On 18/10/2018 13:14, Vitaly Kuznetsov wrote: > > Based on that my suggestion would be to shadow GUEST_SS_AR_BYTES, keep > GUEST_SS_AR_BYTES and unshadow the rest (GUEST_ES_BASE, > GUEST_CS_SELECTOR, GUEST_CS_LIMIT, GUEST_CS_BASE). I can do this as a > separate patch as I see this series is already in kvm/queue.
Yes, it should be a separate patch anyway. GUEST_CS_BASE and GUEST_CS_LIMIT probably matter for 32-bit guests, but I guess it's okay to remove them. GUEST_CS_SELECTOR probably dates back to when we were incorrectly using CPL=CS.RPL instead of CPL=SS.DPL, and can be removed too. GUEST_ES_BASE alone is quite useless, so it can go. Paolo
