On Fri, Mar 13, 2020 at 12:07:55PM -0400, Nathaniel McCallum wrote: > On Thu, Mar 12, 2020 at 8:52 PM Sean Christopherson > <sean.j.christopherson@xxxxxxxxx> wrote: > > FWIW, if the handler reaaaly wanted to pop off the stack, it could do so, > > fixup the stack, and then re-call __vdso_sgx_enter_enclave() instead of > > returning (to the original __vdso_sgx_enter_enclave()). > > My understanding from the documentation issue above was that *if* you > wanted to push parameters back on the stack during enclave exit, you > would *have* to supply a handler so it could pop the parameters and > reset the stack. Which is why restoring %rsp from %rbx didn't make > sense to me. Yep.
