Re: [PATCH v2 03/12] x86/sev: Add an x86 version of prot_guest_has()

From: Christoph Hellwig
Date: Tue Aug 24 2021 - 03:16:26 EST


On Thu, Aug 19, 2021 at 01:33:09PM -0500, Tom Lendacky wrote:
> I did it as inline originally because the presence of the function will be
> decided based on the ARCH_HAS_PROTECTED_GUEST config. For now, that is
> only selected by the AMD memory encryption support, so if I went out of
> line I could put in mem_encrypt.c. But with TDX wanting to also use it, it
> would have to be in an always built file with some #ifdefs or in its own
> file that is conditionally built based on the ARCH_HAS_PROTECTED_GUEST
> setting (they've already tried building with ARCH_HAS_PROTECTED_GUEST=y
> and AMD_MEM_ENCRYPT not set).
>
> To take it out of line, I'm leaning towards the latter, creating a new
> file that is built based on the ARCH_HAS_PROTECTED_GUEST setting.

Yes. In general everytime architectures have to provide the prototype
and not just the implementation of something we end up with a giant mess
sooner or later. In a few cases that is still warranted due to
performance concerns, but i don't think that is the case here.

>
> >
> >> +/* 0x800 - 0x8ff reserved for AMD */
> >> +#define PATTR_SME 0x800
> >> +#define PATTR_SEV 0x801
> >> +#define PATTR_SEV_ES 0x802
> >
> > Why do we need reservations for a purely in-kernel namespace?
> >
> > And why are you overoading a brand new generic API with weird details
> > of a specific implementation like this?
>
> There was some talk about this on the mailing list where TDX and SEV may
> need to be differentiated, so we wanted to reserve a range of values per
> technology. I guess I can remove them until they are actually needed.

In that case add a flag for the differing behavior. And only add them
when actually needed. And either way there is absolutely no need to
reserve ranges.