Subject: kernel crash with 1.5 RELEASE apparently related to xdm/xconsole/etc.
To: None <port-sparc@netbsd.org>
From: Greg Troxel <gdt@ir.bbn.com>
List: port-sparc
Date: 02/02/2001 09:12:18
I haven't experimented well enough to file a PR; I'm posting in case
someone else has had the problem and has a hint.  One of my machines
crashes when logging out on the console (xdm is running).  I've seen a
similar problem on an ELC.

Software: precompiled 1.5 release binaries, including the generic
kernel.  I've updated named along the release branch, but nothing
else.   I'm running a few normal things (postfix, ssh) plus arla.
xdm-krb4 installed but not running.

hardware: sparc 5, 54 Mb ram, 2 scsi disks, cgsix (totally vanilla)

configuration:
/etc/ttys:
console "/usr/libexec/getty suncons"    sun     on secure
ttya    "/usr/libexec/getty std.9600"   unknown off secure
ttyb    "/usr/libexec/getty std.9600"   unknown off secure

xdm="YES" in rc.conf.  /usr/X11R6/lib/X11/xdm files all
have mtime Nov 28th 2045Z, so I'm pretty sure they are from the release.

vague symptoms:

several cases of 'machine panics and reboots when logging out on the
console'

lots of :
Feb  2 08:47:32 [name] init: getty repeating too quickly on port /dev/console, sleeping


I noticed xdm was running xconsole, and figured that getty was losing
on opening /dev/console since it was open by xconsole.  I killed
xconsole, and the system echoed the pid but not my <cr> and then went
dead (I'm remote).  On reboot, I had a kernel core dump:

(gdb) target kcore netbsd.3.core
panic: kernel fault
#0  0xf0056c4c in mi_switch ()
(gdb) bt
#0  0xf0056c4c in mi_switch ()
#1  0xf005645c in ltsleep ()
#2  0xf019ae7c in uvm_scheduler ()
#3  0xf0046448 in check_console ()
#4  0xf00098a0 in cpu_hatch ()
can not access 0x3902fc, invalid address (3902fc)
can not access 0x3902fc, invalid address (3902fc)
can not access 0x3902fc, invalid address (3902fc)
can not access 0x3902fc, invalid address (3902fc)
can not access 0xefffffd8, invalid address (efffffd8)
can not access 0xefffffd8, invalid address (efffffd8)
Cannot access memory at address 0xefffffd8.

Obviously I need to compile a kernel with symbols, and I should try
updating along netbsd-1-5.  Any other hints?