Description of problem: I am running an xorg server (nvidia driver) in 10 bits mode: DefaultDepth 30 When I am starting firefox in a terminal, a lot of errors pop out: rash Annotation GraphicsCriticalError: |[0][GFX1-]: glxtest: ManageChildProcess failed (t=0.157897) [GFX1-]: glxtest: ManageChildProcess failed Crash Annotation GraphicsCriticalError: |[0][GFX1-]: glxtest: ManageChildProcess failed (t=0.157897) |[1][GFX1-]: glxtest: X error, error_code=10, request_code=152, minor_code=5 (t=0.157897) [GFX1-]: glxtest: X error, error_code=10, request_code=152, minor_code=5 Extension error: TypeError: tab is undefined moz-extension://b5004860-4764-4ed1-b430-4a0dc8c5576f/public/js/background.js 25888 And, indeed, firefox is very slow and laggy. Having a look into about:config, searching for hardware accelerated options (containing gl, egl, etc...), they're all disabled with a value being "FAILED_GLXTEST..." Now, forcing hardware accelaration by setting: gfx.webrender.all = true, fixes all issues. I would say there is onl a bug within the GLXTEST, which fails on a 10 bits display. Cheers, chris.
Thank you for the report. glxtest appears in: firefox:/usr/lib64/firefox/glxtest thunderbird:/usr/lib64/thunderbird/glxtest so, a Firefox issue. (In reply to Chris Denice from comment #0) > I am running an xorg server (nvidia driver) in 10 bits mode: > DefaultDepth 30 The signifance of this is beyond me. Did you try other 'depths'? > Now, forcing hardware acceleration by setting: gfx.webrender.all = true, > fixes all issues. Important to note; congrats on finding it. > I would say there is onl a bug within the GLXTEST, which fails on a 10 bits > display. Agree. Not finding anything similar by searching, I suggest raising a bug chez Mozilla: https://bugzilla.mozilla.org/ I guess you would need to give info like: $ inxi -Gxx
CC: (none) => lewyssmith
Do the same problem happen with upstream rpm or flatpak?
CC: (none) => fri
>> DefaultDepth 30 >The signifance of this is beyond me. Did you try other 'depths'? That's the linux way to switch color output to 10 bits mode. Default nowadays should be 24 (8+8+8 bits for RGB). So, 10+10+10 is depth 30. Set in /etc/X11/xorg.conf, in the section "Screen" Section "Screen" Identifier "Screen0" Device "Device0" Monitor "Monitor0" DefaultDepth 30 <----------- ... PS: Default on Mageia is "24", everything runs smooth with the default, but I've gotten a HDR screen and a new RTXA6000 to toy with :) I'll test with upstream flatpak.
Thank you for the explanations. My own /etc/X11/xorg.conf has just: Section "Screen" Identifier "screen1" Device "device1" Monitor "monitor1" EndSection so imagine that the normal depth value 8 is defaulted. In your case, did you manually add the line: DefaultDepth 30 The request for an upstream bug remains...
>In your case, did you manually add the line: > DefaultDepth 30 Yes, exactly, and it works fine. >The request for an upstream bug remains... Yep ;)
https://bugzilla.mozilla.org/show_bug.cgi?id=1935653 Seems that glxtest is failing in many situations, my bug is one among many :-/
From your upstream report, I note the workaround Forcing: gfx.webrender.all = true fixes all issues,
Keywords: (none) => UPSTREAMSummary: Hardware acceleration undetected on 10 bits mode => Firefox: Hardware acceleration undetected on 10 bits modeURL: (none) => https://bugzilla.mozilla.org/show_bug.cgi?id=1935653Status comment: (none) => Workaround see c7
Thank you Chris for the detective work & reporting it upstream. I think it (glxtest) applies to Thunderbird too. Nothing we can do until they fix it. Assigning this to ns80 who nurses Firefox & TB, for when the correction eventually arrives. Nothing he can do either. This is good for ERRATA I think. A definite upstream fault with a DIY cure.
CC: lewyssmith => (none)Summary: Firefox: Hardware acceleration undetected on 10 bits mode => Firefox: Hardware acceleration undetected on 10 bits mode; easy workaround givenAssignee: bugsquad => nicolas.salgueroKeywords: (none) => FOR_ERRATA9
https://wiki.mageia.org/en/Mageia_9_Errata#Firefox_ESR
Keywords: FOR_ERRATA9 => IN_ERRATA9