Bogus temperature readings?

All:

I’m running SUSE 11.3 on an Acer Aspire M5620.

When I cat:
/proc/acpi/thermal_zone/THRM/temperature
I get a constant 30 C which is bogus based on what I see in the BIOS as well as with a local thermometer. This is the same device that the temperature widget reads.

Can someone tell me where we might be going wrong here?

Also, I’d like to read other temperatures and fan speeds but don’t know what glue is missing between the kernal and the hardware.

Thanx,
Lew

Hi
Run the command sensors


Cheers Malcolm °¿° (Linux Counter #276890)
openSUSE 11.3 (x86_64) Kernel 2.6.34.7-0.3-default
up 19:56, 2 users, load average: 0.01, 0.07, 0.03
GPU GeForce 8600 GTS Silent - Driver Version: 256.53

Installed and ran sensors - data must come from the same place as this also reads a constant 30 C and is still wrong (based on bios and other measurements).

Other thoughts?

Thanx…

lgrothe wrote:
> Other thoughts?

i used google to search the web for other users having the same
symptoms (apparently broken reporting of CPU temp) with your
make/model computer…and, found no other reports…

so i wonder if maybe yours is a hardware problem: like maybe the
sensor (in the chip) is broken and/or ‘stuck’ on 30C

do other operating systems report other than 30C?

i see you say the BIOS is reporting something besides 30C, but is it
passing other than 30C to the operating system?

and, those BIOS reports you see–are they near 30? do that vary from
look to look?

i think you might check to see if there is a BIOS upgrade available
(or see if you can find other Linux or openSUSE reports of sensor
reports failures)…


DenverD
CAVEAT: http://is.gd/bpoMD [posted via NNTP w/openSUSE 10.3]
When it comes to chocolate, resistance is futile.

lgrothe adjusted his/her AFDB on Mon 27 September 2010 03:06 to write:

>
> Installed and ran sensors - data must come from the same place as this
> also reads a constant 30 C and is still wrong (based on bios and other
> measurements).
>
> Other thoughts?
>
> Thanx…
>
>

In a console try running ( as root )

/usr/sbin/sensors-detect

This will run a script that will try and detect any sensors that have not
been picked up automatically ( fan, temp, voltages, if they are there ),
just keep pressing [Enter] at each prompt this should find any extra sensors
on you machine and write the needed script to enable reading them ( init.d
). You can either modprobe the modules yourself to see if they work or do a
reboot to get them running.

There is a man page for sensors-detect also if you want to read more.

HTH


Mark
Caveat emptor
Nullus in verba
Nil illegitimi carborundum

I looked at the man pages I think the sensors library is reading the same bogus information from the bios that /proc is.

More info about above…

I’ve also looked for reported problems on Acer and general sites and agree that there have been no reports but, if everyone’s system is happily reporting 30 C (86 F) as a temp under linux, well, then there’s no problem…

This box was running Vista up until last week when it became completely unstable and would not boot. For a number of months, it would suddenly quit (hang) and I could not find a correlation to anything for sure but I was beginning to suspect overheating. The fan used to be pretty noisy but became much quieter about 6 months ago. I’d noticed under Vista that the board temp seemed high (as I remember ~49 C) but the CPU was better (~37 C) and both were well below either critical or alarm temps and the box fan is running (turning) so I presumed other hardware problems.

When I built Vista on this machine 3 years ago, I’d left a partition for the eventuality that I would also run linux occasionally (I have 3 other computers running SUSE 11.0 and 11.2). I figured at the very least I would load and run 11.3 to check out the disk and figure out if the system could be stable and then decide if I want to continue with Vista (this machine’s not qualified for Windows 7 according to Acer).

By booting to bios, I still see that the board temp is around 50 C and the CPU is around 37 C. By blocking the airflow, I can make these temperatures fluctuate. Linux reports a constant 30 C, regardless what I do.

So, I suspect the drivers for all the sensors (/proc and libsensor) are reading the same wrong bios location. Should this be moved to development or reported as a bug or does someone have a quick hack that can make this work?

For my own part, I’m going to replace the box fan and see if that makes the box more stable - it seems to work pretty well under linux as long as I have a fan that is blowing next to the case. It dies occasionally otherwise. Then I’ll decide if I want to go back to Vista. Probably not…

Thanx all…

lgrothe wrote:
> This box was running Vista up until last week when it became
> completely unstable and would not boot. For a number of months, it
> would suddenly quit (hang) and I could not find a correlation to
> anything for sure but I was beginning to suspect overheating.

often such problems without correlation to observable changes are
caused by a weak or dying power supply unit (PSU)…

unfortunately, most box makers buy the very cheapest they can get away
with…and, even don’t mind putting in a PSU they know will just
make it to the end of the warranty period…

and, i guess yours is probably past that…and even slight power
fluctuations can cause strange things to happen…

if you decide to put in a new PSU be sure to get one sufficiently
sized to give good clean power…here is a calculator:
http://extreme.outervision.com/PSUEngine

after you know what you need (i bet it is more than your current PSU)
then when you shop notice the difference in price between (say) 300
and 500 watts…it usually is not so much difference and your machine
might like the larger better…

> For my own part, I’m going to replace the box fan and see if that
> makes the box more stable

while you are in there i’d suggest you carefully and gently wiggle
every connection between the drives (including CD/DVD) and the mother
board…because just a thin layer of corrosion can cause problems
passing data…

and, well…dump out all the chicken bones and cat hair…

if you machine is over about three years you probably need to pull the
heat sink off the CPU, clean the top of of the CPU and bottom of the
heat sink and spread a THIN film of new thermal conductor between the
two…hey, look up (toms hardware is a good place) how to do all this
stuff…and, be sure and ground yourself to not fry your box…

and, well clean the heat sink while you have it off…

now, none of that might help…because, the stuff you describe might
well be cause by a cracked motherboard or intermittent ground circuits…

good luck…NOTE: if you do know your way around inside, do some
reading…it could save you a lot of headaches…

SORRY, i just noticed you are new here: -=welcome=- if i underestimate
your long hardware expertise please forgive, i can’t see the diplomas
on your wall from here!!


DenverD
CAVEAT: http://is.gd/bpoMD [posted via NNTP w/openSUSE 10.3]
When it comes to chocolate, resistance is futile.