protonscr

[BUG] Closing SteamVR breaks USB connections and prevents running VR again

steamvropen bug
ValveSoftware/SteamVR-for-Linux#312 · opened 2020-03-16 by kherchel · updated 2020-06-03 · 3 comments · github
Kkherchel 2020-03-16 github

Describe the bug
I can use SteamVR just fine, but if I close it (e.g. with "Exit VR") I'm unable to start it back again.
First the vrmonitor window hangs for up to a minute (its contents don't get updated at all), then this is displayed in the SteamVR window:

steamvr

I am using Steam Beta with SteamVR Beta on an up-to-date Arch Linux, Radeon RX 5700 XT (RADV) and a Valve Index.

Here are the vrmonitor logs:
https://pastebin.com/Tx02aKxY

To Reproduce
Steps to reproduce the behavior:

  1. Run SteamVR
  2. Do whatever you want, play some games etc.
  3. Use the "Exit VR" option
  4. Try to run SteamVR again
  5. It doesn't run

Expected behavior
SteamVR works.

System Information (please complete the following information):

Kkherchel 2020-03-17 github

Update: Looks like closing SteamVR messes up the whole USB controller - my keyboard stops working completely after a while after SteamVR is closed and replugging it doesn't do anything, only rebooting helps.

While shutting down the PC after using SteamVR I get this message:

[FAILED] Failed unmounting /run/user/1000.
[10139.686668] usb 2-2.1: Failed to suspend device, error -110
[10152.462465] usb 1-9: device descriptor read/64, error -110

The usb 1-9 message then repeats itself once every few seconds until I manually shutdown the PC.

GGoofybud16 2020-06-02 github

I had a similar issue on an old MSI X370 board where the Index would spontaneously crash the onboard USB controller. Eventually a kernel update seemed to have fixed it.

Which motherboard and CPU do you have? Which USB port(s) on the motherboard are you using?

Kkherchel 2020-06-03 github

I have a MSI B450-A PRO MAX and a Ryzen 3700X.

I have since switched the USB port I'm using the Index with (it was on bus 1 or 2, now it's 3 or 4, I'm not sure), and either that, or some random kernel update mostly fixed the issue for me - it happened maybe once in the past month, while previously it was basically every second try.

Nothing extracted yet.