[libvirt] [PATCH v2] qemu: Add check for whether KVM nesting is enabled

[Andrea Bolognani]

On Tue, 2018-11-27 at 14:15 -0500, John Ferlan wrote:
> On 11/27/18 12:05 PM, Andrea Bolognani wrote:
> > Oh, wait, I get it now: 'modprobe -c' doesn't dump the *current* host
> > configuration, but the *static* one! So if you enable nested KVM
> > support by doing
> > 
> >   # modprobe -r kvm_intel
> >   # modprobe kvm_intel nested=1
> > 
> > like I did, then the check above will not report it as enabled even
> > though it is; conversely, if you drop the appropriate config snippet
> > in /etc/modprobe.d/ but don't reload the module it will report it as
> > enabled even though it's not!
> 
> Ugh, sigh... Yep, I was thinking primarily the static config option
> since we had helpers to read. Of course that won't be enough. Joy, more
> code to probe... Maybe it is easier to just say - clear your
> capabilities cache if you alter that particular kernel value.

You should be able to just read the contents of

  /sys/module/kvm_{amd,intel}/parameters/nested

for x86_64; not sure whether a similar trick will work on s390, but
it will definitely *not* work on ppc64, and I haven't even started
thinking about aarch64 yet!

Of course there's the usual caveats about whether this has been
available in the kernel for a long enough time that it's okay for us
to rely on it, and whether the approach is not entirely flawed for
reasons that I'm not aware of :)

-- 
Andrea Bolognani / Red Hat / Virtualization