Hi,
On 1/9/24 06:33, Matthew Wilcox (Oracle) wrote:
> The documentation for this function has become separated from it over
> time; move it to the right place and turn it into kernel-doc. Mild
> editing of the content to make it more about what the function does, and
> less about how it does it.
>
> Signed-off-by: Matthew Wilcox (Oracle) <willy@xxxxxxxxxxxxx>
> ---
> fs/buffer.c | 44 ++++++++++++++++++++++++--------------------
> 1 file changed, 24 insertions(+), 20 deletions(-)
>
> diff --git a/fs/buffer.c b/fs/buffer.c
> index 071f01b28c90..25861241657f 100644
> --- a/fs/buffer.c
> +++ b/fs/buffer.c
> @@ -2864,26 +2864,6 @@ int sync_dirty_buffer(struct buffer_head *bh)
> }
> EXPORT_SYMBOL(sync_dirty_buffer);
>
> -/*
> - * try_to_free_buffers() checks if all the buffers on this particular folio
> - * are unused, and releases them if so.
> - *
> - * Exclusion against try_to_free_buffers may be obtained by either
> - * locking the folio or by holding its mapping's i_private_lock.
> - *
> - * If the folio is dirty but all the buffers are clean then we need to
> - * be sure to mark the folio clean as well. This is because the folio
> - * may be against a block device, and a later reattachment of buffers
> - * to a dirty folio will set *all* buffers dirty. Which would corrupt
> - * filesystem data on the same device.
> - *
> - * The same applies to regular filesystem folios: if all the buffers are
> - * clean then we set the folio clean and proceed. To do that, we require
> - * total exclusion from block_dirty_folio(). That is obtained with
> - * i_private_lock.
> - *
> - * try_to_free_buffers() is non-blocking.
> - */
> static inline int buffer_busy(struct buffer_head *bh)
> {
> return atomic_read(&bh->b_count) |
> @@ -2917,6 +2897,30 @@ drop_buffers(struct folio *folio, struct buffer_head **buffers_to_free)
> return false;
> }
>
> +/**
> + * try_to_free_buffers: Release buffers attached to this folio.
preferably s/_buffers: /_buffers - /
> + * @folio: The folio.
> + *
> + * If any buffers are in use (dirty, under writeback, elevated refcount),
> + * no buffers will be freed.
> + *
> + * If the folio is dirty but all the buffers are clean then we need to
> + * be sure to mark the folio clean as well. This is because the folio
> + * may be against a block device, and a later reattachment of buffers
> + * to a dirty folio will set *all* buffers dirty. Which would corrupt
> + * filesystem data on the same device.
> + *
> + * The same applies to regular filesystem folios: if all the buffers are
> + * clean then we set the folio clean and proceed. To do that, we require
> + * total exclusion from block_dirty_folio(). That is obtained with
> + * i_private_lock.
> + *
> + * Exclusion against try_to_free_buffers may be obtained by either
> + * locking the folio or by holding its mapping's i_private_lock.
> + *
> + * Context: Process context. @folio must be locked. Will not sleep.
> + * Return: true if all buffers attached to this folio were freed.
> + */
> bool try_to_free_buffers(struct folio *folio)
> {
> struct address_space * const mapping = folio->mapping;
--
#Randy
