Wayland/GDM3 graphical mode not starting

[mrspuck]

Hi,

I am on a fresh install of Debian 12 with kernel 6.5 from Debian backports. Sometimes when I boot the machine, the graphical login prompt does not appear. The screen goes black and the monitor goes into sleep because of no signal. My gfx card is a Radeon RX6600. I need kernel 6.5 for my USB WIFI stick to work. So now I can chose between a system without Internet (Kernel 6.1) and one that sometimes does not boot up properly (kernel 6.5).

The CPU has integrated gfx and I tried disabling that in the BIOS but it does not solve the problem.

How can I debug this further. So far what I have is following logs:

Code: Select all

journalctl -u gdm --since "1 hour ago"
First a failed boot, then a good one:

-- Boot bc941149a87b456c9bb01bd2a3957bf2 --
Mar 02 16:30:24 debian systemd[1]: Starting gdm.service - GNOME Display Manager...
Mar 02 16:30:24 debian systemd[1]: Started gdm.service - GNOME Display Manager.
Mar 02 16:30:24 debian gdm-launch-environment][981]: pam_unix(gdm-launch-environment:session): session opened for user Debian-gdm(uid=112) by (uid=0)
Mar 02 16:30:24 debian gdm-launch-environment][981]: pam_unix(gdm-launch-environment:session): session closed for user Debian-gdm
Mar 02 16:30:24 debian gdm-launch-environment][981]: GLib-GObject: g_object_unref: assertion 'G_IS_OBJECT (object)' failed
Mar 02 16:30:24 debian gdm3[964]: Gdm: GdmDisplay: Session never registered, failing
Mar 02 16:30:24 debian gdm3[964]: Gdm: Child process -1069 was already dead.
Mar 02 16:30:24 debian gdm3[964]: Gdm: GdmDisplay: Session never registered, failing
Mar 02 16:30:24 debian gdm3[964]: Gdm: Child process -1069 was already dead.
Mar 02 16:30:24 debian gdm-launch-environment][1446]: pam_unix(gdm-launch-environment:session): session opened for user Debian-gdm(uid=112) by (uid=0)
-- Boot 55f4a77fe6ec43c6bbdedfa0c431d993 --
Mar 02 16:31:22 debian systemd[1]: Starting gdm.service - GNOME Display Manager...
Mar 02 16:31:22 debian systemd[1]: Started gdm.service - GNOME Display Manager.
Mar 02 16:31:22 debian gdm-launch-environment][1023]: pam_unix(gdm-launch-environment:session): session opened for user Debian-gdm(uid=112) by (uid=0)
Mar 02 16:31:30 debian gdm-password][2776]: gkr-pam: unable to locate daemon control file
Mar 02 16:31:30 debian gdm-password][2776]: gkr-pam: stashed password to try later in open session
Mar 02 16:31:30 debian gdm-password][2776]: pam_unix(gdm-password:session): session opened for user puck(uid=1000) by (uid=0)
Mar 02 16:31:30 debian gdm-password][2776]: gkr-pam: unlocked login keyring
Mar 02 16:31:33 debian gdm3[1004]: Gdm: Child process -1120 was already dead.

[ruwolf]

In my opinion, this filter does not show cause of the fail.
Look at all messages between "2024-03-02 16:31:22" and "2024-03-02 16:31:33"

[mrspuck]

The unfiltered log looks like this:

Code: Select all

Mar 02 16:30:24 debian wireplumber[1067]: Failed to set scheduler settings: Operation not permitted
Mar 02 16:30:24 debian wireplumber[1067]: SPA handle 'api.libcamera.enum.manager' could not be loaded; is it installed?
Mar 02 16:30:24 debian wireplumber[1067]: PipeWire's libcamera SPA missing or broken. libcamera not supported.
Mar 02 16:30:24 debian gnome-shell[1200]: Running GNOME Shell (using mutter 43.8) as a Wayland display server
Mar 02 16:30:24 debian gnome-shell[1200]: Device '/dev/dri/card0' prefers shadow buffer
Mar 02 16:30:24 debian gnome-shell[1200]: Added device '/dev/dri/card0' (amdgpu) using atomic mode setting.
Mar 02 16:30:24 debian gnome-shell[1200]: Device '/dev/dri/card1' prefers shadow buffer
Mar 02 16:30:24 debian gnome-shell[1200]: Added device '/dev/dri/card1' (amdgpu) using atomic mode setting.
Mar 02 16:30:24 debian org.gnome.Shell.desktop[1200]: Failed to setup: No GPUs with outputs found
Mar 02 16:30:24 debian gnome-session[1119]: gnome-session-binary[1119]: WARNING: App 'org.gnome.Shell.desktop' exited with code 1
Mar 02 16:30:24 debian gnome-session-binary[1119]: WARNING: App 'org.gnome.Shell.desktop' exited with code 1
Mar 02 16:30:24 debian gnome-session-binary[1119]: Unrecoverable failure in required component org.gnome.Shell.desktop
Mar 02 16:30:24 debian /usr/libexec/gdm-wayland-session[1110]: dbus-daemon[1110]: [session uid=112 pid=1110] Activating service name='ca.desrt.dconf' requested by ':1.2' (uid=112 pid=1119 comm="/usr/libexec/gnome-session-binary --autostart /usr")
Mar 02 16:30:24 debian /usr/libexec/gdm-wayland-session[1110]: dbus-daemon[1110]: [session uid=112 pid=1110] Successfully activated service 'ca.desrt.dconf'
Mar 02 16:30:24 debian gdm-launch-environment][981]: pam_unix(gdm-launch-environment:session): session closed for user Debian-gdm
Mar 02 16:30:24 debian gdm-launch-environment][981]: GLib-GObject: g_object_unref: assertion 'G_IS_OBJECT (object)' failed
Mar 02 16:30:24 debian gdm3[964]: Gdm: GdmDisplay: Session never registered, failing
Mar 02 16:30:24 debian systemd[1]: session-c1.scope: Deactivated successfully.
Mar 02 16:30:24 debian systemd-logind[849]: Session c1 logged out. Waiting for processes to exit.
Mar 02 16:30:24 debian systemd-logind[849]: Removed session c1.
Mar 02 16:30:24 debian gdm3[964]: Gdm: Child process -1069 was already dead.
Mar 02 16:30:24 debian gdm3[964]: Gdm: GdmDisplay: Session never registered, failing
Mar 02 16:30:24 debian gdm3[964]: Gdm: Child process -1069 was already dead.
Mar 02 16:30:24 debian gdm-launch-environment][1446]: pam_unix(gdm-launch-environment:session): session opened for user Debian-gdm(uid=112) by (uid=0)
Mar 02 16:30:24 debian systemd-logind[849]: New session c2 of user Debian-gdm.

