Fix potential cross-vendor landmines due to Intel and AMD having different behavior for MSR_TSC_AUX[63:32], unify the logic across SVM and VMX, and switch MSR_TSC_AUX via user return MSRs on SVM (the original goal). v3: - Fix a comment typo. [Reiji] - Add patches to add missing guest_cpuid_has() check, drop bits 63:32 on AMD, and unify VMX and SVM emulation. - Rebase to kvm/next, commit c4f71901d53b ("Merge tag 'kvmarm-5.13' ... ) v2: - Rebase to kvm/queue (ish), commit 0e91d1992235 ("KVM: SVM: Allocate SEV command structures on local stack") - https://lkml.kernel.org/r/20210422001736.3255735-1-seanjc@xxxxxxxxxx v1: https://lkml.kernel.org/r/20210206003224.302728-1-seanjc@xxxxxxxxxx Sean Christopherson (4): KVM: SVM: Inject #GP on guest MSR_TSC_AUX accesses if RDTSCP unsupported KVM: SVM: Clear MSR_TSC_AUX[63:32] on write KVM: x86: Tie Intel and AMD behavior for MSR_TSC_AUX to guest CPU model KVM: SVM: Delay restoration of host MSR_TSC_AUX until return to userspace arch/x86/kvm/svm/svm.c | 58 +++++++++++++++++------------------------- arch/x86/kvm/svm/svm.h | 7 ----- arch/x86/kvm/vmx/vmx.c | 13 ---------- arch/x86/kvm/x86.c | 34 +++++++++++++++++++++++++ 4 files changed, 58 insertions(+), 54 deletions(-) -- 2.31.1.498.g6c1eba8ee3d-goog