Subject: Interesting getty problem.
To: None <port-macppc@netbsd.org>
From: gabriel rosenkoetter <gr@eclipsed.net>
List: port-macppc
Date: 09/27/1999 00:56:44
So I happened to notice that getty was sitting at the top of my
processor usage with vaguely 90% before it decided it had an excessive
running time, killed itself, and started over again. Kept my load above
1, which was a lot of fun. Really. I swear.

Looking things over, I found that the getty that was misbehaving was
called as "/usr/libexec/getty std.9600 ttyE0".

I shrugged, commented ttyE0 in /etc/ttys, and did a kill -HUP 1, and
my problem went away.

Skimming my authlog, this has been happening since the last time I
rebooted having first gone into MacOS, which I have to do using a
serial terminal (damn broken OF in 7500s gets reset to serial console
if you let the mac ROMs boot). So I had the serial cable plugged in at
the time, so init listened to it and initialized /dev/ttyE0. Great.

Then I unplugged the serial cable and moved the other mac I'd moved
into the closet where I keep the NetBSD box back out again. And getty
kept trying to use the port. Bad.

At least, that's my theory right now.

Looking more closely at /etc/ttys, I opted to uncomment the ttyE0 line
and just switch status from on to off, and another HUP to init proved
that this was okay (getty didn't start sucking up processor again).

Am I breaking anything by doing this? Is there some reason that getty
freaked out, other than that one shouldn't go hot-unplugging serial
consoles?

Thoughts?

       ~ g r @ eclipsed.net