Re: [drochner%netbsd.org@localhost: CVS commit: src/sys/kern]

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Fri, Feb 22, 2008 at 11:28:15AM -0600, David Young wrote:
> How do you propose to handle the console?  Please be more explicit.
> It seems that step (1) potentially detaches the console, which may stop
> it from working, and (2) potentially suspends the console, which we know
> has stopped it from working.

The question is two fold:
(1) Do we want to have the console working from any state for things
like DDB?
(2) Do we only want to support the console for "normal" shutdown and
early boot?

The cnpollc backend should take care of this by distinguishing between
having the parent device registered or not. If it is present, it needs
to at least partly wakeup the parent. This applies e.g. for USB as
console device. This is enough to handle DDB in the early phase of a
wakeup. If there is no parent registered, the backend has to just assume
that it is in a sane state. This is the same situation as faced already
during early boot. The suspend stopped it from working because the
cnpollc didn't bother to enable the keyboard.

Joerg