protonscr

Regression with 1.8.2 Beta + Valve Index, SteamVR Fail

steamvrclosed appid 306bug
ValveSoftware/SteamVR-for-Linux#247 · opened 2019-09-18 by rstrube · updated 2019-09-27 · 16 comments · github · game page · search this game
Rrstrube 2019-09-18 github

Describe the bug
After updating to SteamVR 1.8.2 Beta, I always see the following upon starting up SteamVR:

image

In addition, it's worth noting that for 5-10 seconds the mouse cursor is unresponsive / very slow on my Gnome DE after starting Steam VR.

This behavior does not exist in the 1.7 stable version or the 1.8.1 Beta version of SteamVR

To Reproduce

  1. Install the latest SteamVR Beta 1.8.2
  2. Start SteamVR

Expected behavior
SteamVR should start normally

System Information (please complete the following information):

SSlickMcRunFast 2019-09-18 github
ZZamundaaa 2019-09-18 github

Manjaro KDE, kernel 5.3.0 + Mesa 19.3.0-devel (git-f812cbfd88) (LLVM 10.0.0) here, no such issue.
Looks like a NVidia issue to me.

Jjrtberlin 2019-09-18 github

Same issue with KDE neon (based on Ubuntu 18.04.1) since the latest beta release:

SteamVR Version: 1.8.2 (1568754983)
SteamVR Date: 2019-09-17
Steam Branch: beta
OS: Linux version 5.2.11-steamvr-generic
NVIDIA Driver Version: 435.21

https://gist.github.com/jrtberlin/9f260bc0a659be6ca5081e300f2ac36c

Edit: The stable branch runs fine, The issue is only with the recent beta version of SteamVR

VVash63 2019-09-18 github

I'm also getting this. The SteamVR system report seems to be missing the crash, when I review my system logs I'm seeing the initial crash is in libnvidia-glcore but this is missing from the SteamVR System Report:

audit[4914]: ANOM_ABEND auid=1000 uid=1000 gid=100 ses=2 pid=4914 comm="VulkanVblankThr" exe="/home/vash/.local/share/Steam/steamapps/common/SteamVR/bin/linux64/vrcompositor" sig=11 res=1
steam.desktop[1600]: crash_20190918132109_1.dmp[4975]: Uploading dump (out-of-process)
steam.desktop[1600]: /tmp/dumps/crash_20190918132109_1.dmp
crash_20190918132109_1.dmp[4975]: Uploading dump (out-of-process)
                                  /tmp/dumps/crash_20190918132109_1.dmp
kernel: VulkanVblankThr[4933]: segfault at 0 ip 00007efbf90eecdd sp 00007efbf2f85cf0 error 4 in libnvidia-glcore.so.435.19.03[7efbf82b0000+12bc000]
kernel: Code: 89 c5 53 74 67 48 8b 05 f9 fe e6 00 4d 63 e8 4a 8d 3c ed 00 00 00 00 ff 10 85 ed 48 89 c3 7e 24 31 c9 31 d2 49 8b 04 d4 ff c1 <8b> 00 66 c7 44 d3 04 03 00 66 c7 44 d3 06 00 00 89 04 d3 48 ff c2
kernel: audit: type=1701 audit(1568805670.088:62): auid=1000 uid=1000 gid=100 ses=2 pid=4914 comm="VulkanVblankThr" exe="/home/vash/.local/share/Steam/steamapps/common/SteamVR/bin/linux64/vrcompositor" sig=11 res=1 

steamvr1.8.2.VulkanVblankThr.txt
SteamVR-2019-09-18-PM_01_50_10.txt

Also worth noting is that the symptoms are far worse than I have gsync enabled. Disabling gsync via nvidia-settings -a CurrentMetaMode="DPY-3: 2560x1440_144 @2560x1440 +0+0 {ViewPortIn=2560x1440, ViewPortOut=2560x1440+0+0, AllowGSYNC=off}" improves the situation from locking up X11 and crashing even the SteamVR desktop dashboard to just crashing the vrcompositor (so I could still generate a system report).

DDarkheron 2019-09-18 github

Same.. repost from patch discussion on the steam forums.

