I observed that the camera kill switch of my Librem 5 does not reliably disable the camera. If I put it to the down position and start megapixel most times I get the expected live view of the camera only sometimes the live view stays black, but is reenabled when I restart megapixel a few times.
Is this expected behaviour? Does my Librem 5 has a hardware error?
Absolutely not expected. Are you sure the view is not just the last frozen picture?
I tried again and I cannot find any difference dependent on the position of the kill switch. There even isn’t a message in dmesg when I move the switch.
The screen staying black sometimes seems not to be connected to the state of the kill switch. It just happens sometimes.
Another thing I was wondering is that megapixels is always in a manual mode: I have to set parameters like exposure and color balance. Otherwise it just uses the settings it starts with. No automatic adjustment at all.
Expected. Sorry about that.
Dmesg doesn’t log the switch position, but udev should:
sudo udevadm monitor
Do you get anything there?
Are there any plans for a real camera app, one for unskilled people in camera arts, like me 
In udevadm I can see the change on all of the other kill switches, but no reaction to moving the camera kill switch…
Yep, sadly it’s a lot of work.
I think this might be a hardware issue. If you don’t feel like opening up the device, you could contact support@puri.sm
So are these more like software switches under the hood? I thought the switches were supposed to physically disconnect the power to the associated components?
They disconnect in hardware, and also notify the software about their state.
ISTR that the HKS ‘just’ cuts the control lines, not the actual power to the respective devices.
No. But it is possible that a hardware switch is faulty.
I think the exact behaviour varies between the three HKSs. Study the schematics? …
No, it actually removes the power (using a helper chip to prevent frying). That applies to all switched peripherals.
Yes, in the @ChriChri’s case it sounds like the hardware switch doesn’t actually do the actual switching and most likely needs to be replaced.
How does the helper chip fit into the overall design? For example, could a helper chip with modified firmware render the kill switch ineffective?
I suppose Dorota meant a transistor voltage regulators that are used to actually cut the power off. You can see everything in the schematics.
The chip is basically a handful of basic components to make voltage switching predictable. There’s no space for any firmware. If you decided to replace it with something nefarious, a naked wire would probably suffice to keep the device on at all times. But no switch can fight a bypass wire.
when I get my hands on screwdrivers I’ll try…
So the idea would be that the helper chip is just an integrated circuit without a memory for firmware or software instructions?
Edit: Sorry, I think I’m overthinking this, as i don’t know that much about hardware design. By saying:
I assume that this is just hardware-implemented (using technology I am not familiar with) logic to prevent frying when the switch is used. Nevertheless, any sort of software component to the switch mechanism would be suspect to me without further explanation. Sorry for not getting it—I looked at the schematics, but I don’t have much ability to answer my questions from them.
Those kinds of chips don’t have any digital memory, and they don’t have a CPU on which to execute any kind of firmware.