[ On Saturday, January 29, 2000 at 20:10:47 (-0700), Miles Nordin wrote: ]
> Subject: Re: serial console HOWTO?
>
> I _like_ the FreeBSD BIOS-unassisted keyboard probe idea, but sad as I am
> to say it, in the not-too-distant future when all these bleeding edge folk
> start using USB keyboards it will stop working and won't be fixable, while
> the ugly ``Press a key here to use this console'' will continue to work. I
> know , I know this scenario is silly enough to make almost anyone angry.
> but, unfortunately, i think it's real.  There is no way big PeeCee vendors
> will stay away from USB for long, considering how inexpensive and
> glitzified their keyboards are already with all those CD and answering
> machine controls on them, and the new version of Windows NT supports it.
> This hypothetical keyboard probe would be a rather short-lived hack.

The "press a key here to make this the console" scheme is, in my
opinion, totally unusable for any kind of production use.  Ideally one
would hard-wire the console device but without run-time configuration
support for the boot blocks this isn't a viable option for me.
Therefore the default console selection algorithm must be smart enough
to always pick the most appropriate console given the hardware that's
available.  If a local keyboard is connected then it probably does make
the most sense to make that the console, by default.

I don't see why a (more sophisticated) keyboard probe can't continue to
be used for testing the presence of a USB attached keyboard.

In fact I think there was a thread in tech-kern or somewhere similar
that started to discuss the issues surrounding how to choose which
keyboard or other input device on the USB that should be used as the
console.

Heck the original IBM RS/6000 machines I worked on about seven or so
years ago could even find a console if it was a terminal connected to a
terminal server out on a token-ring LAN!  :-)  It's algorithm was
basically to look for the most directly attached terminal device.

> Still, if Greg Woods has the code, well, I could use it on annabel.
> whatever works out...

Sure do (as a diff against 1.3.3, or a less tested diff against 1.3I).
I'll e-mail the one of your choice to you separately.  I can also supply
ready-to-run /usr/mdec/biosboot.sym.  Hopefully someday soon I'll
upgrade to -current and 1.4.x and pull forward my changes....

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>