Re: [PATCH v2 2/2] Input: atkbd - Correctly map F13 - F24

[Hans de Goede]

Hi,

On 17-Mar-25 5:47 PM, Werner Sembach wrote:
> Hi again,
> 
> Am 17.03.25 um 13:06 schrieb Hans de Goede:
>> Hi,
>>
>> On 11-Mar-25 19:10, Werner Sembach wrote:
>>> Hi Hans, Hi Dimitry,
>>>
>>> resending this too on the v2 to not cause confusion:
>>>
>>> Regarding remapping KEY_ZENKAKUHANKAKU to KEY_TOUCHPAD_TOGGLE:
>>>
>>> Am 11.03.25 um 19:06 schrieb Werner Sembach:
>>>> Currently only F23 is correctly mapped for PS/2 keyboards.
>>>>
>>>> Following to this table:
>>>> https://download.microsoft.com/download/1/6/1/161ba512-40e2-4cc9-843a-923143f3456c/translate.pdf
>>>>
>>>> - F24 and Zenkaku/Hankaku share the same scancode, but since in real world
>>>> Zenkaku/Hankaku keys seem to just use the tilde scancode, this patch binds the
>>>> scancode to F24. Note that on userspace side the KEY_ZENKAKUHANKAKU keycode is
>>>> currently not bound in xkeyboard-config, so it is (mostly*) unused anyway.
>>> I think what the firmware vendor actually wanted to do was to send ctrl+super+f24 upon touchpad toggle. This would somewhat fall in line with, for example, the copilot key being implemented as shift+super+f23.
>> I agree that that seems to be the intent.
>>
>>> Following this, my suggestion is to do this remapping and handle the rest in xkeyboard-config
>> xkeyboard config already contains mappings for F13 - F18 and F20-F23 in
>> /usr/share/X11/xkb/symbols/inet
>>
>> So all that needs to happen there is map FK19 -> F19 and FK24 -> F24.
>>
>> And then teach KDE + GNOME that ctrl+super+f24 means touchpad-toggle.
> 
> Alternative suggestion, again following how the copilot key is implemented:
> 
> key <FK19>   {      [ F19 ]       };
> [...]
> key <FK23>   {      [ XF86TouchpadOff, XF86Assistant ], type[Group1] = "PC_SHIFT_SUPER_LEVEL2" };
> key <FK24>   {      [ F24, XF86TouchpadToggle ], type[Group1] = "PC_CONTROL_SUPER_LEVEL2" };
> 
> Then only xkb has to be touched again, but not KDE and GNOME.

Ah I did not know you could do this. Yes this sounds like a very good
plan wrt the xkbconfig changes and then indeed we can do all the handling
in xkbconfig.


> 
>>
>> We could maybe get away with also dropping the weird mappings for FK13 - FK18
>> and map those straight to F13 - F18, but we need the special mappings
>> for F20 - F23 to stay in place to not break stuff.
> 
> Good question
> 
> XF86Tools launches system settings on KDE.

Right, but XF86Tools is also send for KEY_CONFIG which makes more sense,
the question is are there any devices actually sending KEY_F13 in
a case where they really should be sending KEY_CONFIG instead.

Note this is unrelated to the XF86TouchpadToggle thing though, just
something which I noticed while looking at things.

> Looking at the links in the git log of xkeyboard-config (commit 1e94d48801bf8cb75741aa308d4cdfb63b03c66c and 01d742bc5cd22543d21edb2101fec6558d4075db) these seems to be device specific bindings that got accepted in the default config because the keys where unbound before.

I see, so it might be worthwhile to try and fix these, but in
a separate pull-request from the:

key <FK24>   {      [ F24, XF86TouchpadToggle ], type[Group1] = "PC_CONTROL_SUPER_LEVEL2" };

addition.

Regards,

Hans