On Tue, Mar 29, 2016 at 11:10:42AM -0400, John Ferlan wrote:
On 03/29/2016 09:31 AM, Pavel Hrdina wrote:
> Commit 7068b56c introduced several hyperv features. Not all hyperv
> features are supported by old enough kernels and we shouldn't allow to
> start a guest if kernel doesn't support any of the hyperv feature.
>
> There is one exception, for backward compatibility we cannot error out
> if one of the RELAXED, VAPIC or SPINLOCKS isn't supported, for the same
> reason we ignore invtsc, to not break restoring saved domains with older
> libvirt.
From yesterday's dialog, there's also commit id '59fc0d06' which adds
"hv_crash" and commit id '600bca59' which adds "hv_time".
Neither is handled via these bits, but wouldn't both fall into the same
trap since both were added after commit '2e8f9080'?
Even though libvirt passes those options to QEMU in the same -cpu argument, they
are unrelated to this patch series. And for the same reason as for RELAXED,
VAPIC and SPINLOCKS, we cannot error out to not break things so there is nothing
to do about it.
>
> Signed-off-by: Pavel Hrdina <phrdina(a)redhat.com>
> ---
> src/cpu/cpu_x86.c | 8 ++++++++
> src/cpu/cpu_x86_data.h | 8 ++++++++
> src/qemu/qemu_process.c | 31 +++++++++++++++++++++++++++++++
> 3 files changed, 47 insertions(+)
>
Now I was going to ask about a capability bit for this, but seeing none
in previous commits, I thought I was safe... I guess not.
I don't know what you mean by this, what capability?
Pavel