Return to

2200G and 2400G non-bootable under Linux


I upgraded from an fm2 APU to 2400G with ASRock Mini-ITX Motherboard - AB350 GAMING-ITX/AC but it seems it’s not supported at all under Linux.

I am using Fedora Server 26 with no Desktop but standalone Kodi (disabled now). With 4.15 kernel system freezes completely during boot, no SSH possible. 4.14 does boot but with fixed low resolution, odd refresh rate and periodic fatal freezes ranging from minutes to hours after boot.

This is the exact issue described here:

I bough this for 4k decoding support and now I have no video at all and an unstable server, very sad :frowning:

What can be done to fix this if anything? And more critically (this being primarily a Storage Server) what can be done to prevent freezes even with no video? Anyone had better luck with 2000G series APUs?


Since its new, you might have to run with rawhide (mainline), until it makes its way into the stable releases.


Try a 4.16 release candidate?


Yeah, I suspect this to be the case too. Or he can just wait a week for it to land in latest stable.


According to this post 4.16 and 4.17 have exact same issue.


What would that change exactly? If it’s just the kernel then it won’t help as even 4.17 kernel seems to be broken.


There is no 4.17 kernel yet. Its up to 14.16-rc4

Hard to say, I don’t read the changelogs. But if there was a bug related to this cpu and booting it would be fixed.

Otherwise is just a standard vanilla fedora booting issue. Troubleshoot as normal.


By 4.17 I mean what was mentioned in the article I linked - “Alex Deucher’s 4.17 work-in-progress AMDGPU development branch”.

Regarding Mainline, if the thought behind this fixing is the kernel then I doubt it will fix it since it has been tested already. Perhaps it also includes new drivers or firmware? Not sure where else the root of the bug may lie apart from the kernel.

It was booting just fine before so it’s Hardware related and since others are reporting the exact same issue with exact same hardware I don’t see it being an unrelated issue.


In this case, it would be best to roll back the kernel and stay with it until a newer version has resolved these issues.


Well the 4.14 kernel is not stable, it freezes after minutes or hours so rolling back isn’t a fix either, maybe an even older kernel would work but I haven’t tried yet. Trying nomodset for now, and I might upgrade to fedora 27 while I am at it just in case it’s more stable…


Since 4.14 wasn’t stable for you, then you might have to roll back to next lts kernel.

longterm: 4.9.86


I’m not much of a follower of fedora but I thought you were on the latest version. So yeah that might fix your issue.


Maybe but 2000G sries of APUs are not supported even in 4.14 so it might not work at all with an even older kernel, it’s likely not a bug (in older kernels) but rather no support for then non-existent hardware.

I did try fedora 27 live boot cd and it had similar issue so I doubt it. I have the latest packages, I don’t think 26 vs 27 changes the core elements but will try later.


Yeah, I mean if there isn’t good support OOB on mainline then there’s nothing you can really do.

Keep in mind the live boot isn’t a daily snapshot. There is a possibility that if it works enough on a livecd to install the new version, then updating could fix the rest.


Seems like something AMD can easily get sued over, I mean CPU’s are assumed to be supported in all updated OSs and nowhere did they say it’s not supported in linux…

Yeah live CD had an older kernel too, will try updating shortly… I did see people saying you should do a clean install with any new hardware, seems unreasonable to me especially with a server which would take maybe days to re config again…


And this is not a CPU problem either. It’s a GPU and PCI-e issue. That requires redesign of several aspects of the still growing AMDGPU stack.

You can wait for it to get fixed, or contact the amdgpu stack kernel developers and offer to participate in testing.

AMD is not beholden to Linux or vice versa under any legal obligations regarding new hardware support.
This is an inherent risk you take when buying brand new hardware for use with a OS like Linux.

So if you wish, you may calm down and stop bickering.


GPU is integrated so it’s the same product. I never said AMD had to support linux, please stop making assumptions. While AMD does not have to support any OS I think not saying it does not support it is at least close to false advertisement. I don’t know why you think your opinion is more valid than mine, or do you have some qualifications in consumer law? I mean if a cpu was released that did not work on windows with no mention of it that would be unreasonable and you would be entitled to a refund and have a case for damages if any. Same way that if you buy an apple it is reasonable to assume it’s safe to eat unless explicitly specified. I am also perfectly calm, it’s you who seem to be unable to have a 2 sided discussion.


I think you have said enough.


I love the zero troubleshooting on this thread :smiley:

first port of call always is get access and run fpaste --sysinfo

When your talking about freezes on 4.14… what freezes? the UI or the system its self?

next is always more logs journalctl -xb -p err | fpaste for example

If you want to try out 4.16 use the rawhide kernel from here

If you want to try updated everything graphics related, you may try this copr repo


I think this would be more akin to saying that AMD doesn’t support OSX too, when really it’s Apple at fault here. It’s the responsibility of Linux devs to support AMD hardware. If they didn’t get early samples to people who make that happen, that’s an issue, but whether that happened or not? I have no idea.

I’m waiting for better support too, FWIW.

I do hope you get things figured out, and Linux as a whole moves forward quickly (it tends to do so, I expect it won’t be a long wait).


AMD does support Linux they have their own devs on it among others, what they don’t have a tendency to do since their move to open source is release finished source code before the product is shipped.