Move the card back to the Open state. Additional work is probably better
managed within a wider project around boot-time optimizations.

Continue to chat with Daniel about boot-time-optimization and whether how we
potentially could formalize the work into a project. More discussions will come.

Me and Daniel discussed a solution at Amsterdam. We have a proposal, but
someone needs to prepare a patch and post it. We will keep discussing this
at the next Qcom/KWG sync call.

Discussion on how to move forward at the last KWG/Qcom call. Kernel module
support can't be enforced, we need something else. The typical solution needed
is that we want the idle states to be disabled during boot and at some point
later, user space should be able to re-enable them via sysfs.

It seems like we need to involve cpuidle maintainers to find the best
solution.

Continued to discuss this at LKML and at KWG/Qcom call. I have a draft series
for modularizing cpuidle-psci. It needs some more polishing before I can submit
it as and RFC. Although, this doesn't cover the complete problem, as we have
cases where loadable module support can't be used.