Bug 216134
Summary: | uniq attribute unset for incoming uinput device (AVRCP) | ||
---|---|---|---|
Product: | Drivers | Reporter: | macmpi (spam) |
Component: | Input Devices | Assignee: | drivers_input-devices |
Status: | NEW --- | ||
Severity: | normal | CC: | abhishekpandit, dmitry.torokhov, jikos, kenny.wottrich, luiz.dentz, sujakulk |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | 5.15.41 | Subsystem: | |
Regression: | No | Bisected commit-id: |
Description
macmpi
2022-06-14 08:01:54 UTC
Any feedback (relevance, fit for purpose, won't fix,...)? Thanks Hello, Is there a plan to add ioctl support for getting uinput uniq identifier ? https://patchwork.kernel.org/project/linux-input/patch/20191127185139.65048-1-abhishekpandit@chromium.org/ This previous discussions seems to have gone cold for a long time. One problematic use case : Getting the uniq identifier(EVIOCGUNIQ) for a virtual input device created using libevdev library (libevdev_set_name()) always fails. A lot of VMMs rely on getting the uniq identifier as part of virtualizing it. Uinput needs a similar support as well. I also have a need for this. I am working on exposing the accelerometer and gyroscope sensor in the ASUS ROG Ally device. In parallel to a separate effort to add an iio kernel driver, I am working on userspace solution to adapt the iio data to an evdev device using uinput. One primary development goal is to be able to associate the motion controls of the gyro with the built-in buttons and joysticks, which report as an XBox 360 controller. SDL (which is used by WINE/Proton and many native games and applications) supports linking evdev controllers to their corresponding evdev motion sensors, but ONLY if the uniq attribute matches between the two. This approach works out of the box with existing Playstation and Nintendo controllers. However, without being able to specify the uniq attribute through uinput, my evdev solution is unable to advertise to SDL that my evdev device is the motion control counterpart to the built-in controller. I have read through the whole discussion and I believe the patch proposed remains viable. Please consider taking another look at this, Dmitry et al! |