Lennart Augustsson <augustss@cs.chalmers.se> writes:
> I think that getting the USB keyboard driver functional very early
> in the startup process is going to be hard.  It's rather complicated
> to get USB running, you need PCI attachment, memory allocation,
> bus space allocation, etc.

yup.  doing console keyboards right is a pain.  8-)

my big test is "does 'boot -d' work?"


> What I had in mind was a different solution.  In machines that have a 
> USB aware BIOS there will be some kind of legacy emulation, or at
> least a call to get keyboard input.  So early in the boot process
> we attach a console driver that uses the BIOS code to access the keyboard.
> When the kernel USB driver starts it will take over the console.

Uh, i doubt that's true, especially for non-PC hardare.  actually, it
might be true, but the further you get from a PC, the more like you
are to have the situation where:

	* the console and OS drive the hardware natively and you don't
	  have those callbacks, or

	* the callbacks can't sanely be used from the OS.

on systems with real consoles, and/or real OSes (i.e. not x86 PCs),
using console facilities is often not so easy, and there's often no
need for backward-compatible keyboard emulation.


though it's a pain, it really is probably best to do the console
keyboard code right.



cgd
-- 
Chris Demetriou - cgd@netbsd.org - http://www.netbsd.org/People/Pages/cgd.html
Disclaimer: Not speaking for NetBSD, just expressing my own opinion.