I'm observing random crashes in multi-vCPU L2 guests running on KVM on
Hyper-V. I bisected the issue to the commit 877ad952be3d ("KVM: vmx: Add
tlb_remote_flush callback support"). Hyper-V TLFS states:

"AddressSpace specifies an address space ID (an EPT PML4 table pointer)"

So apparently, Hyper-V doesn't expect us to pass naked EPTP, only PML4
pointer should be used. Strip off EPT configuration information before
calling into vmx_hv_remote_flush_tlb().

Fixes: 877ad952be3d ("KVM: vmx: Add tlb_remote_flush callback support")
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
---
 arch/x86/kvm/vmx.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
index 612fd17be635..e665aa7167cf 100644
--- a/arch/x86/kvm/vmx.c
+++ b/arch/x86/kvm/vmx.c
@@ -1572,8 +1572,12 @@ static int vmx_hv_remote_flush_tlb(struct kvm *kvm)
 		goto out;
 	}
 
+	/*
+	 * FLUSH_GUEST_PHYSICAL_ADDRESS_SPACE hypercall needs the address of the
+	 * base of EPT PML4 table, strip off EPT configuration information.
+	 */
 	ret = hyperv_flush_guest_mapping(
-			to_vmx(kvm_get_vcpu(kvm, 0))->ept_pointer);
+			to_vmx(kvm_get_vcpu(kvm, 0))->ept_pointer & PAGE_MASK);
 
 out:
 	spin_unlock(&to_kvm_vmx(kvm)->ept_pointer_lock);
-- 
2.17.1