> Yeah, it does still need work.  Perry has suggested tossing it and
> taking the FreeBSD driver.  I've not had a chance to look at the FreeBSD
> driver, but that might be a good first step.

I've looked at the FreeBSD driver (before i seriously started looking
at our APM code this last time around, because i'd heard the same
suggestion).  In my opinion, it's not the answer, either.  It may be
better than what our code was before, but even before I liked the
general style of our code better.  It would require at least as much
work as our code does to be a 'appropriately featureful' APM driver,
in my estimation.


> Device drivers (e.g. pccard machinery) definitely need a callback
> mechanism so that they can prepare for things like power-down-at-suspend
> and power-up-after-resume.

Yeah.  However, that's pretty trivial to implement, and certainly not
the largest challenge with the existing code by any stretch of the
imagination.



cgd