[libvirt] [PATCH v4 11/11] qemu: Add swtpm to emulator cgroup

[Daniel P. Berrangé]

On Tue, May 15, 2018 at 11:43:10AM -0400, Stefan Berger wrote:
> On 05/15/2018 11:34 AM, Daniel P. Berrangé wrote:
> > On Tue, May 15, 2018 at 11:25:58AM -0400, Stefan Berger wrote:
> > > On 05/10/2018 05:57 PM, Stefan Berger wrote:
> > > > Add the external swtpm to the emulator cgroup so that upper limits of CPU
> > > > usage can be enforced on the emulated TPM.
> > > I haven't made any changes to this yet. A possibility would be to put swtpm
> > > into its own tpm-emulator cgroup and extend the XML for the TPM to also have
> > > 'period' and 'quota':
> > > 
> > >      <tpm model='tpm-tis'>
> > >        <backend type='emulator'>
> > >          <period>1000</period>
> > >          <quota>500</quota>
> > >        </backend>
> > >      </tpm>
> > > 
> > > Or we add the following to cputune:
> > > 
> > >          <tpm_emulator_period>1000</tpm_emulator_period>
> > >          <tpm_emulator_quota>500</tpm_emulator_quota>
> > > 
> > > The latter would be more consistent, though i would prefer the former.
> > I'm not really seeing a compelling reason to need to set tunables on
> > the swtpm directly. IMHO we should just consider it part of the
> > "emulator" tunables - the fact that it is a separate binary/process
> > rather than inside QEMU is just a private impl detail.
> 
> One reason I could think of is to approximate the real world a little closer
> where a TPM is typically its own chip.

I don't think that's a real use case. You can look at QEMU's machine types
and say they have 10-20 separate chips in the real world, but we don't need
to have control over CPU usage of those chips.

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|