On Wed, 2020-03-25 at 15:05 +0100, Hans de Goede wrote:

Hi,

On 3/25/20 3:02 PM, Bastien Nocera wrote:

quoted

On Wed, 2020-03-25 at 14:55 +0100, Hans de Goede wrote:

quoted

We could do something like that, but TBH I'm not a fan of that

adding extra wrappers makes it harder to see what the code is

actually doing.



I understand your dislike for the extra braces I added and

I'm fine with fixing that by adding __maybe_unused to the

variable declarations at the top. I don't really see what

the problem with the #ifdef-s is given how clean they are,

with the braces thing fixed by using __maybe_unused things

would look like e.g. this:

It's not only the fact that there's extra #ifdef's, it's that the
ifdef's need to be just "that". It's not "#ifdef FOO", it's "#if
defined CONFIG_X86 && defined CONFIG_ACPI".

If that is the problem I would prefer adding:

/* Our special handling for GPIO accesses through ACPI is x86
specific */
#if defined CONFIG_X86 && defined CONFIG_ACPI
#define ACPI_GPIO_SUPPORT
#endif

And use:

#ifdef ACPI_GPIO_SUPPORT

Elsewhere.

Would that work for you?

That's slightly better, but I would still have preferred stubbing out
those ACPI calls directly. Right now, the fact that we expect half of
the commands to be stubbed out and the other half to not be called is
just weird.