KVM: x86: drop TSC offsetting kvm_x86_ops to fix KVM_GET/SET_CLOCK
authorPaolo Bonzini <pbonzini@redhat.com>
Mon, 31 Oct 2016 23:39:48 +0000 (00:39 +0100)
committerPaolo Bonzini <pbonzini@redhat.com>
Wed, 2 Nov 2016 19:03:07 +0000 (20:03 +0100)
commitea26e4ec08d4727e3a9e48a6b74695861effcbd9
treebe3ae8868bd0e93e6029cfd9ce0512fe69dc39ab
parentbd768e146624cbec7122ed15dead8daa137d909d
KVM: x86: drop TSC offsetting kvm_x86_ops to fix KVM_GET/SET_CLOCK

Since commit a545ab6a0085 ("kvm: x86: add tsc_offset field to struct
kvm_vcpu_arch", 2016-09-07) the offset between host and L1 TSC is
cached and need not be fished out of the VMCS or VMCB.  This means
that we can implement adjust_tsc_offset_guest and read_l1_tsc
entirely in generic code.  The simplification is particularly
significant for VMX code, where vmx->nested.vmcs01_tsc_offset
was duplicating what is now in vcpu->arch.tsc_offset.  Therefore
the vmcs01_tsc_offset can be dropped completely.

More importantly, this fixes KVM_GET_CLOCK/KVM_SET_CLOCK
which, after commit 108b249c453d ("KVM: x86: introduce get_kvmclock_ns",
2016-09-01) called read_l1_tsc while the VMCS was not loaded.
It thus returned bogus values on Intel CPUs.

Fixes: 108b249c453dd7132599ab6dc7e435a7036c193f
Reported-by: Roman Kagan <rkagan@virtuozzo.com>
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/include/asm/kvm_host.h
arch/x86/kvm/svm.c
arch/x86/kvm/vmx.c
arch/x86/kvm/x86.c