Arch Linux User Repository

Suggestion: Add a patch that fixes warnings like this:

Nov 16 19:29:34 arch systemd-udevd[290]: nvidia: Process '/usr/bin/bash -c '/usr/bin/mknod -Z -m 666 /dev/nvidiactl c $(grep nvidia-frontend /proc/devices | cut -d \  -f 1) 255'' failed with exit code 1.
Nov 16 19:29:34 arch systemd-udevd[282]: nvidia: Process '/usr/bin/bash -c 'for i in $(cat /proc/driver/nvidia/gpus/*/information | grep Minor | cut -d \  -f 4); do /usr/bin/mknod -Z -m 666 /dev/nvidia${i} c $(grep nvidia-frontend /proc/devices | cut -d \  -f 1) ${i}; done'' failed with exit code 1.

Those come from nvidia-470xx.rules, and they fail because these nodes were already created. I suggest verifying if they already exist before trying to create them. Patch:

diff --git a/nvidia-470xx.rules b/nvidia-470xx.rules
index 97218df..cd2a0a0 100644
--- a/nvidia-470xx.rules
+++ b/nvidia-470xx.rules
@@ -1,6 +1,6 @@
 # Make sure device nodes are present even when the DDX is not started for the Wayland/EGLStream case
-KERNEL=="nvidia", RUN+="/usr/bin/bash -c '/usr/bin/mknod -Z -m 666 /dev/nvidiactl c $$(grep nvidia-frontend /proc/devices | cut -d \  -f 1) 255'"
-KERNEL=="nvidia", RUN+="/usr/bin/bash -c 'for i in $$(cat /proc/driver/nvidia/gpus/*/information | grep Minor | cut -d \  -f 4); do /usr/bin/mknod -Z -m 666 /dev/nvidia$${i} c $$(grep nvidia-frontend /proc/devices | cut -d \  -f 1) $${i}; done'"
-KERNEL=="nvidia_modeset", RUN+="/usr/bin/bash -c '/usr/bin/mknod -Z -m 666 /dev/nvidia-modeset c $$(grep nvidia-frontend /proc/devices | cut -d \  -f 1) 254'"
-KERNEL=="nvidia_uvm", RUN+="/usr/bin/bash -c '/usr/bin/mknod -Z -m 666 /dev/nvidia-uvm c $$(grep nvidia-uvm /proc/devices | cut -d \  -f 1) 0'"
-KERNEL=="nvidia_uvm", RUN+="/usr/bin/bash -c '/usr/bin/mknod -Z -m 666 /dev/nvidia-uvm-tools c $$(grep nvidia-uvm /proc/devices | cut -d \  -f 1) 1'"
+KERNEL=="nvidia", RUN+="/usr/bin/bash -c 'if [ ! -c /dev/nvidiactl ]; then /usr/bin/mknod -Z -m 666 /dev/nvidiactl c $$(grep nvidia-frontend /proc/devices | cut -d \  -f 1) 255; fi'"
+KERNEL=="nvidia", RUN+="/usr/bin/bash -c 'for i in $$(cat /proc/driver/nvidia/gpus/*/information | grep Minor | cut -d \  -f 4); do if [ ! -c /dev/nvidia$${i} ]; then /usr/bin/mknod -Z -m 666 /dev/nvidia$${i} c $$(grep nvidia-frontend /proc/devices | cut -d \  -f 1) $${i}; fi; done'"
+KERNEL=="nvidia_modeset", RUN+="/usr/bin/bash -c 'if [ ! -c /dev/nvidia-modeset ]; then /usr/bin/mknod -Z -m 666 /dev/nvidia-modeset c $$(grep nvidia-frontend /proc/devices | cut -d \  -f 1) 254; fi'"
+KERNEL=="nvidia_uvm", RUN+="/usr/bin/bash -c 'if [ ! -c /dev/nvidia-uvm ]; then /usr/bin/mknod -Z -m 666 /dev/nvidia-uvm c $$(grep nvidia-uvm /proc/devices | cut -d \  -f 1) 0; fi'"
+KERNEL=="nvidia_uvm", RUN+="/usr/bin/bash -c 'if [ ! -c /dev/nvidia-uvm-tools ]; then /usr/bin/mknod -Z -m 666 /dev/nvidia-uvm-tools c $$(grep nvidia-uvm /proc/devices | cut -d \  -f 1) 1; fi'"

As @Serg12344 mentioned I can confirm that nvidia-470xx-utils 470.141.03-2 is working fine with kernel x86_64 Linux 6.0.8-arch1-1 (GPU: NVIDIA GeForce GTX 780 Ti)

@SoftExpert Please remove the out-of-date status. Package nvidia-470xx-* 470.141.03 works fine for me with GeForce GTX 780ti graphics card on linux-zen-6.0.6 and linux-zen-6.0.8 kernels and I don't see package version 470.141.10 on NVIDIA official website. Maybe they removed it

The driver version 470.141.10 is not available in Geforce 730,so please cancel to FLAG TO OUTDATE.

@XAVeRY Yes, I did that exactly locally on my archbox. Only difference is that I did not apply patch named 009*, only four of them (I've strictly followed what fedora nvidia maintainer did) . But so far so good, everything is back to a good shape.

• Stock 470.141.10 does not support Linux 6.0, compilation fails on ACPI stuff.
• Using the kernel-6.0.patch against 470.141.10 results in the module loading on 6.0.6, but failing to run with same messages as previously :

[   17.367232] NVRM: GPU 0000:01:00.0: Failed to copy vbios to system memory.
[   17.367356] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x30:0xffff:874)
[   17.367404] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0

• Applying the Debian patches to 470.141.03 makes the module load and run properly on 6.0.6.
• 470.141.10 with Debian patches seems to work okay as well.

This was tested on my Lenovo Z500 and a GeForce GT 645M. Offload (prime-run glxgears), Vulkan (vkcube) as well as nvidia-smi are working as expected.

PKGBUILDs including the Debian patches are here :

• 470.141.03 : https://github.com/xavery/nvidia-470xx-utils/tree/new-patches
• 470.141.10 : https://github.com/xavery/nvidia-470xx-utils/tree/new-patches-and-tesla

That said, I don't think running 470.141.10 provides any value over the current version. I seem to remember some differences being there between "regular" and "datacenter" (Tesla) drivers so perhaps it's better to stay at .03. My memory might be wrong here, though, so if anybody has more actual information on this, please speak up.

A new version of the driver is out 470.141.10 Also, there are some extra patches that could solve additional issues. With the current patches I was able to run all kernel 6.0.x updates without issues - maybe the new version will be more stable for a larger set of configurations ?

Updated my system only to find out this driver doesn't like 6.0, I guess thats on me for not checking first but its still a little annoying. I was able to fall back to the nouveau driver with my old GT 730 to get back to my desktop, but its not ideal. Hopefully the issue is fixed soon. The GT 730 does NOT like the nouveau driver (main issue is screen tearing, but still)

On my debian sid box everything is fine with the latest patches applied by the debian maintainer. With kernel 6.0 offloading is working fine as well as graphic acceleration. The patches are available here: https://sources.debian.org/src/nvidia-graphics-drivers-tesla-470/470.141.03-3/debian/module/debian/patches/ I'm not an expert thus I don't really know how to apply them correctly.

@BlackIkeEagle yay worked fine for me :)