Diagnosis...
Update supplied by Microsoft?
Worried about The Rise Of The Machines in these troubled times? Cheer yourself up with another example of what computers get up to when humans aren't watching, courtesy of your self-isolated vultures at The Register. Today's computer behaving badly is of the fruity variety as a Raspberry Pi-powered digital sign had a crack at …
A Pi has crashed. In a standalone game thingy. I'm supposing it doesn't get updates, so it doesn't change. What can possibly have brought it to its knees ?
Given that each CPU failed to launch (apparently it has two), the problem shouldn't have anything to do with associated hardware (like spinning rust that failed to respond). Looking at the messages, each CPU has a stack exception, and then fails to complete startup.
So the question is : what threw a stack exception ?
At least the Pi gives us a screen of useful diagnostics, and not just a bluescreen or "an error has occurred" type message. And we know it spent 3.88 seconds trying to get started, before it chilled and relaxed and ended its panic.
Long Live the Pi !
(I suspect a wobbly PSU - give it a cold hard restart and it may be happy once more)
Definitely not a BCM2835 even though the message shown reports it as such.
Such misreporting has caused much confusion for new Pi users who don't understand why their board claims to be something other than what they thought they had purchased. It has left some asking if they have been sold a fake or counterfeit item or a supplier had sent the wrong thing.
The problem is apparently "a Linux issue" which the Raspberry Pi Foundation don't care enough about to fix themselves.
The problem is apparently "a Linux issue" which the Raspberry Pi Foundation don't care enough about to fix themselves.
It was fixed in their version of the Linux kernel, but they are trying to keep the kernel as "vanilla" as possible now that Pi support has been added upstream.
Unfortunately the upstream devs decided that all Raspberry Pis would be recognised as BCM2835 in order to identify the platform. Hopefully one day they will correct that.
At least the system had been up and running. Just imagine what's going to happen when all the little systems that are down right now finally get started up again and wheeze reluctantly back to life. Licences expiring, automatic updates downloading : it will be a bonanza of bork. And we all know about the mysterious increased likelihood of systems breaking themselves the longer they remain untouched, even when no hardware component has failed............
I’m not sure that’s always the case.
From a hardware point of view the shock to a system is at power up.
And even software For example I ran about ten windows 7 “machines” on vSphere on the “one task per machine” requirement. And unless a critical update forced a restart, some of those machines were never rebooted. In an 18 year period some of those machine never fell over. Probably the “one task per machine” is the clue.
A quick and relatively cheap fix has been provided by a couple of the makers of peripherals for the R-Pi. It is a "UPS" that uses NiMH cells and a simple monitoring circuit to charge the battery from the Pi PSU. My improvised version has allowed a "mission critical" Pi (actually just an Office Directory for a big office building I work in) to continue to run continuously for over three years despite brief power outages and dirty mains. It has a little, irritating beeper that sounds if the power is unplugged, and a sign that illuminates requesting that the power is reconnected! This overcomes the "cleaner unplugged it" problem!
Thanks for reminding me about Faro - really ought to ask the family if they'd be up for the F festival again this year (here's to hoping it's still on). Think music festival held in a Portuguese castle complete with e-sports arena for to kill time before the music starts (even had a cresh for the kiddies so you could mooch off and and grab some beers in before it kicked out at 10).