Kubuntu 18.04 LTS or 19.04 (issue occurs on both versions)
Action:
Boot laptop normally
Wait a few minutes
Observed:
Clicking any mouse button does not have any effect anymore
A few minutes later the curser cannot be moved anymore; screen is totally frozen
Laptop fan increases speed
Expected:
Freezing should not occur
Comment:
Kubuntu 18.04 LTS was installed since six months without any issue; issue was first seen on 2019- 08-06
No updates/configuration changes were made before this issue was seen except for disabling Flash plugin in Chromium browser which should have no effect (hopefully)
Issue is 100% reproducible since then
Issue was both seen on driver version "nvidia-driver-390" and "nvidia-driver-418"
Issue is NOT seen when setting PowerMizer to "Prefer Maximum Performance"
However, this setting has to be made on every reboot and the laptop fan speed is high (so this is not an acceptable solution)
Find attached log files from "sudo nvidia-bug-report.sh"
XID 79/fallen off the bus is mostly caused by overheating or insufficient power supply on desktop systems. On notebooks, this rather points to defective hardware. Check if you can replicate the issue using Windows, then RMA (if still under warranty).
Thank you for your quick answer! I am currently running a dualboot system – so this was easy to test for me.
Version: Windows 10 Home 10.0.18362 Build 18362
Nvidia driver version: 431.60
Observation: NO problems at all; NO freezing even when the laptop (DELL Inspiron 7580) is idling.
So, a hardware issue seems unlikely to me.
The issue also occurs in Kubuntu with power supply plugged in.
By opening the NVIDIA X Server Settings application in Kubuntu, it can be observed that the laptop always starts to freeze as soon as the PowerMizer Performance Level reaches 0 for more than ca. 3 seconds.
Any idea or any hint from the log files why this happens?
Unfortunately, XID 79 is just the note that the gpu shut down, no way to debug why.
Since this is an Optimus notebook, in Windows the desktop is driven by the integrated intel gpu only so it is not easy to check if the nvidia gpu is failing since no freezes occur. You will have to explicitely run something on the nvidia gpu, e.g. some unigine demo and some time after exiting it check the system event log. look for messages with yellow exclamation marks telling “the graphics driver failed to respond and had to be restarted” or the like.
I did as you said and downloaded and ran the Unigine Heaven Benchmark Demo from the Nvidia website.
Results:
Overall the demo and graphics are looking good
The system event log did not show any problems at all with the GPU
Thoughts:
When running Kubuntu Linux, the freezing only occured at performance level 0
So maybe the problem was not seen because of the heavy usage of the Nvidia GPU?
I tried setting the Nvidia GPU as global graphics card in the Nvidia control panel and afterwards did nothing so that the laptop could idle (the Nvidia GPU clock even fell down to 139 MHz or something like that according to TechPowerUp GPU-Z)
Again nothing happened; NO freezing at all in Windows 10
So couldn’t it be an Nvidia Linux driver issue? Are the log files really not giving any details?
A driver bug is possible but very unlikely in that case. More likely would be a bios bug, the bios seems to have been updated shortly, is a downgrade to the previous version possible?
You’re absolutely right! A few weeks ago, I triggered a major Windows update and in parallel a BIOS update was installed against my wish. Then I was on vacation and afterwards the first freeze under Linux occured. At that time I did not see any connection between these two events because I totally forgot about the BIOS update.
So I will perform a BIOS downgrade as soon as I can. I will inform you about the outcome.
The BIOS downgrade did not solve the problem (I tested Version 1.7.0 and 1.6.0).
So I did a clean reinstalling of both operating systems (Windows 10 Home and Kubuntu 18.04). The same problem occured again.
But I was able to reproduce the freezing with the nouveau driver (the open source alternative to the Nvidia driver). By default the nouveau driver is using the Intel GPU but I managed it to set the Nvidia GPU as default for a computer game. When I started this game the screen froze immediately.
So the problem analysis goes on but I am now sure that the problem is no longer related to the Nvidia driver.
Update:
The problem is now fixed and the freezing no longer occurs.
I sent the laptop in for repair and they exchanged the mainboard and the SSD (because their tests showed that there were defect sectors – although the hardware test I did at home did not show anything like that).
So in conclusion the freezing seemed to be hardware related.