Drivers & Software

[thanosstr]

Hello everyone,

I installed the AMD GPU driver (amdgpu-pro-18.50-725072-rhel-7.6) in Centos 7.6

(kernel: 3.10.0-957.el7.x86_64) for this GPU model (AMD Vega Series X 64).

The installation was based on the guidelines in the documentation and everything went smoothly. 

The problem is that when I boot the machine I cannot enter the graphical interface where the usernames appear to login.

It looks like it stuck somewhere. However, ssh connection works and OpenCL also works. Then I logged in a terminal and

checked that the display manager is active. When I restart the service it blinks to a black screen and returns to the same window.

However in the dmesg report, the following message appears: 

“gnome-session-c[8235]: segfault at 10 ip 00007f89a31e84e2 sp 00007ffe2b26de80 error 4 in libdrm_amdgpu.so.1.0.0[7f89a31e1000+b000]"

For the record, this message and the same behaviour remains even when the GPU is not plugged in the machine.

I also tried startx command and reports:

"xinit: connection to X server lost".

Can someone give a hint for this problem?

[jdrobinson314]

Try a different login manager, something you can easily access from the repos.

Also, try to alt + f over 5o a terminal and see if you can login via a normal shell.

If you can, then you can launch another login manager and at least have that much.

[thanosstr]

Yes, this fixed the problem and I can log in the graphic environment. I installed lightdm as the display manager and then rebooted the machine. The segfault has gone! It seems like the segfault was caused in the gnome display manager. After all what matters is that I am able to move on. Thanks