At Sun, 15 Jan 2012 22:08:42 -0800, "Greg A. Woods" <woods%planix.ca@localhost>
wrote:
Subject: ongoing major problems with NetBSD-5 and LOCKDEBUG on multi-core system
>
> So I was finally able to get a new server, and its a nice big Dell
> PE2950 with 32GB RAM, lots-o-disk on a PERC-6/i, and a pair of zippy
> Intel Xeon E5440 CPUs (quad-cores x2).
Unfortunately I'm not having much better luck with an HP EliteBook
that's got a pair of Intel Core i5's in it.
Running with 4 hyperthreaded (I think) CPUs I get a very similar
LOCKDEBUG crash when running build.sh on it as well. The stack
backtrace is almost identical on one CPU with rw_vector_exit() appearing
right before the lockdebug_abort1() call.
(savecore failed though, despite what looked like a crash dump having
happened.... now on the second test "reboot" at ddb is hung....)
I'm going to try running build.sh on my VirtualBox image running on my
iMac desktop machine, with its two CPUs, and see what happens.... I've
never had problems with the VB machine before, though I've never really
put it under any significant load (i.e. I've never run a build on it).
One thing to note is that I run "systat -w 1 vm" and sometimes "top"
while running the build -- that might explain the sysctl() in the
backtrace....
Perhaps that's the recipe to repeat: run "systat -w 1 vm" in one
window, and a "nohup build.sh -j 6 & sleep 2 ; tail -f nohup.out" in
another, and maybe "top" in yet another.
--
Greg A. Woods
Planix, Inc.
<woods%planix.com@localhost> +1 250 762-7675 http://www.planix.com/
Attachment:
pgpz4YOdyrtbJ.pgp
Description: PGP signature