Re: [PATCH 05/21] x86/mm/pat: mirror direct map changes to ASI

[Brendan Jackman]

On Wed Jan 21, 2026 at 11:27 AM UTC, Borislav Petkov wrote:
> On Wed, Jan 21, 2026 at 09:45:22AM +0000, Brendan Jackman wrote:
>> 2. The property of some memory being mapped or unmapped in the
>>    "restricted"/"partial"/"sensitive" address space (current patch's
>>    terminology: sensitive/nonsensitive).
>
> Why do we care about the sensitivity of memory? To the kernel it doesn't
> matter whether memory is sensitive or not. You simply map it in the sensitive
> page table or in the nonsensitive or in both.
>
> Having a sensitive and nonsensitive pagetable should be enough. And that'll
> make the whole concept a lot simpler.

Well, something has to manage those pagetables and that thing needs a
term to describe the stuff it's acting on. I.e. we need an adjective
better than "to-be-mapped-into-the-full-pagtable".

Anyway, I think this will be an easier debate to have once we're looking
at actual code etc - I'm still debugging my page allocator changes to
support the guest_memfd thing and that code doesn't need to care about
"sensitivity" at all, there's just one address space and data is
"mapped" or "unmapped". So I propose we defer this conversation until
after that...