ACPI battery state of host hardware stuck after installation of VBox

[Tpau]

After installating Virtualbox on Windows 8 (both 64 bit) the battery charge value (acpi charge level as reported by the machine's bios) gets stuck in one direction and stops updating into the corresponding direction (can be charge or discharge).

Before installing Virtualbox the charge level behaved normally (tested over multiple cycles).

The "one way only/stuck" ACPI charge value persists over reboots - even changing the operating system (multibooting) to Linux (Ubuntu 13.10) does not help!

Therefore this bug-report is most probably machine-specific for my Asus Ultrabook UX51VZ (Zenbook) with bios 203 or bios 204 - nontheless it indicates problems with the ACPI-forwarding portion of Virtualbox.

To resolve the problem I had to uninstall Virtualbox, re-flash the Bios (winflash with /nodate parameter), re-configure my bios-settings and then perform a cycle with the notebook's battery in order to get the battery back to a working state!

Suggestion: please review the acpi-forwaring-code for possible causes of the lockup (writing calls/race conditions?) - maybe add functions to disable the corresponding blocks at driver level (workaround).

Thanks!

PS: The problem has been reported to Asus as well as their bios is the component that's getting permanently stuck.

PPS: Tool used for better battery-metering in windows: CPUID's HWMonitor - normally shows charge updates every 1-2 seconds - the problem first occurred without it beeing installed, so it can be ruled out.

[Frank Mehnert]

Is it sufficient to install VirtualBox without starting any VM to observe the problem or do you need to run a VM?

In the latter case I would not agree with your analysis that the observed behavior should indicate problems of the VirtualBox ACPI code. The VirtualBox code uses the normal Windows API to determine the current ACPI status, see for instance here and other code in that file. It uses GetSystemPowerStatus() to determine the current status. My guess is rather that this is a problem of the BIOS / Host OS. Perhaps the BIOS does not like if that function is called frequently.

[Tpau]

Hm, good point @ frequent calling

I've just finished reviving the battery - will perform a test without any VM running. Only guest-OS ever running as VM was a 32bit Windows 2000

[Tpau]

It seems to depend on running the VM - I've installed Vbox - without running the VM there's no lockup yet. Test of VM without ACPI (vboxmanage modifyvm w2k --acpi off) failed due to ACPI-requirement of VM's HAL - but no lockup without running the w2k guest.

[aeichner]

Please reopen if still relevant with a recent VirtualBox release.