[PATCH v5 0/5] mm: introduce memfd_secret system call to create "secret" memory areas

[Dave Hansen]

On 9/16/20 11:49 AM, Andy Lutomirski wrote:
> I still have serious concerns with uncached mappings.  I'm not saying
> I can't be convinced, but I'm not currently convinced that we should
> allow user code to create UC mappings on x86.

There's another widely-used OS that has a "NOCACHE" flag to be specified
to on what I *think* is the equivalent of anonymous memory mappings
(VirtualAlloc()):

> https://docs.microsoft.com/en-us/windows/win32/memory/memory-protection-constants

I'll offer this not as a reason we should or shouldn't allow it.  But,
we should at least be able to figure out if it's caused problems for
them or not.  I think it's been around a long time.