After installing a new kernel last week my computer stopped displaying anything on reboot until sometime along the Linux boot process, at a point the display mode changed, I think. This made it impossible to see the POST or the BIOS setup. I did the CMOS reset process described in the manual: switch the jumper from 1-2 to 2-3, turn on for 10 seconds, then off for 5, then switch the jumper back. The jumper is clearly labeled.

This made the BIOS setup come on, then automatically reboot. After that the computer does nothing but re-boot infinitely, not even display a screen; the fans spin up for a few seconds. I disconnected the drive, plugged in a boot flashdrive, but the same still happens: no display, no loading of an OS, the capslock switch doesn't toggle its LED.

I can't see anything wrong, get no beeps. I'm using the same keyboard and display I'm using now, so those work.

Later:
In addition to the fans spinning up briefly, the power LED flashes for a fraction of a second, the built-in speaker clicks faintly (not beeps).

I disconnected the power switch and the thermal sensor, removed the RAM, disconnected the SSD and optical drives, tried every position for the CMOS_CLR jumper, to no avail

You may need to go into the bios again to set it up mote completely.

I think it has a setting, "reboot on all Errors".

If you switch that off you might have some progress with seeing any other messages.

Make sure the sata ahci is engaged, etc.

How do I 'go into the BIOS' when the boot process stops before that?

idk, hold the key while resetting.

Another reason it is resetting could be heat...

but my first thought is the bios is trying to boot something that is not there,

perhaps on an old setting.

It doesn't get that far, even far enough to respond to a keyboard.

Well it's 20C and it happens within 3 seconds of turning on. I disconnected the thermal sensor, in case it had gone bad.

I've never seen a BIOS try to boot something that isn't there. They always complain that there's no boot device.

Mine currently does that! I put in a new drive and now the default boot keeps cycling: bleep, reload, bleep, reload...

Can you get a boot menu by pressing F12 or whatever key provides your custom menu? That's how I'm getting in at the moment.

As hazel replied, try the BBS key. If that fails, redo the BIOS reset, but instead of the jumper process, unplug the power cable and remove the motherboard battery for a good while. Eat something, take a nap, be sure you don't have an obstinate NVRAM chip. Also investigate BIOS update availability. Some newer motherboards can automatically update if you put the update USB in the appropriate orifice and push the correct button on the motherboard.

Some of this may be relevant:

https://askubuntu.com/questions/1404...ystem-bootloop

fbx64.efi is the fallback.efi boot loader that runs when the regular boot fails. It looks for a CSV file (like \EFI\ubuntu\bootx64.csv), creates an EFI boot entry for it, and reboots.

fbx64.efi` isn't supposed to be installed on removable media - grub-install --removable skips it. The problem is that removable media is often booted by running the default boot loader, and if that's fbx64.efi then the EFI boot list will be changed every time the USB drive is booted.
…………………………………………………………………………………..

grub-install in Ubuntu 22.04 installs the fbx64.efi into the /EFI/BOOT/ directory and during the boot the shim loader (in debug mode, activated via mokutil --set-verbosity true, I spent a while to debug this) shows Invoked from removable media path, ignoring boot optionsshim.c:912:load_image() attempting to load \EFI\Boot\fbx64.efi and immediately shows Reset System log. Removing a single \EFI\Boot\fbx64.efi file solves the Reset System issue. After the file removing the UEFI secure boot works just fine.

Which is to say there's something there to load. Mine re-boots before it accesses any drives. It reboots if I remove all drives, RAM, connectors to ports, the power switch, the thermal sensor. It comes on for a second or 2, long enough to flash the power LED and spin up the fans.

No.

Already tried. The boot process doesn't get that far.


Tried.

The question I really mean to ask is: what can make a computer reboot before it even loads a drive, or even the BIOS. It reminds me of times I've overloaded a power supply: the device would start up, draw so much current that it drew down the voltage, which caused a shutdown. I've disconnected everything I can from this except the CPU. Of course it could be the power supply no longer capable of providing rated current; that's hard to test. I can measure that the voltages come up to rating when it turns on, but do they drop because they're being turned off by some automatic process or because the PS can't perform?

And why does it start on when I plug it in, even with the power switch disconnected? That's like a short, but I see no damage, nothing gets hot. If I hold the power switch in, it'll turn off, which belies the short theory.

The manual seems to think the thermal sensors set off alarms, not shut down the computer.

What's the failure mode of a failing CPU?

That's normal BIOS behavior on first start with nothing in NVRAM. It wants you to set the clock if nothing else.

How do I set the clock if it re-boots after a second, before it displays anything on a screen, before it recognizes a keyboard. I've measured the voltages: they come up to rated voltage for a second, drop off to near zero, then come back up again for a second, repeatedly.

If under warranty, I would contact the support provider of the PC for guidance and/or an RMA.

Ha!

Reading the above posts I realised one thing we haven't mentioned is the powersupply and connections.

If this is a portable computer it may not be so easy to check.
https://static1.howtogeekimages.com/.../PSUMulti1.jpg
The bios would give a beep code, and/or display cpu not found, or something like that if the motherboard is functional.

Corrosion in the motherboard connectors is common (around here at the beach).