I'm spending way too much time on this, but since I'm debugging Linux things on the Late 2009 Macmini3,1 anyway - found out how to get rid of the "irq 20: nobody cared (try booting with the "irqpoll" option) [..] Disabling IRQ #20" kernel oops on boot:
I looked at everything that sits on IRQ 20 (as it turns out, ohci-pci,snd_hda_intel,wl), and added missing kernel modules for the latter two the the initramfs. Now they get loaded early on, and the message is gone.
(...and as it turns out, snd_hda_intel now gets remapped to IRQ 18.)
Unfortunately, the occasional display corruption problems persist. Guess I'll need to play with nouveau options?
@galaxis Which nvidia GPU do you have? I also have nouveau issues and lots of kernel messages like in the screenshot.
@kuleszdl It's an ancient GeForce 9400 in a 2009 Mac Mini in my case. There was no such problem when I ran on Debian buster / LMDE4 
I don't get anything similar to your messages though. Last one was:
kernel: general protection fault, probably for non-canonical address 0xa5349b41fc600a3f: 0000 [#1] SMP PTI
[..]
kernel: nouveau_gem_object_close+0x118/0x210 [nouveau]
kernel: drm_gem_object_release_handle+0x30/0x90 [drm]
@galaxis Looks like a different issue then. But well, I've had these nouveau issues since a long time, also on older Debian versions. Well, I'm eager to see if the backport kernel (why don't you use the one from unstable instead btw?) works better for you.
@kuleszdl Could try unstable too, it's just that backports are enabled by default in LMDE.
Trying the Nvidia binary drivers instead would probably take some work too, I see just now that nvidia-detect tells me:
Detected NVIDIA GPUs:
02:00.0 VGA compatible controller [0300]: NVIDIA Corporation C79 [GeForce 9400] [10de:0861] (rev b1)
Checking card: NVIDIA Corporation C79 [GeForce 9400] (rev b1)
Your card is only supported by the 340 legacy drivers series, which is only available up to buster.
@galaxis closed source + legacy and unsupported = probably not really secure. :)
Hrm. Added a few kernel options that I'd seen mentioned in regards to problems with old Nvidia graphics chips and Linux (tsc=unstable enable_mtrr_cleanup), and haven't seen any further problems today. Just don't know if that's by chance or actually due to those changes.
Ok, I could revert the config and reboot to see if the display corruption reappeears, but probably won't when everything seems to be running fine right now.
@galaxis There should be a custom emoji of Linus flipping the finger, just for these situations 😅
...aaand another two crashes in nouveau today.
Updated to the latest bullseye-backports kernel (5.16.12-1~bpo11+1) to see if it makes any difference.