The other bureaucratic blocker is that we need to work out how we're doing PSCI
for CPUidle and suspend support. M1/2 platforms cannot support standard PSCI
over SMC/HVC, so we need an alternative for that (if we aren't just putting in
non-PSCI drivers in Linux). So far the most promising solution is doing it over
EFI, and there are patches to that effect, but we aren't shipping them
ourselves nor have they been submitted for upstreaming, and there's a bunch of
device-side work to be done to make suspend work. It's looking like we're going
to start focusing more on sleep/idle support in the coming months, so hopefully
we can get that ball rolling. I do expect quite some talk to come out of this,
because there is no "good" solution. The standards don't and can't work, so we
need to make a new standard.