Black screen with 3 cards on 10 LED Panels

Graphics / Linux Linux
1

We are running a visualization lab with 10 large LED panels. The system is a dell workstation with 3 RTX 6000 Cards, Redhat Linux 9x. The problem is that whenever we switch off the screens and come back later we can’t get the login screen back. All screens remain black until we reboot.

I have disabled nvidia drm in /etc/modprob.d/nvidia-modeset.conf

options nvidia-drm modeset=0

Also Wayland is disabled in /etc/gdm/custom.conf

So far, no luck. Any suggestions?

IMG_8230
IMG_82301920×606 331 KB

2

I assume you can set up ssh access to the setup? After you repro the issue, would you mind running nvidia-bug-report.sh to collect logs we can look at?

3

Uploaded the bug-report. Thank you for your help!

nvidia-bug-report.log.gz (332.5 KB)

4

Hi @axelha ,

Thank you for reporting the bug. I have filed Bug # 5707413 internally to track this issue.

  1. Can you please provide the make and model of the monitor that is in use?
  2. Are you able to observe the same issue with fewer monitors - 1, 2 or 3 for example?
  3. Do you know if this is a driver regression. Is there an older NVIDIA driver that did not hit this issue?
5

The monitors are a Planar System:

[ 22857.408] (–) NVIDIA(GPU-0): Planar Systems, Inc. Planar VC (DFP-4): connected
[ 22857.408] (–) NVIDIA(GPU-0): Planar Systems, Inc. Planar VC (DFP-4): Internal DisplayPort
[ 22857.408] (–) NVIDIA(GPU-0): Planar Systems, Inc. Planar VC (DFP-4): 2670.0 MHz maximum pixel clock

There are three computers attached to it:

One Dell Precision running Windows 11 = No issues

One Mac Pro = No issues

One Dell Precision Running RHEL 9 = That’s the one with the issue. I thing the relavant information is here (from dmesg):

? asm_sysvec_apic_timer_interrupt+0x16/0x20
[79464.126600] ? _ZN11DisplayPort9GroupImpl7expiredEPKv+0x120/0x120 [nvidia_modeset]
[79464.126633] ? _ZN11DisplayPort9GroupImpl11enumDevicesEPNS_6DeviceE+0x1a/0x70 [nvidia_modeset]
[79464.126664] ? _ZN11DisplayPort13ConnectorImpl15notifyDetachEndEb+0xda/0x330 [nvidia_modeset]
[79464.126705] ? _ZN11DisplayPort13ConnectorImpl13dpPostModesetEv+0x81/0x90 [nvidia_modeset]
[79464.126748] ? nvDPPostSetMode+0x73/0x200 [nvidia_modeset]
[79464.126776] ? KickoffModesetUpdateState+0x1a8/0x270 [nvidia_modeset]
[79464.126817] ? nvSetDispModeEvo+0x3ae4/0x4080 [nvidia_modeset]

When this happens Xorg goes into a soft lockup on one cpu core:

watchdog: BUG: soft lockup - CPU#7 stuck for 51789s! [Xorg:19201]

And no, no older driver. We have tried several and it occurs with all of them.