Re: [PATCH v5 4/9] drivers: base cacheinfo: Add support for ACPI based firmware tables

[Sudeep Holla]

(Sorry for the delay, just returning from vacation)

On 12/12/17 23:37, Jeremy Linton wrote:
> On 12/12/2017 05:02 PM, Rafael J. Wysocki wrote:

[...]

>>
>> So call this field "token" or similar. Don't call it "of_node" and
>> don't introduce another "firmware_node" thing in addition to that.
>> That just is a mess, sorry.

I completely agree. Both me and Lorenzo pointed that out in previous
revisions and fair enough you have a valid concern it's use with PPTT.

> 
> I sort of agree, I think I can just change the whole of_node to a
> generic 'void *firmware_unique' which works fine for the PPTT code, it
> should also work for the DT code in cache_leaves_are_shared().
>

Should be fine.

> The slight gocha is there is a bit of DT code which initially runs
> earlier that uses of_node as an indirect parameter to a couple functions
> (by just passing the cacheinfo). Let me see if I can tweak that a bit.
>

May be use a simple inline wrapper functions to convert, might help if
we diverge too.

> Frankly, If I understood completely all the *priv cases I suspect it
> might be possible to collapse *of_node into that as well. That is as
> long as no one decides to flush out DT on x86, or PPTT on x86.
> 

priv is used to save architecture/cache specific details that can't be
generalized. I doubt if this of_node or PPTT pointer/offset falls in
that category.


-- 
Regards,
Sudeep