Re: [PATCH 2/3] x86/intel-family: Resync with Linux

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 16/07/2025 2:56 pm, Jan Beulich wrote:
> On 16.07.2025 15:28, Andrew Cooper wrote:
>> This snapshot is prior to Linux commit db4001f9cc32 ("x86/cpu/vfm: Delete all
>> the *_FAM6_ CPU #defines") at the end of their conversion phase.
>>
>> In addition to non-FAM6 infixed names, defines are added for the Pentium Pro,
>> ArrowLake U, and reintroduced the PHI defines which were incorrectly deleted
>> in the past.
>>
>> In cpufeature.h, provide VFM_* macros to transform constants to/from the
>> cpuinfo_x86 representation.
>>
>> No functional change.
>>
>> Signed-off-by: Andrew Cooper <andrew.cooper3@xxxxxxxxxx>
>> ---
>> CC: Jan Beulich <JBeulich@xxxxxxxx>
>> CC: Roger Pau Monné <roger.pau@xxxxxxxxxx>
>>
>> I meant to object to deleting PHI at the time, but was too late.  Just 
>> because
>> Xen has stopped supporting the PHI doesn't mean the model numbers have ceased
>> existing.
> I don't mind their re-introduction, yet I wonder what value their presence has
> when we don't use them anywhere.

I spend a lot of time cross-checking tables in the speculative papers. 
Knowing which were the released CPUs is important for finding typos.  No
comment on how many tables were correct on their first revision...

That said, there's SoFIA which is not included in intel-family, and for
which we have partial coverage owing to early speculative tables.  IIRC,
it was a custom chip for a mobile base-station, and runs an entirely
custom software stack.

>> --- a/xen/arch/x86/include/asm/intel-family.h
>> +++ b/xen/arch/x86/include/asm/intel-family.h
>> @@ -13,8 +13,8 @@
>>   *  INTEL_FAM6{OPTFAMILY}_{MICROARCH}{OPTDIFF}
>>   * where:
>>   * OPTFAMILY        Describes the family of CPUs that this belongs to. 
>> Default
>> - *          is assumed to be "_CORE" (and should be omitted). The other
>> - *          value currently in use is _ATOM.
>> + *          is assumed to be "_CORE" (and should be omitted). Other values
>> + *          currently in use are _ATOM and _XEON_PHI
> Nit: You lost the full stop.

That seems to be missing in Linux, even at master.

I've got an outstanding task in Linux's intel-family.  I'll try to
remember to fix it upstream.

>
>> @@ -40,131 +40,223 @@
>>   * their own names :-(
>>   */
>>  
>> +#define IFM(_fam, _model)   VFM_MAKE(X86_VENDOR_INTEL, _fam, _model)
>> +
>>  /* Wildcard match for FAM6 so X86_MATCH_INTEL_FAM6_MODEL(ANY) works */
>>  #define INTEL_FAM6_ANY                      X86_MODEL_ANY
>> +/* Wildcard match for FAM6 so X86_MATCH_VFM(ANY) works */
>> +#define INTEL_ANY                   IFM(X86_FAMILY_ANY, X86_MODEL_ANY)
>> +
>> +#define INTEL_PENTIUM_PRO           IFM(6, 0x01)
>>  
>>  #define INTEL_FAM6_CORE_YONAH               0x0E
>> +#define INTEL_CORE_YONAH            IFM(6, 0x0E)
> Here and below I assume it is deliberate that you use a raw number again,
> rather than making use of the immediately preceding #define? Perhaps for
> the sake of brevity?

Because it's transient.

All the FAM* constants get deleted in db4001f9cc32 ("x86/cpu/vfm: Delete all
the *_FAM6_ CPU #defines") and I've got another resync in my local
branch, but it obviously doesn't compile in Xen yet.

> Acked-by: Jan Beulich <jbeulich@xxxxxxxx>

Thanks.

~Andrew