Open LiamDawe opened 3 years ago
I note that X-Plane is the only title I have found thus far where I can recover from this error without restarting the title. This is because the user can toggle VR mode from within X-Plane, not just when X-Plane starts. So when this error occurs (...sometimes startlingly out of the blue...), I can pause X-Plane, toggle out of VR mode, restart SteamVR, then reenter VR mode in X-Plane. It's annoying, but it's doable. Other titles I just have to lose my progress, which is even more annoying.
note that X-Plane is the only title I have found thus far where I can recover from this error without restarting the title.
Same here (but also the only title I play, "waves") Try these in your settings
"disableAsync" : true, "enableLinuxVulkanAsync" : true,
Just been flying all afternoon without incident on those.
@mSparks43
I tried adding:
"disableAsync" : true,
"enableLinuxVulkanAsync" : true,
to the "steamvr" section of the settings file, but I could still reproduce the error.
I noticed other "title-specific" sections in the settings file had the opposite "disableAsync" state set, namely:
"disableAsync" : false,
So I created a "title-specific" section in the settings file for "Paper Beast" and set "disableAsync" to true. With that change, the error did not appear, but async reprojection was also not happening. It was a stutter mess in places.
Async reprojection is so very useful that my current policy is to turn it off when playing titles which trigger the error often, and then turn it back on for titles which trigger the error rarely.
Sure wish a developer with access to source code would take some time to look into this. I note a Windows 11 user has reported the same error, so it might not be just a Linux issue. Investing time into finding out what's wrong may benefit both Windows and Linux users.
but async reprojection was also not happening. It was a stutter mess in places.
yes. but I believe enableLinuxVulkanAsync true fixes that for me (although I am getting 60fps in XP) disableAsync true stops the bug enableLinuxVulkanAsync true gives smooth tracking.
This error -203 happens to me while simply navigating the steamvr menu, steam within steamvr and sometimes when launching a game. I have created a system report here: https://gist.github.com/Fynises/6fbb7aa91213902abbdd4ea74c5f2550
OS: Arch Linux (linux-tkg-pds 5.15.6) GPU: GTX 1070 Graphics Drivers: 495.44 SteamVR: 1.21.3 (Beta) Headset: Valve Index Steam: steam-native-runtime 04/12/2021
I haven't done any extra configurations to steamvr outside of the provided settings menu.
Got a -203 crash in "The Walking Dead: Saints & Sinners" after about 15 minutes up time and finishing a radio conversation in the home camp.
OS: Ubuntu 20.04.3 LTS Steam: built Dec 2 2021 07:35:48 [beta] Proton Version: Proton Experimental SteamVR: 1.21.2 GPU: RTX 3060 NVIDIA driver: 495.44 VR Headset: Valve Index
I examined the "~/.steam/steam/logs/vrmonitor.txt" file in my steam install and I saw these lines around the error:
Wed Dec 08 2021 15:28:25.815541 - [Status Alert] SteamVR Fail (-203) Wed Dec 08 2021 15:28:25.816238 - Large number of events this frame caused the poll loop to exit without draining Wed Dec 08 2021 15:28:25.832634 - Large number of events this frame caused the poll loop to exit without draining Wed Dec 08 2021 15:28:25.853133 - Large number of events this frame caused the poll loop to exit without draining
So it sounds like some event queue is backing up. Why? I have no idea! Debugging without access to source code is near impossible.
Replying to https://github.com/ValveSoftware/SteamVR-for-Linux/issues/452#issuecomment-989295022
Omg, i'm like 98% sure about what is causing it, SteamVR on Linux has this weird issue that causes the runtime to spam vr::VREvent_ActionBindingReloaded
because of failed binding loads on literally every headset. See #307.
You can't see it in the vrmonitor process log (i don't remember if it's visible in vrclient or vrserver), but you can see all logs chronologically as they happen in the developer console, which you can access by going to http://localhost:27062/console/index.html while SteamVR is running.
Its hard to miss this one in the developer console.
I was playing "RUSH" (i.e. the wingsuit racing title) and saw the -203 error preceded by -301 errors in the vrmonitor.txt log. RUSH doesn't error out often, but when it does, it always seems to be at the beginning of a race.
