Re: [PATCH 2/4] dt-bindings: touchscreen: add virtual-touchscreen and virtual-buttons properties

[Michael Riesch]

Hi Krzysztof,

On 5/12/23 09:20, Krzysztof Kozlowski wrote:
> On 12/05/2023 09:08, Michael Riesch wrote:
>> Hi Krzysztof,
>>
>>>> These buttons are actually physical i.e. printed and with a given
>>>> functionality, but still part of the touchscreen as the physical device
>>>> is not aware of them. Therefore they only make sense in the touchscreen
>>>> context.
>>>
>>> So basically you still have the same touchscreen under the buttons and
>>> these are not separate devices. Whether someone put a sticker on
>>> touchscreen, does not really change hardware behavior and it's up to
>>> system to handle this. What you are trying to do here is to create
>>> virtual buttons through Devicetree and DT is not for that.
>>
>> I have already addressed a similar statement here:
>> https://lore.kernel.org/lkml/20230504042927.GA1129520@quokka/t/#m1a93595c36535312df0061459a1da5e062de6c44
>> but let me extend this comment a bit.
>>
>> The notion of "someone putting a sticker on touchscreen" does not really
>> reflect the use case we have in mind. We are talking about devices that
>> are shipped from the factory in a particular configuration, e.g.,
>>
>> +-----------------------+---------+
>> |                       |  power  |
>> |                       |  button |
>> |   touchscreen         +---------+
>> |   (behind: display)   |  return |
>> |                       |  button |
>> +-----------------------+---------+
>>
>> Here, the real touchscreen is larger than the display. The display is
>> behind the "touchscreen" part. Behind the buttons there are symbols
>> engraved in metal or printed foils or what not. I just would like to
>> make it clear that these symbols are not going to change.
>>
>> We believe that the engraved or printed symbols actually define the
>> (expected) hardware behavior. Of course there is a virtual notion to
>> that, and of course it would be possible to let the power button work as
>> return button and vice versa in software. However, the user sees the
>> engraved or printed symbols (which are not going to change) and expects
>> the device to react appropriately.
> 
> OK, I actually missed the concept that display is not equal to the
> touchscreen ("screen" actually suggests display :) ). In your case here
> it sounds good, but please put some parts of this description into this
> common binding. The sketch above is nice, especially if you can point
> where the virtual origin x/y are. Picture is thousands words.

I think this can be arranged :-)

>> Now if you suggest that the system (that is user space, right?) should
>> handle this, then the question would be which component is supposed to
>> handle the touchscreen events and react accordingly. I don't have an
>> answer to that and hope I don't need to find one. But independent of
>> that, a configuration file is required that defines the area of the
>> virtual buttons etc. Wouldn't this be similar to the (mostly) beloved
>> xorg.conf containing the definitions of input devices?
> 
> If the case was a bit different - e.g. display is everywhere - I could
> easily imagine that on the device rotation you want to move
> (re-position) the buttons. Or if user has some accessibility option
> enabled, you want bigger buttons. Then it would be a prove that it must
> be configured and managed from user-space. How, I don't know.

I fully agree here. If user space is able to draw the symbols, then
naturally it is also responsible for handling the events appropriately.
This could happen in an application with a GUI or on display
manager/compositor level (e.g., weston controller or similar).

But I take it that for the situation sketched above the device tree
approach is the correct one. Documentation should be improved, though.

Best regards,
Michael