On Fri, Jan 13, 2017 at 10:34:32PM +0100, Arnd Bergmann wrote:

On Fri, Jan 13, 2017 at 10:15 PM, Dmitry Torokhov
[off-list ref] wrote:

quoted

quoted

This is my fixup, though I'm not too happy with that version.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>

diff --git a/drivers/hid/Kconfig b/drivers/hid/Kconfig
index 1aeb80e52424..3927259a5d5d 100644
--- a/drivers/hid/Kconfig
+++ b/drivers/hid/Kconfig

@@ -785,7 +785,8 @@ config HID_SUNPLUS

config HID_RMI
      tristate "Synaptics RMI4 device support"
      depends on HID
-      select RMI4_CORE
+      depends on SERIO && RMI4_CORE
+      depends on SERIO=y || RMI4_CORE=SERIO

Shouldn't this be simply

select SERIO # needed for F03

Ah, I guess this would work too. I didn't consider it because SERIO is
a user-visible symbol and we generally try not to 'select' them but instead
use 'depends on.

However, SERIO is already used with 'select' all over the place, so adding
another select is actually safer than adding a dependency (which could
cause dependency loops here).

Actually the best solution is probably to have 'select SERIO' in RMI4, like

config RMI4_F03_SERIO
       tristate
       depends on RMI4_CORE
       depends on RMI4_F03
       default RMI4_CORE
       select SERIO

As that avoids the 'depends on SERIO=y || RMI4_CORE=SERIO' statement that
is different from the other SERIO users, it keeps it all in one place,
and it doesn't
prevent you from seeing the RMI4_F03 symbol when SERIO=m.

Hmm, if this works and resilient with user changing symbols after
they've been auto-selected then I like it. How can we run it through
multitude of randconfigs?

Thanks.

-- 
Dmitry