So maybe Purism is overselling the “hardware-ness” of the kill-switches. It seems like it would be the most straightforward and effective design to have the switches literally break the connection of the power supply to the corresponding components–like how a simple lightswitch in a house works.
But it seems that these switches do not work that way. Dlonk would’ve never seen the behavior he mentions if that design was being used.
I have read a Purism employee describe a “helper chip” that allows the power cutoff to be more gradual, to avoid damage to components.
One way Purism can honestly say that they are hardware switches, regardless of how they function under the hood, is that they are physical switches you can feel with the tips of your fingers, and therefore, they are “hardware”, even if they just send a signal saying “shutoff that component please”, that signal being processed by software somewhere in the device.
One related question, are the PinePhone switches more “real” than the Librem 5 ones?
For reference, if you were to completely remove any one of the switches from the main board then that function would be always-on by default (pull-ups on the board would keep the active-HIGH enable pin of the load switch IC at a logic HIGH level). Even though this failure mode should be caught in QC, it is preferred to the opposite being the case, which would manifest in a way where the function (WiFi/BT, WWAN, cam/mic) would not work regardless of the position of the switch (off or on). The switches are not gated by software. The schematic is here for your reference: librem5_mainboard_schematic.pdf · master · Librem5 / l5-schematic · GitLab
Do your other two switches work as intended?
Your Librem 5 would need to be disassembled, inspected, and likely repaired. You should contact support if you haven’t already.
It’s my understanding that the HKS tells the software watchdog at kernel level to disable in software as the connection in the HOS physically disconnects power. Once power is restored, watchdog should re-enable and show the proper icons.
Yes, after this weird “fluke” happened where WiFi and Bluetooth were available while their HKS should have killed them, when I switched the switch on/off again everything went back to working as expected. Other switches are working as expected.
I really want to like Purism products and I use this phone every day and have been since I got it. If I send it in for repairs - for a one-time “hardware bug” the support team might not even know how to reproduce, my understanding is that it may take them many months for the back and forth because Purism usually operates in a human, slow fashion.
This phone is a Liberty Phone so if I stop using it I would probably fall back on my Librem 5 (China version) but its hard drive is full and its RAM is smaller and I don’t know if I want to downgrade if I can avoid it.
A long time ago, maybe in February, I ordered a replacement Liberty Phone because I like the device and use it every day, and I figured I would break it in the months or years before the order shipped. Is it reasonable for me to hold out until that Liberty Phone ships before I sent this one for repairs, or do we know if those are more like sitting on another 4 months before shipping?
thank you for sharing this. I could not affirm this. Here my Phone disable the wireless connection and posh disabled and mark my wireless Network with a grey shadow.
However, if you need a shut down you should verify it by dmesg.
Just for the Nerds of us… you could also use a Software Firewall instruction to dismiss Network packages. The Hardware Kill Switch is only for us folks who do not trust the Kernel and Hardware and Network Driver in the first place.
To have no Phone available and do not record Data is the best way to avoid something. Its just the sad that in this information warfare world where we are, that everyone have to think about this.
No. The schematics are public. Check for yourself.
In the original post, I’m not sure whether you mentioned the state of the cellular modem kill switch. If the WiFi is off and the cellular modem is on then you should still be able to post to this forum (in a general scenario). I can believe that the status icon might get confused occasionally and not appear / disappear when it is supposed to.
It is probably too late to investigate but I would be checking at the shell prompt for the true state of the two networks and looking for journal entries.
In the attached video, I use the modem switch and the wifi swtich to turn off modem and turn off wifi, and for good measure also turn off the mic/camera switch, and then I write and publish a post on this forum.
So it is not necessary nor useful to speculate that “maybe” I wrote the post using the modem. Frankly, speculating like that without reviewing all of the information that I provided puts you at risk of sounding like a chatbot. Given the other incredibly helpful posts you have made on the forum, I know that you are not – but such speculation is unnecessary.
You are however entirely correct that it would have been more informative for me to go log-hunting for more information. But presumably what we would find is that the software that disables the WiFi when the WiFi switch is turned off (moved downward), on this one occasion, did not disable the WiFi. However, toggling the switch again kicked it back into gear so that the software treats WiFi as disabled now every time since then when I turn off the WiFi switch.
I have been using Librem 5 as my primary phone for a year. I am well accustomed to the state of the modem switch, and frequently have only one or the other of wifi or modem enabled.
Okay but what I saw is that sometimes the hardware kill switch doesn’t kill. So what’s even the point of trusting anything?
