KVM already provides guests the ability to use the 'monitor trap flag' VM-execution control. Support for this flag is provided by the fact that KVM unconditionally forwards MTF VM-exits to the guest (if requested), as KVM doesn't utilize MTF. While this provides support during hardware instruction execution, it is insufficient for instruction emulation. Should L0 emulate an instruction on the behalf of L2, L0 should also synthesize an MTF VM-exit into L1, should control be set. The first patch fixes the handling of #DB payloads for both Intel and AMD. To support MTF, KVM must also populate the 'pending debug exceptions' field, rather than directly manipulating the debug register state. Additionally, the exception payload associated with #DB is said to be compatible with the 'pending debug exceptions' field in VMX. This does not map cleanly into an AMD DR6 register, requiring bit 12 (enabled breakpoint on Intel, reserved MBZ on AMD) to be masked off. The second patch implements MTF under instruction emulation by adding vendor-specific hooks to kvm_skip_emulated_instruction(). Should any non-debug exception be pending before this call, MTF will follow event delivery. Otherwise, an MTF VM-exit may be synthesized directly into L1. Third patch introduces tests to kvm-unit-tests. These tests path both under virtualization and on bare-metal. Oliver Upton (2): KVM: x86: Add vendor-specific #DB payload delivery KVM: x86: Emulate MTF when performing instruction emulation arch/x86/include/asm/kvm_host.h | 2 ++ arch/x86/kvm/svm.c | 25 +++++++++++++++++++++ arch/x86/kvm/vmx/nested.c | 2 +- arch/x86/kvm/vmx/nested.h | 5 +++++ arch/x86/kvm/vmx/vmx.c | 39 ++++++++++++++++++++++++++++++++- arch/x86/kvm/x86.c | 27 ++++++----------------- 6 files changed, 78 insertions(+), 22 deletions(-) -- 2.25.0.rc1.283.g88dfdc4193-goog
