Investigate ModelTabletModeNoSuspend=1

[ChocolateLoverRaj]

I'm not sure if this is from using keyd v2 vs v1, or if it's a redrix vs jinlon thing, but ModelTabletModeNoSuspend=1 is now messing up stuff.

With ModelTabletModeNoSuspend=1:

• Side volume buttons never get disabled
• Keyboard never gets disabled

Without ModelTabletModeNoSuspend=1:

• Side volume buttons never get disabled
• Keyboard gets disabled in tablet mode

Tested in NixOS. I'm not using this script, but maybe someone should investigate this option on another Chromebook and undo my earlier pr if needed

[crystalcommunication]

I think this is going to be a per-device problem. I have a few convertible models to test.

(key: ✅ = enabled, ❌ = disabled. note that the 'correct' configuration is ✅ in the second column and ❌ in the third column)

with ModelTabletModeNoSuspend=1...

Device	Volume Rocker	Keyboard
MORPHIUS	✅	✅
VAYNE	✅	❌
LAVA	✅	❌

Notably, the tablet mode switch on my LAVA doesn't even work properly in Linux. GNOME does not activate tablet mode (automatically enable OSK and disable trackpad) but the board's embedded controller must be disabling the keyboard without the operating system's interference because that works fine.

without ModelTabletModeNoSuspend=1...

Device	Volume Rocker	Keyboard
MORPHIUS	❌	❌
VAYNE	✅	❌
LAVA	❌	❌

I know I said the tablet switch on my LAVA doesn't work. It is detected, it's state is just not updated properly when the hinges fold. When the device boots, wakes from sleep, or switches virtual terminals, the state does update. Not only does this break the volume rocker, it also prevents you from shuffling the virtual terminal to get your trackpad back by disabling the keyboard if it gets stuck in tablet mode.

For devices like MORPHIUS on which neither solution produces the desired result, I wrote this script which swaps the keyd config to one that rebinds each individual key to a noop, it works great and adapting it to other Chromebooks should be very easy.

[ChocolateLoverRaj]

Maybe we should use ectool to find out if the EC is detecting tablet mode properly. If it is, then maybe this can properly be fixed with changes to the Linux kernel.