[RFC] powerpc/pseries: Interface to represent PAPR firmware attributes

Michael Ellerman mpe at ellerman.id.au
Wed Jun 16 10:03:11 AEST 2021


Fabiano Rosas <farosas at linux.ibm.com> writes:
> Pratik Sampat <psampat at linux.ibm.com> writes:
...
>>>
>>>> The new H_CALL exports information in direct string value format, hence
>>>> a new interface has been introduced in /sys/firmware/papr to export
>>> Hm.. Maybe this should be something less generic than "papr"?
>>
>> The interface naming was inspired from /sys/firmware/opal's naming convention.
>> We believed the name PAPR could serve as more generic name to be used by both
>> Linux running on PHYP and linux on KVM.
>
> Right, I agree with that rationale, but /opal has identifiable elements
> in it whereas /papr would have the generic "attr_X_name", which does not
> give much hint about what they are.
>
> We also expect people to iterate the "attr_X_*" files, so if we decide
> to add something else under /papr in the future, that would potentially
> cause issues with any tool that just lists the content of the directory.
>
> So maybe we should be proactive and put the hcall stuff inside a
> subdirectory already. /papr/energy_scale_attrs comes to mind, but I
> don't have a strong opinion on the particular name.

Maybe we should use the descriptive part of the hcall.

So H_GET_ENERGY_SCALE_INFO -> ../papr/energy_scale_info/

That should help avoid any naming confusion, because every hcall should
have a unique name.

In future if there's ever a H_GET_ENERGY_SCALE_INFO_2 we would then have
to decide if we expose that as a separate directory, or more likely we
would handle that in the kernel and continue to use the existing sysfs
name.

...

> Based on all the new information you provided, I'd say present all the
> data and group it under the ID:
>
> /sys/firmware/papr/energy_scale_attrs/
>    |-- <id>/
>      |-- desc
>      |-- value
>      |-- value_desc
>    |-- <id>/
>      |-- desc
>      |-- value
>      |-- value_desc

Yeah that seems reasonable.

I'd think we should just omit the value_desc if it's empty.

cheers


More information about the Linuxppc-dev mailing list