Re: behaviour change for keyboards introduced by "HID: generic: create one input report per application type"

[Siarhei Vishniakou <svv@google.com>]

Hi Peter,

It's a bit of both. it seems unnatural to have a single physical
device represented by 2 devices in the system. It also seems not
efficient in terms of performance, because we are now allocating twice
the number of devices, which causes twice the configuration change
events on Android.



On Mon, May 11, 2020 at 10:22 PM Peter Hutterer
<peter.hutterer@who-t.net> wrote:
>
> Hi Siarhei,
>
> On Wed, Apr 01, 2020 at 01:02:03PM -0700, Siarhei Vishniakou wrote:
> > Hello linux-input,
> >
> > We have noticed a recent behavior change introduced by the commit:
> > https://github.com/torvalds/linux/commit/f07b3c1da92db108662f99417a212fc1eddc44d1
> > HID: generic: create one input report per application type.
> >
> > We have been looking into the behaviour of the Microsoft Designer
> > keyboard. Prior to these patches, the keyboard would report as a
> > single input_device to the user space. After these patches, the
> > keyboard is now split into 2 input devices: “Designer Keyboard
> > Consumer Control” and “"Designer Keyboard Keyboard". We noticed that
> > this behavior also applies to the keyboard Dell KB216. In the Dell
> > case, it is split into 3 devices, which include “Consumer Control”
> > (for some of the media keys) and “System Control” (for the sleep
> > button).
> >
> > In both Microsoft and Dell cases, these are pretty standard keyboards,
> > although they do contain media keys and power/sleep keys.
> >
> > Could you please confirm that this new behaviour is indeed
> > intentional, and that you wanted keyboards to be split in this
> > fashion?
>
> Is there an actual issue with the split though or just a cosmetic one? This
> is pretty standard for many keyboards (and other devices) but it shouldn't
> have any negative effect on functionality.
>
> Cheers,
>    Peter