You are not logged in.
Hello! Yesterday I had a problem with amd drivers and kind people on this forum helped me solve it. (https://bbs.archlinux.org/viewtopic.php … 9#p2174059) (linking it in case it somehow related)
However today I can't open steam because of radeonsi_dri. Launching it from terminal outputs the following:
❯ LIBGL_DEBUG=verbose steam
steam.sh[18137]: Running Steam on arch rolling 64-bit
steam.sh[18137]: STEAM_RUNTIME is enabled automatically
setup.sh[18211]: Steam runtime environment up-to-date!
steam.sh[18137]: Steam client's requirements are satisfied
[2024-05-28 12:41:34] Startup - updater built May 24 2024 20:46:19
[2024-05-28 12:41:34] Startup - Steam Client launched with: '/home/ilia21/.local/share/Steam/ubuntu12_32/steam'
05/28 12:41:34 minidumps folder is set to /tmp/dumps
05/28 12:41:34 Init: Installing breakpad exception handler for appid(steam)/version(1716584667)/tid(18250)
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
using driver amdgpu for 9
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
pci id for fd 9: 1002:731f, driver radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/radeonsi_dri.so: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open radeonsi: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
failed to load driver: radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/swrast_dri.so: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open swrast: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
SteamUpdateUI: An X Error occurred
X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 150 (GLX)
Minor opcode of failed request: 3 (X_GLXCreateContext)
Value in failed request: 0x0
Serial number of failed request: 41
xerror_handler: X failed, continuing
Looks like steam didn't shutdown cleanly, scheduling immediate update check
[2024-05-28 12:41:34] Loading cached metrics from disk (/home/ilia21/.local/share/Steam/package/steam_client_metrics.bin)
[2024-05-28 12:41:34] Using the following download hosts for Public, Realm steamglobal
[2024-05-28 12:41:34] 1. https://client-update.akamai.steamstatic.com, /, Realm 'steamglobal', weight was 1000, source = 'update_hosts_cached.vdf'
[2024-05-28 12:41:34] 2. https://cdn.cloudflare.steamstatic.com, /client/, Realm 'steamglobal', weight was 1, source = 'update_hosts_cached.vdf'
[2024-05-28 12:41:34] 3. https://cdn.steamstatic.com, /client/, Realm 'steamglobal', weight was 1, source = 'baked in'
[2024-05-28 12:41:34] Checking for update on startup
[2024-05-28 12:41:34] Checking for available updates...
[2024-05-28 12:41:34] Downloading manifest: https://client-update.akamai.steamstatic.com/steam_client_ubuntu12
[2024-05-28 12:41:34] Manifest download: send request
[2024-05-28 12:41:34] Manifest download: waiting for download to finish
[2024-05-28 12:41:35] Manifest download: finished
[2024-05-28 12:41:35] Download skipped: /steam_client_ubuntu12 version 1716584667, installed version 1716584667, existing pending version 0
[2024-05-28 12:41:35] Nothing to do
[2024-05-28 12:41:35] Verifying installation...
[2024-05-28 12:41:35] Performing checksum verification of executable files
[2024-05-28 12:41:35] Verification complete
UpdateUI: skip show logo
Steam logging initialized: directory: /home/ilia21/.local/share/Steam/logs
XRRGetOutputInfo Workaround: initialized with override: 0 real: 0xef653dc0
XRRGetCrtcInfo Workaround: initialized with override: 0 real: 0xef652500
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
using driver amdgpu for 28
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
pci id for fd 28: 1002:731f, driver radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/radeonsi_dri.so: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open radeonsi: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
failed to load driver: radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/swrast_dri.so: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open swrast: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
crash_20240528124136_5.dmp[18289]: Uploading dump (out-of-process)
/tmp/dumps/crash_20240528124136_5.dmp
/home/ilia21/.local/share/Steam/steam.sh: line 813: 18250 Segmentation fault (core dumped) "$STEAMROOT/$STEAMEXEPATH" "$@"
I tried:
- sudo pacman -Syu
- steam-runtime
- reinstalling steam
- reinstalling lib32-mesa
Any idea what's happening and how to fix it? Maybe a problem on my end?
Any help would be greatly appreciated
Last edited by ilia21 (2024-05-28 15:19:03)
Offline
Please post outputs of
$ LIBGL_DEBUG=verbose glxinfo -B
$ LIBGL_DEBUG=verbose glxinfo32 -B
(install mesa-utils / lib32-mesa-utils if needed)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
Hello, thanks for quick reply!
