Created attachment 182580 [details] journalctl log from wakeup SUMMARY Immediately after upgrading to KDE 6.4.0 experiencing an issue that seems almost like a race/timing issue that occurs every time for me. When monitor turns off (system does not sleep or hibernate, simply turns off screen and then monitor eventually goes to sleep), nvidia + wayland, after 6.4.0 upon waking up the monitor the screen is 1080 30hz and impossible to change (no options listed) requiring a full reboot. Wayland Plasma 6.4.0 on Arch Linux with nvidia card and display port. From logs it seems like plasma is trying to read the screen before the EDID can be read, causing a placeholder 1080 30hz screen to be created and then that gets stuck as it's not giving enough time for the monitor to wakeup. STEPS TO REPRODUCE 1. Turn off monitor or let it go to sleep (not the system, the monitor) 2. Turn back on monitor OBSERVED RESULT 1080p 30hz stuck as only option and impossible to change until restart computer EXPECTED RESULT Resolution should be the native/as I left it SOFTWARE/OS VERSIONS Arch Linux Kernel 6.15.2-arch1-1 (64-bit) KDE Plasma Version: 6.4.0 KDE Frameworks Version: 6.15.0 Qt Version: 6.9.1 Graphics Platform: Wayland nVidia Driver 575.64-1 (also tried on 575.64-2) on GeForce GTX 1060 6GB using DisplayPort 1.2 Monitor Samsung Odyssey Neo G9 Super Ultrawide/UHD (7680x2160 Res Currently) ADDITIONAL INFORMATION Doesn't appear in 6.3.5 or earlier, began immediately with 6.4.0 upgrade Attaching journalctl and xorg logs from wakeup, seems like 1. qt.qpa.wayland is reporting "There are no outputs - creating placeholder screen" which then takes over 2. Seeing powerdevil report Checking EDID failed after 3 tries (B), this seems like was a little too quickly as monitor needed a moment to come up? 3. Seeing powerdevil throw a Locked bus with current call stack (empty) 4. See kwin_wayland throw kwin_wayland_drm: atomic commit failed: Permission denied Meanwhile in xorg log can see it wakes up, nvidia throws warnings could not re-validate the resolution so removes them before the monitor reconnects right after, but at this point it seems like stuck with placeholder and need to reboot full computer Any ideas or help on this one? I can hardcode the EDID into the kernel but of course that's a bandaid not a proper solution, thanks!
Created attachment 182581 [details] Xorg log from wakeup
Sounds very much like Bug 495744.
Unfortunately not, that is a problem where was having issues waking up at all and going back quite a ways. Mine has worked flawlessly for years, and only after 6.4.0 did it suddenly become a problem, it looks very much like a race condition as when I bundle the edid for DP-2 into the firmware and load at kernel command line, problem solved! Maybe something related where the monitor takes a few extra milliseconds to reinit the connections, or possibly something with past configs being involved - I can try as a new user to rule out the latter, but it's not that it's not turning on at all or black screen... It comes back on quickly but the placeholder screen has already initialized and seeing edid read errors before it reconnects milliseconds later, so looks critically different from that case thus far 🤔 Any ideas what can do to provide added info? So far simply copying the edid into usr lib firmware and coding into the init solves fully, and first things by rolling back to 6.3 this was not a problem. I did notice when waking it often made two beeps of discovery then showed the screen, that seems improved but at the expense of using this placeholder screen which turns the monitor on fine, just is stuck at 1080p 30hz until reboot on a dual UHD ha
Hit that a bit quick, sorry for double post - wanted to emphasize don't believe so just from the circumstances (absolutes never a good idea), and logs compared vs that case. First I rolled back (though to be fair, one other change was Nvidia 570 -> Nvidia 575 at same time moved to plasma 6.4, but seemed more likely to be plasma related). Running 6.3 and even 6.2/prior no problem albeit a double audio bloop when waking and took a couple seconds longer than other monitors (this seems like could be a differentiator). In both cases 6.4 and pre-6.4 get to plasma no problem and monitor wakes, just on 6.4 I'm on a placeholder 1080/30hz screen as seems like something changed in power devil or kwin/wayland on the timeouts? Didn't see anything obvious. From the logs alone appears it's throwing up a placeholder quicker than before and then stuck with it, fresh boot no problem and by specifying the edid in the /usr/lib/firmware have no problem so sure seems to be something changed on edid inquiry/timeouts? Again any ideas or ways to gather added info much appreciated so we can get to the right place 🙏 Rest of 6.4 stellar work :)
Created attachment 182635 [details] journalctl log from wakeup on 6.3.5 (no problem) Uploading journalctl log from few days prior of same event (wakeup) where was not a problem (6.3.5 plasma), unfortunately no xorg log going back beyond last day but at least have this one for comparison Before could see powerdevil find the display after a moment, did notice the same placeholder screen creation just after 6.4.0 it seems like throwing an empty stack trace when trying to pull EDID info for some reason, also seeing logging like this beforehand which didn't prior kwin_xwl: Could not find a matching X RandR CRTC/output to set as primary for KWin::PlaceholderOutput(0x562fc7ee2990, name="Placeholder-1", geometry=QRect(0,0 1920x1080), scale=1)
(In reply to Craig from comment #0) > Wayland Plasma 6.4.0 on Arch Linux with nvidia card and display port. From > logs it seems like plasma is trying to read the screen before the EDID can > be read, causing a placeholder 1080 30hz screen to be created and then that > gets stuck as it's not giving enough time for the monitor to wakeup. KWin doesn't read the EDID, it just gets a list of resolutions from the kernel. This is definitely a GPU driver bug. When the display has the wrong resolution, please attach the output of drm_info, with that we can make sure.
Got similar issue on external display connected via USB-C DisplayPort after upgrading to plasma 6.4.0. The resolution change from 1920x1200 to 640x480. Got following logs after reconnecting. Intel tigerlake integrated gpu not nvidia though. ----- Jun 26 11:27:14 leo-x1c kwin_wayland[18660]: kwin_wayland_drm: Could not find edid for connector DrmConnector(id=344, gpu="/dev/dri/card1", name="DP-2", connection="Connected", countMode=5) Jun 26 11:27:16 leo-x1c kernel: usb usb2-port2: Cannot enable. Maybe the USB cable is bad? -----
(In reply to leodream2008 from comment #7) > Got similar issue on external display connected via USB-C DisplayPort after > upgrading to plasma 6.4.0. The resolution change from 1920x1200 to 640x480. > Got following logs after reconnecting. Intel tigerlake integrated gpu not > nvidia though. > ----- > Jun 26 11:27:14 leo-x1c kwin_wayland[18660]: kwin_wayland_drm: Could not > find edid for connector DrmConnector(id=344, gpu="/dev/dri/card1", > name="DP-2", connection="Connected", countMode=5) > Jun 26 11:27:16 leo-x1c kernel: usb usb2-port2: Cannot enable. Maybe the USB > cable is bad? > ----- That's definitely a driver bug too. You can report it at https://gitlab.freedesktop.org/drm/i915/kernel/-/issues
🐛🧹 ⚠️ This bug has been in NEEDSINFO status with no change for at least 15 days. Please provide the requested information, then set the bug status to REPORTED. If there is no change for at least 30 days, it will be automatically closed as RESOLVED WORKSFORME. For more information about our bug triaging procedures, please read https://community.kde.org/Guidelines_and_HOWTOs/Bug_triaging. Thank you for helping us make KDE software even better for everyone!
🐛🧹 This bug has been in NEEDSINFO status with no change for at least 30 days. Closing as RESOLVED WORKSFORME.