You are not logged in.
The "nvidia" package is always compiled against one specific kernel version. 460.67-5 is for the 5.11.11 kernel. You'd have to opt for 460.67-7 for the 5.11.13.arch1 kernel, which is why I'm saying that downgrading the kernel is unnecessary and you could just opt for the relevant nvidia-dkms variant until the bug in the nvidia driver is resolved.
Online
Issue persists with 5.11.16-arch1-1 and nvidia 465.24.02-4. Each day nouveau looks more and more tempting.
Offline
This isn't going to be fixed by akernel update or nvidia rebuild, you're waiting for nvidia 468.37 (<= totally made up version) or so, ie. the next driver update.
Offline
I appear to have this issue as well. However it only happens with a specific monitor and a specific output (1080, Dell 3818DW on display port) switching to HDMI seems to have side stepped it
Offline
I appear to have this issue as well. However it only happens with a specific monitor and a specific output (1080, Dell 3818DW on display port) switching to HDMI seems to have side stepped it
I have the same monitor. I had to downgrade the Nvidia drivers.
Offline
hi, same problem here with the same screen.
When I switch using the display port switchbox, then the screen goes black and there is no way to take it back other than using arandr.
It happends with the nouveau driver, it was working fine with the old nvidia driver.
Offline
It happends with the nouveau driver, it was working fine with the old nvidia driver.
The issue is caused by an error in the nvidia module as shown in the backtraces. You have a kernel panic from the nouveau module referencing _nv015534rm?
Offline
It happends with the nouveau driver, it was working fine with the old nvidia driver.
Might be in error about the meaning of "nouveau", @dummys, what's the output of
lsmod | grep -E '(nvidia|nouveau)'
(w/ the problematic driver)
Last edited by seth (2021-04-27 20:33:27)
Offline
-> $ lsmod | grep -E '(nvidia|nouveau)'
nouveau 2371584 9
mxm_wmi 16384 1 nouveau
drm_ttm_helper 16384 1 nouveau
ttm 81920 2 drm_ttm_helper,nouveau
wmi 36864 4 asus_wmi,wmi_bmof,mxm_wmi,nouveau
video 53248 3 asus_wmi,i915,nouveau
i2c_algo_bit 16384 2 i915,nouveau
drm_kms_helper 282624 2 i915,nouveau
drm 569344 14 drm_kms_helper,drm_ttm_helper,i915,ttm,nouveau
agpgart 40960 4 intel_gtt,ttm,nouveau,drm
There is actually 2 problems.
First the proprietary nvidia driver crash at boot, so I switch back to nouveau. But with nouveau, I get the problem that one screen when switching screen with my display port switchbox, doesn't come to life. I had to use arandr to untick "active", do a apply, then re tick it, and re-apply then my screen is back...
Offline
But with nouveau, I get the problem that…
That has nothing to do w/ the issue of this thread.
Please try either reverting to 460xx as discussed here and/or open a new thread for your nouveau/randr/KVM issue.
Offline
I'm ok that the nouveau things is not, but the crash with the nvidia proprietary is the issue of the thread for sure
Offline
Yes?
Please try […] reverting to 460xx as discussed here
Offline
Guys, a friendly notification that the Nvidia driver is still broken as of today's update: nvidia-465.27-2
I made the mistake of believing that the bug was fixed without evidence and upgraded my system...
And while trying to revert to the last working version, I forgot to downgrade my kernel as well. Had to chroot into my system from a USB to get it to boot again.
I've been searching through the Nvidia forums and this seems to be a setup-specific issue? It doesn't seem to be a widespread issue.
So I think I should give Nvidia the crash logs but I'm not sure how to give them the proper log in case of a kernel panic.
If simply running the script they provide for collecting logs is fine, I'll do it.
Offline
Best would be to run that when the issue happens, check if you can SSH in while the system is "broken" and run the collector script during the broken situation if at all possible, otherwise posting simply the log from a normal run including the backtrace in the journal/dmesg should be sufficient
Online
Had the same issues here with a GT 1030 graphics card connected via DP.
Switching to HDMI solved the problem.
I can now run
Linux 5.11.16-arch1-1
together with
nvidia 465.27-2
running xorg.
Before I always had to downgrade to
linux-5.11.13.arch1-1
with
nvidia-460.67-7-x86_64
in order to start xorg.
