On Tue, Aug 11, 2020 at 09:16:28AM +0200, Jethro Beekman wrote: > On 2020-08-11 00:23, Sean Christopherson wrote: > > Another thought would be to wrap sgx_enclave_exception in a struct to give > > room for supporting additional exit information (if such a thing ever pops > > up) and to allow the caller to opt in to select behavior, e.g. Jethro's > > request to invoke the exit handler on IRQ exits. This is basically the > > equivalent of "struct kvm_run", minus the vCPU/enclave state. > > Actually, the flag I need is “return from the vdso on IRQ exits” (See Andy's > email about preferring returns over callbacks). But maybe the flag should be > “interpret IRQ exit as a normal exit” and let it be handled the same as any > other exit based on whether an exit handler fnptr was passed or not. Ya, slip of the tongue, the behavior would apply to both paths.
