Subject: Re: Serial consoles
To: port-i386@NetBSD.ORG, Michael Richardson <mcr@sandelman.ottawa.on.ca>
From: Matthias Drochner <drochner@zelux6.zel.kfa-juelich.de>
List: port-i386
Date: 04/25/1997 11:18:29
Excerpts from netbsd: 23-Apr-97 Serial consoles Michael Richardson@sande (1762)

>   I submitted a PR today to make pc0 fail to probe if there was no
> keyboard attached.

I have a couple of PCs and PC-compatible single board computers
in embedded-like applications. Most of them are running without
keyboard. I found it useful to connect a monitor and see the last
console printouts in case of trouble.
(If a serial console is not connected all the time, the output
is lost.)
So I think the console output should not depend on the keyboard.

I've never tried to run a PC without a VGA-like graphics
adapter. (There was always something on-board.)
Does it work?
If yes, is there a save way to detect its presence?

>  I needed to change the pccnprobe() to fail rather than pcprobe().

As far as the console selection is concerned:
I think the whole cninit() / constab[] / *cnprobe() design
is not suitable. 2 reasons:
-As things are going now, the console driver may need
  a bus_space_tag/handle to access the hardware.
  It does not have the information to make the tag itself.
  The normal map mechanism might not work before
  mainbus is attached.
-It is desirable that the console selection is passed from
  boot to the kernel. This information should be dealt with
  at a central place.

The i386/stand code contains stubs to pass console
device information to the kernel. It's waiting for a decision
what and how how to pass the data.

best regards
Matthias