Re: [PATCH 6/9] x86/mm: Introduce ptep_set_wrprotect_flush and related functions

[Dave Hansen]

On 06/07/2018 05:59 PM, Andy Lutomirski wrote:
> Can you ask the architecture folks to clarify the situation?  And, if
> your notes are indeed correct, don't we need code to handle spurious
> faults?

I'll double check that I didn't misunderstand the situation and that it
has not changed on processors with shadow stacks.

But, as far as spurious faults, wouldn't it just be a fault because
we've transiently gone to Present=0?  We already do that when clearing
the Dirty bit, so I'm not sure that's new.  We surely already handle
that one.