Windows 1803 and later crashes on KVM

[Bruce Campbell]

For a bionic host, using the current public kvm modules, KVM is not capable of booting a WindowsInsider or msdn Windows 1803 Windows Server iso. In snstalling from an ISO from a started windows 2016 guest results in an unbootable and unrepairable guest.

The hardware is a threadripper 1920x with 32GB of main memory, disk mydigital BPX SSD and WD based 4 column RAID 5 via mdadm.

[Daniel Berrange (berrange)]

This sounds like the same problem we're investigating in Fedora/RHEL land affecting guests with EPYC CPU, or host-passthrough from an EPYC family 17h host. Workaround should be to use "-cpu Opteron_G5" for now

https://bugzilla.redhat.com/show_bug.cgi?id=1592276
https://bugzilla.redhat.com/show_bug.cgi?id=1593190

Looks like windows is tickling an undocumented MSR and we're trying to find out what this MSR is supposed todo and thus how to handle it in KVM/QEMU

[Bruce Campbell (yakman2020)]

That is very helpful. I'll try it.

I standby to try a fix as soon as someone has one.

[Bruce Campbell (yakman2020)]

Regrettably opteron_g5 is not a workaround. I get a complaint that my cpu doesn't provide xop, fma4, tbm.

I've tried a number of other cpus including nehalem, phenom and athlon with similar results.

[Bruce Campbell (yakman2020)]

Opteron didn't work do to some missing features. I was able to get nehalem to come up but that looks like the closest.

[Bruce Campbell (yakman2020)]

Is there any way to define a new CPU model that isn't EPYC (maybe ryzen?) but is feature compatible with the threadripper?

[Emanuele Faranda (black-silver)]

I ran into the same problem on threadripper 1900X. I was using cpu type "host-passthough" and it crashed. I fixed the crash by disabling the MSR with

kvm.ignore_msrs=1

as describe in https://forum.level1techs.com/t/windows-10-1803-as-guest-with-qemu-kvm-bsod-under-install/127425/10

[Thomas Huth (th-huth)]

The QEMU project is currently considering to move its bug tracking to another system. For this we need to know which bugs are still valid and which could be closed already. Thus we are setting older bugs to "Incomplete" now.
If you still think this bug report here is valid, then please switch the state back to "New" within the next 60 days, otherwise this report will be marked as "Expired". Or mark it as "Fix Released" if the problem has been solved with a newer version of QEMU already. Thank you and sorry for the inconvenience.

[Launchpad Janitor (janitor)]

[Expired for QEMU because there has been no activity for 60 days.]