Bug Valve Index Controllers on OpenXR - no data when headset presence is off

I've noticed that when the user takes the headset off, that data from the controllers also stops shortly thereafter. It's possible to see this when using Unity's XR Interaction Debugger. While the headset video functions, the controller data stops - most notably the Pointer Position and Rotation, but any new input system inputs also stop responding - A and B button presses are no longer detected, for example.

We've had some users report issues with this. They believe our game is the issue because, confusingly, the controllers continue to update properly in the Steam VR menu and home, irrespective of the state of the headset's User Presence.

I suspect that there is an issue with some of their headsets - that their presence sensor is not working properly. (One user reported they could make our game work by securing the headset incredibly tightly to their face, for example). But, because Steam VR menus and home behave differently than OpenXR, as does OpenVR (you can see controllers work without user presence in game that still utilize OpenVR), it appears to them that our game is the issue.

Tested in Unity 2021.3.7f1 and OpenXR 1.4.2.

Given the differing behaviour from OpenVR and how the system menus behave - is this a bug? Is there a user workaround for this? Or for us?

 

I would note that this situation is exacerbated by the fact that the headset doesn't actually sleep if it's still moving. But the controller does go missing and stop reporting input under OpenXR within a few seconds of losing user presence, irrespective of headset motion/video output status.

 

OK, at least I can offer users a workaround if their presence sensor is acting up. I don't like it, but yeah at least they can flip the setting "Pause VR when headset is idle" to keep the controllers working.

There's a likely related issue where some users are reporting their Index controllers are not detected after the headset has been on for a long time. Might be a bug related to the hardware processing its presence sensor. In at least one case, a power cycle to the headset fixed the issue; is this something other people have been seeing?