No longer functional on Ubuntu 18.04.3. Receive an error 306. The monitor comes up and shows the connected devices, but it doesn't look like the compositor is starting. Vr_Init seems to be failing, even though the error returned from the call is vr::VRInitError_None

Here's what seems to be the error text:

log snippet

Tue Sep 17 2019 19:48:33.729653 - godot.x11.tools.64 1.8.2 startup with PID=5792, config=/home/darkheron/.local/share/Steam/config, runtime=/home/darkheron/.local/share/Steam/steamapps/common/SteamVR
Tue Sep 17 2019 19:48:33.729699 - vrclient type=VRApplication_Scene
Tue Sep 17 2019 19:48:33.730288 - [Settings] Load Default Json Settings from /home/darkheron/.local/share/Steam/steamapps/common/SteamVR/drivers/htc/resources/settings/default.vrsettings
Tue Sep 17 2019 19:48:33.730350 - [Settings] Load Default Json Settings from /home/darkheron/.local/share/Steam/steamapps/common/SteamVR/drivers/lighthouse/resources/settings/default.vrsettings
Tue Sep 17 2019 19:48:33.730396 - [Settings] Load Default Json Settings from /home/darkheron/.local/share/Steam/steamapps/common/SteamVR/drivers/null/resources/settings/default.vrsettings
Tue Sep 17 2019 19:48:33.730547 - [Settings] Load Default Json Settings from /home/darkheron/.local/share/Steam/steamapps/common/SteamVR/resources/settings/default.vrsettings
Tue Sep 17 2019 19:48:33.730933 - [Settings] Load Json Settings from /home/darkheron/.local/share/Steam/config/steamvr.vrsettings
Tue Sep 17 2019 19:48:33.913145 - CIPCPipe::ConnectPipe(SteamVR_Namespace) attempting connect to steamvr:SteamVR_Namespace
Tue Sep 17 2019 19:48:33.913257 - CSharedResourceNamespaceClient::Init(): received namespace data 4610
Tue Sep 17 2019 19:48:33.913289 - CIPCPipe::ConnectPipe(VR_ServerPipe_4610) attempting connect to steamvr:VR_ServerPipe_4610
Tue Sep 17 2019 19:48:33.915604 - Received success response from vrserver connect
Tue Sep 17 2019 19:48:33.916850 - Not looking for a good app key because Steam didn't start this app
Tue Sep 17 2019 19:48:33.916873 - App key after connect message:system.generated.godot.x11.tools.64
Tue Sep 17 2019 19:48:33.921530 - CIPCPipe::ConnectPipe(VR_CompositorPipe_4610) attempting connect to steamvr:VR_CompositorPipe_4610
Tue Sep 17 2019 19:48:33.921553 - Unable to connect to pipe errno=111
Tue Sep 17 2019 19:48:33.932368 - Starting vrcompositor process: /home/darkheron/.local/share/Steam/steamapps/common/SteamVR/bin/linux64/vrcompositor-launcher 
Using vrcompositor capability proxy
Launching /home/darkheron/.local/share/Steam/steamapps/common/SteamVR/bin/linux64/vrcompositor
Using breakpad crash handler
Setting breakpad minidump AppID = 250820
Forcing breakpad minidump interfaces to load
Tue Sep 17 2019 19:48:33.944276 - CIPCPipe::ConnectPipe(VR_CompositorPipe_4610) attempting connect to steamvr:VR_CompositorPipe_4610
Tue Sep 17 2019 19:48:33.944325 - Unable to connect to pipe errno=111
Looking up breakpad interfaces from steamclient
Calling BreakpadMiniDumpSystemInit
Tue Sep 17 2019 19:48:34.044468 - CIPCPipe::ConnectPipe(VR_CompositorPipe_4610) attempting connect to steamvr:VR_CompositorPipe_4610
Tue Sep 17 2019 19:48:34.044527 - Unable to connect to pipe errno=111
Tue Sep 17 2019 19:48:34.144654 - CIPCPipe::ConnectPipe(VR_CompositorPipe_4610) attempting connect to steamvr:VR_CompositorPipe_4610
Tue Sep 17 2019 19:48:34.144704 - Unable to connect to pipe errno=111
Tue Sep 17 2019 19:48:34.244836 - CIPCPipe::ConnectPipe(VR_CompositorPipe_4610) attempting connect to steamvr:VR_CompositorPipe_4610

is the errno 111 here the standard linux error for connection refused? Did needed permissions change somehow? Worked really well until this patch.

