64-bit VMCS fields were not being handled properly by the VMCS shadowing code. In particular, some full-width fields (TSC_OFFSET and GUEST_PHYSICAL_ADDRESS) were being shadowed, but accesses to the high 32-bits of these fields were not. Interleaved L1 accesses to the full-width fields and the high 32-bits returned inconsistent results. To make this cherry-pick easier, I also pulled in three other commits which are in preparation for virtualized VMCS shadowing support. I plan on submitting the patches for virtualized VMCS shadowing in the very near future. Jim Mattson (4): kvm: nVMX: Add support for "VMWRITE to any supported field" kvm: vmx: Introduce vmcs_read_any and vmcs_write_any kvm: nVMX: Introduce/rename enums for VMCS field encodings kvm: nVMX: Shadow "high" parts of shadowed 64-bit VMCS fields arch/x86/kvm/vmx.c | 231 +++++++++++++++++++++++++++-------------------------- 1 file changed, 119 insertions(+), 112 deletions(-) -- 2.13.2.725.g09c95d1e9-goog