Thu Dec 09 2021 10:05:30.059005 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:05:30.073410 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:05:31.053943 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:08:46.055544 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:08:46.056852 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:08:47.053829 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:11:56.053086 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:11:56.054444 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:11:57.057576 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:12:02.054085 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:12:02.055396 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:12:03.057877 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:14:18.055242 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:14:18.056373 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:14:19.053212 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:14:22.054735 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:14:22.055685 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:14:23.058629 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:14:31.056024 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:14:31.057433 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:14:32.054043 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:14:49.053145 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:14:49.054092 - [Status Alert] SteamVR Fail (-301) Thu Dec 09 2021 10:14:50.052793 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:16:12.044295 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_NotReady'. Thu Dec 09 2021 10:16:12.045894 - [Status Alert] SteamVR Fail (-203) Thu Dec 09 2021 10:16:35.433159 - [Status Alert] SteamVR Fail (-203) Thu Dec 09 2021 10:16:36.033983 - [Status Alert] Headset Error (-202) Thu Dec 09 2021 10:16:37.233044 - [Status Alert] Headset Error (-202) Thu Dec 09 2021 10:16:38.050096 - [System] Transition from 'SteamVRSystemState_NotReady' to 'SteamVRSystemState_Ready'. Thu Dec 09 2021 10:16:41.503419 - [System] Transition from 'SteamVRSystemState_Ready' to 'SteamVRSystemState_ShutdownRequested'. Thu Dec 09 2021 10:16:41.516482 - [System] Quit gracefully: Waiting for process quitting... Thu Dec 09 2021 10:16:42.877738 - [System] Transition from 'SteamVRSystemState_ShutdownRequested' to 'SteamVRSystemState_Shutdown'. Thu Dec 09 2021 10:16:42.877771 - [Audio] Audio shutdown! Thu Dec 09 2021 10:16:43.685008 - Drops over entire run:- 374 single frame- 72 2 frame- 38 3 frame- 45 4 or more- Total 9490 in 42684 frames
I am having this issue with Half Life Alyx since summer 2021. I spoke to Steam support. They told me to downgrade. I thought the issue was fixed, but now I started playing the game again, and it still persists!
@quantumac I can see you have perfectly described the issue I face. Did you find any solutions?
@mato6666663 No solutions yet. I'm just trying to provide as much context as possible for those tasked with fixing the issue. Hopefully that will be sooner rather than later.
The (-203) error persists with SteamVR [beta] 1.21.4.
I saw the same issues with same titles using the same steps I've outlined above. No change.
OS: Ubuntu 20.04 (Linux) SteamVR [beta]: 1.21.4 NVIDIA driver: 495.44 GPU: GTX 1080 VR Headset: Valve Index
Can confirm, that the -203 crash still persists in the latest SteamVR beta 1.21.4
However i also have an interesting finding. I had to make a few test apps with OpenGL and OpenVR, and wouldn't you know it, my apps get -203 too, except this time i am 100% positive it's not the app or the driver, i didn't get a crash/error/exception from either. Here is the interesting part tho: my app and driver get spammed with VREvent_ActionBindingReloaded
events, no idea what's spawning them yet, but my app and drivers are set to ignore them for now. Never seen this event before, it's data is presumably a process, never seen it processed by other open apps/drivers.
Oh and the weirdest thing is that this event is not spammed on windows.
OS: Ubuntu 20.04.3 LTS SteamVR [beta]: 1.21.4 NVIDIA driver: 495.44 GPU: RTX 3060 VR Headset: Valve Index
Wife and I are encountering the same issue on our respective computers/headsets. We crashed playing Phasmophobia. We crashed at different times, but in nearly the same place in game, which makes me wonder if what was rendered is relevant. -203 error for both, headset turns off, same console spam as others are reporting. Been having this issue for a while now and no amount of tinkering or updating has helped, including wiping/reinstalling steam and re-setting up my playspace.
OS: Kubuntu 21.10 SteamVR [beta]: 1.21.4 (& latest nonbeta) NVIDIA driver: 470.74 GPU: 2080 TI VR Headset: Valve Index
Edit: Same issue presented in Bean Stalker, but not until after 30 minutes of play and shortly after loading a new map, whereas Phasmophobia is within the first 30 (or the first few seconds if the new camp map).
Update 12/17: Upgraded both our rigs to 495.44. Phas didn't cause a 203 during a quick test solo, but crashed once everyone loaded in on multiplayer. Beat saber never seems to trigger it, nor does bean stalker.
There was a certain VR game that kept on crashing, throwing this 203 error after some time, no matter what Nvidia driver version I was using. Seems to go away after performing a quick test since updating "Proton Experimental" to the newest version and running the game under this Proton version.
