amdgpu 100% CPU usage causing freeze 1002:15d8

[Alex Deucher]

On Tue, Jan 14, 2025 at 3:49 PM Marco Moock <mm at dorfdsl.de> wrote:
>
> Hello!
>
> This might be related to
> https://lists.freedesktop.org/archives/amd-gfx/2025-January/118759.html
> As I subscribed just now, I can't reply there and can't get the
> Message-I
>
>        description: Motherboard
>        product: Pro A520M-C II
>        vendor: ASUSTeK COMPUTER INC.
>        physical id: 0
>        version: Rev X.0x
>        slot: Default string
>      *-firmware
>           description: BIOS
>           vendor: American Megatrends Inc.
>           physical id: 0
>           version: 3612
>           date: 12/03/2024
>
> I updated the UEFI yesterday to the latest version, problem still
> exists.
>
> 08:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc.
> [AMD/ATI] Picasso/Raven 2 [Radeon Vega Series / Radeon Vega Mobile
> Series] [1002:15d8] (rev c9) Subsystem: ASUSTeK Computer Inc. Device
> [1043:876b] Kernel driver in use: amdgpu Kernel modules: amdgpu
>
> Linux ryz 6.12.9-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.12.9-1
> (2025-01-10) x86_64 GNU/Linux
>
> un  mesa-common-dev           <none>       <none>       (no description available)
> un  mesa-glide2-dev           <none>       <none>       (no description available)
> ii  mesa-libgallium:amd64     24.3.3-1     amd64        shared infrastructure for Mesa drivers
> ii  mesa-libgallium:i386      24.3.3-1     i386         shared infrastructure for Mesa drivers
> un  mesa-opencl-icd           <none>       <none>       (no description available)
> ii  mesa-utils                9.0.0-2+b1   amd64        Miscellaneous Mesa utilities -- symlinks
> ii  mesa-utils-bin:amd64      9.0.0-2+b1   amd64        Miscellaneous Mesa utilities -- native applications
> un  mesa-utils-extra          <none>       <none>       (no description available)
> ii  mesa-va-drivers:amd64     24.3.3-1     amd64        Mesa VA-API video acceleration drivers
> ii  mesa-vdpau-drivers:amd64  24.3.3-1     amd64        Mesa VDPAU video acceleration drivers
> ii  mesa-vulkan-drivers:amd64 24.3.3-1     amd64        Mesa Vulkan graphics drivers
> ii  mesa-vulkan-drivers:i386  24.3.3-1     i386         Mesa Vulkan graphics drivers
> un  mesag-dev                 <none>       <none>       (no description available)
> un  mesag3                    <none>       <none>       (no description available)
> un  mesag3+ggi-dev            <none>       <none>       (no description
> available)
>
>
> I am running Debian Unstable and encounter 100% CPU usage after some
> hours, reproducible. I have to shut off the system with sysrq, I can't
> shut it down the normal way as it is non-responsive.
>
> I tried 6.12.9 and 6.12.8.
>
> 6.12.8 gave some dmesg error messages:
>
> Jan 13 11:09:44 ryz kernel: amdgpu 0000:08:00.0: amdgpu: Dumping IP State
> Jan 13 11:09:48 ryz kernel: [drm:amdgpu_dm_atomic_check [amdgpu]] *ERROR* [CRTC:73:crtc-0] hw_done or flip_done timed out
> Jan 13 11:09:55 ryz kernel: amdgpu 0000:08:00.0: amdgpu: failed to write reg 28b4 wait reg 28c6
> Jan 13 11:10:00 ryz kernel: sysrq: Keyboard mode set to system default
> Jan 13 11:10:00 ryz kernel: sysrq: This sysrq operation is disabled.
> Jan 13 11:10:00 ryz kernel: sysrq: This sysrq operation is disabled.
> Jan 13 11:10:01 ryz kernel: sysrq: Emergency Sync
>
>
> 6.12.9 doesn't gave me them, but it doesn't list sysrq calls either, so
> I assume it didn't manage to store them in dmesg.
>
> I remember the first occurrence last Friday, some mesa packages were
> updated and I assume it was running 6.12.8 according to the apt logs.
>
>
>
>
> Please tell me which further info you need to track down the issue.

What kernel version(s) is it working properly with?  Can you bisect?

Alex