Guest VMs can be provided with a para-virtualized clock source to
perform timekeeping. A KVM guest can map in a PV clock via the
MSR_KVM_SYSTEM_TIME/MSR_KVM_SYSTEM_TIME_NEW virtualized MSRs.
Where as on a Xen guest this can be provided via the vcpu/shared
info pages.
These PV clocks both use a common structure which is mapped between
host <-> guest to provide the PVTI (paravirtual time information)
for the clock. This reference information is a guest TSC timestamp
and a host system time at a SINGULAR point in time.
If KVM decides to update the reference information due to a
KVM_REQ_MASTERCLOCK_UPDATE, a drift between the guest TSC and
the PV clock can be observed, this is exascerbated when the guest
TSC is also scaled too.
If the reference guest TSC & system time within the structure stay
the same there is no potential for a drift between the TSC and PV
clock.
This series adds in two patches, one to add in API/ioctl to allow
a VMM to perform a correction/fixup of the PVTI structure when it
knows that KVM may have updated the KVM clock information and a
second one to verify that the drift is present & corrected.
Jack Allister (2):
KVM: x86: Add KVM_[GS]ET_CLOCK_GUEST for KVM clock drift fixup
KVM: selftests: Add KVM/PV clock selftest to prove timer drift
correction
Documentation/virt/kvm/api.rst | 43 ++++
arch/x86/kvm/x86.c | 87 +++++++
include/uapi/linux/kvm.h | 3 +
tools/testing/selftests/kvm/Makefile | 1 +
.../selftests/kvm/x86_64/pvclock_test.c | 223 ++++++++++++++++++
5 files changed, 357 insertions(+)
create mode 100644 tools/testing/selftests/kvm/x86_64/pvclock_test.c
base-commit: 8cb4a9a82b21623dbb4b3051dd30d98356cf95bc
--
2.40.1