OS : 5.10.84-1-MANJARO SteamVR[Beta]: 1.21.4 NVIDIA Driver : 495.44 GPU : 2060 Super VR Headset : Vive Pro
@eighthourblink I can reproduce the error in native Linux apps as well as Proton Experimental and other versions of Proton, with both the 470 and 495 NVIDIA drivers. It seems to more likely to occur when there's a lot of CPU load, but not always.
@quantumac For me it feels like its more memory bound, there is a CPU load spike, but its caused by the crash not the other way around.
I've still not had a 203 error since switching from Async to LinuxVulkanAsync
I've still not had a 203 error since switching from Async to LinuxVulkanAsync
Implemented these fixes on my my computer as well as my wife's. I crashed immediately upon entering a map in phasmo, she didn't (for the minute she was in the map before quitting).
Both using Proton Experimental, SteamVR nonbeta.
Changing to @mSparks43's suggestions has fixed my sudden crashes to SteamVR, was almost instantaneous crashing without it. Haven't used it enough to confirm this has been a conclusive fix for me.
Proton 6.3-8 SteamVR: non-beta GPU: GTX 1080 Driver: Nvidia 495.44 OS: PopOS 21.04 Kernel: 5.15 Oculus Rift CV1 & OpenHMD (thaytan branch)
Edit: Spoke too soon, have to play the, "restart steam until it works game", again.
I've still not had a 203 error since switching from Async to LinuxVulkanAsync
LinuxVulkanAsync
is on by default on Linux, and i still get -203 crashes same as before.
My latest -203 crash was from just running SteamVR without any apps (even SteamVR Home), it crashed after just running for a few minutes when i was looking at my desktop. It's really inconsistent though, after that crash i restarted it and it did not crash under the same circumstances.
OS: Ubuntu 20.04.3 LTS Steam Build [beta]: Dec 16 2021, at 22:39:26 SteamVR [beta]: 1.21.4 NVIDIA driver: 495.44 GPU: RTX 3060 VR Headset: Valve Index
is on by default on Linux, and i still get -203 crashes same as before.
Maybe it is now, maybe its disabled by disableasync and needs turning back on. maybe it does nothing and I just have high enough fps not to need it.
When I posted afaict disableasync stopped the 203 errors as others reported enablevulkanasync stopped the unusable stuttery mess caused by disableasync.
Not had it since, still get them reliably if i remove the disableasync line. so not fixed, but work aroundable for me.
is on by default on Linux, and i still get -203 crashes same as before.
Maybe it is now, maybe its disabled by disableasync and needs turning back on. maybe it does nothing and I just have high enough fps not to need it.
When I posted afaict disableasync stopped the 203 errors as others reported enablevulkanasync stopped the unusable stuttery mess caused by disableasync.
Not had it since, still get them reliably if i remove the disableasync line. so not fixed, but work aroundable for me.
HA! disableAsync
is different though, or more like its gone. At some point it got removed from the list of all default SteamVR settings -> it could still be doing something and it missing a default could potentially be a cause to some issues as well... I'll need to test this a bit.
some point it got removed from the list of all default SteamVR settings
from the settings menu it moved to per application settings a few months back. but since the settings menu doesnt work for me at all i have to edit the config with a text editor.
some point it got removed from the list of all default SteamVR settings
from the settings menu it moved to per application settings a few months back. but since the settings menu doesnt work for me at all i have to edit the config with a text editor.
That's the thing though, you see applications have access to the same IVRSettings
API drivers do and disableAsync
is still present in the OpenVR API vrsettings key list (because this API didn't get updated for almost a year now...), now why is this a problem? Well you see to catch errors when using IVRSettings
one needs to pass a pointer to an output variable which will get filled with the error code, now here is the kicker: it's not needed to use this API xD
Now that doesn't mean the app will get garbage, the
IVRSettings
API will return sane defaults in case of error, but those still might not be ideal.
So in conclusion ¯\_(ツ)_/¯ its a mess
Edit: Spoke too soon, have to play the, "restart steam until it works game", again.
Are you saying you're experiencing the same phenomenon as okawo80085 seems to be, in which sometimes it'll crash and sometimes it'll just work solid forever?
This issue persisting for so long is definitely the nail in the coffin for me, going to be buying an AMD GPU next time i'm in the market for upgrading.
