ppanchad-amd
commented
2 months ago @Mr-Andersen Internal ticket has been created to investigate this issue. Thanks!
Closed Mr-Andersen closed 1 month ago
ppanchad-amd
commented
2 months ago @Mr-Andersen Internal ticket has been created to investigate this issue. Thanks!
schung-amd
commented
2 months ago Hi @Mr-Andersen, are your kernel and mesa libs up to date? It seems like there are some similar issues caused by out-of-date kernel and mesa versions.
Mr-Andersen
commented
2 months ago @schung-amd I've upgraded to Linux 6.6.47 and Mesa 24.2.0, still having the issue. Can't bump Linux further yet, since I need ZFS. In which versions were those similar issues resolved?
schung-amd
commented
2 months ago Thanks for checking! It's unclear what versions help; one user's issues were fixed by a kernel update in 6.5.x with mesa 23.1.7 (https://www.reddit.com/r/linux_gaming/comments/16jhxnz/starfield_crashes_amd_radeon_rx_6600/), while others are having issues with more recent versions. Some users are reporting RAM problems being related (https://gitlab.freedesktop.org/drm/amd/-/issues/2943). These issues are for different workloads on various hardware, so your underlying issue may be different, but might provide a clue. I'll try to reproduce your issue with wezterm specifically on similar hardware and get back to you.
schung-amd
commented
1 month ago Hi @Mr-Andersen, I was unable to reproduce your issue, but I may be missing something in the NixOS configuration. On a fresh install of NixOS 24.05 on an RX 6400, I installed ROCm and wezterm through the config file by
environment.systemPackages = with pkgs; [
pkgs.rocmPackages.rpp
pkgs.wezterm
];in /etc/nixos/configuration.nix followed by a sudo nixos-rebuild switch, and I can use wezterm without any obvious issue. Is there a crash or hang occurring when you encounter this issue, or are the error messages the only symptom?
I've also tried enabling OpenGL in the config file, but this doesn't cause wezterm to break. Do you have other options enabled which are related to GPU acceleration?
Mr-Andersen
commented
1 month ago Hey @schung-amd, here are all the relevant options from my config:
{ ... }: {
boot.initrd.kernelModules = [ "amdgpu" ];
boot = {
kernelPackages = config.boot.zfs.package.latestCompatibleLinuxPackages;
kernelParams = [ "amdgpu.runpm=0" ]; # <-- this was me trying to fix the issue by reading Arch forums :)
};
hardware = {
graphics = {
enable = true;
enable32Bit = true;
};
};
services = {
displayManager = {
defaultSession = "xfce";
};
xserver = {
enable = true;
displayManager.lightdm.enable = true;
desktopManager.xfce.enable = true;
};
};
}My current nixpkgs commit is 12228ff1752d7b7624a54e9c1af4b222b3c1073b. I am on github:NixOS/nixpkgs/nixos-unstable branch currently, but I've starting seeing the issue while using nixos-24.05.
Here is how I experience it:
I should've provided this config since the beginning, sorry about that. This is my first serious bug report :)
schung-amd
commented
1 month ago No worries, thanks for the config information. A couple follow-up questions so I can try to reproduce the issue:
pkgs.rocmPackages.rpp) or was amdgpu built-in?Mr-Andersen
commented
1 month ago pkgs.rocmPackages.rppschung-amd
commented
1 month ago Sure, a list of packages couldn't hurt. I'm more interested in the other config options, in case we could narrow this down to a config change, but if you haven't tested on a fresh install that's ok.
Mr-Andersen
commented
1 month ago I forgot there is also a hardware-configuration.nix
{ config, lib, pkgs, modulesPath, ... }:
{
imports =
[ (modulesPath + "/installer/scan/not-detected.nix")
];
boot.initrd.availableKernelModules = [ "xhci_pci" "ahci" "usbhid" "usb_storage" "sd_mod" ];
boot.initrd.kernelModules = [ ];
boot.kernelModules = [ "kvm-amd" ];
boot.extraModulePackages = [ ];
nixpkgs.hostPlatform = lib.mkDefault "x86_64-linux";
hardware.cpu.amd.updateMicrocode = lib.mkDefault config.hardware.enableRedistributableFirmware;
}To clarify - by "fresh install" you mean "an install with as little customization as possible"? My install is new - I have had these issues since the first boot. Maybe that's what you meant?
schung-amd
commented
1 month ago Yes, that's what I meant, sorry for any confusion. That will make this much easier, thanks. I'll try running a high load as you suggest and see if I can reproduce the issue.
schung-amd
commented
1 month ago I am unable to reproduce the issue on XFCE, even at high load. Could you upload your configuration.nix file? You can scrub out your user information if you want to. hardware-configuration.nix should be automatically generated, but uploading this might help as well, so I can check for any discrepancies between your system and what I'm trying to repro with. Thanks!
schung-amd
commented
1 month ago Closing this as I can't reproduce the issue. If you'd still like support on this issue, feel free to reopen with your configuration.nix and hardware-configuration.nix files, and ideally with a consistent method of reproducing the issue.
Mr-Andersen
commented
3 weeks ago Sorry for leaving this thread. It seems that the issue was fixed somewhere upstream https://discourse.nixos.org/t/getting-amdgpu-error-that-crashes-desktop/50510/8?u=mr-andersen
schung-amd
commented
3 weeks ago Glad to hear your issue is resolved, thanks for the update!
Problem Description
I get
[gfxhub] page faultand then[drm:amdgpu_cs_ioctl [amdgpu]] *ERROR* Failed to initialize parser -125!I couldn't find my GPU (AMD Radeon RX 6400 / gfx1034) in the list below :( So I chose the first one. Additionally, I didn't find instructions on how to find out my ROCm version; I am guessing it's 6.0.2 since it's the default one on current NixOS.
Operating System
NixOS 24.05 (Uakari)
CPU
AMD Ryzen 5 3600 6-Core Processor
GPU
AMD Instinct MI300X
ROCm Version
ROCm 6.0.0
ROCm Component
No response
Steps to Reproduce
Run wezterm; wait
(Optional for Linux users) Output of /opt/rocm/bin/rocminfo --support
Additional Information