On Wed, 15 Sep 2004, Andrey Petrov wrote:

> On Wed, Sep 15, 2004 at 04:55:07PM +0200, ali (Anders Lindgren) wrote:
> > ..it happened today again, after an uptime of less than 48h (iirc)
> > and quite moderate/small load (building some pkgs, one at a time).
> > I'm not sure if this provides any additional information, but thought
> > I'd throw it in for good measure.
> >
>
> What is a panic message? Can you do ps/l as well.

"panic" in my first mail about 2.0_BETA was referring to the:

----------8<---------8<-------
init: copying out path `/sbin/init' 11
Alignment error: pid=5294.1 comm=sh dsfsr=00000000:00800001 dsfar=0:9
isfsr=0000
0000:00000000 pc=116518
pid 5294 (sh), uid 0: exited on signal 10 (core dumped)
panic: ltsleep: l_stat 3 != LSONPROC
kdb breakpoint at 118b9e0
----------8<---------8<-------

..that started showing up in dmesg.boot after I got the machine to reboot.
By the look of it, it almost looks like the machine has somehow silently
rebooted, panicked in late boot, and dropped into ddb. With DDB_ONPANIC=1
that doesn't make much sense however(?), unless there's a serious hardware
glitch involved.

  Since I only have remote access to a serial console to this box via
a Linux box which has neither screen nor window at the moment, I have to
start the terminal emulator (minicom) each time I log in to check on my
box, which means I get no scrollback. So all I know is the ddb> prompt
I get when I log in. I'll try to get screen installed[0] so I can keep the
serial console permanently connected[1].

  What does ps/l do? ddb(4) only mentions ps/a, /n and /w.

  I'll try to provide some output from ps and other ddb commands next
time it happens.

[0] Not my box.
[1] Dis- and reconnecting minicom doesn't cause serial breaks to happen
    on the tty line, so this should not be an issue. The crashes/deadlocks
    have occurred when no terminal emulator is connected to the serial
    console line.

-- 
/ali
:wq