Re: [PATCH 20/20] honor r/w changes at do_remount() time

[Al Viro]

On Mon, Jul 03, 2006 at 10:30:14AM -0700, Dave Hansen wrote:
> On Wed, 2006-06-28 at 06:19 +0100, Al Viro wrote:
> >         * make the moments when i_nlink hits 0 bump the superblock writers
> > count; drop it when such sucker gets freed on final iput. 
> 
> Could you elaborate on this one a bit?  
> 
> I assume that there are rules that once i_nlink hits 0, it never goes
> back up again.  It seems that a whole bunch (if not all) of the
> individual filesystems do things to it.  Is it really necessary to go
> into all of those looking for the places that i_nlink hits 0?  Seems
> like it would be an awful lot of patching.

Not that much...  That happens in three methods (->unlink(), ->rename(),
->rmdir()) and yes, we really want to track those.  Think for a minute
and you'll see why - we don't want to allow remount ro when there is
a pending truncate/freeing inode.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/