Getting the exact same errrors in my vrmotnitor log as the above Large number of events this frame caused the poll loop to exit without draining
and then a prompt 203/202 error code a few seconds later
Arch, base kernel 5.15.11-arch2-1 latest nvidia drivers RTX 2070 Super
Replying to https://github.com/ValveSoftware/SteamVR-for-Linux/issues/452#issuecomment-1002852365
Don't bother? You'll get the same behavior on AMD. Its a general SteamVR on Linux issue, everyone gets excessive binding load events, which don't get processed and end up killing SteamVR eventually. Why? No clue, it literally should not be an issue.
Edit: Spoke too soon, have to play the, "restart steam until it works game", again.
Are you saying you're experiencing the same phenomenon as okawo80085 seems to be, in which sometimes it'll crash and sometimes it'll just work solid forever?
I don't know about "solid forever" but it did seem to be fine for hours on end and other times will crash intermittently, sometimes within second (when starting the room setup). I think it could be worse on the 495 drivers but that's probably cognitive bias. I haven't had a long session without it crashing recently so I'm going to say 2 thumbs down on the current experience, doesn't bother me too much just have to save the game frequently and restart steam & SteamVR multiple times until it's playable.
I tried contacting support about this issue last week, but just got this:
Unfortunately, SteamVR is only supported on Windows 10, we can only recommend testing this issue out on another computer that can support SteamVR with a Windows 10 operating system.
We do apologize for any inconvenience.
However, should this issue continue to persist on a supported operating system, please let us know within your next response, we'll investigate further.
If graphics cards were reasonably priced I'd also switch to an AMD GPU and put my effort into monado instead of steamvr, because valve don't appear to care any longer. Sad :-(
Unfortunately, SteamVR is only supported on Windows 10, we can only recommend testing this issue out on another computer that can support SteamVR with a Windows 10 operating system.
LMAO what? Just for kicks i went to check the Steam page for SteamVR, and yes for some crazy reason its marked as windows only now. All while they're shipping Linux native drivers with it... I just hope Linux support is not gonna be dropped the same way MacOS was...
Oh damn, so i went digging for a bit, and all SteamVR files in my install are still Linux native and no proton in sight... Why is it marked windows only? xd
There may be a difference between officially declared support and pushing out some files compatible with another OS.
That's what I'd assume, the "support" aspect is what they're going to help out with, given a certain degree of platform predictability.
So whilst they may provide source or binaries for *nix systems, the official position will likely be that the composition of the platform it will be used on is too variable to reasonably provide assistance and rule out edge cases in-house.
There may be a difference between officially declared support and pushing out some files compatible with another OS.
That's what I'd assume, the "support" aspect is what they're going to help out with, given a certain degree of platform predictability.
So whilst they may provide source or binaries for *nix systems, the official position will likely be that the composition of the platform it will be used on is too variable to reasonably provide assistance and rule out edge cases in-house.
Fair enough i guess, still feels weird tho xd
Please stop using this as a random discussion forum and spamming notifications for people following. This is for bug reports only.
I'd say those are at least tangentially related to the discussion. The OS is arch, and the error code is one I only remember getting on linux. Also, Valve officially saying they don't support linux here is definitely something I want to know.
Pretty sure the steam deck is linux though, so I'd be really surprised if the software stopped working on linux, but providing support for Arch Linux, where any packages may or may not be installed, and when most people use windows, seems like it really wouldn't be a good economic decision.
Please stop using this as a random discussion forum and spamming notifications for people following. This is for bug reports only.
I mean one is related to the other, but if you only want bug reports here, sure fair enough. Some people are trying to fix the issue though.
So I noticed two errors on repeat:
The one about
filter 'z' in click_button_actions_legacy_17_user_hand_left_input_thumbstick (and the right hand equivalent)
Which seems to stem from
===== knuckles: state=3, empty, uri=file:~/.local/share/Steam/steamapps/common/SteamVR/drivers/indexcontroller/resources/input/legacy_bindings_index_controller.json
and
===== indexhmd: state=4, NOT empty, uri=file:~/.local/share/Steam/steamapps/common/SteamVR/resources/config/legacy_bindings_generic_hmd.json
I renamed the legacy_bindings_index_controller.json to .bak and restarted steamVR. The logs produced the following new error:
Spinning up controllerbinding because binding_load_failed was sent to bindingui/overlay
But I am no longer getting bombarded by the z filter error. Things seem a little quicker to load in game too, but it's anecdotal.
