Subject: Re: why won't my dual PIV Xeon boxes halt when they are told to?
To: NetBSD/i386 Discussion List <port-i386@NetBSD.org>
From: Greg A. Woods <woods@planix.ca>
List: port-i386
Date: 06/22/2007 19:50:42
--pgp-sign-Multipart_Fri_Jun_22_19:50:41_2007-1
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

At Wed, 13 Jun 2007 15:47:35 -0400, Me-planix.com wrote:
Subject: why won't my dual PIV Xeon boxes halt when they are told to?
>=20
> I've got two different servers here that won't halt when they are told to.
>=20
> They reboot instead:   (this is a serial console, and this is direct
> from the conserver log)
>=20
> [Wed Jun 13 15:40:19 2007]syncing disks... 3 done
> [Wed Jun 13 15:40:19 2007]unmounting file systems... done
> [Wed Jun 13 15:40:19 2007]
> [Wed Jun 13 15:40:19 2007]The operating system has halted.
> [Wed Jun 13 15:40:19 2007]Please press any key to reboot.
> [Wed Jun 13 15:40:19 2007]
> [Wed Jun 13 15:40:20 2007]rebooting...
>=20
>=20
> No key was pressed.
>=20
>=20
> One is an IBM eServer with dual Xeons, and the other is a Dell PE2650
> with dual Xeons.
>=20
> It would look to me almost as if their baseboard management controllers
> somehow trigger a reset as soon as the HLT instruction is executed, but
> I don't see any obvious setting at the BIOS level which might do that.

Further testing shows that this is only a problem when the console is
set to be on a serial port.

This isn't the case on all my i386 boxes with serial consoles.  For
example an ancient PPro box running 1.5W (connected in the same way to
the same console server with the same settings on its termserver port)
halts and stays halted.

I'm wondering if maybe the cngetc() call in the relevant code doesn't go
through the line discipline code and so maybe a <CTRL-S> or <CTRL-Q>
(i.e. XOFF/XON) character from the terminal server might be triggering
the reboot.  I poked through the many layers for a bit, but I didn't
find the answer yet.

Of course the ancient PPro box doesn't have a BIOS capable of doing the
serial console redirection stuff whereas the boxes having the problem do
both support serial console redirection and both have it enabled.....

Ack!  PCs are so horribly "designed".

--=20
						Greg A. Woods
						Planix, Inc.

<woods@planix.com>     +1 416 489-5852 x122     http://www.planix.com/

--pgp-sign-Multipart_Fri_Jun_22_19:50:41_2007-1
Content-Type: application/pgp-signature
Content-Transfer-Encoding: 7bit

-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 5.0i for non-commercial use
MessageID: Wnzz/y0uf/Y8tZOLmNV2qTZ70XBnVhLg

iQA/AwUBRnxgUmZ9cbd4v/R/EQLOXACeObr27hdfjvgTsqnFfRF+WdkiDjMAoJtK
OlHTQlQ5rYTSdOwTNUF4vKn6
=J5AL
-----END PGP SIGNATURE-----

--pgp-sign-Multipart_Fri_Jun_22_19:50:41_2007-1--