Bad design, indeed.
But look at the number of handheld devices where the battery is more or less impossible to remove and there are no switches to surprise the modem by cutting its power. I can understand if modem makers would say “hey, just do a proper shutdown” so they would not have to bother…
Thanks Caliga great info.
Hi Zero,
Please see the youtube video below, watch from 14:40 mins onwards.
This review is of the pinephone, and different OS and their operation on the pinephone, as you will hear at 14:40 onwards the reviewer discusses how he had to restart the phone for the HKS to work?
My point on this post was, HKS should work almost instantly, it should be a mechanical feature not software based, which leads me to suspect that the HKS on the pinephone are not the same as on the Librem 5…
There are comments by other viewers from youtube regarding this issue under the video.
Hi prog-amateur,
I have responded to this comment from another user mate, video and explanation at bottom of the page.
Correct kieran, see video at the bottom of the page.
Hi Iperkins2, video at the bottom of the page.
Just putting it out there but where the functionality is embedded within the SoC that may not be possible.
Where the functionality is on a bus that expects devices to be removable then, yes, the device should cease operation immediately and be hardware-unusable immediately - although it may take longer for software to recognise the situation and make things tidy (e.g. pending requests to the device may have to time out before failing, or be cancelled by the operating system and fail). For V1.0 it probably wouldn’t be a great idea to operate a kill switch while the device is being used. 
Although I am an electrical engineer, I want to keep the principal very simple here just to make sure that communications are clear. If you pull the power plug to a lamp out of the wall, the light always powers off immediately, regardless of whatever else is going on in the room. It goes out because the mechanical connection between the lamp and the power circuit in the wall have become physically separated. Does this analogy apply to the Librem 5 kill switches? I would hate to find out that it’s more like an instruction is sent from the switch on the phone to the CPU that tells a peripheral to pull the plug for you.
Yes it compares.
There are some mechanisms in place to prevent the wifi/bt and baseband modules from frying when you turn them on.
But the your analogy applies
(From the V098 schematic: https://source.puri.sm/Librem5/l5-schematic)
This is the relevant schematic piece. WIFI_POWER_KEY comes from the kill switch. U40 is a load switch where EN is active high. Flipping the Wifi kill switch pulls the EN pin low and disables the load switch which causes 3V3_WIFI to be disabled as well. 3V3_WIFI is the VCC rail for the WIFI/Bluetooth card.
So, as you can see, the kill switch actually turns off power to the WIFI module.
@LinuxNew & @kieran :
Thanks ! I’m reassured. I thought there were fake switches, but these switches look like the ones on PCBs (for example external RAID cases where you have to press “Reset” to confirm the new configuration).
Thanks for the link. 
