On Mon, Jun 19, 2023 at 01:18:32PM +0200, Jan Kara wrote:
> Provide helpers to set and clear sb->s_readonly_remount including
> appropriate memory barriers. Also use this opportunity to document what
> the barriers pair with and why they are needed.
>
> Suggested-by: Dave Chinner <david@xxxxxxxxxxxxx>
> Signed-off-by: Jan Kara <jack@xxxxxxx>
> ---
> fs/internal.h | 34 ++++++++++++++++++++++++++++++++++
> fs/namespace.c | 10 ++++------
> fs/super.c | 17 ++++++-----------
> include/linux/fs.h | 2 +-
> 4 files changed, 45 insertions(+), 18 deletions(-)
>
> diff --git a/fs/internal.h b/fs/internal.h
> index bd3b2810a36b..e206eb58bd3e 100644
> --- a/fs/internal.h
> +++ b/fs/internal.h
> @@ -120,6 +120,40 @@ void put_super(struct super_block *sb);
> extern bool mount_capable(struct fs_context *);
> int sb_init_dio_done_wq(struct super_block *sb);
>
> +/*
> + * Prepare superblock for changing its read-only state (i.e., either remount
> + * read-write superblock read-only or vice versa). After this function returns
> + * mnt_is_readonly() will return true for any mount of the superblock if its
> + * caller is able to observe any changes done by the remount. This holds until
> + * sb_end_ro_state_change() is called.
> + */
> +static inline void sb_start_ro_state_change(struct super_block *sb)
> +{
> + WRITE_ONCE(sb->s_readonly_remount, 1);
> + /*
> + * For RO->RW transition, the barrier pairs with the barrier in
> + * mnt_is_readonly() making sure if mnt_is_readonly() sees SB_RDONLY
> + * cleared, it will see s_readonly_remount set.
> + * For RW->RO transition, the barrier pairs with the barrier in
> + * __mnt_want_write() before the mnt_is_readonly() check. The barrier
> + * makes sure if __mnt_want_write() sees MNT_WRITE_HOLD already
> + * cleared, it will see s_readonly_remount set.
> + */
> + smp_wmb();
> +}
Can you please also update mnt_is_readonly/__mnt_want_write to
indicate that there is a pairing with this helper from those
functions?
> +
> +/*
> + * Ends section changing read-only state of the superblock. After this function
> + * returns if mnt_is_readonly() returns false, the caller will be able to
> + * observe all the changes remount did to the superblock.
> + */
> +static inline void sb_end_ro_state_change(struct super_block *sb)
> +{
> + /* The barrier pairs with the barrier in mnt_is_readonly() */
> + smp_wmb();
> + WRITE_ONCE(sb->s_readonly_remount, 0);
> +}
/*
* This barrier provides release semantics that pair with
* the smp_rmb() acquire semantics in mnt_is_readonly().
* This barrier pair ensure that when mnt_is_readonly() sees
* 0 for sb->s_readonly_remount, it will also see all the
* preceding flag changes that were made during the RO state
* change.
*/
And a comment in mnt_is_readonly() to indicate that it also pairs
with sb_end_ro_state_change() in a different way to the barriers in
sb_start_ro_state_change(), __mnt_want_write(), MNT_WRITE_HOLD, etc.
Memory barriers need clear, concise documentation, otherwise they
are impossible to understand from just reading the code...
Cheers,
Dave.
--
Dave Chinner
david@xxxxxxxxxxxxx
