A hang last night under the conditions mentioned disconfirms my theory that running apmd(8) without any flags and Kernel DMA protection gets rid of hangs. It perhaps helps not to include hw.setperf related flags in apmd(8), given that the frequency of hangs has been lower overall when that condition is fulfilled (though now it feels more like conjecture than fact).
I've also determined that tweaking Enhanced Speedstep in the BIOS (disabling it or changing how it manages the CPU) doesn't seem to help or hurt the situation. Back to the drawing board I guess. -- https://amissing.link
