ChanServ changed the topic of #zink to: official development channel for the mesa3d zink driver || https://docs.mesa3d.org/drivers/zink.html
karolherbst is now known as karolherbst[i]
karolherbst[i] has quit []
karolherbst has joined #zink
odrling has quit [Remote host closed the connection]
odrling has joined #zink
<fdobridge> <O​wo> @ermine1716 what firefox was this?
<fdobridge> <O​wo> flatpak?
<fdobridge> <e​rmine1716> no
<fdobridge> <O​wo> what mesa & firefox version?
<fdobridge> <e​rmine1716> from the package manager (openSUSE tumbleweed)
<fdobridge> <e​rmine1716> Firefox 134.0, mesa 24.3.3
<fdobridge> <O​wo> okay, I have 24.3.1
<fdobridge> <O​wo> I'll see if I can't grab some logs
<fdobridge> <O​wo> I think it also had issues with it crashing (explicit sync issues), did you reproduce that?
<fdobridge> <O​wo> I think I got it through completely hiding the window (workspace, other fullscreen window in focus) and opening firefox again
<fdobridge> <e​rmine1716> for me it just crashes in some time after launching
<fdobridge> <e​rmine1716> Wel, okay it doesn't crash while idle
<fdobridge> <e​rmine1716> Opening discord helps it to crash
<fdobridge> <O​wo> do the logs say anything about wl_surface and already having a drm_syncobj?
<fdobridge> <O​wo> specifically in the firefox crash reporter
<fdobridge> <e​rmine1716> No
<fdobridge> <e​rmine1716> Maybe I can make mesa print some debug data?
<fdobridge> <O​wo> not sure, I only ever got that from WAYLAND_DEBUG=1 and the firefox crash reporter
<fdobridge> <O​wo> I'd be curious to see what it tells you
<fdobridge> <e​rmine1716> Btw i use only MESA_LOADER_DRIVER_OVERRIDE=zink . Am I doing it right?
<fdobridge> <O​wo> yeah
<fdobridge> <O​wo> > GraphicsCriticalError: |[0][GFX1-]: (gnome) Wayland protocol error: wl_surface@69: error 0: DRM Syncobj surface object already created for surface 69
<fdobridge> <O​wo> 16 seconds
<fdobridge> <e​rmine1716> OH
<fdobridge> <e​rmine1716> > GraphicsCriticalError: |[0][GFX1-]: GFX: CanvasTranslator failed creating WebGL shared context (t=14.6444) |[1][GFX1-]: (sway) Wayland protocol error: wp_tearing_control_manager_v1#100: error 0: Tearing control object already exists!
<fdobridge> <O​wo> Interesting
<fdobridge> <O​wo> I'm literally watching a YouTube video with the glitching right now lmao
<fdobridge> <e​rmine1716> seizure warning!
<fdobridge> <O​wo> The right side of the video
<fdobridge> <e​rmine1716> i see
<fdobridge> <O​wo> Heh, full screen did it pretty quick
<fdobridge> <e​rmine1716> Well, okay it doesn't crash while idle (edited)
<fdobridge> <O​wo> > GraphicsCriticalError: |[0][GFX1-]: (gnome) Wayland protocol error: wp_linux_drm_syncobj_surface_v1@81: error 3: Release or Acquire point set but no buffer attached
<fdobridge> <O​wo> > (t=226.657)
<fdobridge> <e​rmine1716> it feels to me that glitches and crash are different issues
<fdobridge> <e​rmine1716> But that's wild guess
<fdobridge> <O​wo> it seems like it's some weird interaction between firefox and mesa wsi tbh
<fdobridge> <O​wo> this might explain a lot
<fdobridge> <O​wo> Firefox is somehow doing things ina way that violate the protocol
<fdobridge> <O​wo> this does also happen with Fedora's shipped Firefox, so the discussion there about them fixing it seems to be wrong
<fdobridge> <O​wo> at least if we're thinking that they already fixed it
<fdobridge> <O​wo> probably something different, not related too much....
<fdobridge> <e​rmine1716> We'll see once they ship the fix
<fdobridge> <O​wo> it looks like it should've already been, though
<fdobridge> <O​wo> 134 is from Jan 7
<fdobridge> <O​wo> this was *august*
<fdobridge> <e​rmine1716> Oh