Re: [PATCH] gpio: Describe interrupt-controller binding

[Thierry Reding <thierry.reding@avionic-design.de>]

[-- Attachment #1: Type: text/plain, Size: 3453 bytes --]

On Tue, Sep 18, 2012 at 08:55:40AM -0600, Stephen Warren wrote:
> On 09/18/2012 07:28 AM, Rob Herring wrote:
> > On 09/18/2012 03:51 AM, Thierry Reding wrote:
> >> In order to use GPIO controllers as interrupt controllers, they need to
> >> be marked with the DT interrupt-controller property. This commit adds
> >> some documentation about this to the general GPIO binding document.
> >>
> >> Cc: Linus Walleij <linus.walleij@stericsson.com>
> >> Cc: Grant Likely <grant.likely@secretlab.ca>
> >> Cc: Rob Herring <rob.herring@calxeda.com>
> >> Cc: devicetree-discuss@lists.ozlabs.org
> >> Cc: linux-kernel@vger.kernel.org
> >> Signed-off-by: Thierry Reding <thierry.reding@avionic-design.de>
> > 
> > Applied for 3.7.
> > 
> > Rob
> > 
> >> ---
> >>  Documentation/devicetree/bindings/gpio/gpio.txt | 33 +++++++++++++++++++++++++
> >>  1 file changed, 33 insertions(+)
> >>
> >> diff --git a/Documentation/devicetree/bindings/gpio/gpio.txt b/Documentation/devicetree/bindings/gpio/gpio.txt
> >> index 4e16ba4..8d125b0 100644
> >> --- a/Documentation/devicetree/bindings/gpio/gpio.txt
> >> +++ b/Documentation/devicetree/bindings/gpio/gpio.txt
> >> @@ -75,4 +75,37 @@ Example of two SOC GPIO banks defined as gpio-controller nodes:
> >>  		gpio-controller;
> >>  	};
> >>  
> >> +If the GPIO controller supports the generation of interrupts, it should
> >> +also contain an empty "interrupt-controller" property as well as an
> >> +"#interrupt-cells" property. This is required in order for other nodes
> >> +to use the GPIO controller as their interrupt parent.
> 
> Surely this is generic information for any interrupt controller, and
> hence doesn't belong in the GPIO binding?

LinusW requested this in order to avoid having to list these properties
in every GPIO controller. I suppose that having it in an extra binding
for interrupt controllers might make sense as well, but in that case we
should probably provide a reference because the GPIO binding is where
people are most likely to look for this information.

There is Documentation/devicetree/bindings/open-pic.txt, which already
lists most of this information, so maybe a reference to that document
will do just as well?

> >> +If #interrupt-cells is 1, the single cell is used to specify the number
> >> +of the GPIO that is to be used as an interrupt.
> >> +
> >> +If #interrupt-cells is 2, the first cell is used to specify the number
> >> +of the GPIO that is to be used as an interrupt, whereas the second cell
> >> +is used to specify any of the following flags:
> >> +  - bits[3:0] trigger type and level flags
> >> +      1 = low-to-high edge triggered
> >> +      2 = high-to-low edge triggered
> >> +      4 = active high level-sensitive
> >> +      8 = active low level-sensitive
> 
> That certainly shouldn't be in the generic GPIO binding; the format of
> the interrupt specifier is determined by the binding for the individual
> device that is the interrupt controller. Just because a device is also a
> GPIO controller doesn't mean that it has to conform to a specific format
> for the interrupt specifier.

I think it does make sense to provide a description of the most commonly
used variants. The above certainly is what the majority is using and
many of those that do not use one of the predefined irq_domain_xlate_*()
functions reimplement them with some additional checks or conversions.

Thierry

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]