On 22/07/14 12:40, Markus Niebel wrote:
> Am 16.07.2014 12:01, wrote Simon Budig:
>> On 15/07/14 19:43, Markus Niebel wrote:
>>> - when adding an int-gpios node to devicetree and parsing this
>>> gpio the gpio will be configured as input - IRQ can be seen by
>>> CPU
>>>
>>> Question:
>>>
>>> - shall we add an int-gpio property to enable gpio pin config as
>>> input or
>>
>> Shouldn't it be possible to configure the pin as input directly
>> from the device tree, indepently from the touch driver section?
>
> I don't see, how to do that, maybe I did miss the right point.
Try looking at
Documentation/devicetree/bindings/pinctrl/pinctrl-bindings.txt
There you have the "input-enable" generic property.
> Global question is: how can we make sure, that the driver that
> request the IRQ requests also the GPIO (implicit or like in the old
> platfrom data days explicit)? But thats beyond input devices and my
> knowledge.
The driver does not depend on the IRQ coming from a "real" CPU gpio. It
*could* be an external gpio expander or whatever (whether that is a good
idea is debatable of course). When I submitted the first version of the
driver it had bits in it to configure the pin properly. The feedback
basically was to remove that, because it made some assumptions that you
can map the IRQ to the relevant GPIO, which is not always possible.
I hope that helps.
(But really, your bootloader really should have a *very* good reason to
configure the touch irq pin as output - I have the impression that right
now there are two outputs driving against each other in your setup - not
healthy for the hardware).
Bye,
Simon
--
Simon Budig kernel concepts GmbH
simon.budig@xxxxxxxxxxxxxxxxx Sieghuetter Hauptweg 48
+49-271-771091-17 D-57072 Siegen
Attachment:
signature.asc
Description: OpenPGP digital signature
