On 18.09.2023 16:02, Andrew Cooper wrote: > On 18/09/2023 12:07 pm, Jan Beulich wrote: >> On 15.09.2023 17:00, Andrew Cooper wrote: >>> --- a/xen/arch/x86/include/asm/amd.h >>> +++ b/xen/arch/x86/include/asm/amd.h >>> @@ -140,6 +140,17 @@ >>> AMD_MODEL_RANGE(0x11, 0x0, 0x0, 0xff, 0xf), \ >>> AMD_MODEL_RANGE(0x12, 0x0, 0x0, 0xff, 0xf)) >>> >>> +/* >>> + * The Zen1 and Zen2 microarchitectures are implemented by AMD (Fam17h) and >>> + * Hygon (Fam18h) but without simple model number rules. Instead, use >>> STIBP >>> + * as a heuristic that distinguishes the two. >>> + * >>> + * The caller is required to perform the appropriate vendor/family checks >>> + * first. >>> + */ >>> +#define is_zen1_uarch() (!boot_cpu_has(X86_FEATURE_AMD_STIBP)) >>> +#define is_zen2_uarch() boot_cpu_has(X86_FEATURE_AMD_STIBP) >>> + >>> struct cpuinfo_x86; >>> int cpu_has_amd_erratum(const struct cpuinfo_x86 *, int, ...); >> With one simply the opposite of the other, and with the requirement of a >> family check up front, do we really need both? Personally I'd prefer if >> we had just the latter. Yet in any event >> Reviewed-by: Jan Beulich <[email protected]> > > We specifically do want both, because they're use is not symmetric at > callsites. > > In particular, having only one would make the following patch illogical > to read.
I don't think it would, but that's perhaps one more of the many areas where we take different perspectives. Jan
