On Tue, 2021-11-16 at 12:38 -0500, Patrick Shanahan wrote:
* Daniel Morris <danielm@ecoscentric.com> [11-15-21 07:40]:
On Sun, Nov 14, 2021 at 03:20:36PM -0500, Patrick Shanahan wrote:
have been using nvidia cards (and others) forever and had few problems, but maybe getting old is now a problem.
Which kernel are you using, which do you think you are using?
I tried the proprietary driver for the first time last week as the nouveau driver keeps locking up every n'th suspend. Initially I got 1024x768 on my 32"/80cm monitor. Then I tried to configure it and just got a black screen after causing an unhelpful xorg.conf file to be created.
This is a Leap 15.2 system (long overdue nuke & paving a fresh install), and a little while later I realised it was running the -prempt kernel. Deleting the xorg.conf and booting from grub to the default kernel had the display spring into life at 2560x1440 without any further intervention. I'll have to give it a few days to see if the lockup after multiple suspend/wake is fixed.
I also notice the X prompt to give the disk decrypt keys for home disks is missing, so I guess there's a dependency/sequencing glitch with the proprietary driver, but simple work-around is to drop to the the virtual console where the kernel's text-mode is waiting for the passphrase.
I think the -prempt kernel was installed due to a virtualbox packaging glitch last year. I stopped digging at that point.
I am definitely running kernel-default-5.14.14.2
NVIDIA stuff is "fun"; I'm hardly a pro with it, but I've had the most, consistent success (when keep TW up-to-date) using the ABSOLUTE latest .run from https://download.nvidia.com/XFree86/Linux-x86_64/ on a laptop that has both Intel & NVIDIA (controlled by "prime"): # prime-select get-current Driver configured: nvidia bbswitch not loaded. NVIDIA modules are loaded I'm currently using 495.44 w/ 5.14.14-2-default (even have secure boot enabled and sign the drivers each time) [paying attention to the missing "sign-file" for each new kernel]. Big things to pay attn to are having nouveau disabled (which the driver installer will do) and ensuring the nvidia_drm kernel module _actually_ loads. There's a file in /etc/modprobe.d that's supposed to do that - but I've had better success just tacking .broken onto the end of it so it doesn't muddy any waters. It seems concerning that no GPU is listed in your Xorg log(s): # grep -i gpu /var/log/Xorg.0.log [ 330.319] (**) | |-->GPUDevice "modesetting" [ 330.319] (**) | |-->GPUDevice "nvidia" [ 330.319] (==) Automatically adding GPU devices [ 330.339] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs [ 330.370] (**) NVIDIA(0): Option "AllowExternalGpus" [ 330.378] (II) NVIDIA(0): NVIDIA GPU Quadro T1000 (TU117GL-A) at PCI:1:0:0 (GPU-0) # uname -a Linux a-dell-laptop 5.14.14-2-default #1 SMP Thu Oct 21 05:05:03 UTC 2021 (2b5383f) x86_64 x86_64 x86_64 GNU/Linux # grep -i T100 /var/log/Xorg.0.log [ 330.378] (II) NVIDIA(0): NVIDIA GPU Quadro T1000 (TU117GL-A) at PCI:1:0:0 (GPU-0) # grep -i drm /var/log/Xorg.0.log [ 330.320] (II) xfree86: Adding drm device (/dev/dri/card1) [ 330.320] (II) xfree86: Adding drm device (/dev/dri/card0) # lsmod | grep -i nvidia nvidia_uvm 1175552 0 nvidia_drm 73728 4 nvidia_modeset 1155072 2 nvidia_drm nvidia 36950016 106 nvidia_uvm,nvidia_modeset drm_kms_helper 299008 2 nvidia_drm,i915 drm 626688 9 drm_kms_helper,nvidia,nvidia_drm,i915,ttm -- ~ Scott Bradnick |- Windows Subsystem for Linux (WSL) Developer |-- Tumbleweed: |--- Dell Precision 5540 [NVIDIA Quadro T1000] (x86_64) |--- O-DROID H2+ [UHD Graphics 600] (x86_64) |--- Raspberry Pi 4 Model B Rev 1.2 (aarch64) |--- WinBook TW100 (x86_64) https://keys.openpgp.org/ :: DBC5AA9A2D2BAEBC