EDK2 + VT-x

EDK2 + VT-x

[Patchmail]

I cannot find anywhere how to enable VT-x in OVMF, if anyone can point me
in the right direction, I'd greatly appreciate it.

thanks,
David Napier

Re: EDK2 + VT-x

[Laszlo Ersek]

CC Ladi and Eduardo

On 09/14/17 02:54, Patchmail wrote:
> I cannot find anywhere how to enable VT-x in OVMF, if anyone can point me
> in the right direction, I'd greatly appreciate it.

If you are looking to do nested virt with OVMF, first of all I recommend
the following blog to your attention:

https://ladipro.wordpress.com/

I think Eduardo must have a bunch of blog posts on CPU models as well:

https://habkost.net/


Second, if you add the +vmx flag to your CPU model on the QEMU command
line, or in the libvirt domain XML, then QEMU will expose that to OVMF
in a dedicated fw_cfg file ("etc/msr_feature_control"), and OVMF will
set the Feature Control MSR accordingly, on all the CPUs in the guest.

Please refer to the following:

- TianoCore BZ <https://bugzilla.tianocore.org/show_bug.cgi?id=86>

- edk2 commit dbab994991c7 ("OvmfPkg/PlatformPei: program
  MSR_IA32_FEATURE_CONTROL from fw_cfg", 2016-07-07).


IOW, with the caveats described in Ladi's and Eduardo's blogs, just add
the +vmx flag to your CPU model, and you should be good to go.

Thanks
Laszlo

Re: EDK2 + VT-x

[Patchmail]

Laszlo,

Thanks a ton, that solution worked.  In addition to that, I added
'nested=1' to the kernel flag for kvm_intel and changed virt-manager to
match the host cpu profile rather than using host-passthrough.  I really
appreciate the help and am thankful people like you guys are around working
on these kinds of projects.

David Napier

On Thu, Sep 14, 2017 at 5:23 AM, Laszlo Ersek <lersek@redhat.com> wrote:

> CC Ladi and Eduardo
>
> On 09/14/17 02:54, Patchmail wrote:
> > I cannot find anywhere how to enable VT-x in OVMF, if anyone can point me
> > in the right direction, I'd greatly appreciate it.
>
> If you are looking to do nested virt with OVMF, first of all I recommend
> the following blog to your attention:
>
> https://ladipro.wordpress.com/
>
> I think Eduardo must have a bunch of blog posts on CPU models as well:
>
> https://habkost.net/
>
>
> Second, if you add the +vmx flag to your CPU model on the QEMU command
> line, or in the libvirt domain XML, then QEMU will expose that to OVMF
> in a dedicated fw_cfg file ("etc/msr_feature_control"), and OVMF will
> set the Feature Control MSR accordingly, on all the CPUs in the guest.
>
> Please refer to the following:
>
> - TianoCore BZ <https://bugzilla.tianocore.org/show_bug.cgi?id=86>
>
> - edk2 commit dbab994991c7 ("OvmfPkg/PlatformPei: program
>   MSR_IA32_FEATURE_CONTROL from fw_cfg", 2016-07-07).
>
>
> IOW, with the caveats described in Ladi's and Eduardo's blogs, just add
> the +vmx flag to your CPU model, and you should be good to go.
>
> Thanks
> Laszlo
>