On Tue, Jan 23, 2024, at 10:11, Christoph Hellwig wrote:
> On Mon, Jan 22, 2024 at 03:57:16PM +0100, Arnd Bergmann wrote:
>> > It would be good to fix the one or two that could use highmem and add a
>> > depends on !HIGHMEM for the others
>>
>> I would prefer a runtime check here, as one might still have a
>> multiplatform kernel where one machine can use highmem and
>> another machine can use one of these drivers, e.g. in
>> imx_v6_v7_defconfig.
>
> Well, if someone can come up with a good runtime check that's fine with me.
I assumed there was a generic way already, but it looks like
there is not, and adding one across nine architectures would be
nontrivial.
Let's use your initial suggestion then and use a Kconfig
dependency. I still don't like how this may impact users that
currently enable highmem and use one of these drivers, but
on a more positive note this might help us kill off HIGHMEM
in the future if users instead choose to go with
CONFIG_VMSPLIT_3G_OPT or similar.
The Marvell driver still needs some other solution of course.
Arnd
