Re: [PATCH v3 9/9] arm64: Documentation - Expose CPU feature registers

[Suzuki K Poulose]

On 06/01/17 12:16, Catalin Marinas wrote:
> On Wed, Jan 04, 2017 at 05:49:07PM +0000, Suzuki K. Poulose wrote:
> > +The following rules are applied to the value returned by the
> > +infrastructure:
> > +
> > + a) The value of an 'IMPLEMENTATION DEFINED' field is set to 0.
> > + b) The value of a reserved field is populated with the reserved
> > +    value as defined by the architecture.
> > + c) The value of a field marked as not 'visible', is set to indicate
> > +    the feature is missing (as defined by the architecture).
> > + d) The value of a 'visible' field holds the system wide safe value
> > +    for the particular feature(except for MIDR_EL1, see section 4).
> > +    See Appendix I for more information on safe value.
> > +
> > +There are only a few registers visible to the userspace. See Section 4,
> > +for the list of 'visible' registers.
> > +
> > +All others are emulated as having 'invisible' features.
>
> BTW, we don't have any statement about whether a visible field may
> become invisible but I guess this wouldn't be a problem as long as the
> feature is reported as missing. I'm thinking about currently RES0 fields
> that are listed as visible but they may report something in the future
> that we don't want exposed to user. At that point, we'll change the
> field to "invisible" while reporting RES0 to user. I don't see an issue
> with this, just I thought worth flagging.

Thanks for raising that. In fact, we treat all the RES0 fields as invisible
and strict for the moment. So, I think it is worth reflecting that in the
documentation. As you mentioned, we could switch them as required based on
the feature without any issues. I will fix this.

> Anyway:
>
> Reviewed-by: Catalin Marinas <catalin.marinas@xxxxxxx>

Thanks for reviewing the entire series. I will resend the series with the tags
and updates to this documentation and a couple of other patches.

Suzuki