6.15-rc6/regression/bisected - after commit f1c6be3999d2 error appeared: *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error

[Pillai, Aurabindo]

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Mike,

It is indeed a bit harder, but we were able to repro the issue on the 6000 series. I'll need to get the DMCUB trace log to confirm, but it looks like an SMU hang from within DMCUB. So we'd need more debugging to find out whats going wrong from SMU side. Meanwhile, I've reverted 219898d29c438d8ec34a5560fac4ea8f6b8d4f20 that triggered the issue for a lot of them.


--

Regards,
Jay
________________________________
From: Mikhail Gavrilov <mikhail.v.gavrilov at gmail.com>
Sent: Friday, May 23, 2025 4:52 PM
To: Pillai, Aurabindo <Aurabindo.Pillai at amd.com>
Cc: Chung, ChiaHsuan (Tom) <ChiaHsuan.Chung at amd.com>; Wu, Ray <Ray.Wu at amd.com>; Wheeler, Daniel <Daniel.Wheeler at amd.com>; Deucher, Alexander <Alexander.Deucher at amd.com>; amd-gfx list <amd-gfx at lists.freedesktop.org>; dri-devel <dri-devel at lists.freedesktop.org>; Linux List Kernel Mailing <linux-kernel at vger.kernel.org>; Linux regressions mailing list <regressions at lists.linux.dev>
Subject: Re: 6.15-rc6/regression/bisected - after commit f1c6be3999d2 error appeared: *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error

On Wed, May 21, 2025 at 10:13 PM Pillai, Aurabindo
<Aurabindo.Pillai at amd.com> wrote:
>
> [AMD Official Use Only - AMD Internal Distribution Only]
>
>
> Hi Mike,
>
> Thanks for the details. We tried to repro the issue at our end on 9000 and 7000 series dgpu, but we're not seeing the dmub errors. We were on Ubunti, so we'll try Fedora.
>

Hi Aurabindo,

Were you able to reproduce the problem?

I also have a 7900XTX, and based on my testing, the issue is
significantly harder to reproduce on this GPU compared to the 6900XT.
However, I did eventually see this in the logs:

amdgpu 0000:03:00.0: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data:
DMCUB error - collecting diagnostic data

It appeared only after a long uptime under load, and unlike with
6900XT, the system did not freeze.
Still, this suggests the problem might be latent or manifest
differently on RDNA 3.

Looking forward to hearing your findings on Fedora.

--
Best Regards,
Mike Gavrilov.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/dri-devel/attachments/20250526/b8557b2d/attachment-0001.htm>