Daniel P. Berrange wrote:
On Thu, Feb 06, 2014 at 12:07:10AM +0400, Roman Bogorodskiy wrote:
> Daniel P. Berrange wrote:
>
> > > + virCommandAddArg(cmd, "-H"); /* vmexit from guest on hlt
*/
> > > + virCommandAddArg(cmd, "-P"); /* vmexit from guest on pause
*/
> >
> > What's the functional effect of having these set, or not ?
>
> Having that set should make bhyve process terminate on these events (htl
> and pause).
Ok, was just wonder if they needed to be expressed in the
XML or not, but sounds quite low level so lets ignore it
for now.
> I guess it'd be better not to use it so bhyve process doesn't suddenly
> vanish from libvirt radar?
I don't know really - as long as the driver detects that it has gone
and marks the guest as shutdown.
I didn't plan to add such detection into an initial driver code,
because it's already rather large and not very easy to review. Is that
fine to leave it as is so far and extend later?
PS BTW, I have updated v4 of this patch which fixes most of the comments
to the current one. I forgot to include checking XML def for apci/apic
(as you have pointed out), but I'm wondering if I should wait for
comments on v4 to submit it along with other fixes or send v5 with
acpi/apic flags right away?
Roman Bogorodskiy