465.24.02 page fault

I ended up going through a full rollback process with the help of this article…

After initially re-configuring my default to 5.8.0-53 I eventually binned 5.8.0-55 entirely. I’ll run the upgrade again once this is all resolved and the newer drivers are being shipped with that upgrade else I’ll be stuffed again and won’t be able to get anywhere to fix anything.

This is running with the 460.73 NVidia drivers - all is well

I’d recommend for anyone on Ubuntu 20.04 to get ahold of 5.8.0-53 if you don’t have it already and try that out - If it’s all good, bin off the newer kernel until Canonical and / or NVidia have resolved this mess (which could take a while).

Same issue here. 1080Ti. Kernel 5.11.22-2-MANJARO #1 SMP PREEMPT. Dual 4k monitors on DP.

Reverting to 460.67 drivers has resolved it for now but this is extremely frustrating. This is the second time nvidia drivers have caused my PC to be unusable. You guys really need to improve your QA processes.

2 Likes

NO. We should pay THEM to do THEIR QA testing. Check your entitlement SMH!

If you’re using 473 then you’re not having the issue in this thread.

I assume you’re replying to the comment above yours, that’s what they’re referring to, the community doesn’t have it’s own QA testing, only nvidia.

Sorry > Brain melt - 460.73 works for me - Anything above that bricks the OS (main reply updated).

Currently working via an older kernel with a hold on 460.73 to stop upgrading. For some reason 473 keeps ringing around my head. Been staring at version numbers far too long…

I was being sarcastic about having to do QA for Nvidia while paying them.

1 Like

Same issue here. When comparing two different Xorg log files (one from a successful boot with driver 390, another with a faulty boot with driver 460.80), I noticed the latter one stops just right before it starts listing monitors, which makes sense since I have a monitor plugged in via DisplayPort.

My solution, as others have pointed, was to use the nvidia-driver-460-server offered by Ubuntu, which is version 460.73.

I hope they fix this soon. Any fresh install of Ubuntu’s 20.04 to 21.04 are guaranteed to black screen on the first boot for DP users. I know because I did two myself.

Setup:
OS: Ubuntu 20.04.02
Kernel: 5.8.0-55-generic
GPU: GTX 1050Ti
Monitor 1: AW2518HF, 240Hz via DisplayPort, 1920x1080
Monitor 2: T24B350, 60Hz via HDMI, 1920x1080

2 Likes

Same issue on 460.80 and 460.84
OS: openSUSE Leap 15.3
Kernel: 5.3.18-57-default
GPU: GTX 1080
Monitor: Dell U3818DW, DP 3840x1600 24bit

Two months for such serious problem. It’s a shame… Just imagine this problem was for Windows - they would fix it in 24 hours…

2 Likes

I agree : nvidia driver is currently broken for most owners using DP connection and using a rolling distro (which are increasingly popular). Considering it’s a recent change, I am very surprised that it takes 2+ months to rollback the culprit change.
My 1080ti is long overdue for a change, but I am seriously considering switching to AMD : last gen is decent, and their linux support is much better.

3 Likes

Not just rolling distros but LTS distros like Ubuntu 20.04.

1 Like

@aplattner you mentioned that this is tracked internally as bug no. 3302807. Is there any progress that you can share?

1 Like

Same problem here, using Manjaro kernel 5.10.42-1-MANJARO and Drivers above 460.73
Also running 3 monitors on DP : 2 HP 27xq 2560x1440 and a Dell U2414H à 1920x1080
Only the Dell is working “alone” on DP, plug one or two HP monitors (with or without the DELL) crash the system
We all are loosing time to figure out the problem !
The lastest drivers working is 460.73

1 Like

I ended up testing my display with ubuntu’s nvidia-driver-460-server and my second monitor remained disabled whereas it is enabled when I use NVIDIA proprietary driver v460.73 (installed manually) - same xorg.conf each time. So no DP problem here as my main display was enabled and connected to DP, but at least not the same behaviour as the non-server driver

I checked in and it looks like they managed to find one of the affected monitor models and there’s a fix in testing for a future release.

I realize it’s taking a while and I appreciate your patience. COVID lockdowns are still affecting our ability to quickly find affected hardware for issues like this.

3 Likes

Though I really appreciate any update on that matter, I as a loyal customer feel pretty confused right now.

This issue was reported April 20th so… almost 2 months ago.

And they managed to find/get an affected monitor about 1 1/2 months later? (corrected date - thanks @danielsuarezcolon )
If it were a small business with a handful of developers, I might be more understanding but we’re dealing with a multi billion company, right? I’m sorry but this sounds to me like linux-driver debugging/development is veeery far down the budget ladder which leaves me a bit uneasy when planning my next GPU upgrade.

Please don’t blame COVID. We’re all working from home office in lockdowns since over a year.

Yes, the first couple of months, customers might be a bit more forgiving given the circumstances - but we all had enough time to change our workflow and logistics so we can continue working almost 100% capacity. So pleeease don’t use the big and ugly covid monster as an excuse anymore.

Please don’t get this the wrong way: I really appreciate an update and you’re most likely not to blame - but I expect more from a giant corporation like fricking NVIDIA!

2 Likes

This is not accurate, they were only ever able to confirm the issue and reproduce it two weeks ago on June 2nd.

1 Like

And people started to include monitor models connected by DP from early may.
I’d guess it is very hard for a big company to get what they need…

1 Like

He said reported.