On Wed, Dec 27, 2017 at 02:51:45PM +0100, Heiko Stuebner wrote:

Hi Linus,

Am Mittwoch, 27. Dezember 2017, 13:15:47 CET schrieb Linus Walleij:

quoted

This driver was merged in 2011 as a tool for detecting the orientation
of a screen. The device driver assumes board file setup using the
platform data from <linux/input/gpio_tilt.h>. But no boards in the
kernel tree defines this platform data.

As I am faced with refactoring drivers to use GPIO descriptors and
pass decriptor tables from boards, or use the device tree device
drivers like these creates a serious problem: I cannot fix them and
cannot test them, not even compile-test them with a system actually
using it (no in-tree boardfile).

I suggest to delete this driver and rewrite it using device tree if
it is still in use on actively maintained systems.

I can also offer to rewrite it out of the blue using device tree if
someone promise to test it and help me iterate it.

Cc: Heiko Stübner <heiko@sntech.de>
Signed-off-by: Linus Walleij <redacted>
---
Heiko: not meaning to be militant here, just contain the situation,
as stated: if you like the driver and can test it, I can reimplement
it from scratch using device tree.

It seems that piece of hardware (gpio-connected orientation-sensors)
was really only used in the one s3c24xx-based device I hacked on in 2011.

I somehow lost focus from trying to do the s3c24xx devicetree migration
when I started hacking on Rockchip stuff, so while I do have the devices
still around, I don't think I'll find the time and energy trying to get a
recent kernel to run on them anyway, so I'm fine with dropping the driver.
It's simple enough to get reintroduced if someone really finds a device
using it or time to redo the ereader support using devicetree.

So long story short
Acked-by: Heiko Stuebner <heiko@sntech.de>

Applied, thank you.

-- 
Dmitry