Laptop dual monitor/gpu intel/nvidia - LVDS black screen with mouse

Hello All,

I passed like one week searching to solve my problem of black screen on internal monitor on my laptop, under xfce 4.16/lightdm.
the external monitor on DP work fine connected to the nvidia 1660 Ti card, the internal display no chance.

I'm on FreeBSD 13 stable, I use the nvidia-driver 460.84 and dkms-mod for the intel gpu with kld_list="i915kms nvidia-modeset fusefs fuse coretemp cpuctl":
Code:
hald_enable="YES"
dbus_enable="YES"
linux_enable="YES"
slim_enable="YES"
lightdm_enable="YES"
powerd_enable="YES"
powerd_flags="-b adp -a max"
fusefs_enable="YES"
fuse_enable="YES"
devfs_system_ruleset="operator_usb"
nvidia_xorg_enable="YES"

Xrandr detect only one provider that is the nvidia.

I don't know what to do anymore. I have been trying so much with xorg.conf, so many tries.
I'm able to have one or the other screen used and working with Xorg depending of the xorg.conf setting but never the both screens at same time.

I have seen that in the xorg.0.log file, at the beginning of the start of xorg, the both gpu are detected and both display too but intel+internal display seems to be disabled after in the process starting.
I don't know if it is related with my problem but there is a warning : (WW) VGA arbiter: cannot open kernel arbiter, no multi-card support

How it is possible that the internal monitor is black but the mouse is shown and the mouse can pass from one to the other screen but no action possible on the internal screen?

If someone has a idea.

I have been searching in the forum a lot but it seems that not a lot of people use FreeBSD with the dual head/gpu on laptop configuration.


I would like to pass to FreeBSD for daily use but i need both displays for work.

Best Regards.
 

Attachments

What is the reason for not having the NVidia driver to control eDP-1-1 ?
Code:
Option "UseDisplayDevice" "DP-0, eDP-1-1"

Also does a minimal xorg configuration which only specify the BUS id of the NVidia card does not work ?
Code:
Section "Device"
	Identifier  "Card0"
	Driver      "nvidia"
	BusID       "PCI:1:0:0"
	Option "PrimaryGPU" "yes"    
EndSection
 
lightdm_enable="YES"
fusefs_enable="YES"
fuse_enable="YES"
nvidia_xorg_enable="YES"

Those don't do anything at all, remove them. They'll only add confusion.

OP says he is using lightdm, so maybe leave lightdm_enable="YES" in and remove slim_enable="YES"
 
it seems that the setup with two monitors and two gpu is not working on FreeBSD when it is nvidia+intel drivers : https://forums.FreeBSD.org/threads/...nd-intel-drivers-freebsd-12.76409/post-471630
I can have the intel gpu and internal display working or the nvidia gpu and external display working but not both at same time or cooperation between gpu to compose the global screen.

On Arch/Fedora, the GPU intel make the gateway to the nvidia gpu in same setup situation, but not here on FreeBSD :(
 
Good point. Enable one or the other, not both.
I use Slim for manage the login and Lightdm is a window manager, both are not incompatible, they are use for different purpose so no problem with lightdm_enable="YES" and slim_enable="YES"
 
I use Slim for manage the login and Lightdm is a window manager, both are not incompatible, they are use for different purpose so no problem with lightdm_enable="YES" and slim_enable="YES"
You are confusing a window manager with an Xorg display manager (a.k.a login manager). "LightDM is a free and open-source X display manager that aims to be lightweight, fast, extensible and multi-desktop."
 
I passed like one week searching to solve my problem of black screen on internal monitor on my laptop, under xfce 4.16/lightdm.
the external monitor on DP work fine connected to the nvidia 1660 Ti card, the internal display no chance.
What is the GPU type connetcted to internal monitor? Is it the same Nvidia, or some other?
Also, remove the xorg.conf. Let the Xorg detect the configuration. What happens?

SLIM needs a configuration file. Do you have it? Personally I like x11/sddm. This is fully automatic and detects all window managers. Give it a try.
 
Hello, i have been testing with some changes as it has been mentioned previously, replacing lightdm by sddm and removing all X.org config file about serverlayout/monitor/device/screen.
But nothing change. I can't have the both monitors working correctly.
Only one provider with xrandr is detected.

In this last configuration, the intel is detected and LVDS monitor used but the nvidia/external monitor couple is not enabled.

