Re: Upstreaming Lenovo tablet switch driver

[Lorenz Brun <lorenz@brun.one>]

Hi Armin,

Thanks for your input. While looking at new WMI drivers which I could 
use as a reference, I stumbled upon 
https://patchwork.kernel.org/project/platform-driver-x86/patch/20221004214332.35934-1-soyer@irl.hu/ 
which was sent last week and implements a driver for the same WMI 
interface. So I guess there is not much left for me to implement here, 
I'll just follow that patch.

Regards,
Lorenz

On Sat, Oct 15 2022 at 03:26:50 +02:00:00, Armin Wolf <W_Armin@gmx.de> 
wrote:
> Am 15.10.22 um 00:12 schrieb Lorenz Brun:
> 
>> Hi x86 platform list,
>> 
>> I'm considering contributing support for the usage mode/tablet mode
>> switch for some Lenovo laptops (including the popular IdeaPad Flex
>> series).
>> There is already an out-of-tree driver at
>> https://github.com/lukas-w/yoga-usage-mode which works well,
>> but I'd like to get this support upstream.
>> 
>> I have a few questions regarding the best approach:
>> 1. There is the ideapad-laptop module already which deals with a 
>> bunch
>> of functionality related to these laptops, but not the mode switch. 
>> Do
>> you prefer a separate driver or should the functionality be added to
>> ideapad-laptop?
>> 2. The existing out-of-tree driver uses a WMI interface to access 
>> mode
>> information. Should I work on the WMI layer or should I figure out
>> which EC registers to use to get the functionality independently? If
>> yes, how do we want to identify support for this type of mode switch?
>> 
>> Regards,
>> Lorenz
>> 
>> 
> Hi,
> 
> 1) In my opinion, the ideapad-laptop driver uses a totally different 
> interface than the out of tree driver, so it would make sense to 
> create a separate driver.
> 2) You should stick to the WMI interface. The vendor could change the 
> internal EC registers at any time, which would drastically limit the 
> portability and
> reliability of the driver in question. Also, there exists a whole WMI 
> subsystem inside the linux kernel, which for example supports 
> automatic discovery of
> WMI devices, etc.
> 
> Looking at the out-of-tree driver in question, it seems that it does 
> not fully utilize the WMI subsystem. If you want to upstream it, i 
> suggest a rewrite
> to fix this. Since the driver is quite small, it should be relatively 
> easy. Optionally, you could also add support for LENOVO_GSENSOR_DATA, 
> since it also
> has some interesting methods. However, the scope of the driver is 
> completely up to you.
> 
> There are already some WMI-based drivers inside the kernel, maybe 
> they can serve as an inspiration?
> 
> Armin Wolf
>