x86/kvm/lapic: preserve gfn_to_hva_cache len on cache reinit
authorVitaly Kuznetsov <vkuznets@redhat.com>
Tue, 16 Oct 2018 16:50:06 +0000 (18:50 +0200)
committerPaolo Bonzini <pbonzini@redhat.com>
Tue, 16 Oct 2018 22:30:17 +0000 (00:30 +0200)
commita7c42bb6da6b1b54b2e7bd567636d72d87b10a79
tree50697d0c76dfff34f090ab30b178ad4bd02f6c06
parent12e0c6186ba44bb6194cf5d2eda8f46880126587
x86/kvm/lapic: preserve gfn_to_hva_cache len on cache reinit

vcpu->arch.pv_eoi is accessible through both HV_X64_MSR_VP_ASSIST_PAGE and
MSR_KVM_PV_EOI_EN so on migration userspace may try to restore them in any
order. Values match, however, kvm_lapic_enable_pv_eoi() uses different
length: for Hyper-V case it's the whole struct hv_vp_assist_page, for KVM
native case it is 8. In case we restore KVM-native MSR last cache will
be reinitialized with len=8 so trying to access VP assist page beyond
8 bytes with kvm_read_guest_cached() will fail.

Check if we re-initializing cache for the same address and preserve length
in case it was greater.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
arch/x86/kvm/lapic.c