I can't use it as this for daily work :(
 
(WW) VGA arbiter: cannot open kernel arbiter, no multi-card support
I don't think it IS supported even under Windows. It's either the one GPU or the other AFAIK. Either Intel OR nvidia.
How it is possible that the internal monitor is black but the mouse is shown and the mouse can pass from one to the other screen but no action possible on the internal screen?
I'd say this means that both displays work. I don't see how it could be otherwise.
Now. Most things in your xorg.conf is garbage (not needed at all). These days you just keep minimalist $section.conf files in the /usr/local/etc/X11/xorg.conf.d directory. Mine has these:
fonts.conf:
Code:
Section "Files"
     FontPath "/usr/local/share/fonts/100dpi"
     FontPath "/usr/local/share/fonts/75dpi"
     FontPath "/usr/local/share/fonts/adobe-cmaps"
     FontPath "/usr/local/share/fonts/artwiz-ru"
   .... (the rest of them) ...
EndSection
Then driver.conf:
Code:
Section "Device"
    Option      "Monitor-HDMI1"  "HDMI1"
    Option      "Monitor-HDMI3"  "HDMI3"
    Identifier  "Card0"
    Driver      "nvidia"
EndSection
And monitor.conf:
Code:
Section "Monitor"
    Identifier  "HDMI1"
    Option      "RightOf"  "HDMI3"
EndSection
That was (in my case) enough to make it place HDMI3 left of HDMI1. But I took those settings from /var/log/Xorg.0.log. Yours, evidently, only shows ONE monitor connected... You will notice that most NVIDIA settings are configured & decided by the driver itself. So, no need (mostly) to put any of them explicitly into any *conf file. You'll find various "documentation" on nvidia configuration stuff, but none of that needs be used in any *conf files, as per my recent experience.

...So I wonder, when you boot your machine, where does your boot message appear? On your builtin monitor or your outer one?
Your Xorg.0.log shows 3 outputs detected: DFP-0, DFP-1 and DFP-2. Now I wonder if this has anything to do with BIOS display settings that would allow to use 2 outputs at least in one of the possible configurations...
 
The config actually looks fine to me. It could be trimmed a bit, but it needs those 2 Device (nvidia and modesetting) sections: Nvidia is connected to an external display, Intel is connected to the laptop's screen. This a bog-standard wiring and, no, it can't be changed. Forget about this stupid advice.

Now, does Xorg actually support displaying a single desktop surface across two different GPUs? I have no idea. Fedora might be using Wayland (well, Gnome, actually) for that.
 
… minimalist $section.conf files in the /usr/local/etc/X11/xorg.conf.d directory. Mine has these:

fonts.conf: …

Maybe no longer necessary to specify font paths and so on. Various package messages are outdated.

My /usr/local/etc/X11/xorg.conf.d is empty.

This morning I found and set aside an old /usr/local/etc/xorg.conf.

Five files here, I don't recall creating or editing them, I'm inclined to leave them:

Code:
% ls -hl /usr/local/share/X11/xorg.conf.d
total 1
-rw-r--r--  1 root  wheel   1.1K 21 Feb  2021 10-evdev.conf
-rw-r--r--  1 root  wheel   1.3K  3 Jul 02:32 10-quirks.conf
-rw-r--r--  1 root  wheel    92B 11 Aug 01:54 10-radeon.conf
-rw-r--r--  1 root  wheel   152B  3 Jul 02:32 20-evdev-kbd.conf
-rw-r--r--  1 root  wheel   1.4K 21 Feb  2021 40-libinput.conf
%

(radeonkms here.)
 
i have done the test, without any file for X.org or with, same problem, only one screen with fully composition.
With the Nvidia driver, the LDVS (internal) monitor show only the mouse moving normally on black screen, no menu or contextual menu on this screen with the mouse. It seems to be like the composition (forcefullcomposition) is not done.
In nvidia-settings, only one screen is shown.

I'm desesperate !
 
… does Xorg actually support displaying a single desktop surface across two different GPUs? …

I don't know about X.org, but here's what might be someone finding comparable difficulty with a dual-GPU extended desktop on Linux: <https://askubuntu.com/q/1301579/25036>

In nvidia-settings, only one screen is shown.

What make and model is the computer?

AUO 409d is found in (for example) an ASUSTek ROG Strix G731GW_G731GW.

There is no solution if the screen is an internal laptop one, wired on Intel GPU without switch possibility.
 
Last edited:
I don't know about X.org, but here's what might be someone finding comparable difficulty with a dual-GPU extended desktop on Linux: <https://askubuntu.com/q/1301579/25036>



What make and model is the computer?

AUO 409d is found in (for example) an ASUSTek ROG Strix G731GW_G731GW.
Hello, i use actually a MSI GP75 Leopard 9SD :https://www.msi.com/Laptop/GP75-Leopard-9SD/Specification

I use EndeavourOS (Arch Linux based) without any problem with X.org and Xfce with dual monitor using the joined config file.
I have been using before Arch Linux and it was the same config for X.org server.

i used the exact config purpose (nvidia only section ) in : https://wiki.archlinux.org/title/NVIDIA_Optimus#Use_Intel_graphics_only

I don't understand why it is not possible with BSD :(
 

Attachments

I like the multi-coloured keyboard.
1024.png
 
… I use EndeavourOS (Arch Linux based) without any problem with X.org and Xfce with dual monitor using the joined config file.

I have been using before Arch Linux and it was the same config for X.org server. …

Do you still have the configuration file? (Is that what's attached, above?)
 

Attachments

Back
Top