Re: [PATCH v2] thermal/drivers/rcar: fix error checking in probe()

[Dan Carpenter <error27@gmail.com>]

On Wed, Jun 24, 2026 at 03:12:47PM +0200, Geert Uytterhoeven wrote:
> Hi Dan,
> 
> On Wed, 24 Jun 2026 at 15:03, Dan Carpenter <error27@gmail.com> wrote:
> > This code accidentally calls thermal_zone_device_enable() before checking
> > whether thermal_zone_device_register_with_trips() failed.  Move the call
> > until later to avoid an error pointer dereference of "priv->zone".
> >
> > The driver works differently depending on if we are using OF thermal or
> > not.  We use thermal_add_hwmon_sysfs() if we are using OF thermal and
> > call thermal_zone_device_enable() if not.
> >
> > Moving the thermal_zone_device_enable() call is a bit cleaner as well.
> > The original code used a three step process to cleanup:
> > 1. Call thermal_zone_device_unregister() to cleanup.
> > 2. Set priv->zone to an error pointer to preserve the error code.
> > 3. Set priv->zone to NULL to avoid a second call to
> >    thermal_zone_device_unregister() in the rcar_thermal_remove()
> >    function.
> >
> > Now we can just do a direct goto error_unregister and rcar_thermal_remove()
> > handles the cleanup properly.
> >
> > Fixes: bbcf90c0646a ("thermal: Explicitly enable non-changing thermal zone devices")
> > Signed-off-by: Dan Carpenter <error27@gmail.com>
> > ---
> > v2: Use the correct fixes tag and re-write the check in a cleaner way.
> 
> Thanks for the update!
> 
> Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
> 
> > --- a/drivers/thermal/renesas/rcar_thermal.c
> > +++ b/drivers/thermal/renesas/rcar_thermal.c
> 
> > @@ -510,6 +504,10 @@ static int rcar_thermal_probe(struct platform_device *pdev)
> >                         ret = thermal_add_hwmon_sysfs(priv->zone);
> >                         if (ret)
> >                                 goto error_unregister;
> > +               } else {
> > +                       ret = thermal_zone_device_enable(priv->zone);
> > +                       if (ret)
> > +                               goto error_unregister;
> 
> This error path is the same in the other branch, so it could be shared
> after the if/else block.
> 

Even better.  :)  v3 coming up.

regards,
dan carpenter