Two minor patches for stuff I found while reviewing Eric's VGIC fixes [1]. First patch should be straightforward documentation update. I thought of the second patch when I was hacking kvmtool to read GICR_TYPER from userspace. I was passing the raw MPIDR_EL1 value (the result of KVM_GET_ONE_REG(MPIDR_EL1)) to KVM_DEV_ARM_VGIC_GRP_REDIST_REGS, which obviously returned an error code. I started investigating and I thought that the problem was caused by VGIC_TO_MPIDR not setting the RES1 bit (bit 31) from MPIDR_EL1. I actually wrote a patch to change that when I realized that the macro wasn't supposed to create a valid MPIDR_EL1 value at all, because the result of the macro is then compared to kvm_vcpu_get_mpidr_aff(). I wrote patch #2 hoping that the change in the macro name will prevent someone else from making the same mistake. On the other hand, using mpidr to refer to the GICR_TYPER.Affinity_Value fields looks to be well established in the kernel and in the documentation, so I'm not sure that changing the macro name is really necessary. [1] https://lore.kernel.org/kvmarm/20201212185010.26579-1-eric.auger@xxxxxxxxxx/ Alexandru Elisei (2): Documentation: kvm: vgic-v3: Update return code for KVM_DEV_ARM_VGIC_GRP_CTRL KVM: arm64: vgic-v3: Rename VGIC_TO_MPIDR() to VGIC_TO_MPIDR_AFF() Documentation/virt/kvm/devices/arm-vgic-v3.rst | 4 +++- arch/arm64/kvm/vgic/vgic-kvm-device.c | 12 ++++++------ arch/arm64/kvm/vgic/vgic.h | 10 +++++----- 3 files changed, 14 insertions(+), 12 deletions(-) -- 2.30.0 _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
