Subject: Ping to bad address with redirection hangs forever?
To: None <port-macppc@netbsd.org>
From: Donald Lee <donlee_ppc@icompute.com>
List: port-macppc
Date: 07/18/2001 08:32:53
I found a cool behavior this morning. Looks like a kernel bug?
It does the same thing with my slightly modified 1.5 production kernel,
and also with a stock 1.5.1 kernel, both running on "early" machines.
(Powercenter 132 and PM 7600)
Try this script:
#!/bin/ksh
set -x
BADIP=209.46.8.79
ping -c 1 $BADIP
echo Done.
ping -c 1 $BADIP | wc
echo This one works fine.
ping -c 1 $BADIP > /dev/null
echo Never gets here
ping -c 1 $BADIP > /tmp/rand_file
echo Also does not finish
While waiting for this third (or fourth) ping to complete, if you do a
ping from another telnet session, the ping command will finish.
Pinging to the machine from somewhere else does not help. Other network
activity on the machine does not trigger the relevant event. So far,
only an outgoing ping seems to do the trick.
This smells strongly to me like the problem I found a few months back
in macppc/extintr.c with lost soft interrupts. Can anyone point me in the
right direction to track this down?
-dgl-