Re: [PATCH 1/3] gpio / ACPI: add ACPI support

[Mathias Nyman <mathias.nyman@linux.intel.com>]

On 11/09/2012 04:18 PM, Grant Likely wrote:
> On Fri, Nov 9, 2012 at 2:11 PM, Mathias Nyman
> <mathias.nyman@linux.intel.com>  wrote:
>> On 11/08/2012 09:38 PM, Mika Westerberg wrote:
>> ...
>>
>>>>> +#include<linux/errno.h>
>>>>> +#include<linux/gpio.h>
>>>>> +#include<linux/module.h>
>>>>> +#include<linux/acpi_gpio.h>
>>>>> +#include<linux/acpi.h>
>>>>> +
>>>>> +static int acpi_gpiochip_find(struct gpio_chip *gc, void *data)
>>>>> +{
>>>>> +       acpi_handle handle = data;
>>>>> +       acpi_handle gc_handle;
>>>>> +
>>>>> +       if (!gc->dev)
>>>>> +               return false;
>>>>> +
>>>>> +       gc_handle = gc->dev->acpi_handle;
>>>>> +       if (!gc_handle)
>>>>> +               return false;
>>>>
>>>>
>>>> This test is redundant with the next one... unless 'handle' is also NULL
>>>> :-)
>>>>
>>>> Is it at all possible for multiple gpiochips to be used for a single
>>>> ACPI gpio controller node? Such as if the gpio controller has multiple
>>>> banks that should be controlled separately? If so then this won't be
>>>> sufficient. I've got the same issue with DT support where the find
>>>> function needs to also check if the pin is provided by that specific
>>>> gpiochip.
>>>
>>>
>>> AFAIK no but I'll let Mathias to answer that as he knows this better.
>>
>>
>> I'm interpreting it the same way as Mika, max one actual controller per ACPI
>> device node
>>
>> The path (called ResourceSource in ACPI5 specs) in GpioIO/GpioInt resources
>> is a "string which uniquely identifies the GPIO controller referred to by
>> this descriptor."  The pin number is zero based controller relative.
>>
>> The  ACPI device controller node includes all other resources needed by the
>> controller driver (ioport/mem base, range, interrupt, and Hardware ID used
>> to pair with a driver)
>>
>> Checked a board with two identical gpio controllers on it and it had two
>> separate ACPI device node entries. (with only different io address base and
>> interrupt resources)
>
> That's not really the situation that I'm thinking about. What I mean
> is for a gpio controller that is more convenient for Linux to support
> using multiple gpiochips (Linux internal detail), even though there it
> is described with a single ACPI node.
>

Ok, now I get it.

Yes, in case a driver uses several gpiochips internally for different 
banks of a controller then all would have the same acpi_handle.
acpi_get_gpio() would use the gpiobase of the first gpiochip that 
matches the handle, even if it's the wrong one.

I guess It's possible to write a driver like that.
The only acpi enumerated driver with the acpi_handle set (soon coming to 
upstream) is not done like that.

Do you think this is a case that should be solved now?  or just expect 
acpi gpio device driver to not use several gpiochips in one driver?

-Mathias