Dear All,
I am getting the following error message when trying to start OBS Studio:
info: Initializing OpenGL... libEGL warning: DRI2: failed to authenticate MESA: error: ZINK: failed to choose pdev libEGL warning: egl: failed to create dri2 screen error: Unable to create EGL context: EGL_BAD_ATTRIBUTE error: Failed to create context! error: device_create (GL) failed error: Failed to initialize video. Your GPU may not be supported, or your graphics drivers may need to be updated.
This error began after:
Packages Altered: Upgrade mplayer-1.5.1-0.12.20240317svn.fc39.x86_64 @rpmfusion-free-updates Upgraded mplayer-1.5.1-0.8.20230811svn.fc39.x86_64 @@System Upgrade mplayer-common-1.5.1-0.12.20240317svn.fc39.x86_64 @rpmfusion-free-updates Upgraded mplayer-common-1.5.1-0.8.20230811svn.fc39.x86_64 @@System Upgrade akmod-nvidia-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded akmod-nvidia-3:550.54.14-2.fc39.x86_64 @@System Upgrade nvidia-modprobe-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded nvidia-modprobe-3:550.54.14-1.fc39.x86_64 @@System Upgrade nvidia-persistenced-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded nvidia-persistenced-3:550.54.14-1.fc39.x86_64 @@System Upgrade nvidia-settings-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded nvidia-settings-3:550.54.14-1.fc39.x86_64 @@System Upgrade xorg-x11-drv-nvidia-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-3:550.54.14-1.fc39.x86_64 @@System Upgrade xorg-x11-drv-nvidia-cuda-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-cuda-3:550.54.14-1.fc39.x86_64 @@System Upgrade xorg-x11-drv-nvidia-cuda-libs-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-cuda-libs-3:550.54.14-1.fc39.x86_64 @@System Upgrade xorg-x11-drv-nvidia-kmodsrc-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-kmodsrc-3:550.54.14-1.fc39.x86_64 @@System Upgrade xorg-x11-drv-nvidia-libs-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-libs-3:550.54.14-1.fc39.x86_64 @@System Upgrade xorg-x11-drv-nvidia-power-3:550.67-1.fc39.x86_64 @rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-power-3:550.54.14-1.fc39.x86_64 @@System
Could you please help me?
Thanks in advance,
Paul
Hi.
On Sun, 24 Mar 2024 22:19:51 +0000 Paul Smith wrote:
This error began after:
Packages Altered:
...
@rpmfusion-nonfree-updates Upgraded akmod-nvidia-3:550.54.14-2.fc39.x86_64 @@System
...
@rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-3:550.54.14-1.fc39.x86_64 @@System
Have you wait for akmods to compile the nvidia kernel modules before rebooting ?
"lsmod | grep nvidia" should list the nvidia module (and some others) loaded. Example:
lsmod | grep nvidia nvidia_drm 81920 5 nvidia_modeset 1511424 3 nvidia_drm nvidia_uvm 2760704 0 nvidia 40706048 99 nvidia_uvm,nvidia_modeset video 77824 2 dell_wmi,nvidia_modeset
"modinfo nvidia | grep version" should give the same version as xorg-x11-drv-nvidia, ie: 550.54.14
If this is not the case, try to rebuild the kernel modules with:
sudo akmods --rebuild
On 03/30/2024 10:43 AM, Francis.Montagnac@inria.fr wrote:
Have you wait for akmods to compile the nvidia kernel modules before rebooting ?
Unless things have changed considerably since I used those drivers, the akmod module runs at boot, and checks to see if it needs to do anything. If it does, it rebuilds the kmod; if not, it simply exits. Running it after an update and before reboot is just making sure it gets done and can be skipped if you don't mind the extra few seconds during boot. I never bothered and never had any problems letting them do their jobs as they were designed to do.
On 30/03/2024 16:55, Joe Zeff wrote:
On 03/30/2024 10:43 AM, Francis.Montagnac@inria.fr wrote:
Have you wait for akmods to compile the nvidia kernel modules before rebooting ?
