Re: [RFC PATCH] gpio: support for GPIO forwarding

["Rafael J. Wysocki" <rjw@rjwysocki.net>]

On Tuesday, February 10, 2015 06:32:46 PM Alexandre Courbot wrote:
> On Fri, Jan 23, 2015 at 8:21 PM, Heikki Krogerus
> <heikki.krogerus@linux.intel.com> wrote:
> > Hi guys,
> >
> > On Thu, Jan 22, 2015 at 05:14:22PM +0100, Rafael J. Wysocki wrote:
> >> On Thursday, January 22, 2015 11:57:55 AM Alexandre Courbot wrote:
> >> > If we decide to go ahead with the solution proposed by this patch for
> >> > practical reasons (which are good reasons indeed), I still have one
> >> > problem with its current form.
> >> >
> >> > As the discussion highlighted, this is an ACPI problem, so I'd very
> >> > much like it to be confined to the ACPI GPIO code, to be enabled only
> >> > when ACPI is, and to use function names that start with acpi_gpio.
> >>
> >> I can agree with that.
> >>
> >> > The current implementation leverages platform lookup, making said lookup
> >> > less efficient in the process and bringing confusion about its
> >> > purpose. Although the two processes are indeed similar, they are
> >> > separate things: one is a legitimate way to map GPIOs, the other is a
> >> > fixup for broken firmware.
> >> >
> >> > I suppose we all agree this is a hackish fix, so let's confine it as
> >> > much as we can.
> >>
> >> OK
> >>
> >> Heikki, any comments?
> >
> > I'm fine with that.
> >
> > That actually makes me think that we could then drop the lookup tables
> > completely and use device properties instead with the help of "generic
> > property" (attached):
> >
> > We would just need to agree on the format how to describe a gpio
> > property, document it and of course convert the current users as
> > usual. The format could be something like this as an example (I'm
> > writing this out of my head so don't shoot me if you can see it would
> > not work. Just an example):
> >
> > static const u32 example_gpio[] = { <gpio>, <flags>,爙;
> >
> > static struct dev_gen_prop example_prop[] =
> >         {
> >                 .type = DEV_PROP_U32,
> >                 .name = "gpio,<con_id>",
> >                 .nval = 2,
> >                 .num = &example_gpio,
> >         },
> >         { },
> > };
> >
> > static struct platform_device example_pdev = {
> >         ...
> >         .dev = {
> >                 .gen_prop = &example_prop,
> >         },
> > }
> >
> >
> > In gpiolib.c we would then, instead of going through the lookups,
> > simply ask for that property:
> >
> >         ...
> >         sprintf(propname, "gpio,%s", con_id);
> >         device_property_read_u32_array(dev, propname, &val, 2);
> >         ...
> >         desc = gpio_to_desc(val[0]);
> >         flags = val[1];
> >         ...
> >
> >
> > So this is just and idea. I think it would be relatively easy to
> > implement. What do you guys think?
> 
> At first sight, that looks like a very good idea and a great use of
> the device properties API. Are you willing to explore it further?

I guess Heikki is waiting for my feedback on his core patch that I'm
going to provide later today.

And yes, this is an interesting idea.


-- 
I speak only for myself.
Rafael J. Wysocki, Intel Open Source Technology Center.