Re: [PATCH 5.19 005/192] net: mvpp2: debugfs: fix memory leak when using debugfs_lookup()

[Greg Kroah-Hartman <gregkh@linuxfoundation.org>]

On Tue, Sep 13, 2022 at 05:48:58PM +0100, Russell King (Oracle) wrote:
> On Tue, Sep 13, 2022 at 04:01:51PM +0200, Greg Kroah-Hartman wrote:
> > From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > 
> > commit fe2c9c61f668cde28dac2b188028c5299cedcc1e upstream.
> > 
> > When calling debugfs_lookup() the result must have dput() called on it,
> > otherwise the memory will leak over time.  Fix this up to be much
> > simpler logic and only create the root debugfs directory once when the
> > driver is first accessed.  That resolves the memory leak and makes
> > things more obvious as to what the intent is.
> > 
> > Cc: Marcin Wojtas <mw@semihalf.com>
> > Cc: Russell King <linux@armlinux.org.uk>
> > Cc: "David S. Miller" <davem@davemloft.net>
> > Cc: Eric Dumazet <edumazet@google.com>
> > Cc: Jakub Kicinski <kuba@kernel.org>
> > Cc: Paolo Abeni <pabeni@redhat.com>
> > Cc: netdev@vger.kernel.org
> > Cc: stable <stable@kernel.org>
> > Fixes: 21da57a23125 ("net: mvpp2: add a debugfs interface for the Header Parser")
> > Signed-off-by: David S. Miller <davem@davemloft.net>
> > Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > ---
> >  drivers/net/ethernet/marvell/mvpp2/mvpp2_debugfs.c |    4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > --- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_debugfs.c
> > +++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_debugfs.c
> > @@ -700,10 +700,10 @@ void mvpp2_dbgfs_cleanup(struct mvpp2 *p
> >  
> >  void mvpp2_dbgfs_init(struct mvpp2 *priv, const char *name)
> >  {
> > -	struct dentry *mvpp2_dir, *mvpp2_root;
> > +	static struct dentry *mvpp2_root;
> > +	struct dentry *mvpp2_dir;
> >  	int ret, i;
> >  
> > -	mvpp2_root = debugfs_lookup(MVPP2_DRIVER_NAME, NULL);
> >  	if (!mvpp2_root)
> >  		mvpp2_root = debugfs_create_dir(MVPP2_DRIVER_NAME, NULL);
> 
> This looks broken to me.
> 
> What happens if this is built as a module, and the module is loaded,
> binds (and creates the directory), then is removed, and then re-
> inserted?  Nothing removes the old directory, so doesn't
> debugfs_create_dir() fail, resulting in subsequent failure to add
> any subsequent debugfs entries?
> 
> I don't think this patch should be backported to stable trees until
> this point is addressed.

I'll drop this for now and get back to the fixup next week.

thanks,

greg k-h