On 2021-10-14 10:46, Andrew Jones wrote:
On Wed, Oct 13, 2021 at 01:03:41PM +0100, Marc Zyngier wrote:
Forward accesses to the ICV_*SGI*_EL1 registers to EL1, and
emulate ICV_SRE_EL1 by returning a fixed value.
This should be enough to support GICv3 in a protected guest.
Out of curiosity, has the RVIC work / plans been dropped?
ARM has dropped the architecture, and it makes no sense
to move KVM to support non-architectural stuff.
Which means we will eventually have to harden the guest itself
to cope with the fact that it cannot trust the interrupt controller.
Yes, this is crap.
M.
--
Jazz is not dead. It just smells funny...
