VBox 4.0.10 → 4.1.0 migration: can't restore some guests' state (ICH9 only)

[anonymous]

I had VBox 4.0.10 installed. Suspended all running Windows VMs before VirtualBox package upgrade. Upgraded my Linux kernel (2.6.32 → 2.6.39), then rebooted. Started VirtualBox package upgrade, everything was fine. Then tried to restore all my saved VMs. Most are started up fine, but 2 has been throwing an error: ich9pci#0: Device in slot 0xf0 has been removed! vendor=0x8086 device=0x244e [ver=2 pass=final] (VERR_SSM_LOAD_CONFIG_MISMATCH) A reset has helped.

[Jim Avera]

Same here, VBox.log ends with an identical error (except for aIID).
However my host kernel is 2.6.35-30-generic (Ubuntu 10.10) and the suspended client was Ubuntu.

[Frank Mehnert]

This comes from using the experimental ICH9 chipset.

[aeichner]

Please reopen if still relevant with a recent VirtualBox release.

[pantweb]

Got the same issue when updating from VBox 5.1.26 to 5.2.6.

Apertura di una sessione per la Macchina Virtuale Xubuntu 16.04 Backup non riuscita.

ich9pci#0: Device in slot 0xc0 has been removed! vendor=0x8086 device=0x2448 [ver=2 pass=final] (VERR_SSM_LOAD_CONFIG_MISMATCH).

Codice di uscita:E_FAIL (0x80004005) Componente: ConsoleWrap Interfaccia: IConsole {872da645-4a9b-1727-bee2-5585105b9eed}

[VVD]

Replying to pantweb:

Got the same issue when updating from VBox 5.1.26 to 5.2.6.

…

ich9pci#0: Device in slot 0xc0 has been removed! vendor=0x8086 device=0x2448 [ver=2 pass=final] (VERR_SSM_LOAD_CONFIG_MISMATCH).

Same here. There is bug report in FreeBSD's bugzilla: ​https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=225288

[Socratis]

Do not use ICH9. It is experimental. Which means it has bugs that they're aware of, that's why it's called experimental. Or else not even "alpha"...

[Klaus Espenlaub]

This is a side effect of a code cleanup in 5.2.6. In the past with ICH9 there were unconditionally 2 bridges present, and now they're only present if the VM config needs them. The result is what you observed - a saved state incompatibility. It would be possible to fix this, but it's very hard to justify this effort for a feature which is marked as experimental, without having any explanation why ICH9 was picked.

If someone volunteers (and the error message is a good hint where the change would need to be) then of course we'd integrate a sensible fix/workaround, but otherwise it'll simply stay this way.