On Thu, Oct 27, 2022 at 5:31 PM Sean Christopherson <seanjc@xxxxxxxxxx> wrote: > > On Tue, Oct 18, 2022, David Matlack wrote: > > @@ -50,6 +73,9 @@ int main(int argc, char *argv[]) > > TEST_REQUIRE(kvm_has_cap(KVM_CAP_SMALLER_MAXPHYADDR)); > > > > vm = vm_create_with_one_vcpu(&vcpu, guest_code); > > + vm_init_descriptor_tables(vm); > > + vcpu_init_descriptor_tables(vcpu); > > + vm_install_exception_handler(vm, PF_VECTOR, guest_page_fault_handler); > > Instead of installing an exception handler, > > u8 vector = kvm_asm_safe(KVM_ASM_SAFE(FLDS_MEM_EAX), > "a"(MEM_REGION_GVA)); > > then the guest/test can provide more precise information if a #PF doesn't occur. I gave this a shot and realized this would prevent checking that it is a reserved #PF, since kvm_asm_safe() only returns the vector. It's probably more important to have more precise testing rather than more precise failure reporting. But what did you have in mind specifically? Maybe there's another way.
