Page MenuHomeSolus

Solus critically froze, damaged moniter firmware
Closed, ResolvedPublic

Description

I have no clue what happened here, but it had me pretty worried that I'd have to reinstall solus and that my display was a write off, so I'll just step you through what happened.

Prelude to Issue

  • installed solus
  • ran eopkg up
  • installed usual packages (eopkg log here)
  • installed nvidia driver (v 415.27) from Hardware Drivers
  • restarted computer
  • opened up lollypop and started listening to music
  • opened up firefox and vscode
  • started installing vscode extensions

Issue

  • A few hours after clean installing solus
  • Simultaneously, display froze and music playing became stuck in a 1-2 sec loop
  • Tried pressing ctrl+alt+F1 nothing happened
  • waited 30s
  • force shutdown computer
  • rebooted
  • moniter looked similar to this image from the web (below)
  • unplugging display cable had no effect, could no access monitor on screen settings


(more specifically, what I had wasn't so much the image above but scrolling lines from the display output)

Upon restarting the monitor looked almost exactly like this (except different looking lines + view sonic)

Nothing like this has every happened with (a) this computer, or (b) this monitor before.

Notes

  1. I checked the integrity of both the iso and the live usb used to install before doing so.
  2. All I have done so far is literally install software from eopkg, one snap (bitwarden) and then run said software.
  3. After this happened I got a second monitor out, booted from live usb, and ran fsck, then booted into (non-usb) solus and ran eopkg check (all OK apparently)
  4. Music was being played from a folder in the windows install also on the disk (C:\Users\<me>\Music)
  5. I imagine you want logfiles, let me know which ones you want.

The Big Concern

This time, after seemingly being unable to perform a factory reset with the monitor, then leaving it unplugged for a few minutes it seems to work again. However I am genuinely afraid that this may happen again, and next time I may not be so lucky.

tecosaur created this task.Feb 9 2019, 12:18 AM
tecosaur updated the task description. (Show Details)Feb 9 2019, 12:21 AM

What GPU and do you have the possibilit to test with another GPU?

Due the fact your monitor is working again I guess your gpu is broken and I never heard of a PC crash damaging monitor firmware

I'm using an MSI 980ti, and I am not able to test another GPU (ryzen CPU, no integrated graphics).

I got it new early last year, and up till now I haven't had any display issues (unless you count Windows becoming corrupt and dying on me).

What monitor is this?

It's a Viewsonic VA2719-smhd (27" 1440p IPS).

In use since early last year, and has behaved well up until this point.

Alright, so no freesync or anything like it. Best guess is a bug in either the nvidia driver, mesalib, or the kernel. This might be a little scary, but I sincerely doubt your monitor is being damaged. There's probably just something goofy with the signaling (HDMI/Displayport).

There's a couple of things you could try to work around this:

  • Hope it doesn't happen again and wait for all the updates next week
  • Uninstall the nvidia-glx-driver and try the older nvidia-390-glx-driver or the newer nvidia-beta-driver
  • Try swapping HDMI for Displayport or vice-versa
tecosaur added a comment.EditedFeb 9 2019, 2:05 AM

So I'm guessing my best bet currently is:

  • Run sudo eopkg remove nvidia-glx-driver && sudo eopkg install nvidia-390-glx-driver
  • Hope it doesn't happen again

A few other points:

  1. What made me particularly concerned wasn't so much what happened before I turned off the monitor, but that after (a) unplugging the monitor HDMI cable, (b) restarting the monitor (c) turning monitor mains power off and on again I still had issues
  2. Is it worth downgrading the kernel too?
  3. What's happening with the updates next week that would be likely to resolve this issue?
  1. Most monitors nowadays have some non-volatile memory to store settings. You'd be surprised how much survives an over night power outtage
  1. I think you'll be better off upgrading next Friday, tbh.
  1. I'll be updating the kernel, mesalib, and some xorg stuff tomorrow. You'd be surprised how much gets fixed with just updating those.

What I did is go from the nvidia driver to nouveau. Has been fine up till now, and then in a minor glitch, I got some patchy squares to appear around my screen (they went away after I clicked a window).

Does this indicate anything about the original problem, or would this be a separate one?

mati865 added a subscriber: mati865.Feb 9 2019, 1:38 PM

Has been fine up till now, and then in a minor glitch, I got some patchy squares to appear around my screen (they went away after I clicked a window).

I'd say it's nouveau bug.
On my Nvidia 1080 there are so many issues with nouveau it's unsuitable even for web browsing, official drivers are the only way to go.

@mati865 the original, and imo more serious issue was with the official drivers.

As an owner of a 980ti I can tell you there isn't any issue with the drivers, for me personally it looks more like a hardware issue so your GPU is dying in short, because the first picture remembers me of my dying 960 I had.

Oh dear. That does not sound good. I've been using this card for around one year (acquired unused) and never had any issues with it before. That would be rather annoying.

Would doing anything like putting it through the Unigin Heaven benchmark in Windows help confirm whether it is/isn't a GPU issue?

Update

On Windows 10 I ran both MSI Kombuster for 15 mins, and Unigin Heaven for 30 mins; these are the results I got:

MSI Kombuster (15 min)

FramerateGPU LoadGPU PowerResolutionArtifacts
170 FPS96%97% TDP1080p0

Unigin Heaven (30 min)

Benchmark FPSMin FPSMax FPSResolutionQualityGPU Freq.Mem. Freq
11092351440pHigh~1550 MHz~3500 MHz

Overall

Throughout both benchmarks, the GPU reached a max temp of ~70°C (I have aftermarket cooling), and displayed no issues.

Conclusion

While this is by no means conclusive, this does not suggest any hardware issue with my graphics card.

tecosaur added a comment.EditedFeb 17 2019, 3:29 AM

Another Update

I applied the Friday updates and installed the Nvidia driver. Within two hours I had two issues. I've switched back to the noeveau driver.

This is the second graphical issue.

I have similar issues when I wake up from standby mode with lattedock icons, till I hover over it.

Yea that issue is also fairly common for Electron / Chromium based applications, it is not indicative of anything nor is it tied to any one specific update. Been having it with Atom, VS Code, etc. since the dawn of time.

Thanks Girtablulu and Josh. That's reassuring to hear. Interesting that I haven't had that happen with neoveau though.

Update

I've had a similar issue to that I had initially, again.

Symptoms

  1. Desktop screen completely froze
  2. Ctrl+Alt+F1 did nothing
  3. I had music in the background, the last second or so started looping

At the time I had the following applications open:

  • Firefox
  • Bookworm
  • Lollypop
  • VS Code

And was simply working through a book in bookworm.

DataDrake edited projects, added Hardware; removed Lacks Project.Mar 2 2019, 3:35 AM

I still have yet to see any indication that this is actually a Solus-specific issue. It sounds more like your GPU and Monitor don't play nice with the proprietary Nvidia driver which is way more common than you think.

From what I've heard in this thread, that seems like the most likely issue. Oh well. Feel free to close this for now and I'll comment again if anything happens differently to that I have already spoken about.

JoshStrobl closed this task as Resolved.Mar 31 2019, 4:51 PM
JoshStrobl claimed this task.