Re: [PATCH] Input: da7280 - protect OF match table with CONFIG_OF

[Jeff LaBundy]

Hi Dmitry,

On Sun, Jan 03, 2021 at 05:58:41PM -0800, Dmitry Torokhov wrote:
> Hi Jeff,
> 
> On Sat, Dec 19, 2020 at 08:01:09PM -0600, Jeff LaBundy wrote:
> > Hi Dmitry,
> > 
> > On Fri, Dec 18, 2020 at 04:49:48PM +0000, Roy Im wrote:
> > > On Friday, December 18, 2020 3:50 PM, Dmitry Torokhov wrote:
> > > 
> > > > The OF match table is only used when OF is enabled.
> > > > 
> > > > Fixes: cd3f609823a5 ("Input: new da7280 haptic driver")
> > > > Reported-by: kernel test robot <lkp@xxxxxxxxx>
> > > > Signed-off-by: Dmitry Torokhov <dmitry.torokhov@xxxxxxxxx>
> > > > ---
> > > >  drivers/input/misc/da7280.c | 2 ++
> > > >  1 file changed, 2 insertions(+)
> > > > 
> > > > diff --git a/drivers/input/misc/da7280.c b/drivers/input/misc/da7280.c index 2f698a8c1d65..b08610d6e575 100644
> > > > --- a/drivers/input/misc/da7280.c
> > > > +++ b/drivers/input/misc/da7280.c
> > > > @@ -1300,11 +1300,13 @@ static int __maybe_unused da7280_resume(struct device *dev)
> > > >  	return retval;
> > > >  }
> > > > 
> > > > +#ifdef CONFIG_OF
> > > >  static const struct of_device_id da7280_of_match[] = {
> > > >  	{ .compatible = "dlg,da7280", },
> > > >  	{ }
> > > >  };
> > > >  MODULE_DEVICE_TABLE(of, da7280_of_match);
> > > > +#endif
> > 
> > Just for my own understanding, would it not work just as well
> > to include of_device.h? This includes mod_devicetable.h which
> > in turn defines the of_device_id struct (even if CONFIG_OF is
> > not set).
> 
> The issue here is not that the structure is undefined, but the variable
> is unused. We could also fix this by not using of_match_ptr() when
> assigning the match table to the driver structure, making the variable
> referenced even if CONFIG_OF is off.

ACK. The call to of_match_ptr() is what I was missing; the other
drivers I was looking at do not use it which must be why the bot
has not complained.

> 
> > 
> > The reason for asking is because it seems many drivers do not
> > include these guards.
> 
> It could be that they are either only compiled with OF, or they decided
> it is not worth saving a few bytes, or maybe they are used on ACPI-based
> systems with PRP0001 bindings in which case the match table in the
> driver might still be needed.

Makes perfect sense; thank you for the follow-up.

> 
> Thanks.
> 
> -- 
> Dmitry

Kind regards,
Jeff LaBundy