Yo,
I'm working on a patch (needed for the Real Time tree) to cleanup use of
the pm_idle pointer, something each architecture defines and optionally
calls from within the CPU idle loop, if it's defined. The problem is
that there's a classical bug (which has existed for quite some time)
whenever a function providing that pointer is unloaded from under it.
So, I'm proposing an ordered list of pm_idle functions with a
register/unregister function to ensure safe access. Anyway, before I
post that, does anyone have a good reason for why things are as they are
currently in the kernel with respect to the pm_idle stuff? Cruftiness?
Jon.