I can provide more information and dig around if necessary, but this is all I have time for today.

Llostgoat 2019-09-18 github

Hey Everyone,

Thanks for reporting this issue. As far as I can tell we have two separate crashes here.

Crash # 1 is from users who have Failed Watchdog timeout in thread Render in Update after 5.806722 seconds. Aborting. in their system reports. This hang is 1.8.2 specific and reverting back to 1.8.1 should be a temporary workaround.

Crash # 2 is in VulkanBlankThread. This is due to an Nvidia driver regression in their beta drivers. Moving to a non-beta driver branch should help resolve this issue.

If neither of the above works for you let me know.

P.S. I'll be using this thread to track crash # 1 as that is the one specified by OP and the one under our control. For # 2 please see:
https://devtalk.nvidia.com/default/topic/1062856/linux/435-19-02-crashes-steamvr/

Rrstrube 2019-09-18 github

Crash # 2 is in VulkanBlankThread. This is due to an Nvidia driver regression in their beta drivers. Moving to a non-beta driver branch should help resolve this issue.

Thank you for your reply. I do want to clarify that the Nvidia 435.21 driver is not the beta driver, it's the latest "short lived branch" stable driver.

https://www.nvidia.com/en-us/drivers/unix/

Either way, it sounds like it's a nvidia driver bug...

Edit: now I understand, the reported Crash 2 is related to the Beta Nvidia drivers, sorry for the confusion. Based on the comments I thought Crash 2 was happening even with the stable 435.21 drivers, but apparently the culprit is Crash 1.

Rrstrube 2019-09-18 github

I'm also getting this. The SteamVR system report seems to be missing the crash, when I review my system logs I'm seeing the initial crash is in libnvidia-glcore but this is missing from the SteamVR System Report:

audit[4914]: ANOM_ABEND auid=1000 uid=1000 gid=100 ses=2 pid=4914 comm="VulkanVblankThr" exe="/home/vash/.local/share/Steam/steamapps/common/SteamVR/bin/linux64/vrcompositor" sig=11 res=1
steam.desktop[1600]: crash_20190918132109_1.dmp[4975]: Uploading dump (out-of-process)
steam.desktop[1600]: /tmp/dumps/crash_20190918132109_1.dmp
crash_20190918132109_1.dmp[4975]: Uploading dump (out-of-process)
                                  /tmp/dumps/crash_20190918132109_1.dmp
kernel: VulkanVblankThr[4933]: segfault at 0 ip 00007efbf90eecdd sp 00007efbf2f85cf0 error 4 in libnvidia-glcore.so.435.19.03[7efbf82b0000+12bc000]
kernel: Code: 89 c5 53 74 67 48 8b 05 f9 fe e6 00 4d 63 e8 4a 8d 3c ed 00 00 00 00 ff 10 85 ed 48 89 c3 7e 24 31 c9 31 d2 49 8b 04 d4 ff c1 <8b> 00 66 c7 44 d3 04 03 00 66 c7 44 d3 06 00 00 89 04 d3 48 ff c2
kernel: audit: type=1701 audit(1568805670.088:62): auid=1000 uid=1000 gid=100 ses=2 pid=4914 comm="VulkanVblankThr" exe="/home/vash/.local/share/Steam/steamapps/common/SteamVR/bin/linux64/vrcompositor" sig=11 res=1 

steamvr1.8.2.VulkanVblankThr.txt
SteamVR-2019-09-18-PM_01_50_10.txt

Also worth noting is that the symptoms are far worse than I have gsync enabled. Disabling gsync via nvidia-settings -a CurrentMetaMode="DPY-3: 2560x1440_144 @2560x1440 +0+0 {ViewPortIn=2560x1440, ViewPortOut=2560x1440+0+0, AllowGSYNC=off}" improves the situation from locking up X11 and crashing even the SteamVR desktop dashboard to just crashing the vrcompositor (so I could still generate a system report).

