Re: [PATCH] pinctrl: baytrail: show output gpio state correctly on Intel Baytrail

[David Cohen <david.a.cohen@linux.intel.com>]

On Fri, Oct 31, 2014 at 08:20:05AM -0500, Felipe Balbi wrote:
> Hi,
> 
> On Fri, Oct 31, 2014 at 09:12:16AM +0100, Linus Walleij wrote:
> > On Tue, Oct 28, 2014 at 3:42 PM, Felipe Balbi <balbi@ti.com> wrote:
> > > On Tue, Oct 28, 2014 at 11:15:20AM +0100, Linus Walleij wrote:
> > >> On Mon, Oct 13, 2014 at 9:36 PM, Felipe Balbi <balbi@ti.com> wrote:
> > >> > On Mon, Oct 13, 2014 at 02:26:32PM -0500, Felipe Balbi wrote:
> > >>
> > >> > I also noticed that this is missing:
> > >> >
> > >> > diff --git a/drivers/pinctrl/pinctrl-baytrail.c b/drivers/pinctrl/pinctrl-baytrail.c
> > >> > index e12e5b0..7db5ab9 100644
> > >> > --- a/drivers/pinctrl/pinctrl-baytrail.c
> > >> > +++ b/drivers/pinctrl/pinctrl-baytrail.c
> > >> > @@ -614,3 +614,9 @@ static int __init byt_gpio_init(void)
> > >> >  }
> > >> >
> > >> >  subsys_initcall(byt_gpio_init);
> > >> > +
> > >> > +static void __exit byt_gpio_exit(void)
> > >> > +{
> > >> > +       platform_driver_unregister(&byt_gpio_driver);
> > >> > +}
> > >> > +module_exit(byt_gpio_exit);
> > >>
> > >> But the Baytrail driver is not a loadable module, it is bool:
> > >>
> > >> config PINCTRL_BAYTRAIL
> > >>         bool "Intel Baytrail GPIO pin control"
> > >>         depends on GPIOLIB && ACPI && X86
> > >>
> > >> (...)
> > >>
> > >> So I guess it won't need handling for removal, as it can only
> > >> be compiled-in.
> > >
> > > you can still unbind it through sysfs, right ? The thing also already
> > > provides a ->remove() method anyway.
> > 
> > Yes you're right of course...
> > 
> > But another way to get rid of the dilemma is to set
> > .suppress_bind_attrs = true on the .driver field of the
> > device driver. The one can't unbind it through sysfs anymore.
> > 
> >         .driver = {
> >                 .name   = "foo",
> >                 .suppress_bind_attrs = true,
> >         },
> > 
> > So one of them need to be done.
> > 
> > I suspect this is a kind of common problem...
> 
> so instead of taking of taking a three-liner which just makes sure this
> can be used as "intended" you prefer to:
> 
> diff --git a/drivers/pinctrl/pinctrl-baytrail.c b/drivers/pinctrl/pinctrl-baytrail.c
> index e12e5b0..254ba81 100644
> --- a/drivers/pinctrl/pinctrl-baytrail.c
> +++ b/drivers/pinctrl/pinctrl-baytrail.c
> @@ -587,16 +587,6 @@ static const struct acpi_device_id byt_gpio_acpi_match[] = {
>  };
>  MODULE_DEVICE_TABLE(acpi, byt_gpio_acpi_match);
>  
> -static int byt_gpio_remove(struct platform_device *pdev)
> -{
> -	struct byt_gpio *vg = platform_get_drvdata(pdev);
> -
> -	pm_runtime_disable(&pdev->dev);
> -	gpiochip_remove(&vg->chip);
> -
> -	return 0;
> -}
> -
>  static struct platform_driver byt_gpio_driver = {
>  	.probe          = byt_gpio_probe,
>  	.remove         = byt_gpio_remove,
> @@ -605,6 +595,7 @@ static struct platform_driver byt_gpio_driver = {
>  		.owner  = THIS_MODULE,
>  		.pm	= &byt_gpio_pm_ops,
>  		.acpi_match_table = ACPI_PTR(byt_gpio_acpi_match),
> +		.suppress_bind_attrs = true,
>  	},
>  };
>  
> 
> I don't quite care since this is not an architecture I work for, but I
> prefer drivers which can be unbound one way or another. Not to mention
> that there's already a ->remove callback on the platform_driver anyway.

I think adding the module exit + allowing this driver to be a module
would be a good approach. Then we don't need to force generic x86 kernel
binaries to always have this driver. Unless Mathias or Mika knows a
constraint to force this driver to be builtin only.

Br, David

> 
> -- 
> balbi