What is Kernel parameter amdgpu.dc=0 doing and why do I suddenly need it?

Hi everyone,

after one of the last Stable updates ( the small one that included the new version of gdm ) I restarted and after seeing display output during early boot, I got stranded in a black screen without a cursor.

After trying several different things I switched my desktops GPU, an AMD RX550, with a different GPU an AMD 5700XT. With the 5700 I was able to boot into Manjaro without problems, switching them back resulted in the issue I just described before.

In the end it turned out that I can use my RX550 only if I apply the amdgpu.dc=0 parameter to GRUB_CMDLINE_LINUX_DEFAULT.

I was however unable to find out why I do need that now all of a sudden and what exactly it does.

If anyone can point me in the right direction or give an explanation I would be grateful!

Thanks in advance!

1 Like

Hello,

More about it here
https://www.kernel.org/doc/html/latest/gpu/amdgpu-dc.html
Probably you switched from 5.8 to 5.9 kernel … or some code got changed in both.

It’s a really new card that should be supported by the new DC layer.

It’s most likely a bug in the kernel, in the amd DC part.

Think of DC (previously called DAL) as build in translating layer wich enables AMD to share lots of code from the Windows driver.

You putting DC=0, you are disabling that layer because the switch that decides if DC supported card is present is turning it on.

I did not switch kernels and neither updated it at first. I updated exactly two packages gdm and another one. After a reboot I was seeing the black screen. After I booted with the working 5700 I did another update that provided new versions of the 5.8 and 5.4 kernels. While I am running the 5.8 kernel I also tested that the same issue persists in kernel 5.4,

Edit: Since I saw the 5.9 Kernel is out I installed and tried that one, the problem unfortunately persists.

Thank you for the explanation. I hoped kernel version 5.9 might fix this then, but it did not help.

Does seem a bit unlikely to me why such an ancient bug should haunt me now. Also people there describe systems freezing, but my computer runs just fine only outputting no signal. When I start without amdgpu.dc=0, I can still ssh into my machine.

VGA cable? What do you mean? I do not have a vga port!

Hmm, then it’s ptobably newer bug. I have no clue then.

You can report it :slight_smile: Or if you have a working kernel and want to learn something new, you can do the tedious Kernel BISECT, to find the offending commit.

Then you would have very hight chance to have it fixed, because the bisect would tell you Who broke it!


Other option is to use older kernel that works with your card and wait until it get fixed :slight_smile:

A friend just send me this 209721 – No output after AMDGPU loads.
It seems this issue will be fixed soon!

1 Like

This topic was automatically closed 15 days after the last reply. New replies are no longer allowed.