I ran zypper dup today and now have the situation where the newest/default kernel (6.3.6-1-default) will not load the display manager. After a minute or so I get dumped at a TTY login. The older kernel (6.3.4-1-default) continues to work fine.
I’ve collected logs from both boots and the interesting detail is in dmesg. The problem appears to be with the nVidia proprietary driver for my RTX 3060. In the new kernel I get:
[ 26.082651] nvidia-nvlink: Nvlink Core is being initialized, major device number 239
[ 26.082673] traps: Missing ENDBR: _nv010989rm+0x0/0x10 [nvidia]
[ 26.082866] ------------[ cut here ]------------
[ 26.082867] kernel BUG at arch/x86/kernel/traps.c:255!
[ 26.082869] invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
[ 26.083558] CPU: 10 PID: 1476 Comm: modprobe Tainted: P OE 6.3.6-1-default #1 openSUSE Tumbleweed d92ec5864371d7852882cd4aa0a220829340020d
... snip ...
[ 55.954140] nvidia_modeset: gave up waiting for init of module nvidia.
[ 55.957388] nvidia_modeset: Unknown symbol nvidia_register_module (err -16)
[ 55.957548] nvidia_modeset: gave up waiting for init of module nvidia.
[ 55.964158] nvidia_modeset: Unknown symbol nvidia_get_rm_ops (err -16)
[ 55.964917] nvidia_modeset: gave up waiting for init of module nvidia.
[ 55.965627] nvidia_modeset: Unknown symbol nvidia_unregister_module (err -16)
With the older working kernel I get:
[ 18.745377] nvidia-nvlink: Nvlink Core is being initialized, major device number 239
[ 18.745822] nvidia 0000:01:00.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=none:owns=io+mem
[ 18.787526] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 470.182.03 Fri Feb 24 03:29:56 UTC 2023
Ah I’m not sure how I ended up choosing G05 in the first place. I’ve now switched over to G06, which appeared to install happily. It doesn’t appear to have made a difference to the issue unfortunately.
Both of these are different from yours so I’m not quite sure what to make of it.
Oddly, my system has also gotten into a state where I need to run prime-select next-boot nvidia once before every single boot or else it doesn’t even try to start the display manager, regardless of kernel selection. I think something went awry while changing from G05 to G06 (I was prompted to use prime-select service restore).
I tried this earlier after seeing it mentioned when searching online but it made no difference.
Y’know I didn’t really think hard about the integrated intel graphics and I was just setting this up on autopilot. It occurred to me that my current workload doesn’t require the high-powered GPU so I could just yank it out, so I did. After a protracted adventure where this forced my firmware into UEFI mode, where my installation didn’t even have EFI packages or an EFI partition, eventually everything seems to have come right on the newest kernel.
It’s a bit disappointing not to find the root cause - thanks for the suggestions - but I will have to assume it was something I screwed up on my machine or else this thread would be full of people.
@thombles So it’s a desktop system? If so check the system BIOS as likely just a setting to switch of the integrated GPU, or like I indicated use Prime Render Offload (which I do here on my desktop).
I have the same problem with my Nvidia driver. Also with the newest kernel 6.3.7. Do you have find a solution? Current I have switch to my internal Intel graphic with prime-select.
I use also the G05 driver for my nvidia card.I have an NVIDIA GeForce GTX 1650 with Max-Q Design/PCIe/SSE2. Run this Nvidia Card with the G06 driver? Is this card from the series 700 or newer? I can not find information about that.
“G06 = driver for GeForce 700 series (Kepler) and up (Maxwell, Pascal…)”