Re: + r-o-bind-mount-clean-up-ocfs2-nlink-handling.patch added to -mm tree

Re: + r-o-bind-mount-clean-up-ocfs2-nlink-handling.patch added to -mm tree

[Daniel Walker]

On Wed, 2006-08-09 at 12:12 -0700, akpm@osdl.org wrote:

> -	/* We can set nlink on the dinode now. clear the saved version
> -	 * so that it doesn't get set later. */
> +	if (S_ISDIR(inode->i_mode))
> +		drop_nlink(inode);
> +	drop_nlink(inode);
>  	fe->i_links_count = cpu_to_le16(inode->i_nlink);
> -	saved_nlink = 0;
>  
>  	status = ocfs2_journal_dirty(handle, fe_bh);


There's one too many drop_nlink()'s in this block, unless I'm not
reading this right.

Daniel

Re: + r-o-bind-mount-clean-up-ocfs2-nlink-handling.patch added to -mm tree

[Dave Hansen]

On Thu, 2006-08-10 at 06:56 -0700, Daniel Walker wrote:
> On Wed, 2006-08-09 at 12:12 -0700, akpm@osdl.org wrote:
> > -	/* We can set nlink on the dinode now. clear the saved version
> > -	 * so that it doesn't get set later. */
> > +	if (S_ISDIR(inode->i_mode))
> > +		drop_nlink(inode);
> > +	drop_nlink(inode);
> >  	fe->i_links_count = cpu_to_le16(inode->i_nlink);
> > -	saved_nlink = 0;
> >  
> >  	status = ocfs2_journal_dirty(handle, fe_bh);
>
> There's one too many drop_nlink()'s in this block, unless I'm not
> reading this right.

It needs to be double-dropped for any directory inodes.  Some of the
older code just did i_nlink-=2, and I chose to just call drop_nlink()
twice instead of making another helper to do arbitrary arithmetic on
i_nlink.

I believe this made up for the 

        if (S_ISDIR(inode->i_mode))
                inode->i_nlink = 0;

in the old code.

-- Dave

Re: + r-o-bind-mount-clean-up-ocfs2-nlink-handling.patch added to -mm tree

[Daniel Walker]

On Thu, 2006-08-10 at 09:02 -0700, Dave Hansen wrote:

> > There's one too many drop_nlink()'s in this block, unless I'm not
> > reading this right.
> 
> It needs to be double-dropped for any directory inodes.  Some of the
> older code just did i_nlink-=2, and I chose to just call drop_nlink()
> twice instead of making another helper to do arbitrary arithmetic on
> i_nlink.
> 
> I believe this made up for the 
> 
>         if (S_ISDIR(inode->i_mode))
>                 inode->i_nlink = 0;
> 
> in the old code.


Ok, I had a feeling it was something like that .. A comment around there
might be helpful though, so people don't query you on in the future.

Daniel