RL 8.10, 570.124.060, RTX A4000, Barco projector causes Xid 16

Graphics / Linux Linux
1

It is a complex setup, so I will incrementally refine the issue description.

We have a Dell Inc. Precision 7960 Rack/05H0JD, BIOS 2.4.4 09/27/2024 with a RTX A4000 with four DP outputs and want to display on two highend Barco projectors UDM 4K15. We run RockyLinux 8.10 with the 570.124.060 drivers. It is a fresh install. We use a Barco E2 switcher between inputs and the output to the projectors.

The good news is that the same setup works well with an older Dell Precision with Nvidia GPU, including stereo output. So we can assume the cabling and the hardware is ok. I can provide details.

Furthermore, the new system also works but only when booting without DP cables connected to the machine, and only connecting the DP cables after the X server is already running. This is the state the attached nvidia bug report collected.

However, when booting with the cables connected, the driver keeps generating Xid 16 and nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000c67e:6:0:1230, and does not provide a signal to the DP outputs. Let me provide a full bug report for this state as well.

The system also works normally when two regular Dell monitors are connected to the same DP outputs, instead of the E2 (to the projectors).

We have tried the customEDID option but without success sofar. I saw that is all recorded in the logs.

Additional details include matrox motherboard graphics (not recognized by X, nor should it). It can be disabled in the BIOS but that does not seem to make a difference in any of our testing.

Any feedback or ideas welcome,

-Andreas

nvidia-bug-report.log.gz (486.3 KB)

Collected with ssh login.

2

And the bug report in the error state with Xid 16 messages, after booting with connected DP cables:

nvidia-bug-report_xid16.log.gz (537.0 KB)

Collected with ssh login.

3

Details on the old, working system. Unfortunately, I currently do not have root access, so no bug report.

Dell Inc. Precision Rack 7910/01J90F, BIOS 2.1.6 06/02/2016
Linux version 2.6.32-642.3.1.el6.centos.plus.x86_64 (mockbuild@worker1.bsys.centos.org) (gcc version 4.4.7 20120313 (Red Hat 4.4.7-16) (GCC) ) #1 SMP Tue Jul 12 18:28:07 UTC 2016
Command line: ro root=/dev/mapper/vg_root-lv_root rd_NO_LUKS rd_LVM_LV=vg_root/lv_root LANG=en_US.UTF-8 rd_NO_MD SYSFONT=latarcyrheb-sun16 crashkernel=auto KEYBOARDTYPE=pc KEYTABLE=us vga=0x31B rd_NO
_DM rdblacklist=nouveau biosdevname=0
Quadro M6000
Driver Version: 361.28

xorg.conf:

Section "ServerLayout"
    Identifier     "Layout0"
    Screen      0  "Screen0" 0 0
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"
    Option         "Xinerama" "0"
EndSection

Section "Files"
    FontPath        "/usr/share/fonts/default/Type1"
EndSection

Section "InputDevice"

    # generated from default
    Identifier     "Mouse0"
    Driver         "mouse"
    Option         "Protocol" "auto"
    Option         "Device" "/dev/input/mice"
    Option         "Emulate3Buttons" "no"
    Option         "ZAxisMapping" "4 5"
EndSection

Section "InputDevice"

    # generated from data in "/etc/sysconfig/keyboard"
    Identifier     "Keyboard0"
    Driver         "kbd"
    Option         "XkbLayout" "us"
    Option         "XkbModel" "pc105"
EndSection

Section "Monitor"
    Identifier     "Monitor0"
    VendorName     "Unknown"
    ModelName      "Unknown"
    HorizSync       28.0 - 33.0
    VertRefresh     43.0 - 72.0
    Option         "DPMS"
EndSection

Section "Device"
    Identifier     "Device0"
    Driver         "nvidia"
    VendorName     "NVIDIA Corporation"
    BoardName      "quadro M6000"
    BusID          "PCI:131:0:0"
    Option         "Stereo" "3"
    Option         "AllowDFPStereo" "1"
    Option         "InbandStereoSignaling" "1"
#    Option         "TwinView" "1"
EndSection

Section "Screen"
    Identifier     "Screen0"
    Device         "Device0"
    Monitor        "Monitor0"
    DefaultDepth    24
#    Option         "Stereo" "3"
#    Option         "AllowDFPStereo" "1"
#    Option         "TwinView" "1"
    Option         "nvidiaXineramaInfoOrder" "DFP-3"
    Option         "metamodes" "2560x1600 +2318+0, 2560x1600 +0+0"
#    Option         "metamodes" "GPU-729d4627-d0ab-3592-3286-4590ee346129.DP-0: 2560x1600 +2240+0, GPU-729d4627-d0ab-3592-3286-4590ee346129.DP-6: 2560x1600 +0+0"
    Option         "BaseMosaic" "off"
    Option         "nvidiaXineramaInfo" "False" 
    SubSection     "Display"
        Depth       24
    EndSubSection
EndSection

Section "Extensions"
    Option         "Composite" "Disable"
EndSection
4

A RL 8.10 system update by
$ dnf update
updated the driver to v. 575.51.03 (‘latest-dkms’) and the kernel slightly to 4.18.0-553.51.1.el8_10.x86_64 (from 4.18.0-553.50)

With v. 575.51.03 the system still does not produce output to the connected E2 after boot but at least does not generate the Xid 16 and Idling display engine errors.

Then unplugging the cables, stopping gdm (and Xorg), starting gdm (and Xorg) without cables, and then replugging the cables works for Xorg to properly detect modes and generate output.

Dell blesses driver v. 510 for the machine, so I will try that next.

5

The Dell blessed version v. 510 did not work any better.

In the mean time I installed Windows 11 Pro to and the latest Nvidia desktop driver to test but could not get very far. Only the Matrox motherboard graphics works sofar. Somehow the Barco E2 switcher really seems to confuse the machine. Another Windows machine with an old card works fine with the E2.