[SOLVED]Plasma crashes on loop until computer has warmed up

Redbyte · 2022-12-08 03:13:32

Hello, I've recently noticed an issue where upon boot plasma keeps saying desktop effects have been reset. It continues for a few minutes until the system seems to have warmed up. I haven't yet tried other DEs but I can confirm that a completely clean install of Arch with just the Nvidia drivers and KDE installed still has this issue. I have tried both linux-lts and nvidia-lts of which neither seems to make any difference. It's difficult for me to troubleshoot considering that I have to wait for my computer to come down to room temperature to try a different idea.

I can also say that I've tested with both my daily rtx 3080ti and my older gtx 1080 that I keep around just in case. I have also tested a different cpu and motherboard (identical) but that hasn't made any difference either. The issue is 'NVRM Xid 32' and occasionally 'NVRM Xid 69'. I'm not really sure what I can do to try and fix this.

I've also found that before the error starts to go away if I open a game the rendering is a total mess until the system stabilizes. Originally I thought I had a coredump I could try to debug but I believe it was just the WallpaperEngine plugin for KDE crashing each time this error occured, it stopped being generated once I switched backgrounds away from the plugin. I will include a journalctl log. I'm not sure what other log files I should examine or share to help with this process. Frankly I'm at a loss of what I can do to fix this issue as it doesn't seem to be hardware related since everything works fine in isolation.

I would appreciate any help or suggestions. I'm still fairly new to Linux and Arch but I've been trying to daily drive for the last while to help myself learn more. I've been on and off for a couple years so my knowledge isn't very deep.

https://pastebin.com/cL5Q1GD1 << Shortened log

Last edited by Redbyte (2022-12-09 16:43:52)

seth · 2022-12-08 07:56:11

https://docs.nvidia.com/deploy/xid-errors/index.html
XID 32: Invalid or corrupted push buffer stream
XID 69: Graphics Engine class error

I have to wait for my computer to come down to room temperature to try a different idea

That's for sure? A warm (even a cold reboot into a warm system) reboot does not cause this?
Mechanical tension in either the MoBo or the GPU?

Redbyte · 2022-12-08 15:49:07

seth wrote:

https://docs.nvidia.com/deploy/xid-errors/index.html
XID 32: Invalid or corrupted push buffer stream
XID 69: Graphics Engine class error
I have to wait for my computer to come down to room temperature to try a different idea
That's for sure? A warm (even a cold reboot into a warm system) reboot does not cause this?
Mechanical tension in either the MoBo or the GPU?

I've reseated everything, even the cpu and reapplied thermal paste.

Yeah, rebooting doesn't cause the issue to crop up again once it's gone. If I reboot immediately it will persist until the system has warmed up.

ewaller · 2022-12-08 15:54:12

I would not rule out a massive memory leak either. Before the system gets 'too hot', try putting it to sleep, let it cool, and then wake it up. If it fails quickly at that point look closer at memory or other state issues. If it works until it comes back up to temperature, then you have a thermal issue.

seth · 2022-12-08 16:23:30

@ewaller, afaiu the OP the system fails *before* it's warmed up
@Redbyte, including the motherboard?
Does it help if you pre-warm the GPU in isolation (eg. w/ a hair dryer - be super careful w/ heat guns!)

Redbyte · 2022-12-08 16:53:49

seth wrote:

@ewaller, afaiu the OP the system fails *before* it's warmed up

Yeah that's correct. Once it's been up for a few minutes plasma suddenly stops crashing and rendering returns to normal.
Additionally, I note that if I boot into windows (backup for a few things I haven't switched over from, seperate drive though, not a partition) there's no issue whatsoever.

seth wrote:

@Redbyte, including the motherboard?
Does it help if you pre-warm the GPU in isolation (eg. w/ a hair dryer - be super careful w/ heat guns!)

I've actually replaced both the CPU and motherboard in their entirety. Just RMA'd them. This was the behavour I had before I replaced them though.
I haven't tried a hairdryer yet, that's a good idea. I will attempt to try that.

Redbyte · 2022-12-09 16:39:28

So I believe I've identified the problem. Despite them all passing memtest86 it seems to be caused by one of my memory modules. Regardless of the configuration as long as I remove that specific module the issue goes away completely.

seth · 2022-12-09 16:45:12

The test will probably heat the system up enough before it addresses that bank?
Is it the specific module or the specific slot?

Redbyte · 2022-12-09 16:57:01

It's not the slot, as soon as I introduce the module into the system the issues return independent of which slot I use.
So yeah, by the time memtest starts up it's already warmed up

Last edited by Redbyte (2022-12-09 16:57:34)

Arch Linux

#1 2022-12-08 03:13:32

[SOLVED]Plasma crashes on loop until computer has warmed up

#2 2022-12-08 07:56:11

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#3 2022-12-08 15:49:07

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#4 2022-12-08 15:54:12

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#5 2022-12-08 16:23:30

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#6 2022-12-08 16:53:49

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#7 2022-12-09 16:39:28

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#8 2022-12-09 16:45:12

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

#9 2022-12-09 16:57:01

Re: [SOLVED]Plasma crashes on loop until computer has warmed up

Board footer