to_claim must point to a super_block object.
> */
> extern const struct blk_holder_ops fs_holder_ops;
Note that this is not strictly speaking true, it is only true for
file systems actually using fs_holder_ops, not file systems in general.
That being said a comment to that extent is indeed useful.
