CPU temp warning after 2.6 kernel upgrade
 

I recently began booting on a new Debian packaged 2.6.7 kernel, Everything was fine except that my PC automatically restarted after 30 seconds ( :rolleyes: ), but I fixed this by not loading the acpi kernel module (power management). Now I have a different problem:

This pops up very frequently (at an interval of about 15 seconds) on any command prompt I open up. I somehow doubt that my CPU temp is above the threshold, because I have a very nice Zalman CNPS7000-Cu cooling unit sitting ontop of my PIV, and I am usually not consuming more than 25% of the CPU clock cycles. I will note though that I'm in Phoenix and its pretty hot in my room, but this computer has survived in here for a year and I never had a CPU overheating problem back on my Windows monitoring software.


Anyone have an idea of what's wrong and how I can get rid of this annoying message? Thanks.

Can't you edit the threshold value in syslogd conf files? If you really know the cpu is ok, raising this value will do no harm to it.

If that's not it, you can raise the warning temperature in the BIOS, and see if that helps. What does cat /proc/acpi/thermal_zone/THRM/temperature say?

/proc/acpi doesn't even exist. Thanks for the tips though, I will try both and hope that they work. :)

Well I went into my BIOS and didn't see any options for playing with CPU temperature settings. Although I did see that, according to my BIOS, my PIV 2.8Ghz CPU was running at 93C/199F. How the HECK can it be running that hot already when I just turned on the computer? :confused: I think something is amiss...

Sorry to semi-hijack here, but what are you using to monitor your CPU temp? I've been looking for such a thing!

ACPI can measure temperature. As for the original question, try turning on ACPI in the kernel, along with the Thermal Zone option.

I can't. When I turn on ACPI on for my kernel (2.6.7), my computer automatically restarts itself after 30 seconds. :( If you have any clue how to fix this it would be most helpful.

I have what hopefully is a similar problem. I have a dual boot machine (Windows and Debian Testing). Installed debian some time ago. According to my BIOS my CPU temp is 95C (Celcius) and it has been that way for about 2-3 years. So I disabled the CPU Temp warning in the BIOS (long time ago), and have never had a problem with it.

Couple of days ago, I appled the debian s/w updates (apt-get upgrade). Now every time I boot into debian, it automatically reboots after some specified time (may be a minute or two at best). Windows however has no problems, and has been working nice (this email comes from the windows side).

I tried passing "nomce" to the kernel but that did not help. After I read this thread I am going to try "acpi=off". In the long run, I dont want this to prevent me from using Debian (was working nicely till yesterday).

Any help will be appreciated. BTW I am using one of the 2.4.2x debian stock kernels (comes with debian)

- Murali

I just checked. I am using 2.4.26-1-686 Debian Stock kernel. Passing acpi=off to the kernel (through lilo) did not help.
I am running out of ideas. I will get a Knoppix CD, so I can have some more time with the machine between reboots.
Also, "apt-get upgrade" did not complain about anything what so ever.

- Murali

Well at least you can rule out that its not acpi which is causing the shutdown for your PC. I doubt it, but could it possibly be the other power module (I believe its called apm)? I don't know what else could be causing a forced system restart.

R00ts, that's an idea, try disabling APM when enabling ACPI, If it still doesn't work, try using APM instead.

In one of my numerous attempts. I booted with apci=off and as soon as I got a root shell, I did a "/etc/init.d/apmd stop". This was successful and about 5 seconds later it rebooted. I dont know this is equivalent to runing off apm. What are the option to turn off apm? I will try that together with "acpi=off". I have a feeling that the machine will still reboot.

If nothing else, I will have to re-install debian :-(

- Murali

Did you apply any thermal grease to the processor when you applied the heat sink. If so, how long ago? You many need to reapply. Also are you overclocking your PC? If so back it off to the standard Ghz...

Walt

I got the machine assembled. I have not applied any thermal grease since I got the machine (2-3 years ago).
I am not overclocking, running at regular speed (2.4 GHz). Can you point me to where I can learn more about thermal grease. Mainly, what to buy, and how to apply it.

- Murali

Thermal grease greatly enhances heat transfer between two surfaces. It is generally a white paste and is availible at electronic stores and radio-shack. Just apply a thin layer to the cpu and heatsink (where the two touch) and put the heatsink back on, no need to add a lot as you don't want it sqeezing out everywhere.

Thanks a lot for all the help, I will do that and let you know what happened.

- Murali

I will be getting the grease tomorrow, and will apply it then. In the mean time, I have a question. This grease will help reduce the temperature of the CPU (with better heat dissipation). But I am able to use windows on that machine without any problems. So if the cause as overheating of the CPU, then the machine should have rebooted in Windows as well (Intel CPU). But that does not happen.

So is it possible that Linux uses the CPU more than Windows leading to the CPU overheating? I have even watched DVD's in Windows and I have had no problems. It gets weirder and wierder....

- Murali

Finally the problem was solved. The issue was nothing to do with thermal grease. It was due to a typo, in /etc/hotplug/blacklist. Apparently the module "i810-tco" causes this 30 second reboot. The file /etc/hotplug/blacklist has the list of modules which should not be loaded by the hotplug script. The blacklist file has it as "i810_tco" instead of "i810-tco" (something to do with the way 2.6 and 2.4 kernels name their modules). Once I changed the _ to a -, it was working again.

Now all I need to do to get my machine working is to figure why X wont see my mouse (but that is a question for a different forum).

Thanks for the help guys.

- Murali

Thermal grease is a substance that helps the CPU dissipate more heat to your heatsink. I've never seen any in retailer stores like CompUSA or BestBuy, but you can buy it online for pretty cheap (I think mine was ~$3,$4). Just google for it, I'm sure you can find an online store like NewEgg that can give you a supply. When you apply it, basically its just like squeezing out toothpaste on your CPU. Then take your finger or a cloth or something and rub it around the top of the CPU so you have a nice even coat of grease on it. Then plop down your heat sink unit and you're good to go.


Anyways, I found out that my CPU really WAS overheating (99C, and the max threshold on the specifications is 75C). I think my CPU got damaged because of it, because now and then my computer will unexpectedly freeze. :( Well it doesn't freeze completely actually. The mouse cursor still works, but I can't click the mouse buttons or send any input thru my keyboard. But it actually seems like its getting better though; My uptime is about 27 hours now. :D I found out the reason it was overheating was because when my brother had my computer he fried it somehow and the mobo, psu, and CPU had to be replaced. He got a heatsink with the CPU too (for some reason) and put it on there and its a crappy Intel one. I took it off and put on my old Zalman CNPS7000, but I didn't notice that the heat sink unit had lost most of the thermal grease applied to it and thus there wasn't good heat conductivity between it and my CPU. Oh well, I learned something at least I guess. :o