You are not logged in.
Pages: 1
Hi!
For about 3-4 days i now experienced a weird issue with the naming of my displays.
I use 2 monitors and in my autostart file for openbox i use the following line (worked smoothly for months until now):
xrandr --output DisplayPort-0 --primary --mode 2560x1440 --rate 143.86 --output DVI-D-0 --mode 1920x1200 --left-of DisplayPort-0 --rotate leftBut now this doesnt work anymore, i have to adjust the naming at every boot.
Monitor 1 changes between "DisplayPort-0" and "DP-1"
Monitor 2 changes between "DVI-D-0" and "DVI-D-1"
There are no changes on the monitors, the GPU or the cables.
Thanks for all hints and help!
Offline
Optimus system?
Yo're probably randomly selecting GPU or at least driver the X11 server runs on.
Please post at least one xorg log.
Online
Hi,
nope, its a desktop PC without iGPU, so only the dGPU is present
Here is the Xorg.0.log:
https://pastebin.com/PcJWxh8z
Offline
Ok, then we'll need to see a log where the otput name is "DP-1" but my hunch would be that the GPU isn't fully initialized and you somehow end up w/ the modesetting driver.
In this case try https://wiki.archlinux.org/index.php/Ke … _KMS_start (though the usual symptom is that X11 simply doesn't start)
Online
Alright, then i'm gonna wait for next shuffling of the displayname to DP-1 and post the log. The last 2 boots were fine though.
Offline
Hello.
I have the same situation (2 screens and reliability on some configuration script; more to say, xfce4 switches monitors left to right and doesn't shows my beautiful wallpapers). To get it I don't even need to reboot - enough to logout and then login.
I'm using xfce4 with lightdm and lightdm-gtk-greeter.
Here you can see two logs - one for bad case (modeset), another for normal (amdgpu). I removed timestamps so you can use any comparing utility.
https://pastebin.com/ax1bERxn - bad
https://pastebin.com/mM8TMNLN - good
Check line 49 (both files), you will see that the bad log misses this line
(II) xfree86: Adding drm device (/dev/dri/card0)And around line 57 (both files)
The bad
(==) Matched ati as autoconfigured driver 0
(==) Matched modesetting as autoconfigured driver 1
(==) Matched fbdev as autoconfigured driver 2
(==) Matched vesa as autoconfigured driver 3The good
(II) Applying OutputClass "AMDgpu" to /dev/dri/card0
loading driver: amdgpu
(==) Matched amdgpu as autoconfigured driver 0
(==) Matched ati as autoconfigured driver 1
(==) Matched modesetting as autoconfigured driver 2
(==) Matched fbdev as autoconfigured driver 3
(==) Matched vesa as autoconfigured driver 4And all way down on the bad log we have modest instead of AMDGPU for the good log.
Thanks for your attention.
Offline
In this case try https://wiki.archlinux.org/index.php/Ke … _KMS_start (though the usual symptom is that X11 simply doesn't start)
Online
myself wrote:In this case try https://wiki.archlinux.org/index.php/Ke … _KMS_start (though the usual symptom is that X11 simply doesn't start)
Yes, I just was in process of making it (I tried to put the module into HOOKS...). Thank you.
And seems like it works great, I even got fullHD resolution for systemd messages during booting.
Yet, seems like this is some bug anyway and this isn't the best solution. Feel free to ask me for some logs or experiments.
Last edited by furioness (2021-01-10 16:21:37)
Offline
It's not a "bug" but a race condition - the X11 server starts before the GPU is fully initialized.
Moving the GPU driver into the initramfs circumvents that.
The alternative is to stall the lauch of the X11 server until the GPU is set up (afaiu lightdm's https://wiki.archlinux.org/index.php/Li … TTY_output does that)
The systemd graphical.target could probably do the same but in any event you'll slow down the boot process.
Online
Hej!
Atfer quite some time it happened again - here is the fresh xorg.0.log from this very boot: https://pastebin.com/8uZF3MiQ
I'm gonna try the early KMS start too, but wanted to give another data point before doing so.
Offline
Pages: 1