Re: [EXTERNAL] Re: [PATCH v1] ASoC: tas2781: mark const variables tas2563_dvc_table as __maybe_unused

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Sat, Aug 24, 2024 at 06:33:41AM +0000, Ding, Shenghao wrote:
> > From: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> > Sent: Friday, August 23, 2024 9:34 PM
> > On Fri, Aug 23, 2024 at 02:24:17AM +0000, Ding, Shenghao wrote:
> > > > From: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> > > > Sent: Thursday, August 22, 2024 9:37 PM
> > > > To: Ding, Shenghao <shenghao-ding@ti.com> On Thu, Aug 22, 2024 at
> > > > 02:32:02PM +0800, Shenghao Ding wrote:

...

> > > > >  /* pow(10, db/20) * pow(2,30) */
> > > > > -static const unsigned char tas2563_dvc_table[][4] = {
> > > > > +static const __maybe_unused unsigned char tas2563_dvc_table[][4]
> > > > > += {
> > > > >  	{ 0X00, 0X00, 0X00, 0X00 }, /* -121.5db */
> > > > >  	{ 0X00, 0X00, 0X03, 0XBC }, /* -121.0db */
> > > > >  	{ 0X00, 0X00, 0X03, 0XF5 }, /* -120.5db */
> > > >
> > > > As far as I can see in the latest Linux Next the above mentioned
> > > > table is used solely in sound/soc/codecs/tas2781-i2c.c. Why not
> > > > moving it to the C file instead?
> > > We have a big plan to implement the digital gain for tas2563 in the
> > > side codec driver too, so abstracted this table into a separated header file.
> > 
> > So, this __maybe_unused is a temporary stub, correct? Just don't forget to
> > remove it when the mentioned functionality will be ready.
> > 
> > Thanks for the explanation.
> To add __maybe_unused is not temporary, because this driver support both 
> tas2563 and tas2781, if the hardware only contains tas2781 chips, 
> tas2563_dvc_table is obviously useless, so marked it as __maybe_unused.

You got me confused.

__maybe_unused is a compile-time (okay, link-time) attribute. If what you are
telling is true, why do you need it and can't use respective #ifdef?

Otherwise, why is it in the _header_ file and not in the C-file?

-- 
With Best Regards,
Andy Shevchenko