Last edited by fermikante (2021-04-30 08:02:24)
Offline
Chiming in, me too on GTX2080+DisplayPort+Dell U3818, was unable to boot so no SSH server started to go steal logs
Downgrade command:
sudo pacman -U linux-5.11.12.arch1-1-x86_64.pkg.tar.zst nvidia-460.67-6-x86_64.pkg.tar.zst nvidia-utils-460.67-1-x86_64.pkg.tar.zst nvidia-settings-460.67-1-x86_64.pkg.tar.zst
However, I can no longer launch `steam`:
steam
Running Steam on arch rolling 64-bit
STEAM_RUNTIME is enabled automatically
Pins up-to-date!
Can't find 'steam-runtime-check-requirements', continuing anyway
WARNING: Using default/fallback debugger launch
/home/greg/.local/share/Steam/ubuntu12_32/steam
[2021-04-30 08:02:21] Startup - updater built Mar 23 2021 20:11:00
ILocalize::AddFile() failed to load file "public/steambootstrapper_english.txt".
Installing breakpad exception handler for appid(steam)/version(1618256785)
SteamUpdateUI: An X Error occurred
X Error of failed request: BadValue (integer parameter out of range for operation)
glxgears launches fine, anyone else run into this after the downgrade? I've tried reinstalling those packages and manually calling `mkinitcpio -P` and no dice either
appreciate the wisdom of all in this thread, Thank You
Offline
lib32-nvidia-utils …
Offline
Still happening for me.
nvidia-465.27-2
linux 5.11.16
1650 GTX Mobile
Offline
lib32-nvidia-utils …
Aha! damn I shoulda just grepped for "nvidia" in my package list.
Thanks for the answer, sorry it was a softball
Offline
I can confirm this problem as well - a blank screen on bootup with a kernel OOPS. It started with the nvidia-465.24.02 driver update and persists with the latest nvidia-465.27-2 update as well. The only way I could recover was to downgrade to the nvidia-460.67-10 package with it's corresponding linux kernel.
I'm using a Nvidia GeForce GTX 1060 graphics card.
Offline
There's a thread in nvidia's forum, Kernel panic at startup since 465 in Arch Linux, linking back here.
Last edited by keibak (2021-05-01 13:18:49)
Offline
I appear to have this issue as well. However it only happens with a specific monitor and a specific output (1080, Dell 3818DW on display port) switching to HDMI seems to have side stepped it
That's it! I just made a fresh install and tested with the latest latest Nvidia drivers 465.27 + latest Kernel and over HDMI everything worked fine but as soon as I tried to plug second monitor (Dell P2418D) via Display Port the Desktop Enviroment crashed immediately, rebooting gave the same results with the driver crashing even running the `nvidia-smi` command before the display manager even loading. I did tried all the other DP outputs and none worked
Can anyone else test this please? This would also explain why some people don't have this issue, that's because they're probably running HDMI and/or not a Dell monitor.
Offline
I can confirm: I know of 4 PC not working. All have some kind of a NVIDIA GPU (GT, Quattro, GTX, RTX) and some kind of DELL monitor (24", curved 34") all connected via display port. All get the equivalent of a "blue screen" as soon as as the nvidia driver gets loaded.
Some are backported to 460xx-dkms, some are switched to HDMI (no issue with HDMI so far).
Offline
I have this issue with an Acer RC271U connected by DisplayPort, so it is not exclusive to Dell monitors. My other Acer monitor works fine.
Offline
I have this issue with an Acer RC271U connected by DisplayPort, so it is not exclusive to Dell monitors. My other Acer monitor works fine.
I can confirm: I know of 4 PC not working. All have some kind of a NVIDIA GPU (GT, Quattro, GTX, RTX) and some kind of DELL monitor (24", curved 34") all connected via display port. All get the equivalent of a "blue screen" as soon as as the nvidia driver gets loaded.
Some are backported to 460xx-dkms, some are switched to HDMI (no issue with HDMI so far).
Thanks! So the issue is the Display Port, but not always apparently: So I tested again now using a "DP to HDMI converter" (using the DP-out from the GPU and converting it to HDMI) and again it worked fine as second and as main monitor. Unfortunately this adapter/converter I used is limited to 1080p and the monitor is 1440p but that's another +
Could it be the resolution or the signal? When I look at the nvidia-settings on the DP-X specific page it says that the signal is TMDS, is this different from normal?
Offline