On 1/14/19 4:10 PM, Greg Kurz wrote:
> On Mon, 14 Jan 2019 09:19:50 +0100
> Cédric Le Goater wrote:
>
>> On 1/11/19 10:04 PM, Greg Kurz wrote:
>>> When using the in-kernel interrupt controller, the state of all irqs is
>>> synchronized to KVM at machine reset time. In the case of PHB hotplug, we
On Mon, 14 Jan 2019 17:29:48 +0100
Cédric Le Goater wrote:
> On 1/14/19 4:10 PM, Greg Kurz wrote:
> > On Mon, 14 Jan 2019 09:19:50 +0100
> > Cédric Le Goater wrote:
> >
> >> On 1/11/19 10:04 PM, Greg Kurz wrote:
> >>> When using the in-kernel interrupt controller, the state of all irqs is
On Mon, 14 Jan 2019 09:19:50 +0100
Cédric Le Goater wrote:
> On 1/11/19 10:04 PM, Greg Kurz wrote:
> > When using the in-kernel interrupt controller, the state of all irqs is
> > synchronized to KVM at machine reset time. In the case of PHB hotplug, we
> > will need to synchronize LSIs manually.
On 1/11/19 10:04 PM, Greg Kurz wrote:
> When using the in-kernel interrupt controller, the state of all irqs is
> synchronized to KVM at machine reset time. In the case of PHB hotplug, we
> will need to synchronize LSIs manually.
Yes. This is because the interrupt sources in the KVM XICS device
When using the in-kernel interrupt controller, the state of all irqs is
synchronized to KVM at machine reset time. In the case of PHB hotplug, we
will need to synchronize LSIs manually.
Do this for the existing KVM XICS implementation and put a placeholder for
the upcoming KVM XIVE.