❯ LIBGL_DEBUG=verbose glxinfo -B
name of display: :1
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
using driver amdgpu for 4
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
pci id for fd 4: 1002:731f, driver radeonsi
MESA-LOADER: failed to open /usr/lib/dri/radeonsi_dri.so: /usr/lib/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open radeonsi: /usr/lib/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib/dri, suffix _dri)
failed to load driver: radeonsi
MESA-LOADER: failed to open /usr/lib/dri/swrast_dri.so: /usr/lib/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open swrast: /usr/lib/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib/dri, suffix _dri)
X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 150 (GLX)
Minor opcode of failed request: 24 (X_GLXCreateNewContext)
Value in failed request: 0x0
Serial number of failed request: 42
Current serial number in output stream: 43
❯ LIBGL_DEBUG=verbose glxinfo32 -B
name of display: :1
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
using driver amdgpu for 4
libGL: Can't open configuration file /etc/drirc: No such file or directory.
libGL: Can't open configuration file /home/ilia21/.drirc: No such file or directory.
pci id for fd 4: 1002:731f, driver radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/radeonsi_dri.so: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open radeonsi: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
failed to load driver: radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/swrast_dri.so: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open swrast: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
X Error of failed request: BadValue (integer parameter out of range for operation)
Major opcode of failed request: 150 (GLX)
Minor opcode of failed request: 24 (X_GLXCreateNewContext)
Value in failed request: 0x0
Serial number of failed request: 42
Current serial number in output stream: 43
mesa-utils and lib32-mesa-utils are already installed
Offline
https://bbs.archlinux.org/viewtopic.php … 7#p2036377 triggered me to check , and the symbol amdgpu_cs_ctx_stable_pstat does occur in libdrm .
Please post the output of
pacman -Qikk libdrm
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
❯ pacman -Qikk libdrm
Name : libdrm
Version : 2.4.120-1
Description : Userspace interface to kernel DRM services
Architecture : x86_64
URL : https://dri.freedesktop.org/
Licenses : custom
Groups : None
Provides : None
Depends On : libpciaccess glibc
Optional Deps : cairo: needed for modetest tool [installed]
Required By : amdgpu_top-bin ddcutil egl-wayland electron28 ffmpeg ffmpeg4.4 gamescope gst-plugins-bad-libs gst-plugins-base-libs hsa-rocr hsakmt-roct kpipewire kwin
lib32-libdrm libliftoff libva libva-mesa-driver mesa mesa-utils minecraft-launcher mpv qt6-base qt6-webengine vulkan-mesa-layers vulkan-radeon
vulkan-swrast wayland-utils webkit2gtk-4.1 xorg-server xorg-xwayland-osu
Optional For : fastfetch
Conflicts With : None
Replaces : libdrm-new libdrm-nouveau
Installed Size : 1250.41 KiB
Packager : Andreas Radke <andyrtr@archlinux.org>
Build Date : Sat 13 Jan 2024 10:19:14 PM MSK
Install Date : Tue 28 May 2024 12:32:05 PM MSK
Install Reason : Installed as a dependency for another package
Install Script : No
Validated By : Signature
warning: libdrm: /usr/lib/libdrm_amdgpu.so.1 (Symlink path mismatch)
warning: libdrm: /usr/lib/libdrm_amdgpu.so.1 (Modification time mismatch)
libdrm: 92 total files, 1 altered file
Offline
What is the output of
$ ls -l /usr/lib/libdrm_amdgpu*
?
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
❯ ls -l /usr/lib/libdrm_amdgpu*
lrwxrwxrwx - root 13 Jan 22:19 /usr/lib/libdrm_amdgpu.so -> libdrm_amdgpu.so.1
lrwxrwxrwx - root 28 May 12:32 /usr/lib/libdrm_amdgpu.so.1 -> libdrm_amdpro.so.1.0.0
.rwxr-xr-x 43k root 13 Jan 22:19 /usr/lib/libdrm_amdgpu.so.1.0.0
lrwxrwxrwx - root 27 May 20:28 /usr/lib/libdrm_amdgpu.so.1.bak -> libdrm_amdpro.so.1.0.0
(squares are supposed to be nerdfont icons)
I think this is the problem? Somehow /usr/lib/libdrm_amdgpu.so.1 is pointing to libdrm_amdpro.so.1.0.0 instead of libdrm_amdgpu.so.1.0.0. This was a problem that caused sddm not to work, and I thought I fixed it by linking /usr/lib/libdrm_amdgpu.so.1 -> libdrm_amdgpu.so.1.0.0 . But somehow today, after rebooting, it linked back to libdrm_amdpro.so.1.0.0.
