Ubuntu 24.04 Wayland RTX 5070Ti 580.82.09 drive doesn't initialise daisy chained monitors

nvidia-bug-report.log.gz (420.6 KB)

I am daisy chaining two monitors using DP port on RTX5070Ti. After Bootup and login only second screen is functional. First screen remains black. Dmesg logs say
[ 18.825431] [drm:nv_drm_dev_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000100] Failed to add connector for NvKmsKapiDisplay 0x00000800
[ 28.913577] [drm:nv_drm_dev_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000100] Failed to get dynamic displays
[ 39.017579] [drm:nv_drm_handle_dynamic_display_connected [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000100] Failed to add encoder for NvKmsKapiDisplay 0x00000001
I have set drm modeset 1. However, unplugging and changing DP port on GPU fixes the issue.
Bug report after changing port

nvidia-bug-report.log.gz (516.3 KB)

Hi @swarankar.ani24

Thank you for highlighting issue, I have filed a bug 5649576 internally for tracking purpose.

Shall get back to you if required any additional information.

@swarankar.ani24 it looks like your GPU was failing to initialize correctly.

2025-10-26T13:15:28.146151-04:00 arthavpc kernel: NVRM: The NVIDIA GPU 0000:01:00.0 (PCI ID: 10de:2c05)
2025-10-26T13:15:28.146153-04:00 arthavpc kernel: NVRM: installed in this system requires use of the NVIDIA open kernel modules.
2025-10-26T13:15:28.146153-04:00 arthavpc kernel: NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x22:0x56:886)
2025-10-26T13:15:28.146154-04:00 arthavpc kernel: NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0

Looking earlier in the logs, it looks like the nouveau kernel module is being loaded and interfering with the initialization of the out-of-tree NVIDIA kernel driver, leading to the experience you are seeing.

[    6.540771] nouveau 0000:01:00.0: unknown chipset (1b3000a1)

But I have blacklist nouveau

I have blacklist-nouveau.conf in /etc/modprobe.d

blacklist nouveau
options nouveau modeset=0

Could you get rid of the second line in blacklist-nouveau.conf and try rebuilding your initramfs on your distribution? According to the logs, nouveau seems to be loading.

I’m seeing the same problem with daisy chained monitors on EndeavorOS (Arch) with an RTX 3070 and have never had nouveau installed.

From dmesg:

[   17.394678] [drm:nv_drm_dev_load [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000a00] Failed to add connector for NvKmsKapiDisplay 0x00001000
[   27.482679] [drm:nv_drm_dev_load [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000a00] Failed to get dynamic displays
[   27.544844] [drm] Initialized nvidia-drm 0.0.0 for 0000:0a:00.0 on minor 1
[   27.561365] Console: switching to colour dummy device 80x25
[   27.569591] nvidia 0000:0a:00.0: vgaarb: deactivate vga console
[   27.731716] fbcon: nvidia-drmdrmfb (fb0) is primary device
[   27.793117] Console: switching to colour frame buffer device 240x67
[   30.818682] [drm:nv_drm_atomic_commit [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000a00] Flip event timeout on head 0
[   30.820492] nvidia 0000:0a:00.0: [drm] fb0: nvidia-drmdrmfb frame buffer device

Funnily enough, switching to a different DP port works until I need to reboot.

When I live boot from a USB with 570.124.04 version drivers I have no issues with my setup. Reading similar reports of the same issue elsewhere, it appears to have been an issue since at least 575 drivers:

  1. Driver v575, v580: second external display missing and NvKmsKapiDisplay errors · Issue #879 · NVIDIA/open-gpu-kernel-modules · GitHub
  2. Reddit - The heart of the internet

I tried doing that. Unfortunately, I couldn’t even see any login screen post boot. Just black screen.
Above observation when booted with Display port MST on.

Any indication as to when this bug will be resolved?

Hi @shay.brynes

Could you please share nvidia bug report from repro state when connected displays in daisy chained configuration.

Hi @swarankar.ani24

Could you please try adding “nvidia-drm.modeset=1” kernel parameter and share bug report if issue still persists.

Hi @amrits ,

Thanks for getting back to me. I’m not sure what the “repro state” is (reproduced state?), but I’ve generated two bug reports and attached them here.

  • In the “reproduced state” the first monitor in the daisy chain receives no display port output.
  • In the “corrected state” I moved the display port connection to a different slot on the RTX3070 at the 5 minute mark, at that point the first display in the daisy chain started working again.

I did have a look through the open source kernel and noticed this commit, which is in the area reported in the dmesg log. Could this be the cause?

The commit changes “kernel-open/nvidia-drm/nvidia-drm-connector.c” between the versions described here: Driver 575, 580, 590: second external display missing and NvKmsKapiDisplay errors · Issue #879 · NVIDIA/open-gpu-kernel-modules · GitHub

Many thanks,

Shay

nvidia-bug-report.log.gz (691.9 KB)

nvidia-bug-report.log_reproduced_state.gz (682.0 KB)

I believe nvidia drm modeset was set to 1 in the initial bug reports I added but I can try and attach again. The issue is still there on driver version 580.95.05

Same propblem here, 2 monitors connected via MST to DisplayPort.
Fedora 53, latest kernel, nvidia version: 580.119.02 with Wayland

the connections are computer → Monitor-1 → Monitor-2

The blank monitor is Monitor-1

My workaround is to power off Monitor-2 when booting and turn it on later, when the gdm login screen appears. Then the Monitor-1 works too.

Same setup works with Windows 11 (dualboot) ithout problem, so it’s not a cable/monitor problem.

I am still not able to duplicate issue in our lab, shall continue to run tests on other configs and update.

Can confirm the issue, got the same problem. GPU → Monitor 1 (Black) → Monitor 2 (gets Images). I need to cut off the connection between Monitor 1 and Monitor 2 while booting to make sure Monitor 1 gets Images. After then, I can replug the DP Cable to Monitor 2 and both Monitors get images (until next reboot). Using Ubuntu latest kernel with Wayland.

Unplugging the cable while the OS Is running on Monitor 2 doesn’t fix the issue on Monitor 1, as it doesn’t get detected in any way (until I use my manual fix described above).

Also, it is a Nvidia-driver issue, as using xserver-xorg-video-nouveau fixes the issue.

While booting, I’m also getting the Ubuntu loading screen with the xorg-driver and with Nvidia not. I need to wait for a few seconds (after I wiggled my mouse and pressed some keys on my keyboard) until Monitor 2 gets some images with Nvidia (when on login screen).

What kind of things are you trying? Seems like quite a few reports of this issue coming through now so not sure what else to suggest.

My first monitor is a Dell U2725QE and the downstream is a Samsung U28E590D both are 4k but the first has an integrated USB-C hub.

Upgraded to nvidia open driver 590.48.01 and kernel 6.18.2-arch2-1.
I’m seeing the same issue (as expected). I’ve attached the result of nvidia-bug-report.sh for the reproduced state again here.

nvidia-bug-report.log.gz (669.8 KB)

Hi All,

I have been able to duplicate issue locally.

Shall share further updates when available from engineering team.

How are the team getting on with this one @amrits - are we going to get a fox for this one in the next release.

I have a similar problem on X11. Is there any update on this?