Debian bullseye (11.6) dmesg error

I have an error in dmesg, is this a bug? I am asking this because I’m not able to use the hardware acceleration in ffplay for playing 4k HEVC videos.
Here is the relevant part from dmesg:

[ 8.193304] BUG: spinlock bad magic on CPU#0, systemd-udevd/385
[ 8.193321] lock: 0xffffffc012d85080, .magic: 00000000, .owner: /-1, .owner_cpu: 0
[ 8.193328] CPU: 0 PID: 385 Comm: systemd-udevd Not tainted 5.10.110-37-rockchip-g74457be0716d #rockchip
[ 8.193331] Hardware name: Radxa ROCK 5B (DT)
[ 8.193336] Call trace:
[ 8.193346] dump_backtrace+0x0/0x1a8
[ 8.193351] show_stack+0x2c/0x38
[ 8.193358] dump_stack_lvl+0xd0/0xfc
[ 8.193362] dump_stack+0x14/0x2c
[ 8.193369] spin_bug+0x8c/0xac
[ 8.193374] do_raw_spin_lock+0x40/0xd8
[ 8.193380] _raw_spin_lock+0x20/0x2c
[ 8.193695] pq_push+0xb4/0xf8 [8852be]
[ 8.193972] _phl_ser_mdl_init+0x128/0x1ec [8852be]
[ 8.194246] bk_module_init+0x38/0x68 [8852be]
[ 8.194512] dispr_register_module+0x14c/0x2b4 [8852be]
[ 8.194780] phl_disp_eng_register_module+0x58/0x70 [8852be]
[ 8.195046] phl_register_ser_module+0x30/0x3c [8852be]
[ 8.195314] rtw_phl_init+0x4c0/0x680 [8852be]
[ 8.195584] rtw_hw_init+0x58/0x190 [8852be]
[ 8.195852] rtw_dev_probe+0x364/0x4f0 [8852be]
[ 8.195863] pci_device_probe+0xbc/0x12c
[ 8.195871] really_probe+0x2e0/0x47c
[ 8.195876] driver_probe_device+0x130/0x140
[ 8.195881] device_driver_attach+0x48/0x70
[ 8.195885] __driver_attach+0x140/0x144
[ 8.195889] bus_for_each_dev+0x8c/0xd4
[ 8.195893] driver_attach+0x30/0x3c

In this log can be seen that the kernel version is: 5.10.110-37-rockchip-g74457be0716d
Is this the correct one to use with Debian/Bullseye?

Matches my Rock5. I’ve also been troubleshooting a lack of hw graphics acceleration.

Excerpt from neofetch command
OS: Debian GNU/Linux 11 (bullseye) aarch64
Host: Radxa ROCK 5B
Kernel: 5.10.110-37-rockchip-g74457be0716d

I’ve confirmed the panfrost file was already in /lib/firmware (it was) and glmark2-es2 shows the following for 3D rendering and has been returning 600-700 scores.
GL_RENDERER: Mali-LODX
GL_VERSION: OpenGL ES 3.2 v1.g6p0-01eac0.efb75e2978d783a80fe78be1bfb0efc1

I’ll respond again if I find a fix, otherwise I’ll change distros.

This looks promising:
https://forum.radxa.com/t/archlinux-on-rock5b/13851

@linuxlion Just reflashed my eMMC to Debian from RebornOS, I don’t know what I am missing, but this boards linux support driving me crazy. 10 years before when I was on AMD64 Debian distribution it was way better, I think the driver support is the issue here both kernel and user programs (like ffmpeg, mpv, chromium, etc…).
Many times Chromium doesn’t even start to play a youtube video, mpv lack hardware acceleration, hard to change keyboard layout, 4k resolution so tiny that I am unable read anything, etc…

@linuxlion I figured out that my X.org was using llvmpipe (which is software) and not Mali-G610 driver.
You can check it in your installation: glxinfo -B.
I have some minor progress on this issue, but it is still buggy and really not that trivial way.
So I installed https://gitlab.com/panfork/mesa, fallowed the instructions at the sections Depdendencies, Build & install, but not installed the mentioned different firmware binary (my current installation had it and it worked).
I experienced that my KDE installation is not yet compatible with it (or I still don’t see the full picture :slight_smile: )
Hence I stopped and disabled sddm and started plasma manually from linux console command prompt.
After this I can see that, for example if I move a window content, than it was way faster, not laggy, and had the glassy “see through” effect over its area while moving it.
But I’m having some side effects, like some (but not all) huge icons on pinned tasks in taskbar; and I am unable to configure it (this actually freezes KDE).