I have a HP Z4 Workstation with a Quadro P2200 under RedHat 8.
The machine is connected to 2 monitors (24-inch screen)
Our goal is to have 2 different X servers, one on each screen.
For that we have installed the last version of the NVIDIA package (available on the website Index of /compute/cuda/repos/rhel8/x86_64):
[root@IBL_03 ~]# rpm -qa | grep nvidia
On the tool nvidia-settings, we can configure the second screen to use a new X server.
But when we apply the modification and restart the computer, the second screen is in a strange state:
We can display a specific application but there is no background, no menu bar, and even the mouse is not displayed correctly (mouse is replaced by a “cross”)
Moreover after the installation of the NVIDIA driver, many errors appear at start-up.
ACPI Error: Aborting method _SB.WMIV.WMPV due to a previous error (AE_AML_OPERAND_TYPE) (20200925/psparse-531)
ACPI Error: Needed [Buffer/String/Package], found [Integer] 000000009d45f434
ACPI Error: AE_AML_OPERAND_TYPE, While resolving operants for [OpcodeName unavailable]
ACPI Error: Aborting method _SB.WMIV.WMPO due to a previous error (AE_AML_OPERAND_TYPE)
ACPI Error: Aborting method _SB.WMIV.WMPV due to a previous error (AE_AML_OPERAND_TYPE)
ACPI BIOS Error (bug) Attempt to CreateField of length zero (20200925/dsopcode-134)
Note: During the NVIDIA installation, the Nouveau driver are removed, and Wayland has been disabled.
We regularly use the $DISPLAY environment variable to know the current screen and to choose on which screen to display our applications.
Before redhat 8, we had the display “:0” (and for the first screen “:0.0” and for the second screen “:0.1”)
But now, by deactivating Wayland, we have only the display “:1”.
The display “:0” is not retrieved, and we no longer get the screen unit
nvidia-bug-report.log.gz (677.9 KB)