Re: radeon_fbdev_river_fbdev: failed to initialize framebuffer and setup emulation

[Mario Limonciello <superm1@kernel.org>]

On 8/8/25 6:47 AM, Christian Zigotzky wrote:
> 
> 
>> On 08 August 2025 um 01:23 pm, Mario Limonciello <superm1@kernel.org> wrote:
>>
>> 
>>
>>> On 8/8/25 2:36 AM, Christian Zigotzky wrote:
>>> On 07 August 2025 at 04:21 pm, Limonciello, Mario <Mario.Limonciello@amd.com> wrote:
>>> Does applying
>>> https://github.com/torvalds/linux/commit/907a7a2e5bf40c6a359b2f6cc53d6fdca04009e0
>>> help?
>>> - - -
>>> Hello Mario,
>>> Thanks a lot for your patch.
>>> I tested it today but unfortunately it doesn’t solve the issue with the Radeon framebuffer.
>>> I have created two kernels with and without the drm-next-2025-07-30 updates [1] because of the issue with the Radeon graphics framebuffer device #15. [2]
>>> Download and further information: https://github.com/chzigotzky/kernels/releases/tag/v6.17.0-alpha5
>>> I have tested both kernels this week and I can definitely confirm, that the issue is somewhere in the commit drm-next-2025-07-30 updates [1].
>>> The Radeon graphics framebuffer doesn't work with the kernel with the drm-next-2025-07-30 updates [1]. Without these updates, the framebuffer works.
>>> I bisected yesterday. [2]
>>> There are some other user reports:
>>> - https://forum.hyperion-entertainment.com/viewtopic.php?p=60606#p60606
>>> - https://forum.hyperion-entertainment.com/viewtopic.php?p=60595#p60595
>>> They use other Radeon graphics chips.
>>> @All
>>> Please check the drm-next-2025-07-30 updates [1]
>>> Thanks,
>>> Christian
>>> [1] https://web.git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=260f6f4fda93c8485c8037865c941b42b9cba5d2
>>> [2] https://github.com/chzigotzky/kernels/issues/15
>>
>> I think there is a mistake in your bisect, which I notice from looking at your other user reports.
>>
>> The original commit you identified in your bisect and thus CC'ed me came in 6.16-rc1.
>>
>> ❯ git describe --contains 4d4c10f763d7808fbade28d83d237411603bca05
>> v6.16-rc1~50^2~19^2
>>
>> There absolutely was a regression for that was raised, but this was fixed in 6.16-rc3.
>>
>> ❯ git describe --contains 907a7a2e5bf40c6a359b2f6cc53d6fdca04009e0
>> v6.16-rc3~13^2~1
>>
>> Users in your forums posts talk about 6.16 final working fine.  So I think that when you bisected you ran into multiple confounding issues and replicated in a similar fashion.  You should check every single bisect step to see if 4d4c10f763d7808fbade28d83d237411603bca05 is applied.  If it's applied but 907a7a2e5bf40c6a359b2f6cc53d6fdca04009e0 is not applied in that step you should apply 907a7a2e5bf40c6a359b2f6cc53d6fdca04009e0.
>>
>> Hopefully that will get you down to the real cause.
> 
> Thanks for the hint. Unfortunately I don’t have more time for figuring out the issue.
> 
> Two facts:
> 
> - The Radeon driver can no longer be used.
> The Radeon driver is broken for us.
> 
> - The issue is somewhere in the commit drm-next-2025-07-30 updates [1].
> 
> I cannot provide any more due to time constraints.
> 
> I've already spent time that I didn't actually have.
> 
> Thanks for your help.
> 
> Christian
> 
> 

If you're sure that the issue was raised in that merge commit, I suggest 
you (or someone else affected by this issue) do the bisect using the drm 
kernel tree [1] and bisect between the problematic tag [2] and previous 
tag [3].

[1] https://gitlab.freedesktop.org/drm/kernel
[2] https://gitlab.freedesktop.org/drm/kernel/-/tags/drm-next-2025-07-30
[3] https://gitlab.freedesktop.org/drm/kernel/-/tags/drm-next-2025-06-06