[StarBook MkVI - Intel] Wonky system power states

[bryanvobo]

I've had power state problems for a while but am just now getting around to opening this issue about it.

I was previously running 8.95/1.18 which had a problem where it would power on/start the laptop any time I plugged in the barrel power connector, regardless of whether the coreboot setting for that was Enabled or Disabled. I upgraded to 8.97/1.19 which did fix the plug-in-power-on problem but also introduced a new problem where any time I rebooted or powered off the laptop it wouldn't "take" and I had to manually hold down the power button for several seconds.

I downgraded back to 8.95/1.18 because I'd rather deal with plug-in-power-on than manual-intervention-to-reboot-or-poweroff, but instead of only having the plug-in-power-on problem it kept the new problem as well.

Hoping to get rid of both problems, I switched to the AMI firmware (I flashed 1.3.0.bin but F2'ing into it says BIOS Version 1.2.0 and EC Version 01.05) but whatever 8.97/1.19 introduced has mostly persisted. Now the laptop does eventually reboot on its own after a ~15 second delay but powering off still requires that I hold down the power button for several seconds.

I noticed that 8.97/1.19 has been removed from this git repo (though it's still available from fwupd) so I'm guessing that somebody is already aware that it included something bad. How do I unbreak what has become broken?

---

Details on how I upgrade/downgrade the firmware/ITE/EC:

I've never been able to use fwupdmgr from my internal-SSD OS (Arch Linux with flashrom 1.2-4 and fwupd 1.9.6-1), it errors with:

failed to open device for prepare: failed to read layout from Intel ICH descriptor

A Star Labs support page warns that the "disk cannot be encrypted with LUKS", which my disk is, so I assumed that may somehow be the reason. I haven't troubleshot that any further because what does work is booting from an Ubuntu installation USB:

$ sudo dd if=/path/to/ubuntu-23.04-desktop-amd64.iso of=/dev/sdX bs=1M conv=fsync oflag=sync status=progress

installing flashrom packages from Canonical's repo, and running:

# flashrom -p internal -w <firmware.file> -i bios --ifd -N

To update the ITE/EC, I followed the directions from this Star Labs page - unpack the contents of efi-B6-I.zip (1.18.rom/1.19.rom, ifu.efi, startup.nsh) to the root of another USB drive, boot to EFI Shell while it's inserted, and let it countdown to run startup.nsh.

I'm not sure if I need to update the firmware and ITC/EC individually but that's what I've been doing. Until I switched over to the AMI firmware, where all I did was flash 1.3.0.bin without also running an EFI shell script.

[divico]

I've never been able to use fwupdmgr from my internal-SSD OS (Arch Linux with flashrom 1.2-4 and fwupd 1.9.6-1), it errors with:

failed to open device for prepare: failed to read layout from Intel ICH descriptor

This is because flashrom in arch is too old. As explained in the Starlabs documentation you need a more recent version. I'm also on Arch and with this it works

pacman -Qs flashrom
local/flashrom-starlabs 1.3.0rc2-1

I used the PKGBUILD from Manjaro to install it

[Sean-StarLabs]

A Star Labs support page warns that the "disk cannot be encrypted with LUKS", which my disk is, so I assumed that may somehow be the reason.

That's for standalone EC updates - the Mk VI doesn't use those.

[bryanvobo]

Thank you for the flashrom build script. It will be more convenient to update without booting from a USB stick. Somehow my brain didn't register the lower version even though flashrom 1.2-4 is right there staring me in the face.

And I'll stop doing the standalone EFI shell updates for EC and only flash the main firmware.

How do you recommend I proceed, to make the laptop behave normally? Should I try flashing 8.95.rom again? Or an older version? Or wait for a newer version?