TL;DR - this is intentionally off by default and opt in.
To use it you need:
- in the host set kvm.nested=1
- depending on your other virt stack set you might need to set host-passthrough or the sief2 feature (qemu -cpu host or ...,sief2=1)
Then the guest will have virt features "sie" in /proc/cpuinfo | grep features
If anything we should consider backporting the "needs opt in" to Xenial.
That would be the following patch:
commit a411edf1320ed8fa3b4560902ac4e033c4a72bcf
Author: David Hildenbrand <email address hidden>
Date: Tue Feb 2 15:41:22 2016 +0100
KVM: s390: vsie: add module parameter "nested"
Let's be careful first and allow nested virtualization only if enabled
by the system administrator. In addition, user space still has to
explicitly enable it via SCLP features for it to work.
Hi,
I investigated this a bit.
TL;DR - this is intentionally off by default and opt in.
To use it you need:
- in the host set kvm.nested=1
- depending on your other virt stack set you might need to set host-passthrough or the sief2 feature (qemu -cpu host or ...,sief2=1)
Then the guest will have virt features "sie" in /proc/cpuinfo | grep features
If anything we should consider backporting the "needs opt in" to Xenial. a3b4560902ac4e0 33c4a72bcf
That would be the following patch:
commit a411edf1320ed8f
Author: David Hildenbrand <email address hidden>
Date: Tue Feb 2 15:41:22 2016 +0100
KVM: s390: vsie: add module parameter "nested"
Let's be careful first and allow nested virtualization only if enabled
by the system administrator. In addition, user space still has to
explicitly enable it via SCLP features for it to work.