Date: Sat, 05 Dec 2009 19:49:55 -0800 From: "Kevin Oberman" <oberman@es.net> To: Steven Friedrich <freebsd@insightbb.com> Cc: freebsd-acpi@freebsd.org Subject: Re: ACPI temperature Message-ID: <20091206034955.EF3331CC0B@ptavv.es.net> In-Reply-To: Your message of "Sat, 05 Dec 2009 12:42:42 EST." <200912051242.42894.freebsd@insightbb.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> From: Steven Friedrich <freebsd@insightbb.com>
> Date: Sat, 5 Dec 2009 12:42:42 -0500
>
> On Saturday 05 December 2009 12:24:34 am you wrote:
> > > From: Steven Friedrich <freebsd@insightbb.com>
> > > Date: Fri, 4 Dec 2009 23:37:04 -0500
> > > Sender: owner-freebsd-acpi@freebsd.org
> > >
> > > I sent this to questions last Sunday, but only one person responded. He's
> > > running FreeBSD 8 and I think his system is reporting bogus temps too.
> > > I think there might be a missing scaling factor. I'm a hardware guy, but
> > > I don't currently have temperature measuring equipment and I would want
> > > to do it on one of my towers (which are currently in storage), not my
> > > laptop anyway.
> > >
> > > I booted my HP Pavilion zd8215us and I immediately invoked
> > > chkCPUTemperature. The first temp reported was 52C, which is 125.6F. This
> > > leads me to believe that acpi has an anomaly regarding temperature
> > > measurement. The ambient temp was 71F (21.6C). The machine had been off
> > > for over eight hours.
> > >
> > > Here's chkCPUTemperature:
> > >
> > > #!/bin/sh
> > > # $Id:$
> > > #
> > >
> > > # CPU Temperature Information from ACPI
> > > POLLING_RATE=`sysctl hw.acpi.thermal.polling_rate|awk '{print $2}'`
> > > while [ 1 ]
> > > do
> > > sysctl hw.acpi.thermal.tz0.temperature
> > > sleep $POLLING_RATE
> > > done
> > >
> > > uname -a
> > > FreeBSD laptop2.StevenFriedrich.org 7.2-RELEASE-p4 FreeBSD 7.2-RELEASE-p4
> > > #1:
> >
> > Why do you not believe the report? The temperature reported is usually
> > measured on the die, not the package. (You couldn't measure it externally,
> > if you wanted to.) Due to the VERY low thermal mass of the die, it heats
> > up very quickly.
> >
> I've been running FreeBSD on this laptop since 2005 and only in the past month
> has it started shutting down when the temp it 81C. So I found the sysctl where
> it reports the temp and I wrote chkCPUTemperature, a bourne script to check
> the temp every 10 seconds.
>
> I have placed 1/2 inch spacers, ok, bottle caps from 2 litre bottles,
> under the four corners and it's not shutting down now. I'm an old
> hardware guy and I understand the die vs package issue, but what's the
> temp diff between the two? I was hoping to spark some interest in this
> issue with someone who has the ability to verify the actual temp with
> the reported temp. I was trying to find a linux user that might be
> seeing something different, possibly indicating that FreeBSD's ACPI
> port has a bug not in the linux code.
I hate to suggest this to an old hardware guy, but have you blown out
your heat sink lately? On my laptop I do that about annually. The first
time I did it the temperature of the CPU when the system was idle
dropped by 12C. The temperature during a buildworld dropped from 91C to
72C. Of course, you may have already done this.
While it is possible that your system is tickling a bug, it really seems
unlikely. Again, the temperature read is one of the most trivial ACPI
operations. It is not uncommon to not get a temperature at all and it is
possible that a system is reporting it incorrectly. It's even possible
that the thermal sensor on the die is defective, I really have a hard
time believing that the ACPI code in FreeBSD is having a problem with
this. There is no scaling. Just take the reported temperature and add
273.
You can try booting up a Knoppix CD and see if it reports something
different, but FreeBSD and Linux share the same ACPI code which is
actually written and supported by Intel, though both do adjust it for
their systems. I'm betting that Linux will show the same results as
FreeBSD, whether it's right on not.
--
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman@es.net Phone: +1 510 486-8634
Key fingerprint:059B 2DDF 031C 9BA3 14A4 EADA 927D EBB3 987B 3751
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20091206034955.EF3331CC0B>