Christophe Leroy <christophe.le...@c-s.fr> writes:
> Le 04/02/2019 à 11:24, Michael Ellerman a écrit :
>> Christophe Leroy <christophe.le...@c-s.fr> writes:
>>
>>> Since commit c40dd2f76644 ("powerpc: Add System RAM to /proc/iomem")
>>> it is possible to use the generic walk_system_ram_range() and
>>> the generic page_is_ram().
>>>
>>> To enable the use of walk_system_ram_range() by the IBM EHEA
>>> ethernet driver, the generic function has to be exported.
>>
>> I'm not sure if we have a policy on that, but I suspect we'd rather not
>> add a new export on all arches unless we need to. Especially seeing as
>> the only user is the EHEA code which is heavily in maintenance mode.
>
> If you take the exemple of function walk_iomem_res_desc(), that's
> similar. It is only used by x86 it seems and exported for nvdimm/e820
> driver only.
>
> See commit d76401ade0bb6ab0a7 ("libnvdimm, e820: Register all pmem
> resources")
OK. Which begs the question whether we need both exported. It looks like
you could probably use walk_iomem_res_desc() with the right flags to do
the same thing as walk_system_ram_range().
>> I'll put the export in powerpc code and make sure that builds.
>
> I thought there was a rule that EXPORT_SYMBOL has to immediately follow
> the function it exports. At least checkpatch checks for that.
Yeah that is a rule. But rules are made to be broken :)
I'll merge it for now with the export in powerpc code, if we want to we
can do a separate patch to move that export into generic code and get
acks for that.
cheers
