Hi Jean,
On Mon, Oct 03, 2022 at 05:29:53PM +0200, Jean Delvare wrote:
> The atmel_captouch driver does not actually depend on OF, it includes
> a non-OF device ID which could be used to instantiate the device, and
> the driver code is already prepared to be built with or without OF. So
> drop the unneeded dependency.
>
> Signed-off-by: Jean Delvare <jdelvare@xxxxxxx>
> Cc: Daniel Hung-yu Wu <hywu@xxxxxxxxxx>
> Cc: Dmitry Torokhov <dmitry.torokhov@xxxxxxxxx>
> ---
> The problem I'm trying to solve here is that "depends on OF ||
> COMPILE_TEST" does not make sense since OF can now be enabled on all
> architectures. One way to fix this is by removing the dependency
> altogether (this patch).
>
> If the driver is known to be needed only on OF-enabled systems then we
> could leave the dependency on OF and only drop COMPILE_TEST (and
> simplify the driver code accordingly). I have an alternative patch
> doing that already. Tell me what you prefer, I'm fine either way.
The driver is not operable on systems without OF:
static int atmel_captouch_probe(...
...
node = dev->of_node;
if (!node) {
dev_err(dev, "failed to find matching node in device tree\n");
return -EINVAL;
}
if (of_property_read_bool(node, "autorepeat"))
So the reason for "depends on OF || COMPILE_TEST" is to avoid prompting
users who actually try to configure real systems for drivers that make
no sense for them while still allowing people interested in compile
coverage to select COMPILE_TEST and enable more drivers.
Thanks.
--
Dmitry
