Re: [REGRESSION] amdgpu error -22 first encountered in kernel 6.19, not fixed in 7.0

[Thorsten Leemhuis]

[note: please remove Timo Lindfors when replying to this mail and
include Timur Kristóf – sorry everyone and especially Timo, my mailers
address book tricked me]

On 4/27/26 11:35, Linux regression tracking (Thorsten Leemhuis) wrote:
> On 4/26/26 01:22, Brandon Taylor wrote:
>> I have an AMD Radeon R9 270X GPU (yes, I know it's old) which used to
>> default to the `radeon` firmware
> 
> FYI, as using the wrong terms can lead to confusion while dealing with
> bugs: that's not a firmware, that's the driver (sometimes also called
> kernel module)
> 
>> on kernel versions up to and
>> including 6.18. I could manually change this behavior to use the new
>> `amdgpu` firmware by setting `radeon.si_support=0 amdgpu.si_support=1`
>> as kernel parameters in the GRUB bootloader. Everything worked fine,
>> and I was able to play my Windows games without any problems.
>>
>> That was, until kernel version 6.19 came out.>
>> 6.19, from what I was able to Google, was supposed to have AMD GPUs to
>> default to the new `amdgpu` firmware. Unfortunately, when I updated
>> the kernel and rebooted, I got a black screen.
>>
>> Further Google searches led me to reboot, set `nomodeset` in the
>> kernel parameters, switch to a TTY, login, and investigate the output
>> of a `dmesg` command — and what I saw ... Well, let's just say it's
>> the kind of thing that would make Linus go Chernobyl:
>>
>> `amdgpu: probe with driver amdgpu failed with error -22`
> 
> Please provide a full kernel log, this is needed to full understand the
> problem
> 
>> And this issue has not been fixed in kernel version 7.0, which I guess
>> is set to be the next LTS kernel.
> 
> No, it's not.
> 
>> Now, I hope to God that this email finds whoever is responsible for
>> breaking this `amdgpu` firmware before the same thing happens to Linus
>> via Fedora 44, and he goes berserk. I realize that's not a heck of a
>> lot of time, but God only knows HOW many Linux users with AMD GPUs are
>> pissed off by this kind of issue — and I for one can only imagine how
>> Linus' reaction will compare when it happens to HIM!
> 
> FWIW, you seem to assume that the problem hits everyone with a AMD GPUs.
> That's not the case, we'd know this by now. Maybe it's not even
> something that hits everyone with a card/gpu like yours, as some falires
> only show up in certain environments, for example in combination with a
> specific mainboard chipset or firmware.
> 
> Anyway: I CCed a few people that might be able to help. But they likely
> will need at least the full kernel log mentioned above .
> 
> Ciao, Thorsten