Re: [RFC PATCH] xen/arm: check on domain type against hardware support

[Julien Grall <julien.grall@linaro.org>]

On 09/29/2014 01:35 PM, Ian Campbell wrote:
>>>> I propose cpu_has_aarch32 to check for both A32 and T32. If any one
>>>> (A32/T32) is set we
>>>> can assume that aarch32 is supported. For ARM64 that supports only
>>>> Aarch64 ID_PFR0_EL1 is RES0. so this check for T32 | A32 should work.
>>>> Something like this
>>>>
>>>> diff --git a/xen/include/asm-arm/cpufeature.h b/xen/include/asm-arm/cpufeature.h
>>>> index 7a6d3de..379e366 100644
>>>> --- a/xen/include/asm-arm/cpufeature.h
>>>> +++ b/xen/include/asm-arm/cpufeature.h
>>>> @@ -22,7 +22,9 @@
>>>>  #define boot_cpu_feature32(feat)       (boot_cpu_data.pfr32.feat)
>>>>
>>>> -#define cpu_has_aarch32   (boot_cpu_feature32(arm) == 1)
>>>> +#define cpu_has_a32       ((boot_cpu_feature32(arm) == 1)
>>>>  #define cpu_has_thumb     (boot_cpu_feature32(thumb) >= 1)
>>>> +#define cpu_has_aarch32   (cpu_has_a32 || cpu_has_thumb)
>>>
>>> Yes, this looks like what is needed, thanks.
>>
>> Following my comment above, I don't think this change is necessary.
> 
> Iff you are right about v8. You are probably right but it's not 100%
> clear. On the other hand checking for what we actually mean has no
> downside.

The downside is name readability for the developer as the term A32 has
been introduced on ARMv8.

With only the ARMv7 spec the name is very confusing. I had to read the
ARMv8 spec to completely understand the name and the different between
aarch32 and a32.

I would at least add a comment on top of cpu_has_a32.

Regards,

Regards,

-- 
Julien Grall