Re: [PATCH mmotm] vfs: optimize touch_time too fix

[Christoph Hellwig <hch@infradead.org>]

On Mon, Aug 31, 2009 at 04:03:02PM +0100, Hugh Dickins wrote:
> Recent mmotms give me WARNING: at fs/namespace.c:612 mntput_no_expire()+...
> when unmounting: __mntput()'s WARN_ON(count_mnt_writers(mnt)).
> 
> That's because vfs-optimize-touch_time-too.patch inverted the sense
> of mnt_want_write_file(), which is error-returning, not a boolean.
> 
> Presumably filetime updates went missing too, but I didn't notice those.
> 
> Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
> ---
> 
>  fs/inode.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> --- mmotm/fs/inode.c	2009-08-28 10:07:56.000000000 +0100
> +++ linux/fs/inode.c	2009-08-31 14:08:24.000000000 +0100
> @@ -1481,7 +1481,7 @@ void file_update_time(struct file *file)
>  		return;
>  
>  	/* Finally allowed to write? Takes lock. */
> -	if (!mnt_want_write_file(file))
> +	if (mnt_want_write_file(file))

Eeek.  having an error return from a function called want_foo is rather
strange.  It's a name the implies a bolean return value.