Unless things have changed considerably since I used those drivers, the akmod module runs at boot, and checks to see if it needs to do anything. If it does, it rebuilds the kmod; if not, it simply exits. Running it after an update and before reboot is just making sure it gets done and can be skipped if you don't mind the extra few seconds during boot. I never bothered and never had any problems letting them do their jobs as they were designed to do.
The rpmfusion nvidia howto says it may take 5 minutes to build. That's a long time to wait with no obvious progress during a reboot. It's much more comfortable to wait until the builds (maybe for several kernels) are complete before triggering the reboot. And ISTR that graphically selecting a restart early can give problems too.
systemctl show-jobs can show progress, and then systemctl reboot.
John P
On Sat, 2024-03-30 at 17:15 +0000, John Pilkington wrote:
The rpmfusion nvidia howto says it may take 5 minutes to build. That's a long time to wait with no obvious progress during a reboot.
Many years ago I had to put up with that (different hardware now). But, I didn't hide the textual boot progress messages, so I could see what I was waiting for. I don't recall if I got progress info as it compiled, or I could just see that the stall was due to a compile being started.
Also, back then, we didn't have parallel boot processes, so the system did wait for it to finish before carrying on.
On 31 Mar 2024, at 02:33, Tim via users users@lists.fedoraproject.org wrote:
Also, back then, we didn't have parallel boot processes, so the system did wait for it to finish before carrying on.
The boot in today’s Fedora waits for akmods to build anything that is out of date. But there is a known problem which is likely depmods not running if you reboot at the wrong time
Barry
On Sat, 30 Mar 2024 10:55:35 -0600 Joe Zeff wrote:
On 03/30/2024 10:43 AM, Francis.Montagnac@inria.fr wrote:
Have you wait for akmods to compile the nvidia kernel modules before rebooting ?
Unless things have changed considerably since I used those drivers,
Probably not.
the akmod module runs at boot, and checks to see if it needs to do anything. If it does, it rebuilds the kmod; if not, it simply exits.
Right, but it also runs asynchonously to the dnf transaction, before the reboot.
The check at boot is incomplete: it only checks that the kmod-nvidia-KVER rpm is installed, but this rpm does a mandatory "depmod" in postinstall and this "depmod" may have been killed during shutdown.
I made a proposal to prevent that:
kmod failed to load after upgrade Fedora using dnf system-upgrade https://bugzilla.redhat.com/show_bug.cgi?id=2011120
still waiting for approval:
Running it after an update and before reboot is just making sure it gets done and can be skipped if you don't mind the extra few seconds during boot.
Not only. AFAIK, the nvidia kernel module is mandatory at boot otherwise Xorg may fail since it loads the nvidia_drv.so driver from xorg-x11-drv-nvidia.that has to have the exact version as the kernel module.
Also, on some machines, nvidia-drm.modeset=1 is mandatory in the kernel command line.
I never bothered and never had any problems letting them do their jobs as they were designed to do.
IMO you were lucky.
On Sat, Mar 30, 2024 at 4:43 PM Francis.Montagnac@inria.fr wrote:
Hi.
On Sun, 24 Mar 2024 22:19:51 +0000 Paul Smith wrote:
This error began after:
Packages Altered:
...
@rpmfusion-nonfree-updates Upgraded akmod-nvidia-3:550.54.14-2.fc39.x86_64 @@System
...
@rpmfusion-nonfree-updates Upgraded xorg-x11-drv-nvidia-3:550.54.14-1.fc39.x86_64 @@System
Have you wait for akmods to compile the nvidia kernel modules before rebooting ?
"lsmod | grep nvidia" should list the nvidia module (and some others) loaded. Example:
lsmod | grep nvidia nvidia_drm 81920 5 nvidia_modeset 1511424 3 nvidia_drm nvidia_uvm 2760704 0 nvidia 40706048 99 nvidia_uvm,nvidia_modeset video 77824 2 dell_wmi,nvidia_modeset
"modinfo nvidia | grep version" should give the same version as xorg-x11-drv-nvidia, ie: 550.54.14
If this is not the case, try to rebuild the kernel modules with:
sudo akmods --rebuild
I thank all who have answered!
By following Francis's advice of
sudo akmods --rebuild
I was able to fix the problem.
Paul