Allen Briggs wrote:

> Aren't the akbd entries in OFW even if there is no apparent ADB
> keyboard?

Seen the code in akbd.c:akbdattach(), it looks like the adb driver 
somehow gets enough information to get the type of keyboard. Isn't this 
enough to make sure there is a true akbd? You mean even machines 
without akbd's get some ghost akbd0 attached to them?
Or is the point that we need to detect the console only using OFW 
information?

> Another option is to start reading the system name and build a
> capabilities table in the kernel to work from.  We could then say
> that all PowerBook systems use ADB, or some such.  I had some hacks
> in a local tree to do that, but they got eaten by a hungry fsck.

If we could avoid this, it would be better IMHO.

  Aymeric