What follows then looks like a startup of X11, but ends in a black screen.

A good bootup looks like this:

Code: Select all

Mar 08 19:50:31 debian gnome-shell[1237]: Running GNOME Shell (using mutter 43.8) as a Wayland display server
Mar 08 19:50:31 debian wireplumber[1104]: Failed to set scheduler settings: Operation not permitted
Mar 08 19:50:31 debian wireplumber[1104]: SPA handle 'api.libcamera.enum.manager' could not be loaded; is it installed?
Mar 08 19:50:31 debian wireplumber[1104]: PipeWire's libcamera SPA missing or broken. libcamera not supported.
Mar 08 19:50:32 debian gnome-shell[1237]: Device '/dev/dri/card0' prefers shadow buffer
Mar 08 19:50:32 debian gnome-shell[1237]: Added device '/dev/dri/card0' (amdgpu) using atomic mode setting.
Mar 08 19:50:32 debian gnome-shell[1237]: Device '/dev/dri/card1' prefers shadow buffer
Mar 08 19:50:32 debian gnome-shell[1237]: Added device '/dev/dri/card1' (amdgpu) using atomic mode setting.
Mar 08 19:50:32 debian gnome-shell[1237]: Created gbm renderer for '/dev/dri/card0'
Mar 08 19:50:32 debian gnome-shell[1237]: Created gbm renderer for '/dev/dri/card1'
Mar 08 19:50:32 debian gnome-shell[1237]: Boot VGA GPU /dev/dri/card0 selected as primary

I still cannot see what is going wrong. Searched the web but what I found was related to NVidia cards; mine is AMD.

[ruwolf]

Yes, this is sign of the cause:

Code: Select all

Mar 02 16:30:24 debian org.gnome.Shell.desktop[1200]: Failed to setup: No GPUs with outputs found

I am not an expert, but I think, that this can be the reason:

The amdgpu kernel driver needs to be loaded before the radeon one.

Output of this command could be useful:

Code: Select all

inxi -Gaxxxz --edid

[mrspuck]

This is an AM5 CPU; I think it also uses the amdgpu driver. Radeon is not loaded.

Code: Select all

Graphics:
  Device-1: AMD Navi 23 [Radeon RX 6600/6600 XT/6600M] vendor: ASRock
    driver: amdgpu v: kernel arch: RDNA-2 code: Navi-2x process: TSMC n7 (7nm)
    built: 2020-22 pcie: gen: 4 speed: 16 GT/s lanes: 16 ports: active: DP-4
    empty: DP-1, DP-2, DP-3, DP-5, HDMI-A-1 bus-ID: 03:00.0 chip-ID: 1002:73ff
    class-ID: 0300
  Device-2: AMD Raphael vendor: Micro-Star MSI driver: N/A alternate: amdgpu
    arch: RDNA-2 code: Navi-2x process: TSMC n7 (7nm) built: 2020-22 pcie:
    gen: 4 speed: 16 GT/s lanes: 16 bus-ID: 11:00.0 chip-ID: 1002:164e
    class-ID: 0300
  Display: wayland server: X.org v: 1.21.1.7 with: Xwayland v: 22.1.9
    compositor: gnome-shell v: 43.9 driver: X: loaded: vesa
    unloaded: fbdev,modesetting,radeon gpu: amdgpu display-ID: 0
  Monitor-1: DP-4 model: Dell U2413 serial: <filter> built: 2013
    res: 1920x1200 dpi: 94 gamma: 1.2 chroma: red: x: 0.682 y: 0.310 green:
    x: 0.208 y: 0.702 blue: x: 0.145 y: 0.051 white: x: 0.314 y: 0.329
    size: 518x324mm (20.39x12.76") diag: 611mm (24.1") ratio: 16:10
    modes: 1920x1200, 1920x1080, 1600x1200, 1280x1024, 1152x864, 1280x720,
    1024x768, 800x600, 720x576, 720x480, 640x480, 720x400
  API: OpenGL v: 4.6 Mesa 22.3.6 renderer: AMD Radeon RX 6600 (navi23 LLVM
    15.0.6 DRM 3.54 6.5.0-0.deb12.4-amd64) direct-render: Yes

Now I cannot reproduce the problem any more. All I did was to try different kernel parameters such as "amdgpu.dc=0" dynamically from the grub boot screen. The problem still occurred with this parameter. But now, booted many times without it and the problem is gone. Weird.

[ruwolf]

Radeon is unloaded, so IMHO it means, that it was loaded.
But I do not say, that it was cause for sure, it was my opinion only.
As it is not used, I would try to disable it.