On Tue, 19 Dec 2023 06:58:52 +0000, Oliver Upton wrote:
> Here's the alternative approach I had suggested in response to Kunkun's
> bug report, building the GICv3 ISPENDR accessor on top of the existing
> ISPENDR / ICPENDR accessors. With these changes userspace should
> consistently read / configure the hardware pending state for GICv4.1
> vSGIs.
>
> Oliver Upton (3):
> KVM: arm64: vgic: Use common accessor for writes to ISPENDR
> KVM: arm64: vgic: Use common accessor for writes to ICPENDR
> KVM: arm64: vgic-v3: Reinterpret user ISPENDR writes as I{C,S}PENDR
>
> [...]
I have queued this for 6.8, together with my GICv4.1 fix.
[1/3] KVM: arm64: vgic: Use common accessor for writes to ISPENDR
commit: 13886f34444596e6eca124677cd8362a941b585b
[2/3] KVM: arm64: vgic: Use common accessor for writes to ICPENDR
commit: 561851424d93e91083df4071781b68dc4ba1fc5a
[3/3] KVM: arm64: vgic-v3: Reinterpret user ISPENDR writes as I{C,S}PENDR
commit: 39084ba8d0fceb477a264e2bb8dfd3553876b84c
Cheers,
M.
--
Without deviation from the norm, progress is not possible.
