Linux DRM Adds Quirk For Dealing With Valve's Index VR Headset

[phoronix]

Phoronix: Linux DRM Adds Quirk For Dealing With Valve's Index VR Headset

With Valve's Index VR headset slated to ship starting in July, this premium $999 USD headset has now been quirk'ed to properly behave under Linux...

http://www.phoronix.com/scan.php?page=news_item&px=Valve-Index-Non-Desktop-Quirk

[Britoid]

It's kinda silly for each VR headset to need to have its product ID in the kernel for the kernel to not to treat it as a display.

[zoomblab]

It's kinda silly for each VR headset to need to have its product ID in the kernel for the kernel to not to treat it as a display.

My thoughts exactly.

[Guest]

It's kinda silly for each VR headset to need to have its product ID in the kernel for the kernel to not to treat it as a display.

Not a big deal though. It's just few constants...

What non-silly solution would you propose, and especially be willing to implement?

[Britoid]

Not a big deal though. It's just few constants...

What non-silly solution would you propose, and especially be willing to implement?

A method for the EDID to tell the display manager that it's not a display.

[andrei_me]

I thought it was a udev rule, not a kernel code

[anth]

Anyone know why there are so many? 17, including 16 in a contiguous block, compared with 16 for all other VR headsets combined.

I guess it is for future products and maybe past prototypes and Valve are the only VR vendor to really commit to this OS. Perhaps the two screens are presented individually rather than as one so need separate EDIDs but I don't think any other headset has done that.

[coder]

It's kinda silly for each VR headset to need to have its product ID in the kernel for the kernel to not to treat it as a display.

Yeah, there should be some standard EDID-type query you can do to see what class of display it is.

[coder]

and especially be willing to implement?

Presumably, the classification bits would need to be part of some the DisplayPort (in this case) or HDMI standard.

[johanb]

Not a big deal though. It's just few constants...

It actually is a pretty big deal though, this means that you would need a very new Linux kernel while feature-wise it has been supported for multiple versions.

When more VR headsets gets released in the future it will be even more silly if for example you need the latest kernel while feature-wise it's been supported for 3 years.