Bugzilla – Bug 1212954
amdgpu: Monitor displaying wrong resolution
Last modified: 2024-02-16 13:34:02 UTC
Created attachment 867944 [details] IMG_20230703_164538751_HDR.jpg Having a 4K monitor (SAMSUNG LU28R55) attached to the docking station of my ThinkPad via HDMI I can easily 1:1 mirror the screen output of the laptop and the monitor with openSUSE Leap 15.4. Doing the same with openSUSE Leap 15.5 will result in monitor displaying a resolution 1980x2160 instead of the set 1980x1080. Using a 4x different screen (Philips) is working fine. Photo of monitor attached.
Created attachment 867945 [details] hwinfo-samsung.txt
Created attachment 867946 [details] xrandr-samsung.txt
Created attachment 867947 [details] dmesg-samsung.txt
Created attachment 867948 [details] xrandr-verbose-samsung.txt
Created attachment 867949 [details] hwinfo-philips.txt
Created attachment 867950 [details] xrandr-philips.txt
Created attachment 867951 [details] dmesg-philips.txt
Adding a modline created with > cvt -r 1920 1080 with: > xrandr --newmode "1920x1080R" 138.50 1920 1968 2000 2080 1080 1083 1088 1111 +hsync -vsync > xrandr --addmode DP-3 "1920x1080R" > xrandr --output DP-3 --mode 1920x1080R will result in having monitor displaying the correct resolution
Thanks for reporting, Holgi!
Looks like it uses some weird DoubleScan mode here ... maybe that's a feature of latest driver(s) on Leap 15.5 kernels. 1920x1080 60.01* 59.97 59.96 60.00 50.00 59.94 59.93 1920x1080 (0x4b) 356.375MHz -HSync +VSync DoubleScan h: width 1920 start 2080 end 2288 total 2656 skew 0 clock 134.18KHz v: height 1080 start 1081 end 1084 total 1118 clock 60.01Hz 1920x1080 (0x4c) 266.500MHz +HSync -VSync DoubleScan h: width 1920 start 1944 end 1960 total 2000 skew 0 clock 133.25KHz v: height 1080 start 1081 end 1084 total 1111 clock 59.97Hz 1920x1080 (0x4d) 173.000MHz -HSync +VSync h: width 1920 start 2048 end 2248 total 2576 skew 0 clock 67.16KHz v: height 1080 start 1083 end 1088 total 1120 clock 59.96Hz 1920x1080 (0xc1) 148.500MHz +HSync +VSync h: width 1920 start 2008 end 2052 total 2200 skew 0 clock 67.50KHz v: height 1080 start 1084 end 1089 total 1125 clock 60.00Hz 1920x1080 (0xc2) 148.500MHz +HSync +VSync h: width 1920 start 2448 end 2492 total 2640 skew 0 clock 56.25KHz v: height 1080 start 1084 end 1089 total 1125 clock 50.00Hz 1920x1080 (0xc3) 148.352MHz +HSync +VSync h: width 1920 start 2008 end 2052 total 2200 skew 0 clock 67.43KHz v: height 1080 start 1084 end 1089 total 1125 clock 59.94Hz 1920x1080 (0x4e) 138.500MHz +HSync -VSync *current h: width 1920 start 1968 end 2000 total 2080 skew 0 clock 66.59KHz v: height 1080 start 1083 end 1088 total 1111 clock 59.93Hz
Could you check the behavior with the upstream 6.4.x kernel? Test with the one in OBS Kernel:stable:Backport repo. If the upstream kernel is confirmed to work, test with the one in OBS home:tiwai:kernel:6.3 and home:tiwai:kernel:6.2 (use "backport" repo for Leap -- "standard" is for TW), so that we can narrow down a regression fix range.
Created attachment 867952 [details] IMG_20230703_180130037.jpg Same behavior as with regular 15.5 kernel - see photo
OK, then it means that it's an upstream bug :) We need to report it to the upstream bug tracker, gitlab.freedesktop.org DRM/AMD Issues.
BTW, does this happen with Wayland, too? I stumbled on the existing report: https://gitlab.freedesktop.org/xorg/driver/xf86-video-amdgpu/-/issues/68 and it says it's X who adds DoubleScan modes.
Ok. But that would mean that Holger runs amdgpu X driver. Holger, have you installed xf86-video-amdgpu package. Could you attach your X logfile, please?
If Holger really uses the amdgpu X driver and removing xf86-video-amdgpu packages fixes the issue, it's indeed the amdgpu driver.
Sorry, it's reported against amdgpu X driver, but the issue is in X server. Not sure why this didn't happen yet on Leap 15.4. We still have xorg-server 1.20.3 there, whereas we have 21.1.4 on Leap 15.5, but I think we didn't see relevant changes to this aspect.
Also, I built a test kernel with a hackish patch to limit the doublescan for HDMI. Could you try the kernel in OBS home:tiwai:bsc1212954 repo?
Created attachment 868854 [details] dmesg-5.14.21-150500.1.g08214d9-default.txt Unfortunately, still same behavior with your test kernel
As the issue happens only with X amdgpu driver, I reassign now. Please take back to kernel if it turned out to be a kernel problem.
(In reply to Stefan Dirsch from comment #15) > Ok. But that would mean that Holger runs amdgpu X driver. Holger, have you > installed xf86-video-amdgpu package. Could you attach your X logfile, > please? Could you still check this, please? Thanks!
No, package "xf86-video-amdgpu" is not installed: holger@t14s:~> rpm -qa | grep -i amdgpu kernel-firmware-amdgpu-20230724-150500.3.6.1.noarch libdrm_amdgpu1-2.4.114-150500.3.2.x86_64
Ok. So obviously it's not the amdgpu X driver. So either Xserver itself or amdgpu kernel driver. If it doesn't happen with Wayland we can safely assume it's the Xserver. Can you try this, Holger. Easiest would be to install gdm+GNOME and try with a GNOME session. Check what you're runnng with ... # echo $XDG_SESSION_TYPE x11
(In reply to Stefan Dirsch from comment #23) > Ok. So obviously it's not the amdgpu X driver. So either Xserver itself or > amdgpu kernel driver. If it doesn't happen with Wayland we can safely assume > it's the Xserver. Can you try this, Holger. Easiest would be to install > gdm+GNOME and try with a GNOME session. Check what you're runnng with ... > > # echo $XDG_SESSION_TYPE > x11 Could you still try this, please?
Any news on that one?
Unfortunately enabling of Wayland did not work as expected, even using gdm.
(In reply to Holger Sickenberg from comment #26) > Unfortunately enabling of Wayland did not work as expected, even using gdm. Yes, we don't know why. gdm/Gnome for some reasons the hardware isn't capable of handling a Wayland desktop.