Machine Reboots opening Welcome or Software Boutique

Hi,
Trying to leave windows and to use Mate 19.04. (Aorus Z390 Master +I9-9900K+Samsung M2), no separate video card. It worked fine for a couple of days (doing silly things just to test the setup) , but then the machine started to reboot if I open Software Boutique or if the Mate Welcome screen opens.
AFAI can tell the logs under /var/log do not reflect any crashes.
I have rotated the 4 memory sticks (using only 1 right now) with no luck.
The problem can be reproduced if using the Mate USB and then executing the welcome screen.
I installed Windows 10 in a separate HDD to run the Intel CPU tests, they come clean.
memtest86+ comes clean with the single stick I have now.
Windows 10 seems to work fine
kubuntu seems to work fine from the USB (tried to install software boutique on it without luck).
Mate seems to work fine if I do not use the software boutique or run the welcome screen.

Nothing is overclocked.

I am tempted to fault the mainboard, but the error is so specific that I am not sure if I am barking to the right tree. I am running out of ideas....

thanks for any help you can provide

Gustavo

I've never heard anything like that before, but if the problem is so specific to those two apps only, here's what I would try:

  • use the computer without opening the apps to make sure it's otherwise stable
  • try installing another random snap app and seeing if the problem happens with that new app (welcome/boutique are snaps, the only of this type shipped with Ubuntu MATE, all other apps are debs)
  • if the computer is fine with the new snap, then I'm out of ideas, but if the problem can be reproduced with another random snap, then my guess is something in the BIOS, like Secure Boot, and the snap confinement are not working well together)

At this point this is all a guess. I don't know how snaps work nor how the BIOS would care about that, but those would be my first steps to try to rule out.... Hope this helps

Hi vkareh,

thanks a lot for your help. I tried other snaps (classic and not) without problem. I tried many other things like uninstalling other minor things, no results.

Then since you mentioned BIOS, I switched the BIOS from position 1 to 2 (I have the SB switch in position 2, which means independent BIOS) and now everything is working (so far): IN order to install ubuntu I had to upgrade the BIOS to F9, and the mainboard seems to try to copy the old over the new one. At the end I set them up separately and did 2 flashes (1 for each BIOS) it seems there is something wrong with BIOS 1.

I had no less than 20 different machines through out my life (Pre made and assembled, at work and personal) all of them had a single BIOS chip, and they have NEVER failed to me on any flash or after it, and here I have a machine with 2 BIOS which is supposed to prevent BIOS corruption problems and is giving me all this trouble!!!!!.

thanks

I am happy to hear you have found a workaround. But like vkareh I have not seen a this issue before, could you please run inxi -F and paste the output here please?

That's a NICE machine. Hopefully the "not overclocked" part is only while dealing with this problem: it would be almost criminal to leave it that way. :slight_smile:

Like vkareh, I think the fact that the only two problem apps are snaps has to be relevant, regardless of how strange it seems, and also despite other snaps not triggering the issue.

Personally, I would probably never install the same BIOS to both slots on a dual-BIOS system: there's no real advantage to doing so, and potentially a significant loss from not being able to check for regressions or recover from a bad release. I've been through even more machines than you over the years, and only once had a BIOS update completely brick the machine (which was Not Fun: I literally had to have the mobo manufacturer send me a new chip and replace it on the board!) but it certainly is rare.
The BIOS realistically CAN'T silently fail to flash properly, as it should be checksummed afterwards and any issues with the EEPROM detected at that point, so I would think it has to be a setting that's causing the problem. Next time you're in there, look for anything like "Windows 10 Fast Boot", which stops the BIOS from actually initialising the hardware properly, and turn that off.

Good luck - hopefully you'll find the answer soon, or there'll be another BIOS update that fixes it: I see there's a 10b version already, to fix an issue with XMP. Since you DO have the luxury of a dual-BIOS system, I'd think it's well worth updating one of them to that and seeing if that improves things.

One thing the 2 snaps have in common is that they're powered by WebKit. It's possible it's trying to use hardware acceleration for rendering the "web page" and it crashes. There's been a weird bug with these applications rendering a white window under VirtualBox if 3D acceleration was turned on, normally because WebKit suffered a segfault.

That said, a reboot is very unusual :face_with_raised_eyebrow: ... presumably it just instantly hard reset? Does the window appear? Normally, the worst case scenario would be a kernel panic with Num/Caps lock flashing on the keyboard. :skull_and_crossbones:

The Welcome/Software Boutique snaps currently force software rendering, but it could be worth seeing if it happens with hardware acceleration - to rule out integrated graphics bugs:

LIBGL_ALWAYS_SOFTWARE=0 ubuntu-mate-welcome

Something to possibly check: Open Additional Drivers (Software & Updates) and see if enabling (or disabling) Intel microcode makes a difference (after a restart). See this answer for details on what microcode is:

Further reading - https://wiki.debian.org/Microcode

Microcode are like patches for inner workings of the CPU. It isn't retained across reboots since it's applied at BIOS/UEFI start-up and/or again when the OS boots.