Re: [patch] of: check for IS_ERR()

[Dan Carpenter <error27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>]

On Fri, Feb 26, 2010 at 01:54:20AM -0800, David Miller wrote:
> From: Dan Carpenter <error27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> Date: Fri, 26 Feb 2010 12:49:41 +0300
> 
> > get_phy_device() can return an ERR_PTR()
> > 
> > Signed-off-by: Dan Carpenter <error27-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
> > ---
> > I don't have a cross compile environment set up so I can't even compile 
> > test this.  :/  But err.h is included so it should be OK.
> 
> It should return ERR_PTR() consistently.  Checking for both
> NULL and ERR_PTR() is undesirable.

Hi Giulio,

get_phy_device() currently returns NULL because of: 3ee82383f0098a2 "phy: 
fix phy address bug".  If I change it to return ERR_PTR(-ENODEV) that
will mean we break out of the loop with an error in mdiobus_register()
where before we would just continue on.

drivers/net/phy/mdio_bus.c
   119                          phydev = mdiobus_scan(bus, i);
   120                          if (IS_ERR(phydev)) {
   121                                  err = PTR_ERR(phydev);
   122                                  goto error;
   123                          }

Is that OK?

I'm not really familiar with this hardware at all, I'm just going based
on static analysis. :/

regards,
dan carpenter