Re: runtime regression with "x86/mm/pat: Emulate PAT when it is disabled"

[One Thousand Gnomes]

On Thu, 10 Mar 2016 20:19:33 +0100
Borislav Petkov <bp@xxxxxxx> wrote:

> On Thu, Mar 10, 2016 at 02:04:29PM -0500, Paul Gortmaker wrote:
> > So, I guess that is a qemu bug?  If there is no real silicon out there
> > that has no MTRR but does claim PAT, then qemu32 is a flawed CPU type?  
> 
> Well, AFAICT, "qemu32" is emulating something PPRO-like:
> 
> #define PPRO_FEATURES (CPUID_FP87 | CPUID_DE | CPUID_PSE | CPUID_TSC | \
>           CPUID_MSR | CPUID_MCE | CPUID_CX8 | CPUID_PGE | CPUID_CMOV | \
>           CPUID_PAT | CPUID_FXSR | CPUID_MMX | CPUID_SSE | CPUID_SSE2 | \
> 	  ^^^^^^^^^
> 
> and that one advertizes PAT but not MTRRs.
> 
> I need to go dig into history to find out what PPRO actually supported.

Pentium Pro has MTRR, PAT came later.

I believe the qemu32 CPU isn't a "real" CPU type therefore.

Alan