Re: [PATCH v2] mfd: syscon: allow to register syscon with a device

[Philipp Zabel]

Hi Arnd,

Am Mittwoch, den 24.02.2016, 17:26 +0100 schrieb Arnd Bergmann:
> On Wednesday 24 February 2016 16:02:28 Philipp Zabel wrote:
> > > > @@ -473,6 +474,12 @@ static const struct of_device_id imx6q_pinctrl_of_match[] = {
> > > >  
> > > >  static int imx6q_pinctrl_probe(struct platform_device *pdev)
> > > >  {
> > > > +       struct device_node *syscon_np;
> > > > +
> > > > +       syscon_np = of_find_compatible_node(NULL, NULL, "fsl,imx6q-iomuxc-gpr");
> > > > +       if (syscon_np)
> > > > +               syscon_register(&pdev->dev, syscon_np);
> > > > +
> > > >         return imx_pinctrl_probe(pdev, &imx6q_pinctrl_info);
> > > >  }
> > > >  
> > > > The pinctrl driver is probed at arch_initcall time.
> > > 
> > > I still don't see it where you are getting with this. Is this just for
> > > the debugfs interface or is there something else this does on top of
> > > that?
> > 
> > Sorry, yes, this is about the missing debugfs functionality.
> 
> Ok, so maybe there is a different way of doing this: If I understood
> it correctly, we now lost the debugfs handling on all syscon nodes?
> 
> Should we try to come up with a way to bring it back for all of them
> then, rather than having to do it per device?
> 
> I can see two possible ways of doing that:
> 
> a) change the regmap code to provide a debugfs interface for all
>    regmaps, including those without a device
> b) change the syscon driver to explicitly register the debugfs
>    interface at a later point in boot when the devices become
>    available

Thanks, b) seems to fit well for the i.MX IOMUXC. I have sent a patch to
use regmap_attach_dev for that.

regards
Philipp