Subject: random core dumps while swapping
To: None <email@example.com>
From: David Laight <firstname.lastname@example.org>
Date: 03/17/2003 22:01:45
My sparc system is generating random core file while swapping
heavily. I don't know if this is a sparc specific fault, or a
generic one - I haven't tried the same test on any other systems.
NetBSD 1.6P (GENERIC) #14: Mon Mar 17 17:55:52 GMT 2003
total memory = 65216 KB
avail memory = 56848 KB
using 840 buffers containing 3360 KB of memory
mainbus0 (root): SUNW,TP90: hostid 80260032
cpu0 at mainbus0: TMS390S10 @ 50 MHz, on-chip FPU
cpu0: physical 4K instruction (32 b/l), 2K data (16 b/l): cache enabled
The system is booting diskless, but has swap on a SCSI disk.
I was running the following:
$ ulimit -p 532
$ while sleep 60 & do :; done
and noticed that the shell was reporting that processes had dumped core.
I've also had a 'ps' command dump while the code was active.
The core dumps are typically 'Segmentation fault', or 'Illegal instrcution'
and usually confuse gdb. I don't get a core dump every time, but have
seen quite few. I've also had the system lock solid (wouldn't even
break to the prom).
There may be a related problem that /bin/sh sometimes decides that the
while loop should not be executed, but abort after starting 1 sleep.
Anyone seen anything like this?
David Laight: email@example.com