> > On Tue, Feb 16, 2021, Dave Hansen wrote: > > > Having separate device nodes for SGX driver and KVM virtual EPC also > > > allows separate permission control for running host SGX enclaves and > > > KVM SGX guests. > > > > Specifically, 'sgx_vepc' is a less restrictive interface. It would > > make a lot of sense to more tightly control access compared to 'sgx_enclave'. > > The opposite is just as likely, i.e. exposing SGX to a guest but not allowing > enclaves in the host. Not from a "sgx_enclave is easier to abuse" perspective, > but from a "enclaves should never be runnable in the host in our environment". Agreed. CSP may want to provide SGX service in VMs, but not to run SGX app in host.