I started Phasmophobia and loaded into the campsite multiple times without an error, something that reliably 203s me.
I tried to rename /legacy_bindings_generic_hmd.json as well but it caused showstopping problems.
TLDR, can others test and report back: Rename ~/.local/share/Steam/steamapps/common/SteamVR/drivers/indexcontroller/resources/input/legacy_bindings_index_controller.json` to anything else, and see if you still -203.
TLDR, can others test and report back: Rename ~/.local/share/Steam/steamapps/common/SteamVR/drivers/indexcontroller/resources/input/legacy_bindings_index_controller.json` to anything else, and see if you still -203.
I renamed the .json file, had an immediate crash to desktop error in half life alyx almost immediately, did the restart everything musical chairs game a few times and then it was fine for approx 30 minutes until it came up with the -203 error agai.
This hasn't helped in my situation unfortunately, good try though!
The only other thing I did was try another binding from within the game. It crashed after that (but before renaming the json) so I didn't include it in the report. Figured maybe an old community-contributed profile may work around a bug complaining that something's missing.
Replying to #452 (Comment)
I tried that, and well its not that simple, I have more devices and i get like 5 different legacy binding loads from different devices depending on which one is set as the active one.... Last time i had enough time to test it i missed 1 controller legacy binding file, although even with 1 binding still missing loads it got better, "Half-Life: Alyx" didn't crash for way longer. Feels like a step in the right direction, still needs more testing though, for example applications that rely on legacy bindings might miss behave (e.g. VRChat).
I'm out of time for the week, so if anyone is up for testing that'd help a lot!
From my limited testing this week:
@okawo80085 Any chance you verified you got the same error as I did at some point about "spinning up controllerbinding", or did the one config you missed prevent that?
I still haven't had a crash since implementing the steps I listed above, I'm hoping the hard-coded "controllerbinding" is formed in such a way as to be exempt from this problem.
Ultimately it's looking like the fix is as simple as valve preventing the binding load infinite loop, at least for this case so it can error gracefully and be done with it.
@earldbjr I'm not really sure, I ran out of time and didn't go through the entirety of the log. But until I connected a new device type (which still had legacy bindings) I didn't see any serious errors.
Again I need to test it more, in my last half-life alyx test I didn't notice that I still had 3 device types with legacy bindings still active.
I need to disable all legacy bindings and try again really.
That lines up with my experience as well. Turn on vr but not knuckles and no relevant errors. As soon as a controller light goes green the flood starts.
I finally weeded out all legacy bindings (including controllers), no more infinite failed binding load loops! 🥳
I used beat saber for testing this time, before it usually crashed like once in a few hours, so far its been up 2 hours without any crashes. It also feels a bit snappier, but thats most likely me just imaging things.
Gotta note this tho, I had to manually load new bindings for beat saber using the legacy binding UI, I broke the default... So yeah gotta fix that xD
I'll keep a few heavier things running to stress test it more, hopefully it gets better.
It also feels a bit snappier, but thats most likely me just imaging things.
I noticed this as well. It's easy to dismiss as bias, but it there is a logic to it. I noticed if I let those errors go for even 10 seconds I could stop steamvr and the console would continue to catch up for a long time.
Can you list all the files you disabled?
Can you list all the files you disabled?
No, there were too many.
Also further testing doesn't look good... Half Life: Alyx crashed with -203 on loading a save... Could be caused by my mess of devices, I'll try to disconnect some.
Ok after disconnecting all the trackers i had lying around it started fine and the default binding for controllers loaded fine so thats good (but since HLA doesn't use legacy bindings i expected that), didn't stutter when loading a save.
Lets see how long it stays stable. (I can't really play the game rn, so if it doesn't crash i'll try that a few hours later, before HLA usually crashed after 40 minutes)
And it crashed after an hour... It did perform a bit better, but it still crashes so back to square 1...
Was the error the same? Any instances of "Large number of events this frame caused the poll loop to exit without draining"?
Describe the bug I am facing an issue where SteamVR will just fail on me constantly, don't get even 5 minutes out of it before it just freezes up and dies on me.
To Reproduce Steps to reproduce the behavior: Use NVIDIA with Arch, try and play SteamVR and watch it fail?
Expected behavior For it to work like it usually does?
System Information (please complete the following information):
Note: Commenters who are also experiencing this issue are encouraged to include the "System Information" section in their replies.