[PATCH] Input: mtk-pmic-keys: Fix null pointer dereference when no compatible data

[Dmitry Torokhov]

On Mon, Jun 30, 2025 at 03:59:46PM -0400, Nícolas F. R. A. Prado wrote:
> On Mon, 2025-06-30 at 08:25 -0700, Dmitry Torokhov wrote:
> > On Mon, Jun 30, 2025 at 04:06:53PM +0200, AngeloGioacchino Del Regno
> > wrote:
> > > Il 30/06/25 16:03, Louis-Alexis Eyraud ha scritto:
> > 
> > [... snip ...]
> > 
> > > > @@ -316,6 +316,9 @@ static int mtk_pmic_keys_probe(struct
> > > > platform_device *pdev)
> > > >         const struct of_device_id *of_id =
> > > >                 of_match_device(of_mtk_pmic_keys_match_tbl,
> > > > &pdev->dev);
> > > > +       if (!of_id)
> > > > +               return -EINVAL;
> > > 
> > > Please, change this to `return -ENODEV;`
> > 
> > No, this definitely should not be a "silent" error because it
> > indicates
> > there is something wrong with the kernel.
> > 
> > Stepping back, why do we even enter mtk_pmic_keys_probe() if there is
> > not a matching OF ID? Are there any other patches that are not
> > upstream?
> 
> I'm guessing it's because the driver can be probed by a parent MFD
> driver, drivers/mfd/mt6397-core.c, and the compatibles defined in the
> MFD don't necessarily match the ones in the pmic-keys driver, for
> instance 'mediatek,mt6359-keys' is only listed in the MFD. Adding the
> missing compatibles to the pmic-keys driver should fix this.

I think if we stop using the generic "mtk-pmic-keys" in MFD core it
should solve the issue. I just sent out a patch...

Thanks.

-- 
Dmitry