Booting Ubuntu with Nvidia-340 gets stuck on a plain purple screen

[Mirko]

When I install the proprietary NVidia drivers, even if I set:
- nosplash noveau.modeset=0 in grub
- WaylandEnable=false in /etc/gdm3/custom.conf

My system keeps staying on the purple screen before asking for login.
The System is accessible and working via SSH.

After a thread in bug 1889377, @vanvugt suggested to me the following procedure:
[QUOTE]
We would need to see the system log from one of those failed boots. To do that please:

1. Reproduce a failed boot with nvidia-340.

2. Uninstall nvidia-340 and after rebooting collect the log from the previous failed boot. For example, to collect the last three boots run:

   journalctl -b-1 > prevboot-1.txt
   journalctl -b-2 > prevboot-2.txt
   journalctl -b-3 > prevboot-3.txt

and then please open a new bug with those logs attached.
[/QUOTE]

Attached the three files.

Aditional info:
1) Ubuntu 20.04.1 LTS
2) nvidia-340
3) system boots
4) system is not booting

Thanks for helping.
Mirko

[Ubuntu Foundations Team Bug Bot (crichton)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. It seems that your bug report is not filed about a specific source package though, rather it is just filed against Ubuntu in general. It is important that bug reports be filed about source packages so that people interested in the package can find the bugs about it. You can find some hints about determining what package your bug might be about at https://wiki.ubuntu.com/Bugs/FindRightPackage. You might also ask for help in the #ubuntu-bugs irc channel on Freenode.

To change the source package that this bug is filed about visit https://bugs.launchpad.net/ubuntu/+bug/1889541/+editstatus and add the package name in the text box next to the word Package.

[This is an automated message. I apologize if it reached you inappropriately; please just reply to this message indicating so.]

[Mirko (mirkos1973)]

I'm not sure this is the proper package.
I installe nvidia-340 which I cannot select

[Daniel van Vugt (vanvugt)]

Please attach a photo (or video) of the problem, because I am not sure what the "purple screen" is that you refer to.

Please also run this command to collect more system information

  apport-collect 1889541

[Daniel van Vugt (vanvugt)]

Here is the log relevant to this bug.

[Mirko (mirkos1973)]

Hi Daniel,
I run apport-collect 1889541 but I have removed nvidia drivers otherwise I cannot login anymore (as we have discussed together in bug 1889377).

For what concerns the purple screen, please find attached a picture I've taken from the web with the exact problem I'm getting.
My login process (usually lasting no more than 30 seconds) gets stuck on this page forever (at least much longer than I'm keen to wait for - I left it in this state also for one hour before restarting).

Thanks for your help

[Daniel van Vugt (vanvugt)]

Thanks. That's very interesting. I've only ever seen Ubuntu get stuck on that purple screen when there was a graphics driver bug (it was the VMware driver last time - bug 1832138). So that makes me think there's a problem with the nvidia-340 driver, OR it could mean some fallback codepath in mutter to support old drivers is buggy.

The bad news is I don't think I can test the nvidia-340 driver anymore because the only Nvidia cards new enough to support my monitors are too new for that driver :S

[Mirko (mirkos1973)]

I actually think it must be somehow related to my configuration, because in my previous installation in the same machine (I had to install a fresh new Ubuntu around 2 weeks ago after an HDD major crash), I was able to install nvidia-340 drivers.

The only difference between the old and the new machine configurations are the storage:
>>> PAST: I had a Seagate Barracuda 1TB HDD

>>> NEW: I have a Crucial 500 GB SDD (where OS is installed) and a Toshiba 3 TB HDD

But I don't think this could make any difference on nvidia-340 behaviour

[Daniel van Vugt (vanvugt)]

I don't think your choice of hard drive is causing this problem. And your kernel log reports no errors with the Crucial SSD...

When it was working 2 weeks ago, was that Ubuntu 20.04 or an older version?

[Mirko (mirkos1973)]

It was Ubuntu 20.04

[Daniel van Vugt (vanvugt)]

That's actually good news then. If you can safely wipe the machine and reinstall then it sounds like a fresh install might eliminate whatever the cause of the problem is.

[Mirko (mirkos1973)]

I installed the OS 3 times in the last two weeks to get rid of the nvidia problem without any success :-(

I was hoping to find a way to identify in the logs what's going wrong because a fresh install won't solve the problem.

[Daniel van Vugt (vanvugt)]

I would be curious to see what a fresh install looks like. Because when you first started reporting these bugs you had disabled the open source driver in the kernel, didn't have the proprietary driver installed, and had installed a custom unsupported kernel. So I am still skeptical that some of those changes, or other changes, may have broken the system.

Sorry it's so complicated... Nvidia retires their "old" hardware very quickly. Even when many of us would not consider it "old". Yours is in the process of being retired as it's only "supported" via an old driver. And that old driver is too old for newer kernels. It would be easy to blame Nvidia but you have to remember that these problems don't happen on Windows. So I would also blame the design of the Linux kernel for not providing a stable ABI (one that doesn't require rebuilding drivers).

[Mirko (mirkos1973)]

Im considering, as your suggestion in the other post, to buy a new GPU (GeForce GT 1030) and for example I've seen the following on Amazon, with also Prime so I would get it tomorrow:
- Gigabyte GT 1030 Low Profile 2G - graphics cards (NVIDIA, GeForce GT 1030, 4096 x 2160 pixels, 1257 MHz, 1506 MHz, 2 GB), GV-N1030D5-2GL

I've checked it should be compatible with my old P5QL PRO mother board, therefore I assume it should address the problem

Do you think a new GPU with the most recent drivers would solve my issue?

Thanks.
Regards,

[Daniel van Vugt (vanvugt)]

I can't be sure that buying a new GPU will solve the problem so that is at your own risk and expense... But such a new(er) GPU does support Nvidia's latest drivers. In fact I've been using a GT 1030 in recent months for my own Nvidia-related work in Ubuntu 20.10. And that card is roughly 7x more powerful than your current one. But I can't predict the future and can't guarantee you won't face new problems or the same problems.

I've just logged on again (it's Friday night here) to see what recent Ubuntu updates might have caused this issue... It can't be nvidia-graphics-drivers-340 because that's never changed. So perhaps it was a kernel update or xorg-server update. To test this properly, without buying anything, I suggest reinstalling Ubuntu with networking *disabled*, but the 3rd part components option *enabled*. That will give you a fresh Ubuntu system without any updates installed, and hopefully with nvidia-340 installed right from the beginning (it's included in the Ubuntu installer). If that works then we can figure out more easily what software updates (if any) have caused the problem.

[Mirko (mirkos1973)]

OK, I didn't mean to have a guarantee but just a flavour.
I bougth the new GPU, also because in the process of renewing a bit my old machine, replacing the GPU was quite needed anyhow.

In case it won't work properly I'll give it back, but I'm confident it'll work.

Thanks for your support.

Regards