Comment 19 for bug 1925126

Discussion over at Arch led me to try adding the kernel parameter intel_iommu=on.
https://bbs.archlinux.org/viewtopic.php?id=256520

That appears to have solved the problem (I hope I'm not speaking too soon), but now (as discussed in another Arch forum post) there's a new one:
https://bbs.archlinux.org/viewtopic.php?id=228604

I also understand this causes power use to go up.

$ dmesg -T | grep -i i915
[Mon Apr 19 23:57:41 2021] i915 0000:00:02.0: [drm] VT-d active for gfx access
[Mon Apr 19 23:57:41 2021] i915 0000:00:02.0: vgaarb: deactivate vga console
[Mon Apr 19 23:57:41 2021] i915 0000:00:02.0: [drm] DMAR active, disabling use of stolen memory
[Mon Apr 19 23:57:41 2021] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[Mon Apr 19 23:57:41 2021] [drm] Initialized i915 1.6.0 20200515 for 0000:00:02.0 on minor 0
[Mon Apr 19 23:57:41 2021] fbcon: i915drmfb (fb0) is primary device
[Mon Apr 19 23:57:42 2021] i915 0000:00:02.0: [drm] *ERROR* uncleared fifo underrun on pipe A
[Mon Apr 19 23:57:42 2021] i915 0000:00:02.0: [drm] *ERROR* CPU pipe A FIFO underrun
[Mon Apr 19 23:57:42 2021] i915 0000:00:02.0: [drm] *ERROR* uncleared pch fifo underrun on pch transcoder A
[Mon Apr 19 23:57:42 2021] i915 0000:00:02.0: [drm] *ERROR* PCH transcoder A FIFO underrun
[Mon Apr 19 23:57:42 2021] i915 0000:00:02.0: fb0: i915drmfb frame buffer device
[Mon Apr 19 23:59:47 2021] snd_hda_intel 0000:00:1b.0: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])