Re: [PATCH v3 06/10] gpio: sysfs: don't use driver data in sysfs callbacks for line attributes

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

On Mon, Jun 30, 2025 at 03:39:20PM +0200, Bartosz Golaszewski wrote:
> On Mon, Jun 30, 2025 at 3:28 PM Andy Shevchenko
> <andriy.shevchenko@intel.com> wrote:
> > On Mon, Jun 30, 2025 at 02:37:13PM +0200, Bartosz Golaszewski wrote:

...

> > > -     struct gpiod_data *data = dev_get_drvdata(dev);
> > > +     struct gpiod_data *data = container_of(attr, struct gpiod_data,
> > > +                                            dir_attr);
> >
> > > -     struct gpiod_data *data = dev_get_drvdata(dev);
> > > +     struct gpiod_data *data = container_of(attr, struct gpiod_data,
> > > +                                            dir_attr);
> >
> > I still think that bunch of definitions to wrap contaner_of():s will help.
> > Note, two of them you are using even in the code below, besides the attribute
> > show()/store() cases.
> 
> I think this is overkill, the container_of() macro is pretty explicit
> and having 4 new macros will result in more LOC in the end.

I don't know how you counted. I counted as first one takes the same +4/-4 LoCs,
second one will gain already 1 LoC if we stack them as

#define foo_attr_to_gpiod_data(a)	container_of(...)
#define bar_attr_to_gpiod_data(a)	container_of(...)

...and so on...

But yes, some people are still fans of 80, and this might distort
the calculations I made.

> I'll allow myself to keep this as is.

Fair enough.

-- 
With Best Regards,
Andy Shevchenko