Should I link it? And what is the correct way to link system libraries? I know how ln works in Linux, but is it the right way to link libraries (I ask because I've never done it and maybe there's a more "correct" way to do it)?Never mind, there's probably a wiki page
Last edited by ilia21 (2024-05-28 11:20:49)
Offline
I want to add that any pacman operation resets my link back to /usr/lib/libdrm_amdgpu.so.1 -> libdrm_amdpro.so.1.0.0... I linked it using the following commands:
sudo mv /usr/lib/libdrm_amdgpu.so.1 /usr/lib/libdrm_amdgpu.so.1.bak
sudo ln /usr/lib/libdrm_amdgpu.so.1.0.0 /usr/lib/libdrm_amdgpu.so.1
Offline
This shouldn't happen automatically.
pacman -Qs amd
grep amdpro /etc/pacman.d/hooks /usr/share/libalpm/hooks
Online
Hello, thanks for quick reply!
pacman -Qs amd
local/amd-ucode 20240510.b9d2bf23-1
Microcode update image for AMD CPUs
local/amd-vulkan-prefixes 1-1
Select needed vulkan implementation with vk_radv, vk_amdvlk or vk_pro prefix
local/amdgpu_top-bin 0.8.3-1
Tool to display AMDGPU usage (binary release)
local/amdgpu_top-bin-debug 0.8.3-1
Detached debugging symbols for amdgpu_top-bin
local/amdvlk 2024.Q2.1-1
AMD's standalone Vulkan driver
local/lib32-vulkan-radeon 1:24.0.8-1
Open-source Vulkan driver for AMD GPUs - 32-bit
local/libteam 1.32-1
Library for controlling team network device
local/rocm-core 6.0.2-2
AMD ROCm core package (version files)
local/rocm-opencl-runtime 6.0.2-1
OpenCL implementation for AMD
local/vulkan-radeon 1:24.0.8-1
Open-source Vulkan driver for AMD GPUs
❯ grep amdpro /etc/pacman.d/hooks /usr/share/libalpm/hooks
grep: /etc/pacman.d/hooks: No such file or directory
grep: /usr/share/libalpm/hooks: Is a directory
Offline
V1del had a typo in the grep command, use
grep amdpro /etc/pacman.d/hooks/* /usr/share/libalpm/hooks/*
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
Everyone makes mistakes
❯ grep amdpro /etc/pacman.d/hooks/* /usr/share/libalpm/hooks/*
zsh: no matches found: /etc/pacman.d/hooks/*
Offline
Let's try to simplify your setup a bit :
remove amd-vulkan-prefixes , amdgpu_top-bin , amdgpu_top-bin-debug, amdvlk
re-install libdrm
reboot , post pacman -Qikk libdrm .
Also the journal (with root-rights) of the boot.
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
I guess it deleted my link again and now I don't have access to my desktop. (Less logs = better, I guess?) Not touching anything for now, afraid to make it worse.
Output of both:
https://0x0.st/XNKQ.txt
Offline
No clue how those symlink are wrong again, but the graphical target is reached before the videocard has finished initialising .
This can have all kinds of bad consequences and is very likely the reason why sddm coredumps. Configuring early KMS is simplest solution for that.
Add amdgpu to the modules= line in /etc/mkinitcpio.conf , re-generate initramfs and reboot.
edit: typos
Last edited by Lone_Wolf (2024-05-28 13:23:53)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
This has solved my issue with link breaking, thanks! (posting the directory in case it is relevant)
❯ ls /usr/lib/libdrm_amdgpu*
/usr/lib/libdrm_amdgpu.so -> libdrm_amdgpu.so.1
/usr/lib/libdrm_amdgpu.so.1
/usr/lib/libdrm_amdgpu.so.1.0.0
/usr/lib/libdrm_amdgpu.so.1.bak -> libdrm_amdpro.so.1.0.0
/usr/lib/libdrm_amdgpu.so.1.bak2 -> libdrm_amdpro.so.1.0.0
But steam still refuses to open with the same error as before
MESA-LOADER: failed to open /usr/lib32/dri/radeonsi_dri.so: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open radeonsi: /usr/lib32/dri/radeonsi_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
failed to load driver: radeonsi
MESA-LOADER: failed to open /usr/lib32/dri/swrast_dri.so: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate
MESA-LOADER: failed to open swrast: /usr/lib32/dri/swrast_dri.so: undefined symbol: amdgpu_cs_ctx_stable_pstate (search paths /usr/lib32/dri, suffix _dri)
Offline
Focus was on libdrm until now, we still need to look at lib32-libdrm .
