Trouble with EDID DisplayID and an ultrawide monitor

Hi,

I got an ultrawide monitor, capable of DQHD (5120x1440) @ 240 Hz, a Samsung Odyssey G9. With Xorg in autoconfig mode and driver version 455.45.01, the monitor reverts to 3840x1080 @ 120Hz. I can force it to 5120x1440, 120Hz with
nvidia-settings (or Xorg settings), but when the display returns from
powersave it reverts. GPU is a GeForce RTX 2060 SUPER.

As far as I can tell (and bolstered in my supposition by edid-decode’s author), the way the driver handles EDID data is likely the culprit. Quote:

The problem is likely that the preferred video timing is declared in the
DisplayID Extension Block, and not in the first Detailed Timing Descriptor (DTD)
in Block 0, as is usually the case. It can’t use a DTD since DTDs are limited to
4095x4095.

And I suspect that the nvidia driver doesn’t handle that correctly.

I tried overriding EDID with modelines + UseEDID False in Xorg config, but I am unsuccessful so far - the resolution drops even lower when I try that. I’m wildly guessing that kernel modesetting trips me up here.

If you need details, you may find my monitor’s EDID info here: https://linux-hardware.org/index.php?probe=2745a37c0a&log=edid

Please, Nvidia devs, let me use my monitor to its full potential!

Hi,
did you try to use nvidia-settings --load-config-only?
https://download.nvidia.com/XFree86/Linux-x86/1.0-7174/nvidia-settings-user-guide.txt

you could eventually if that doesn’t help with waking from sleep put that into /etc/pm/sleep.d.

Well, the resetting problem is annoying but I can work around that with XFCE display settings profiles. Not ideal, but it works.

I can’t work around it not using 240Hz though. According to vsynctester.com in Chrome, it’s using 120Hz.

I’d report that to: linux-bugs@nvidia.com
Make sure to run nvidia-bug-report.sh (as root) and include that in your report.

Thanks for reporting this. I filed internal tracking bug 3198533. Please attach a bug report log or send one to linux-bugs like you suggested and I’ll be sure to attach it to the internal bug tracker.

Thanks, I sent the requested data to the given email address and included the bug number you specified.

For the part about the resolution reverting after power save mode: it’s possible that your monitor is generating a hotplug event after powersave mode, which could be triggering your desktop environment to ask the driver to change the mode to the one it thinks is correct. If possible, can you please try changing the mode through your desktop environment’s control panel instead of nvidia-settings?

I tried reproducing this with fluxbox on a second Xorg display, I couldn’t reproduce the resolution reverting on resume from powersave, so I agree that XFCE is likely to blame for the reset on resume. I managed beating XFCE into submission by forcing it to apply a display settings profile every time it detects a new display.

Find screenshots of the display settings attached.