RE: Lock order reversal in 5.2-CURRENT

From: Terrence Koeman <root_at_mediamonks.net>
Date: Thu, 12 Aug 2004 00:09:57 +0200
> -----Original Message-----
> From: owner-freebsd-current_at_freebsd.org
> [mailto:owner-freebsd-current_at_freebsd.org] On Behalf Of Doug White
> Sent: Wednesday, August 11, 2004 19:22
> To: Terrence Koeman
> Cc: freebsd-current_at_FreeBSD.org; 'John Baldwin'
> Subject: RE: Lock order reversal in 5.2-CURRENT
>
> On Wed, 11 Aug 2004, Terrence Koeman wrote:
>
> > I think something else is wrong, as I get different lock
> order reversals and
> > some other errors that all lockup the box. Earlier I had a
> corrupted cc
> > binary after a buildworld.
> >
> > Everything points to a hardware failure somewhere, but I
> already switched
> > the hardware before this happened, I swapped RAID arrays in
> identical
> > machines, and the machine where -CURRENT runs on now was a
> production server
> > that ran 4.9/4.10-STABLE for months under heavy load
> without any problems
> > whatsoever.
> >
> > The following is what I got today:
> >
> > Second bad
> > /: bad dir ino 16110954 at offset 24: mangled entry
> > panic: ufs_dirbad: bad dir
>
> Your RAID is doing a really good job of corrupting your data. :)  What
> RAID controller and volume layout are you using?

It's a Promise FastTrak TX2000 with two mirrored 160Gb Maxtor drives.

> > Fatal trap 18: integer divide fault while in kernel mode
>
> This looks more serious .. you may have a bad CPU, memory, or
> some other
> critical component.

I thought so too, because multiple weird errors usually point to the hardware.

But I have three identical systems with the only difference being the contents 
of the disks. The other two systems are running 4.10-STABLE with heavy load 
without any problems. I swapped the disks (only the disks) with a working 
system twice now and it locks up just the same.

I think the chance of three systems having the same hardware problem is really 
small, especially because 4.10-STABLE hasn't had a single problem on those 
systems in the couple of months they run.

Maybe 5.2-CURRENT has a specific problem with the hardware in the systems? But 
it's not like it is exotic hardware, they are SuperMicro 1U barebones with a 
Celeron 2600, 512MB of RAM and a FastTrak TX2000.

-- 
Regards,
Terrence Koeman

MediaMonks B.V. (www.mediamonks.com)
Please quote all replies in correspondence. 

Received on Wed Aug 11 2004 - 20:08:36 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:05 UTC