Launching server with SteamVR + ALVR results in OpenGL logspam

When I try to launch the actual server on a real device using SteamVR + ALVR + AVP, I just get this printed to the console, and it just stays loading in SteamVR:
[SK info] Before init >>>
libEGL warning: egl: failed to create dri2 screen
libEGL warning: egl: failed to create dri2 screen
2025-09-01T04:11:36.802235Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_INVALID_ENUM in glGetIntegerv(pname=GL_TEXTURE_2D)
2025-09-01T04:11:36.802297Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_OUT_OF_MEMORY in
2025-09-01T04:11:36.804001Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_INVALID_ENUM in glGetIntegerv(pname=GL_TEXTURE_2D)
2025-09-01T04:11:36.804024Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_OUT_OF_MEMORY in
2025-09-01T04:11:36.817999Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_INVALID_ENUM in glGetIntegerv(pname=GL_TEXTURE_2D)
2025-09-01T04:11:36.818116Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_OUT_OF_MEMORY in
<repeating...>
[SK info] Before init >>>
libEGL warning: egl: failed to create dri2 screen
libEGL warning: egl: failed to create dri2 screen
2025-09-01T04:11:36.802235Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_INVALID_ENUM in glGetIntegerv(pname=GL_TEXTURE_2D)
2025-09-01T04:11:36.802297Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_OUT_OF_MEMORY in
2025-09-01T04:11:36.804001Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_INVALID_ENUM in glGetIntegerv(pname=GL_TEXTURE_2D)
2025-09-01T04:11:36.804024Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_OUT_OF_MEMORY in
2025-09-01T04:11:36.817999Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_INVALID_ENUM in glGetIntegerv(pname=GL_TEXTURE_2D)
2025-09-01T04:11:36.818116Z ERROR tokio-runtime-worker smithay::backend::renderer::gles: 358: [GL] GL_OUT_OF_MEMORY in
<repeating...>
I initially tried with the Fedora packaged version, and that had the same issue as above (except ending in a segfault), and I just get that spam above with no segfault when built in dev mode locally. Flatscreen works fine, though. Is this possibly due to Nvidia proprietary drivers? I'm using Nvidia driver version 575.64.05.
27 Replies
Nova
Nova2w ago
yep, it's probably nvidia we're refactoring the server to use opengl so this hopefully shouldn''t be an issue much longer
dinosauce313
dinosauce313OP2w ago
Ah got it
Nova
Nova2w ago
wait i meant to not use opengl to use vulkan my brain does NOT wanna work today :blobcatgoogly:
dinosauce313
dinosauce313OP2w ago
Okay that makes a bit more sense
Nova
Nova2w ago
anyway if you wanna get an update when that's finished i can give you stardust notification role
dinosauce313
dinosauce313OP2w ago
yeah that'd be dope thanks
Nova
Nova2w ago
here's a sneak peek btw
Nova
Nova2w ago
the server is using bevy now, with a custom wayland impl from scratch to take advantage of vulkan with the dmabufs
Nova
Nova2w ago
dinosauce313
dinosauce313OP2w ago
are those things that would have been laggy in opengl, or just useful tests for the vulkan implementation? sorry i'm not super familiar with graphics programming lol
Nova
Nova2w ago
ohhh no that's just a fun client i mocked up in a few hours and it looked neat
dinosauce313
dinosauce313OP2w ago
ah gotcha lol
Nova
Nova2w ago
realistically for now vulkan is just for compatibility with everything, using the gpu that openxr is using (your dedicated one if present), working on nvidia, and eventually gaussian splatting
dinosauce313
dinosauce313OP2w ago
is the nvidia opengl implementation not quite the same as the open source driver implementations?
Nova
Nova2w ago
no, they impl their own except if you use mesa which is glitchy in other ways vulkan lets us guarantee we can pick the right gpu that openxr does 100% of the time and so far it just works on nvidia will have to ensure that's the case with wayland dmabuf tho
dinosauce313
dinosauce313OP2w ago
by pick the right gpu, you mean for people who have an igpu + a dedicated gpu, sometimes it will render stardust on a different gpu than openxr?
Nova
Nova2w ago
the old stereokit one is hardcoded to use /dev/dri/renderD128 no matter what because openxr never tells us what gpu is used when using opengl!!! and i couldn't find a reliable way to guesstimate which it would be but openxr tells us (via vulkan_enable2) what vkdevice is used when using vulkan
dinosauce313
dinosauce313OP2w ago
lol that sounds painful
Nova
Nova2w ago
yes indeed that's why it's been 10 MONTHS since the last release loooll
dinosauce313
dinosauce313OP2w ago
lol anyways, thanks for the project! i'll keep an eye out for that update. also, not sure if useful, but i'd be happy to contribute any low hanging fruit that may exist - i kind of know my way around systems programming, though I've never really dealt with graphics. otherwise, happy to beta test whenever this gets there
Nova
Nova2w ago
!! yea i think there's a few things that'd be a huge help you know rust? i don't really need as much graphics help rn (i mean i do but schmarni knows enough) but there's some abstracting of things that'd make code wayyy easier to maintain using dbus and tokio and all
dinosauce313
dinosauce313OP2w ago
I made a toy wgpu project a while back (like a year or two at least lol), but I'm decently experienced with C++, though nowadays I do more JS than anything else :P I'm a bit rusty with lower level stuff, but I think I could probably pick it back up without too much difficulty C++ != Rust though, so not sure if there's some basic stuff that's alright for someone to get their feet wet - I don't want to be a net burden with review as opposed to a net benefit lol
Nova
Nova2w ago
wgpu isn't c++ though right?
dinosauce313
dinosauce313OP2w ago
yeah i made a toy rust app
Nova
Nova2w ago
nice nice
dinosauce313
dinosauce313OP2w ago
was trying to make a race sim with proper kinematics and such, then got busy and didn't finish lol
Nova
Nova2w ago
mood

Did you find this page helpful?