Please post
$ pacman -Qikk lib32-libdrm
$ ls -l /usr/lib32/libdrm_amdgpu*
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
Okay, thank you for guiding me on the right direction.
❯ pacman -Qikk lib32-libdrm
Name : lib32-libdrm
Version : 2.4.120-1
Description : Userspace interface to kernel DRM services (32-bit)
Architecture : x86_64
URL : https://dri.freedesktop.org/
Licenses : custom
Groups : None
Provides : None
Depends On : lib32-libpciaccess lib32-glibc libdrm
Optional Deps : None
Required By : lib32-gst-plugins-base-libs lib32-libva lib32-mesa lib32-vulkan-radeon lib32-vulkan-swrast
Optional For : None
Conflicts With : None
Replaces : None
Installed Size : 384.77 KiB
Packager : Laurent Carlier <lordheavym@archlinux.org>
Build Date : Wed 06 Mar 2024 08:27:57 AM MSK
Install Date : Tue 28 May 2024 12:32:05 PM MSK
Install Reason : Installed as a dependency for another package
Install Script : No
Validated By : Signature
warning: lib32-libdrm: /usr/lib32/libdrm_amdgpu.so.1 (Symlink path mismatch)
warning: lib32-libdrm: /usr/lib32/libdrm_amdgpu.so.1 (Modification time mismatch)
lib32-libdrm: 26 total files, 1 altered file
❯ ls -l /usr/lib32/libdrm_amdgpu*
lrwxrwxrwx - root 6 Mar 08:27 /usr/lib32/libdrm_amdgpu.so -> libdrm_amdgpu.so.1
lrwxrwxrwx - root 28 May 12:32 /usr/lib32/libdrm_amdgpu.so.1 -> libdrm_amdpro.so.1.0.0
.rwxr-xr-x 46k root 6 Mar 08:27 /usr/lib32/libdrm_amdgpu.so.1.0.0
Offline
The amdpro grep was a very try random shit attempt (sorry for the flub with the * wildcard, whoops) Generally what I would think you should check especially if this happens after every pacman operation whether you added some hook that would reintroduce these symlinks. You did some very non standard experiment there when switching/using the prop drivers. Or did you maybe alias pacman to execute a script that creates the symlinks?
Last edited by V1del (2024-05-28 13:53:21)
Online
Just checked: no pacman aliases defined
❯ type pacman
pacman is /usr/bin/pacman
My hooks are: "base udev autodetect microcode keyboard keymap modconf block filesystems fsck", according to wiki nothing unusual
(Note to self: never mess with proprietary drivers ever again)
Offline
I'm talking pacman hooks, the ones that lie in the two directories I tried to grep for amdpro for. e.g. /etc/pacman.d/hooks should "normally" be empty, do you have files in there? What somewhat kills that theory is that I haven't seen any suspicious hook running in the pacman.log you posted in the previous thread. Other than that, maybe some udev rule that didn't get cleaned up fully when you removed the amdpro package.
Last edited by V1del (2024-05-28 14:32:05)
Online
/etc/pacman.d/hooks does not even exist on my filesystem, but there are some files in /usr/share/libalpm/hooks/, but none of them mention amd (if I grep'd correctly).
grep -r amd /usr/share/libalpm/hooks/*
(empty result)
Checking udev hooks still no mentions of amd
❯ grep -r amd /usr/lib/udev/rules.d/*
(emty result)
Offline
FWIW, are you sure this still happens and this wasn't simply "not yet" done for the lib32 path which should be fixed now/can be fixed manually now as well?
Last edited by V1del (2024-05-28 14:45:42)
Online
This was happening after any pacman operation. After adding amdgpu to the modules= line in /etc/mkinitcpio.conf I installed amdgpu-top to see if it would break the link again, but it didn't, I rebooted, and the link still exists, so based on that I can say that pacman operations have not broken the link so far.
Am I safe to link /usr/lib32/libdrm_amdgpu.so.1 -> /usr/lib32/libdrm_amdgpu.so.1.0.0 now?
Offline
I could see the progl "properly" adding a hook to reapply the override, which got removed as well when removing the package, if the mainlink didn't break then doing this link now should be the general fix yes.
Online