On Mon, Feb 07, 2022 at 09:37:39AM -0600, Michael Roth wrote:

Absolutely, I know a thorough review is grueling work, and would never
want to give the impression that I don't appreciate it. Was just hoping
to revisit these in the context of v9 since there were some concerning
things in flight WRT the spec handling and I was sort of focused on
getting ahead of those in case they involved firmware/spec changes. But
I realize that's resulted in a waste of your time and I should have at
least provided some indication of where I was with these before your
review. Won't happen again.

Thanks, that's appreciated.

And in case you're wondering, the kernel is the most flexible thing from
all parties involved so even if you have to change the spec/fw, fixing
the kernel is a lot easier than any of the other things. So make sure
you do a good job with the spec/fw - the kernel will be fine. :-)

Ok, will work this in for v10. My plan is to introduce this struct:

  struct cpuid_leaf {
      u32 fn;
      u32 subfn;
      u32 eax;
      u32 ebx;
      u32 ecx;
      u32 edx;
  }

Ok.

as part of the patch which introduces sev_cpuid_hv():

  x86/sev: Move MSR-based VMGEXITs for CPUID to helper

and then utilize that for the function parameters there, here, and any
other patches in the SNP code involved with fetching/manipulating cpuid
values before returning them to the #VC handler.

Sounds good.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette