Nvidia prime update gpu problems

[zimudec]

Hi

For months I have used my Nvidia gaming rig and had absolutely no problem using prime to launch games and play them for hours.

Today an update appeared (attached) where one of the programs is nvidia-prime. After I upgraded, I rebooted and a series of problems started that I had never had before. First, I realized that I couldn't start programs with nvidia-optimus-offload-glx. And indeed the gpu did not start.

I went into "Nvidia configuration" in the "Nvidia optimus" applet, and I didn't get info, temperature and GPU speed options. I tried restarting the computer and it wouldn't shut down. I was getting an error stating that it was waiting for the nvidia GPU to stop. I had to forcibly shut down the equipment.

I reinstalled the nvidia driver and everything seemed to work again. But when closing a game, I realized from Nvidia configuration, that even though there was no program using nvidia, the gpu appeared at 100%. When shutting down the computer, it got stuck again.

When starting, again in Nvidia Configuration Nvidia options disappeared.

Clearly the problem was triggered by the update. A bug?

Code: Select all

System:
  Kernel: 5.8.0-40-generic x86_64 bits: 64 compiler: N/A 
  Desktop: Xfce 4.14.2 tk: Gtk 3.24.13 info: xfce4-panel, plank wm: xfwm4 
  dm: LightDM 1.30.0 Distro: Linux Mint 20.1 Ulyssa base: Ubuntu 20.04 focal 
Machine:
  Type: Laptop System: LENOVO product: 81Y4 v: IdeaPad Gaming 3 15IMH05 
  serial: <filter> Chassis: type: 10 v: IdeaPad Gaming 3 15IMH05 
  serial: <filter> 
  Mobo: LENOVO model: LNVNB161216 v: SDK0R33126 WIN serial: <filter> 
  UEFI: LENOVO v: EGCN27WW date: 07/21/2020 
Battery:
  ID-1: BAT0 charge: 44.3 Wh condition: 44.8/45.5 Wh (98%) volts: 13.0/11.4 
  model: Sunwoda L19D3PF4 type: Li-poly serial: <filter> status: Charging 
  cycles: 52 
CPU:
  Topology: 6-Core model: Intel Core i7-10750H bits: 64 type: MT MCP 
  arch: N/A L2 cache: 12.0 MiB 
  flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx 
  bogomips: 62399 
  Speed: 800 MHz min/max: 800/5000 MHz Core speeds (MHz): 1: 800 2: 800 
  3: 800 4: 800 5: 800 6: 800 7: 801 8: 800 9: 800 10: 800 11: 800 12: 800 
Graphics:
  Device-1: Intel UHD Graphics vendor: Lenovo driver: i915 v: kernel 
  bus ID: 00:02.0 chip ID: 8086:9bc4 
  Device-2: NVIDIA driver: nvidia v: 450.102.04 bus ID: 01:00.0 
  chip ID: 10de:1f99 
  Display: x11 server: X.Org 1.20.9 driver: modesetting,nvidia 
  unloaded: fbdev,nouveau,vesa tty: N/A 
  OpenGL: renderer: Mesa Intel UHD Graphics (CML GT2) v: 4.6 Mesa 20.2.6 
  direct render: Yes 
Audio:
  Device-1: Intel Comet Lake PCH cAVS vendor: Lenovo driver: snd_hda_intel 
  v: kernel bus ID: 00:1f.3 chip ID: 8086:06c8 
  Device-2: NVIDIA driver: snd_hda_intel v: kernel bus ID: 01:00.1 
  chip ID: 10de:10fa 
  Sound Server: ALSA v: k5.8.0-40-generic 
Network:
  Device-1: Intel Wi-Fi 6 AX201 driver: iwlwifi v: kernel port: 6000 
  bus ID: 00:14.3 chip ID: 8086:06f0 
  IF: wlp0s20f3 state: down mac: <filter> 
  Device-2: Realtek RTL8111/8168/8411 PCI Express Gigabit Ethernet 
  vendor: Lenovo driver: r8169 v: kernel port: 3000 bus ID: 0c:00.0 
  chip ID: 10ec:8168 
  IF: enp12s0 state: up speed: 1000 Mbps duplex: full mac: <filter> 
Drives:
  Local Storage: total: 1.03 TiB used: 804.07 GiB (76.5%) 
  ID-1: /dev/nvme0n1 vendor: Toshiba model: KBG40ZMT128G MEMORY 
  size: 119.24 GiB speed: 15.8 Gb/s lanes: 2 serial: <filter> rev: 0105AELA 
  scheme: GPT 
  ID-2: /dev/sda vendor: Seagate model: ST1000LM035-1RK172 size: 931.51 GiB 
  speed: 6.0 Gb/s rotation: 5400 rpm serial: <filter> rev: LFM3 scheme: GPT 
Partition:
  ID-1: / size: 68.71 GiB used: 21.05 GiB (30.6%) fs: ext4 
  dev: /dev/nvme0n1p6 
  ID-2: /home size: 915.88 GiB used: 782.98 GiB (85.5%) fs: ext4 
  dev: /dev/sda2 
  ID-3: swap-1 size: 5.86 GiB used: 0 KiB (0.0%) fs: swap 
  dev: /dev/nvme0n1p5 
Sensors:
  System Temperatures: cpu: 53.0 C mobo: N/A 
  Fan Speeds (RPM): N/A 
Repos:
  No active apt repos in: /etc/apt/sources.list 
  Active apt repos in: /etc/apt/sources.list.d/additional-repositories.list 
  1: deb https://repo.vivaldi.com/archive/deb/ stable main
  Active apt repos in: /etc/apt/sources.list.d/lutris-team-lutris-focal.list 
  1: deb http://ppa.launchpad.net/lutris-team/lutris/ubuntu focal main
  Active apt repos in: /etc/apt/sources.list.d/official-package-repositories.list 
  1: deb http://packages.linuxmint.com ulyssa main upstream import backport
  2: deb https://mirror.ufro.cl/ubuntu focal main restricted universe multiverse
  3: deb https://mirror.ufro.cl/ubuntu focal-updates main restricted universe multiverse
  4: deb https://mirror.ufro.cl/ubuntu focal-backports main restricted universe multiverse
  5: deb http://security.ubuntu.com/ubuntu/ focal-security main restricted universe multiverse
  6: deb http://archive.canonical.com/ubuntu/ focal partner
Info:
  Processes: 328 Uptime: 3m Memory: 7.64 GiB used: 1.66 GiB (21.7%) 
  Init: systemd v: 245 runlevel: 5 Compilers: gcc: 9.3.0 alt: 9 Shell: bash 
  v: 5.0.17 running in: xfce4-terminal inxi: 3.0.38

------

Edit:

When gpu startup fails, dmesg shows me the following:

Code: Select all

[   40.216201] nvidia 0000:01:00.0: can't change power state from D3cold to D0 (config space inaccessible)
[   40.216285] NVRM: Xid (PCI:0000:01:00): 79, pid=1996, GPU has fallen off the bus.
[   40.219383] snd_hda_intel 0000:01:00.1: can't change power state from D3cold to D0 (config space inaccessible)
[   40.604058] snd_hda_codec_hdmi hdaudioC1D0: Unable to sync register 0x4f0800. -5

And when running an app with nvidia it shows me the following:

Code: Select all

$ nvidia-optimus-offload-glx glxgears -info
X Error of failed request:  BadValue (integer parameter out of range for operation)
  Major opcode of failed request:  152 (GLX)
  Minor opcode of failed request:  3 (X_GLXCreateContext)
  Value in failed request:  0x0
  Serial number of failed request:  36
  Current serial number in output stream:  37

[roblm]

NVRM: Xid (PCI:0000:01:00): 79, pid=1996, GPU has fallen off the bus.

That message is most often caused by an insufficient/broken power supply or overheating.

https://docs.nvidia.com/pdf/XID_Errors.pdf

What is the output of this command: lspci -nnk | grep -EA3 "3D|VGA"

[zimudec]

That message is most often caused by an insufficient/broken power supply or overheating.

Since I installed linux mint and use it to play games almost every day. I did normal OS updates every time they came out. I even went from mint 20 to 20.1, and no problem.

And just yesterday when this update comes out, I reboot and these problems start. Too much of a match to be a hardware problem.

What is the output of this command: lspci -nnk | grep -EA3 "3D|VGA"

Code: Select all

00:02.0 VGA compatible controller [0300]: Intel Corporation UHD Graphics [8086:9bc4] (rev 05)
	Subsystem: Lenovo Device [17aa:3fa2]
	Kernel driver in use: i915
	Kernel modules: i915
--
01:00.0 VGA compatible controller [0300]: NVIDIA Corporation Device [10de:1f99] (rev a1)
	Subsystem: Lenovo Device [17aa:3fc4]
	Kernel driver in use: nvidia
	Kernel modules: nvidiafb, nouveau, nvidia_drm, nvidia

When the gpu stops detecting me, when I try to turn off the computer I get the following (attached):

[roblm]

You could try downgrading the nvidia-prime package back to version 8.14. To do this, open the Synaptic Package Manager. Search for nvidia-prime and hightlight that package. At the top select Package > Force Version and select 8.14 (focal). Click the Force Version button. At the top click Apply. When you exit Synaptic, there will be a message about Quit and discard marked changes. Click Quit.

[zimudec]

You could try downgrading the nvidia-prime package back to version 8.14. To do this, open the Synaptic Package Manager. Search for nvidia-prime and hightlight that package. At the top select Package > Force Version and select 8.14 (focal). Click the Force Version button. At the top click Apply. When you exit Synaptic, there will be a message about Quit and discard marked changes. Click Quit.

I did the downgrade.

However, now my computer has frozen multiple times in 2 places:

1- When I am selecting the operating system in grub.
2- When the grub passes and arrives before the graphic environment appears.

Freezes completely. The keyboard doesn't even light up when pressing the "Caps Lock" and "Num Lock" button.}

How can I go back to exactly the same versions of the packages I had before updating? For example "grub-common 2.04-1ubuntu26.7" does not appear in the synaptic list to select.

[roblm]

Downgrading the nvidia-prime package doesn’t effect any other package. I just tested that again on my Optimus laptop. I then updated it back to version 8.15.3-0.

I hope you have a Timeshift snapshot from an earlier time. Those need to be created periodically, especially before a lot of updates.

The grub-common update affects other packages. I have Mint 20 installed, instead of 20.1. My installed version of grub-common was 2.04-1ubuntu26. I tested updating it to 2.04-1ubuntu26.8 and then was able to downgrade back to 2.04-1ubuntu26 in Synaptic, but I noticed that 7 other packages were removed in the process. I don’t know if those same packages were affected during the update. I haven’t experienced any problem yet. In Synaptic, version 2.04-1ubuntu26.2 was also listed for downgrade.

[zimudec]

When I shut down the computer at one point I managed to get the error that appears before freezing (Attached)

I have some old restore points in timeshift. I will do a test to rule out if problems keep occurring.

[zimudec]

I restored to a restore point that I had created before upgrading from version 20 to 20.1

I did several tests and apparently the problem was solved with this. Before restoring, when it failed, I notice the following behavior changes that now do not occur:

• Before when doing boot, when I correctly took the nvidia gpu, before starting the graphical environment, it was stuck for a couple of seconds in a login (automatic login by console before starting the graphical environment?)

• When starting a game (for example tomb raider on steam), I closed the game and opened the nvidia panel. In the information section it said that the "GPU utilization" was at 100%, even though the game process was completely closed (even steam).

• When starting the nvidia panel, on several occasions it took a couple of seconds to open, and in the "PowerMizer" section it appeared that it activated the gpu overclock to the maximum for a few seconds.

All of this no longer occurs after restore (I'll keep testing).

I do not know the cause of the problem. After I updated to 20.1 I had no problem until the last update mentioned in the first post.

How should I proceed now? keep version 20.0?

[roblm]

How should I proceed now? keep version 20.0?

You can upgrade to 20.1 but create a Timeshift snapshot first if you made any additional changes since using the previous snapshot, and update the packages that you think gave the problem separately, one at a time.

I personally never upgrade one version to another but do a fresh installation. However, most upgrades are successful. One major advantage of upgrading to 20.1 is that some of the bugs introduced in 20 were eliminated.

[zimudec]

Okay

I started updating the system again, keeping version 20.0.

First I updated the security patches and kernel updates. I did several tests, rebooting several times and had no problems or strange behavior change. Later I updated all the programs, except the ones I updated in the first post. No problem.

When I updated the package "ubuntu-drivers-common" and rebooted, it immediately started various behaviors that I had before when I had the problem before restoring, but the system did not crash. I did a lot of tests.

Later I installed the 2 grub packages, and the behaviors remained the same.

Finally I updated "nvidia-prime", and after rebooting a couple of times, I started to have the first problems of freezing in the boot process, before starting the graphical environment.

Apparently the problem is in "ubuntu-drivers-common" in conjunction with "nvidia-prime", and I also doubt the grub packages, because in previous post I mentioned a freeze selecting the operating system.

I'm going to do a restore again and will avoid updating those packages until there is a new update that fixes these issues.

[zimudec]

For now I have kept updating everything except the 3 that give me problems. The system has worked perfectly so far, with mint 20.1

Meanwhile, what should I do so that these problems disappear and I can update without problems the 3 that I have ignored?
How can I know if these errors have been reported and are they working on the solution?

[roblm]

Meanwhile, what should I do so that these problems disappear and I can update without problems the 3 that I have ignored?

Installing the newer versions of ubuntu-drivers-common and nvidia-prime in the Update Manager don’t correct the missing nvidia-prime applet in Intel mode. I also installed an even newer version of nvidia-prime, which is version 0.8.15.3~0.20.10.1 from here and it also didn’t help:
https://www.ubuntuupdates.org/package_m ... idia-prime

You will have to wait until the Ubuntu developers release updates that work correctly. Or you have the option of installing the newer packages and using the solution by burntex in this topic. He also submitted a bug report:
viewtopic.php?p=1964360#p1964360

However, the nvidia-prime applet is not missing in Nvidia Performance mode and On-Demand mode. I think many users don’t need to switch to Intel mode any longer if they have an Nvidia GPU with the Turing architecture or newer, because of the newer power management feature that shuts off power to the GPU when idle. So battery power usage will be the same in Intel and On-Demand modes, if the Nvidia GPU is not used. This can be verified by testing with Powertop. Once the developers figure out how to use external monitors that are connected to the Nvidia GPU in On-Demand mode, then the other two modes won’t be needed.

One question I have for you is when you open the Menu and right clicked on an app, such as VLC, does the right click menu have the option Run with Nvidia GPU. That feature was not working for me with a fresh installation of Mint 20.1. It was working in Mint 20.

[zimudec]

However, the nvidia-prime applet is not missing in Nvidia Performance mode and On-Demand mode. I think many users don’t need to switch to Intel mode any longer if they have an Nvidia GPU with the Turing architecture or newer, because of the newer power management feature that shuts off power to the GPU when idle. So battery power usage will be the same in Intel and On-Demand modes, if the Nvidia GPU is not used. This can be verified by testing with Powertop. Once the developers figure out how to use external monitors that are connected to the Nvidia GPU in On-Demand mode, then the other two modes won’t be needed.

I think there's a misunderstanding. The problems that happen to me and that I have mentioned in this thread, happen to me in On demand mode and Nvidia Performance, I do not use Intel mode. In Intel mode it does not happen, as the nvidia GPU is completely unused.

It seems to me that they are completely different problems.

One question I have for you is when you open the Menu and right clicked on an app, such as VLC, does the right click menu have the option Run with Nvidia GPU. That feature was not working for me with a fresh installation of Mint 20.1. It was working in Mint 20.

That option has never appeared to me. It seems to me that it is only available for cinnamon. I use XFCE.


With the above mentioned, then what should I do?

[roblm]

I got your problem mixed up with the one in this topic, which was also related to the nvidia-prime and ubuntu-drivers-common package updates. I forgot to re-read your first post to refresh my memory:
viewtopic.php?f=59&t=341289

Your problem may be related to what’s described here;
https://bugs.launchpad.net/ubuntu/+sour ... ug/1904583

The nvidia-prime package also has a change to fix a problem where /lib/udev/rules.d/50-pm-nvidia.rules was left behind, after a system upgrade 20.04 LTS (focal) to 20.10 (groovy), and yet prime-select had lost the ability to remove the udev rules file, leaving the end user with no means to re-enable his/her Nvidia GPU.

Does this file exist: /lib/udev/rules.d/50-pm-nvidia.rules

Did you try updating ubuntu-drivers-common to the newer version in Update Manager, which is 1:0.8.6.4~0.20.04.1. The previous version that gave you problems was 1:0.8.6.3~0.20.04.2

If that doesn’t work, then you could try the newest nvidia-prime package, version 8.15.3~0.20.10.1 from here:
https://www.ubuntuupdates.org/package_m ... idia-prime

I checked my system and I could use On-Demand mode and the nvidia-optimus-offload-glx command with any of the ubuntu-drivers-common updates. The only minor problem was the missing nvidia-prime applet in Intel mode.

[zimudec]

I got your problem mixed up with the one in this topic, which was also related to the nvidia-prime and ubuntu-drivers-common package updates. I forgot to re-read your first post to refresh my memory:
viewtopic.php?f=59&t=341289

Your problem may be related to what’s described here;
https://bugs.launchpad.net/ubuntu/+sour ... ug/1904583

Oh I understand, so they are different problems but from the same source.

Does this file exist: /lib/udev/rules.d/50-pm-nvidia.rules

Yes, I have that file on my system. Should I do something with it?

Did you try updating ubuntu-drivers-common to the newer version in Update Manager, which is 1:0.8.6.4~0.20.04.1. The previous version that gave you problems was 1:0.8.6.3~0.20.04.2

Okay, I'll test the latest version. I noticed 2 new updates came out after updating with the problem.

If that doesn’t work, then you could try the newest nvidia-prime package, version 8.15.3~0.20.10.1 from here:
https://www.ubuntuupdates.org/package_m ... idia-prime

That is an unofficial version on linux mint?

I checked my system and I could use On-Demand mode and the nvidia-optimus-offload-glx command with any of the ubuntu-drivers-common updates. The only minor problem was the missing nvidia-prime applet in Intel mode.

Does that mean they officially removed that option? or is it a mistake?

[roblm]

According to the info from the link in my last post, the file /lib/udev/rules.d/50-pm-nvidia.rules should be removed. It can prevent the Nvidia driver from loading. I don’t have it on my Optimus-based laptop. You can disable it by renaming it:

Code: Select all

sudo mv /lib/udev/rules.d/50-pm-nvidia.rules /lib/udev/rules.d/50-pm-nvidia.rules.old

The nvidia-prime package, version 8.15.3~0.20.10.1, is an official Ubuntu package for the next Ubuntu disto release, which is groovy (20.10). Mint 20.x is based on Ubuntu focal (20.04). I think that package will be the next one available for Mint, as a focal release package. Installing that package is a last resort, only if the other ones don’t work.

The On-Demand feature has not been removed.

[zimudec]

I updated the packages that were giving me problems to the latest version. Unfortunately the exact same problems mentioned from the beginning of this thread came back. I ran some programs using my nvidia gpu, and seconds later the screen went black and I got a new error (attached).

According to the info from the link in my last post, the file /lib/udev/rules.d/50-pm-nvidia.rules should be removed. It can prevent the Nvidia driver from loading. I don’t have it on my Optimus-based laptop. You can disable it by renaming it:
CODE: SELECT ALL

sudo mv /lib/udev/rules.d/50-pm-nvidia.rules /lib/udev/rules.d/50-pm-nvidia.rules.old

I confirm that once updated this file disappeared.


When the card was deactivated, when trying to execute it by terminal, it returns this:

Code: Select all

$ nvidia-optimus-offload-glx glxinfo
name of display: :0.0
X Error of failed request:  BadValue (integer parameter out of range for operation)
  Major opcode of failed request:  152 (GLX)
  Minor opcode of failed request:  24 (X_GLXCreateNewContext)
  Value in failed request:  0x0
  Serial number of failed request:  100
  Current serial number in output stream:  101

I even updated the nvidia-prime 8.15.3 ~ 0.20.10.1 package and rebooted. The same problems remain.

I go back to restore to restore point. Meanwhile I skip updating the packages that give me problems. How I should proceed?

[roblm]

I think a fresh installation of Mint 20.1 is best. If you want to try troubleshooting the problem further, then post the output of these commands:

Code: Select all

lspci -nnk | grep -EA3 "3D|VGA"   			
ls /lib/modprobe.d/		
dpkg -l | grep -i nvidia
mokutil --sb-state
ls /usr/share/X11/xorg.conf.d/
grep nvidia /lib/udev/rules.d/*
nvidia-smi

Use these commands to upload the Xorg.0.log & jouranlctl log to terminal.com pastebin:

Code: Select all

cat /var/log/Xorg.0.log | nc termbin.com 9999 
journalctl -b | nc termbin.com 9999

[zimudec]

I think a fresh installation of Mint 20.1 is best

It would be strange. All these problems happen to me in version 20.0 too, and it is a clean version.

If you want to try troubleshooting the problem further, then post the output of these commands:
CODE: SELECT ALL

lspci -nnk | grep -EA3 "3D|VGA"
ls /lib/modprobe.d/
dpkg -l | grep -i nvidia
mokutil --sb-state
ls /usr/share/X11/xorg.conf.d/
grep nvidia /lib/udev/rules.d/*
nvidia-smi

Should I trigger the errors again, and after that publish the results of these commands?

Use these commands to upload the Xorg.0.log & jouranlctl log to terminal.com pastebin:
CODE: SELECT ALL

cat /var/log/Xorg.0.log | nc termbin.com 9999
journalctl -b | nc termbin.com 9999

If I get an error that freezes my computer, I must reboot. After that, what log should I get and post here?

[roblm]

Should I trigger the errors again, and after that publish the results of these commands?

Yes.

If I get an error that freezes my computer, I must reboot. After that, what log should I get and post here?

Same logs. You didn't mention any freezing happening in your first post. If you can't get to the desktop, then press Ctrl+Alt+F1 to enter tty1 and enter the logs there.