x86: NMI vs. MCE handling
Currently, care is being taken to deal with the potential of an NMI being signalled at the very moment when the debug fault is correcting the stack for the sysenter path, or when in at the stack pointer load in the sysenter path itself. The same care, however, is not taken in the MCE code path, and since I'd consider this equally important (especially when the event is correctable) I wonder whether this is an oversight or I'm misunderstanding something.
Thanks for a clarification,
Jan