[ On Mon, March 18, 1996 at 13:59:11 (EST), Gordon W. Ross wrote: ]
> Subject: missing /dev/console
>
> The /dev/kd device exists as a real device so that the /dev/console
> "redirector" has a real device to "point at."  The /etc/ttys file,
> as distributed, runs getty on /dev/kd and /dev/ttya only.  The only
> use for /dev/console should be output messages, except for init which
> opens it read/write in single-user mode.  This avoids the possibility
> of two gettys on the same tty line (i.e. /dev/console and /dev/ttya).

What do you mean by ``the console "redirector"''?  Do you mean the
smarts in the kernel which look at the EEPROM settings and decide what
to talk to?

If the dualing gettys problem is the user-land purpose of /dev/kd, then
I indeed approve.  I just recently "upgraded" my SunOS-4 Sun-3/260
server to being headless, and ran into just that problem (I had always
had terminal connected to it, since it sits in the basement).  [[ The
server runs much better without X11 too! ]]

Unless someone beats me to the job (I won't likely get a start on it
until I've shaken out enough of my environment to get NetBSD/sun3
running on my server), I'll send-pr some manual pages for these devices.

(And I'm also eager to get NetBSD/sun3 disk performance up to at least
somewhere on par with SunOS-4.1.1_U1 -- more details in a new thread.)

-- 
							Greg A. Woods

+1 416 443-1734			VE3TCP			robohack!woods
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>