Subject: Re: =?us-ascii?Q?SUPPORT=5FSERIAL=3D=3F?=
To: NetBSD/i386 Discussion List <port-i386@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: port-i386
Date: 06/21/2000 17:50:26
[ On Tuesday, June 20, 2000 at 11:43:19 (+0200), Manuel Bouyer wrote: ]
> Subject: Re: SUPPORT_SERIAL=?
>
> A reliable test to see if a keyboard is connected isn't that easy.

It's not that hard -- but there are a lot of variants and it does
require a fairly deep understanding of the hardware behaviour in each
variant, so yeah, OK, a common test is somewhat hard.  Also, see my
other messages about using the test done by the BIOS....

> Also, I have applications where I don't want the boot loader to switch
> to serial console when there's no keyboard (consoles switchs in the wrong
> position when the machine boots).

I handle that problem by always having a serial console and then
offering a getty on the hardware keyboard & display so that the
switch-box user can still login (and even run X11 if there's a mouse
too, though you need a real fancy switch-box that emulates a mouse when
not enabled to fool the BIOS on some machines with ps2 mice). The serial
consoles are always connected permanently to a terminal server.

Obviously this does need to be configurable though, and it would be very
nice to have the ability to do all kinds of run-time configuration at
every stage too.  Such ability would solve the final problem I have with
my boot blocks -- i.e. the one opposite to what you mention, because I
want to switch the kernel's idea of where the console is over to the
serial port after someone has accidentally booted with a keyboard
attached...

-- 
							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>