Where should I get support for kernel bugs?

happysmash27 · Apprentice Joined: 28 Mar 2016 Posts: 220

I have an extremely harmful, persistent issue where my RX 480 GPU will crash and fail to reset. This happens consistently when launching SteamVR after X11 has been on for a while, and has happened a couple times when running acominer: https://github.com/ishitatsuyuki/acominer/issues/9.

I am sick and tired of these reset issues. I can't use virtual TTYs, shut down, or even use the reset button; all fail. I have to close all applications, then hold the power button manually in order to shut down, so I can reboot and recover.

I believe this is a kernel bug.

The problem: I am using the gentoo-sources. Furthermore, the problems are also related to software such as SteamVR, and Mesa.

Should I file this as a Gentoo kernel bug?

Rebuild with vanilla kernel and try to purposefully reproduce this family of issues? I'm sure trying to run SteamVR and Acominer at once could reproduce it pretty consistently, since both seem to have problems when anything else is using the GPU.

Who should I file this bug too? It is extremely severe and is causing me many problems and I can't find anything else about it online.

eccerr0r · Posted: Sat Jan 22, 2022 4:34 pm Post subject:

Yes see if a vanilla kernel works or not to narrow it down whether the Gentoo patches caused the issue or not. If it also fails with a vanilla kernel, you should file a bug upstream. Submitting to the mesa developers might be a good starting point since they have to deal with the OpenCL distribution.

You may have to also consider your hardware is failing or is broken which could explain why you're the only one seeing the problem. Sorry I can't afford a RX480 so I can't test/debug this issue.
_________________
Intel Core i7 2700K/Radeon R7 250/24GB DDR3/256GB SSD
What am I supposed watching?

Goverp · Advocate Joined: 07 Mar 2007 Posts: 2217

RX480 means it might be a bug in amdgpu, (or Steam ...).
Before doing anything else - has it ever worked OK? Was there a kernel after which the problem occurs? If so, you probably need to do a "git bisect" to locate the change that broke it.
Have you tried playing with amdgpu.gpu_recovery=1 (boot/module parameter)? I'm not sure if it's useful in these circumstances or not.

The kernel MAINTAINERS file says to file bugs in bugs.freedesktop.org. That said, I believe the developers don't look there unless someone gets to them via the mailing list amd-gfx@lists.freedesktop.org, I guess you might have to help by collecting diagnostic info.
_________________
Greybeard

figueroa · Posted: Sun Jan 23, 2022 4:58 am Post subject:

Unless you can provide meaningful logs and repeatable behaviors specifically linked to a kernel issue, filing a bug is flailing against the wind. You need to troubleshoot which means eliminate variables, test alternatives, swapping out different software and hardware components including versions of kernel sources, binary kernels, and maybe even other distributions, in order to ratchet in on the source of the problem.

I feel your frustration, but as outlined the problem is as likely to be buggy software, failing hardware or mis-configuration; maybe even heat.
_________________
Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi