Re: [PATCH 06/23] gpio: sysfs: clean up chip class-device handling

[Johan Hovold <johan@kernel.org>]

On Mon, Apr 27, 2015 at 12:54:41PM +0900, Alexandre Courbot wrote:
> On Wed, Apr 22, 2015 at 12:42 AM, Johan Hovold <johan@kernel.org> wrote:
> > Clean gpio-chip class device registration and deregistration.
> >
> > The class device is registered when a gpio-chip is added (or from
> > gpiolib_sysfs_init post-core init call), and deregistered when the chip
> > is removed.
> >
> > Store the class device in struct gpio_chip directly rather than do a
> > class-device lookup on deregistration. This also removes the need for
> > the exported flag.
> >
> > Signed-off-by: Johan Hovold <johan@kernel.org>

> > diff --git a/include/linux/gpio/driver.h b/include/linux/gpio/driver.h
> > index f1b36593ec9f..8c26855fc6ec 100644
> > --- a/include/linux/gpio/driver.h
> > +++ b/include/linux/gpio/driver.h
> > @@ -20,6 +20,7 @@ struct seq_file;
> >   * struct gpio_chip - abstract a GPIO controller
> >   * @label: for diagnostics
> >   * @dev: optional device providing the GPIOs
> > + * @cdev: class device (may be NULL)
> 
> Maybe a comment explaining that this field is non-NULL when a chip is
> exported would be useful to understand how it is used in the code?

I've added comments where the field is used. I didn't want to get into
explaining sysfs implementation details in the header file, but the "may
be NULL" is there as a hint to actually look at the code.

And a gpio chip will always be registered with driver core (rather than
"exported" ;) ) until it is removed. [ Currently we also allow for
"late" registration, though. ]

This is related to the issue discussed in my last mail, and again the
plan is to let chip registration be used for more than the sysfs
interface.

Johan