Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!
Bug 938952 - sys-kernel/gentoo-kernel-bin-6.10 series amdgpu igpu crashes
Summary: sys-kernel/gentoo-kernel-bin-6.10 series amdgpu igpu crashes
Status: UNCONFIRMED
Alias: None
Product: Gentoo Linux
Classification: Unclassified
Component: Current packages (show other bugs)
Hardware: All Linux
: Normal normal
Assignee: Distribution Kernel Project
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2024-09-03 00:17 UTC by thezombiehunter
Modified: 2024-10-07 03:06 UTC (History)
3 users (show)

See Also:
Package list:
Runtime testing required: ---


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description thezombiehunter 2024-09-03 00:17:54 UTC
When using the 7950x igpu with the latest 6.10.x kernel I experience random GUI crashes (not kernel crashes, the GUI just needs to reload). This can happen in five minutes to five hours.

I tried everything I can think of short of RMA'ing the CPU. I changed a ton of bios settings, etc.

Then I read that some users were experiencing crashes with the amdgpu driver on the 6.10.x kernel and the latest linux-firmware. I downgraded to 6.6.48 and so far everything seems smooth...

Unfortunately gnome is also not using wayland on 6.6.48 (it did on 6.10.x) so thats driving me a bit crazy.
Comment 1 Ionen Wolkens gentoo-dev 2024-09-03 03:47:34 UTC
(In reply to thezombiehunter from comment #0)
> Unfortunately gnome is also not using wayland on 6.6.48 (it did on 6.10.x)
> so thats driving me a bit crazy.
What's to say that no longer using wayland is what solved your crashes rather than the kernel version? Had you tried X11 with 6.10.x too?
Comment 2 Ionen Wolkens gentoo-dev 2024-09-03 04:12:09 UTC
May want to try different mesa versions too, regressions in mesa affecting only some specific hardware aren't all that rare.
Comment 3 thezombiehunter 2024-09-03 10:07:02 UTC
Yes, I've tried X11 on 6.10.x.
Comment 4 thezombiehunter 2024-09-03 10:33:07 UTC
The GPU driver crashing was also happening on both gnome and KDE.
Comment 5 Nowa Ammerlaan gentoo-dev 2024-09-03 19:38:22 UTC
We will need at least a dmesg log from when there is a crash to figure out in which direction to look.
Comment 6 thezombiehunter 2024-09-05 17:01:17 UTC
Hi,

Wanted to give you a quick update. I’m nearing the end of the return period for this cpu so I installed windows 11 to make sure it’s not a hardware issue. No crashing so far. I’m going to revert some bios changes to make sure it’s still stable.

If I can’t replicate the crashing on windows 11 illl know it’s a Linux issue.

Will reinstall gentoo once my testing is done and get back to you with the error logs.
Comment 7 thezombiehunter 2024-09-07 00:56:40 UTC
after extensive testing it is indeed a linux issue.

I will be reinstalling gentoo shortly and will hopefully have a bug for you by the end of next week.
Comment 8 thezombiehunter 2024-09-07 17:34:58 UTC
Mmm. Either windows 11 updated hardware firmware or there was something fubar with my old install.

New install (so far) doesnt have this issue.

I will close this ticket and re-open if the issue returns.
Comment 9 thezombiehunter 2024-09-10 01:34:50 UTC
I'm reopening this, because somehow installing windows 11 and updating drivers and then reinstalling gentoo fixed the problem.

I guess its possible the issue was the gentoo install, but I think its much more likely there was a hardware/firmware/driver update on windows 11 that fixed it.

Do those driver updates carry over to other operating systems?
Comment 10 Nowa Ammerlaan gentoo-dev 2024-09-10 07:54:10 UTC
> Do those driver updates carry over to other operating systems?

Driver updates no, but updating the windows driver does sometimes also update some firmware and this may or may not affect other operating systems. Linux has it's own firmware loading mechanism, but the way this works is different from Windows (don't ask me for the details because this is not my area of expertise, I have no clue). As I understand it, the Windows firmware updates are actually written to the GPU, whereas in Linux firmware is loaded when the kernel loads the driver. This would mean that when the system is booting into Linux, the actual initialization of the GPU (before the OS loads) may be performed by older firmware that you may only be able to update via Windows. This may explain why your issue is resolved after updating the Windows driver.

This is actually one of the reasons I still have Windows installed on a secondary disk, even though I never really use it, it allows me to update firmware that I cannot update via Linux.
Comment 11 thezombiehunter 2024-09-11 00:21:55 UTC
Well, it's been rock solid on the new install.

If installing the firmware/drivers through windows did fix the issue, how would one update those through linux?
Comment 12 Nowa Ammerlaan gentoo-dev 2024-09-11 05:45:06 UTC
(In reply to thezombiehunter from comment #11) 
> If installing the firmware/drivers through windows did fix the issue, how
> would one update those through linux?

You often cannot, firmmware updates are not always supported through Linux.
Comment 13 thezombiehunter 2024-10-07 03:06:34 UTC
Weird, well, its been rock solid since the windows updates.

I'm not sure its hitting 60FPS refresh rate though on gnome with a 5k monitor, seems a little laggy?

I know on windows moving around windows on the 5k monitor causes 90%-100% GPU load, so it probably just cant handle it.

Otherwise, its been great.

I just would prefer to not keep windows installed for future firmware updates.