Fedora 43 booted to a black screen

Ask Fedora
, , ,
1

Hi something very similar to Fedora 43 KDE sometimes boots to a black screen happened to me today.

Observations:

  • sporadic, first time this happens daily driving, updated to F43 last month
  • mini PC with Display Port monitor, I do not remember this happening with the same monitor on HDMI but that was before updating to F43
  • AMD GPU, see below
  • I could log in to console using Ctrl-Alt-F2 and the system was otherwise fine

The following from /var/log/messages seems relevant:

sddm-helper-start-wayland[2498]: "No backend specified, automatically choosing drm\n"
sddm-helper-start-wayland[2498]: "Accepting client connections on sockets: QList(\"wayland-0\")\n"
sddm-helper-start-wayland[2498]: "qt.qpa.wayland: qtvirtualkeyboard currently is not supported at client-side,
...
kernel: amdgpu 0000:76:00.0: [drm] enabling link 0 failed: 15
...
kernel: fbcon: Taking over console  <- Ctrl-Alt-F2 I suspect
...
kernel: Console: switching to colour frame buffer device 240x67
...
sddm-helper-start-wayland[2498]: "qt.qpa.wayland: There are no outputs - creating placeholder screen\n"
sddm-greeter-qt6[2504]: There are no outputs - creating placeholder screen
sddm-greeter-qt6[2504]: Loading file:///usr/share/sddm/themes/01-breeze-fedora/Main.qml...
sddm-greeter-qt6[2504]: QQuickView(0x556b8eb6e140) already has a shell integration. Call QWindow::close() first and show it again
sddm-helper-start-wayland[2498]: Stopping... "kwin_wayland"

I can provide the full log if necessary.

2

I see some reports about the amdgpu driver failing on 6.19 kernels.[1] Is that what you are using? Does using an older (or newer) kernel work around the problem?

  1. ↩︎

3

Versions …

$ uname -r
6.19.13-200.fc43.x86_64
$ lsmod | grep amdgpu
amdgpu              21454848  95
amdxcp                 12288  1 amdgpu
i2c_algo_bit           20480  1 amdgpu
drm_ttm_helper         20480  1 amdgpu
ttm                   135168  2 amdgpu,drm_ttm_helper
drm_exec               12288  1 amdgpu
drm_panel_backlight_quirks    12288  1 amdgpu
gpu_sched              69632  1 amdgpu
drm_suballoc_helper    20480  1 amdgpu
drm_buddy              32768  1 amdgpu
drm_display_helper    344064  1 amdgpu
video                  81920  1 amdgpu
cec                   106496  2 drm_display_helper,amdgpu

The blank screen happened once out of about ten boots, all with the above kernel.

4

Forgot …

$ rpm -qa | grep ^amd-
amd-gpu-firmware-20260410-1.fc43.noarch
amd-ucode-firmware-20260410-1.fc43.noarch
5

Hardware …

# lspci | grep VGA
76:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Rembrandt [Radeon 680M] (rev c8)
6

https://bbs.archlinux.org/viewtopic.php?pid=2254925#p2254925 has a potential workaround:

I’ve found a temp fix. Just downgrading your Refresh rates to a lower value than the max make the job. I’m currently at 120Hz instead of 165Hz. Reload BIOS to default, Disabel secure boot (in my case), Enabled EXPO1. All is working perfectly. I’m suspecting that since DisplayPort 1.4 lot of manufacturer use DSC to have lower latency and better refresh rates. May be the firmware doesnt handle very well the retrocompatibility with 1.2 version. Like I was wondering how I dont have more log like my computer/firmware didnt handle any problem and found on reddit that G27QCA monitor have terrible problems with DP2.1 Graphic Cards. To be more clear the problem is happening on both Windows and Linux and when switching to HDMI the problem doesnt exist anymore. Same when I plugged my monitor to my old 3070 the problem is gone with Display Port cable. So yeah there are a problem with the link and I guess we have no way to fix it

7

Additional context …

Updated to Fedora 43 on April 8 (amd-gpu-firmware-0:20260309-1).

Black screen happened on April 22.

Did not happen again.

Updated to amd-gpu-firmware-0:20260410-1 on April 23.

Never happened again (daily driving).

I guess it was just a rare race condition.

8

Just happened again, switching to a virtual console and rebooting gave me a screen again.

Comparing logs, this seems to be the common symptom:

kernel: amdgpu 0000:76:00.0: [drm] enabling link 0 failed: 15

Context:
kernel-7.0.4-100
amd-gpu-firmware-20260410-1
amd-ucode-firmware-20260410-1

9

It might be. It looks like there is an interesting power management patch that has been pushed to the kernel recently (14 hours ago): Fix CPU hotplug activation race in the timer migration code. I wonder if that could be related to this problem?

10

With the problem occurring twice over about 20 days, once that patch is in effect I would have to observe the system for another 20 before calling it effective, wouldn’t I?

11

Yeah, you might run something like sudo journalctl --since=-3weeks -g "enabling link 0 failed" to get an idea about how often the problem is occurring now, then rerun that command a few weeks after that power management patch has arrived and see if the problem is still occurring with the same frequency.