Can you clarify what Nvidia driver version your running?

VVash63 2019-09-18 github

I'm running 435.19.03, but I got a reply from Nvidia that it should be fixed in the newly released 435.24.02. I'll test this when back at my PC if no one else here has gotten a chance.

Rrstrube 2019-09-18 github

I'm running 435.19.03, but I got a reply from Nvidia that it should be fixed in the newly released 435.24.02. I'll test this when back at my PC if no one else here has gotten a chance.

Thank you, I see the confusion now, I conflated the two issues.
Crash 2 (your crash) is an independent problem related to the current Nvidia Beta drivers.

VVash63 2019-09-19 github

New Nvidia drivers 435.24.02 did fix my previous crash in VulkanVblankThr, but it's still crashing. I can no longer generate a full system report but here is the vrcompositor.txt log:
vrcompositor.txt

Edit: Also from my syslog at the exact time of the SteamVR crash:
sep 19 01:40:30 vashnix kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830

I suspect the nvidia drivers are still the root cause

Hh1z1 2019-09-19 github

Can confirm the same.

[70878.662550] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:1:0:830
[70880.662949] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:1:0:830
[70882.722677] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:1:0:830
[70884.723482] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:1:0:830
Thu Sep 19 2019 17:36:13.417977 - Failed Watchdog timeout in thread Render in Up
date after 5.894147 seconds. Aborting.
# cat /proc/driver/nvidia/version 
NVRM version: NVIDIA UNIX x86_64 Kernel Module  435.21  Sun Aug 25 08:17:57 CDT 2019
GCC version:  gcc version 4.8.5 20150623 (Red Hat 4.8.5-36) (GCC) 

Latest stable 435.21 doesn't have crash 2 but 1 still remains.

Latest Vulkan Beta release today 435.24.02 doesn't have crash 2 but 1 still remains.

Interesting thing is NVIDIA considers 2 seconds to be an error condition while Steam waits 5. Not sure how that effects actual state between them.

Either way both of these are ultimately NVIDIA problems. Steam shouldn't be able to hang a GPU.

Ffabiankranewitter 2019-09-21 github

I get almost the same error with the beta of steam vr (1.8.5). Only with me the error is no longer 306 but 307. Everything is running normally on the stable version.

steamvr

dmesg output:
[12300.213722] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:2:0:830
[12302.214347] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:2:0:830
[12310.244155] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:2:0:830
[12312.244612] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:2:0:830
[12314.393243] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:2:0:830
[12316.394122] nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:2:0:830
[12319.648691] show_signal_msg: 53 callbacks suppressed
[12319.648693] Web Thread[7210]: segfault at 0 ip 00000000006115b4 sp 00007fa5acd05c40 error 4 in vrserver[400000+406000]

cat /proc/driver/nvidia/version 
NVRM version: NVIDIA UNIX x86_64 Kernel Module  435.21  Sun Aug 25 08:17:57 CDT 2019
GCC version:  gcc version 8.3.0 (Ubuntu 8.3.0-6ubuntu1) 
uname -s -r -p
Linux 5.0.0-29-generic x86_64
Mmimattr 2019-09-21 github

Can confirm the same SteamVR launch issue as @fabiankranewitter, @h1z1 and @Vash63:

14:11:46 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:11:46 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:11:44 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:11:42 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:11:40 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:10:13 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:10:11 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:10:09 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:10:07 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:09:31 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:09:29 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:09:27 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:09:25 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:08:59 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830
14:08:57 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c57e:2:0:830

System Information

I confirm:

  • [x] that I have checked whether there are updates for my system available.

SteamVR System Report
vrcompositor.txt

Reproduction

  1. Select the SteamVR icon within the Steam client (top-right)
Llostgoat 2019-09-26 github

A fix for this issue has been pushed out with SteamVR Beta 1.8.6

Rrstrube 2019-09-27 github

I can confirm that 1.8.6 resolved this problem, thanks!

Nothing extracted yet.