Created attachment 147533 [details] Screenshot of fixed result SUMMARY *** NOTE: If you are reporting a crash, please try to attach a backtrace with debug symbols. See https://community.kde.org/Guidelines_and_HOWTOs/Debugging/How_to_create_useful_crash_reports *** When using more than one display with the same EDID name and multiple different rotations, display configuration will apply your changes but instantly think a change was made so there are TWO apply steps. Due to a hardware issue, sometimes my displays reconnect and the portrait rotation from the right monitor is applied to the left making the left upside down. I have to go in each time and reapply the rotation, apply twice, and then close out. Confirmed on two different AMD GPU systems with the same monitor layout (One HDMI and two Displayport) My fix was to create a specific EDID for each of the portrait monitors with their own name. Image attached showing the result of the fix. STEPS TO REPRODUCE 1. Have three monitors 2. Arrange in one landscape, one portrait, and one reverse portrait 3. Rotation applied incorrectly to one of the portrait montiors OBSERVED RESULT On display reconnect/resume, the wrong rotation is applied to one of the portrait monitors due to the naming of the display being the same between all. EXPECTED RESULT Displays to be treated separately (possibly by serial/unique identifier based on connection name) SOFTWARE/OS VERSIONS Linux/KDE Plasma: Arch Linux KDE Plasma Version: 5.24.3 KDE Frameworks Version: 5.92.0 Qt Version: 5.15.3+kde+r134-1 ADDITIONAL INFORMATION
Forgot to mention session type X11. I think the connector based layout is working but there's possibly a bug with two of the same name in portrait rotation.
Thanks to a wonderful contributor over on Reddit, finally realized what exposed this issue. My custom EDID setup was already in place when this started. The DisplayPort dock that I have will not drive to monitors at the full refresh rate. So I already had a custom EDID with a single mode single refresh rate set for both of those. Guess which two those were, both portrait. I used a basic EDID generator for the mode and refresh rate which doesn't set a serial number. On top of that I used the same file for both of them in the kernel command line, so to the system they were essentially the same monitor with an nulled out serial. My "solution" of creating to EDID with different names worked because Plasma then have something to differentiate them by even with a blank serial. Wayland with the same EDID works as it already names them as "unknown <connector type>" programmatically. For the X11 issue at hand, I get that it was my EDID with the same nulled serial causing the issue. I know those get used a lot so probably not the first to run into it. Will be interesting to see if there's anything that can be done there. Thanks.
Fascinating.
*** Bug 451462 has been marked as a duplicate of this bug. ***
I'm experiencing the same issue. I have two identical external displays connected to my laptop. One of them is mounted horizontally (connected directly via hdmi), the other is rotated by 90° (connected via mini displayport -> hdmi adapter). After rebooting or disconnecting the displays, the rotation of the vertically mounted display is reset to horizontal. The configured display position stays the same.
This may have been fixed by a change to the duplicate detection algorithm in Plasma 5.26. Would you mind testing that and reporting back on whether it fixes the issue or not? You may want to delete your kscreen configs in ~/.local/share/kscreen first to test it fresh.
I have been on 5.26 with the same setup without any modified EDID and it appears to be working as expected. I believe I changed the bottom option in display settings to remember it per monitor setup if that helps. I was on 5.23 shortly before this (Steam Deck) and I was experiencing the same rotation issue on the left hand monitor. After Steam Deck Main branch updated to 5.26, smooth sailing.
Great news! Thanks for following up.