# Re: Why page w/o buffers can not be released?

- List: linux-ext4
- 1 message, 1 author
- Page: https://ratatoskr.run/linux-ext4/2012/09/66849/t

## Re: Why page w/o buffers can not be released?

- From: Jan Kara \<jack@suse.cz\>
- Date: 2012-09-05 12:28:11 
- Message: https://ratatoskr.run/linux-ext4/2012/09/66849.md

```text
On Wed 05-09-12 15:30:29, Dmitry Monakhov wrote:
> 
> ext3, ext4 and ocfs2 do not allow to release page w/o buffers.
> Can some one please explain me why?
> 
> ext4_releasepage(struct page *page, gfp_t wait)
> {
>         if (!page_has_buffers(page))
>                 return 0;
>         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^( zero means fail to release)
>         if (journal)
>            return jbd2_journal_try_to_free_buffers(journal, page, wait);
>         else
>                 return try_to_free_buffers(page);
> }
> 
> This is looks very strange because if page has not buffers then
> it has no fs-specific data associated with it. Am I right?
> IMHO code should looks like follows:
> xxx_releasepage() {
> 
>      if(PageWriteback(page))
>         return 0;
>      if (!page_has_buffers(page))
>         return 1;
>  
>     return do_fs_specific_stuff()
> }
  Yeah, I agree. Although this is mostly a harmless thing because all call
sites test for page_has_private(page) before calling try_to_release_page(page)
so !page_has_buffers(page) is impossible. But a cleanup (remove the test) would
be nice.

  Also PageWriteback() is checked in try_to_release_page() so filesystems
don't have to care...

								Honza

-- 
Jan Kara [ref]
SUSE Labs, CR
```