"CPU temperature above threshold" readings are wrong i think

akudewan · 10-13-2005, 01:46 AM

I was checking some log files and I noticed this error message in /var/log/syslog

Code:

Oct 13 11:59:29 ranjan404 kernel: CPU0: Temperature above threshold
Oct 13 11:59:29 ranjan404 kernel: CPU0: Running in modulated clock mode
Oct 13 11:59:50 ranjan404 kernel: CPU0: Temperature above threshold
Oct 13 11:59:50 ranjan404 kernel: CPU0: Running in modulated clock mode
Oct 13 12:00:45 ranjan404 kernel: CPU0: Temperature above threshold
Oct 13 12:00:45 ranjan404 kernel: CPU0: Running in modulated clock mode
Oct 13 12:01:07 ranjan404 kernel: CPU0: Temperature above threshold
Oct 13 12:01:07 ranjan404 kernel: CPU0: Running in modulated clock mode
Oct 13 12:01:17 ranjan404 kernel: CPU0: Temperature above threshold
Oct 13 12:01:17 ranjan404 kernel: CPU0: Running in modulated clock mode

I looked at my CPU temperature in gkrellm. It showed 22 C. It always shows 22 C, it never changes. So I installed lm-sensors, it gives me this output when I run "sensors"

Code:

eeprom-i2c-0-52
Adapter: SMBus I801 adapter at 0500
Memory type:            SDR SDRAM DIMM
Memory size (MB):       128

eeprom-i2c-0-51
Adapter: SMBus I801 adapter at 0500
Memory type:            SDR SDRAM DIMM
Memory size (MB):       128

it87-isa-0290
Adapter: ISA adapter
VCore 1:   +4.08 V  (min =  +1.42 V, max =  +1.57 V)
VCore 2:   +4.08 V  (min =  +2.40 V, max =  +2.61 V)
+3.3V:     +8.16 V  (min =  +3.14 V, max =  +3.46 V)
+5V:       +6.85 V  (min =  +4.76 V, max =  +5.24 V)
+12V:     +16.32 V  (min = +11.39 V, max = +12.61 V)
-12V:      +3.93 V  (min = -12.63 V, max = -11.41 V)
-5V:       +4.03 V  (min =  -5.26 V, max =  -4.77 V)
Stdby:     +6.85 V  (min =  +4.76 V, max =  +5.24 V)
VBat:      +4.08 V
fan1:        0 RPM  (min =    0 RPM, div = 2)
fan2:        0 RPM  (min = 2657 RPM, div = 2)          ALARM
fan3:        0 RPM  (min = 2657 RPM, div = 2)          ALARM
M/B Temp:    -55 C  (low  =   +15�C, high =   +40�C)   sensor = thermistor
CPU Temp:    +87 C  (low  =   +15�C, high =   +45�C)   sensor = diode
Temp3:       -55 C  (low  =   +15�C, high =   +45�C)   sensor = thermistor

it87-i2c-0-2d
Adapter: SMBus I801 adapter at 0500
VCore 1:   +4.08 V  (min =  +1.42 V, max =  +1.57 V)
VCore 2:   +4.08 V  (min =  +2.40 V, max =  +2.61 V)
+3.3V:     +8.16 V  (min =  +3.14 V, max =  +3.46 V)
+5V:       +6.85 V  (min =  +4.76 V, max =  +5.24 V)
+12V:     +16.32 V  (min = +11.39 V, max = +12.61 V)
-12V:      +3.93 V  (min = -12.63 V, max = -11.41 V)
-5V:       +4.03 V  (min =  -5.26 V, max =  -4.77 V)
Stdby:     +6.85 V  (min =  +4.76 V, max =  +5.24 V)
VBat:      +4.08 V
fan1:        0 RPM  (min =    0 RPM, div = 2)
fan2:        0 RPM  (min = 2657 RPM, div = 2)          ALARM
fan3:        0 RPM  (min = 2657 RPM, div = 2)          ALARM
M/B Temp:    -55 C  (low  =   +15�C, high =   +40�C)   sensor = thermistor
CPU Temp:    +87 C  (low  =   +15�C, high =   +45�C)   sensor = diode
Temp3:       -55 C  (low  =   +15�C, high =   +45�C)   sensor = thermistor

Notice the last temperature is sub-zero LOL. Also, these temperarutes always remain the same. Never fluctuate. So I'm assuming they must be wrong, and 87 C is a bit too hot, when I touch the cabinet, it seems only little warm, as usual.

Next, I installed a software in windows. It said CPU temperature 22 C, same as gkrellm and "cat /proc/acpi/thermal_zone/THRM/temperature"

I have an intel i845 chipset. Does anybody know what the heck is going on? Does my mobo even support temerature monitoring? The fan speed is also zero.

vharishankar · 10-13-2005, 02:18 AM

Most likely it appears that there is no monitoring. Check your BIOS Setup while booting and see. That usually gives a reliable result. If there's no bios panel for this, you should assume that there's no thermal monitoring.

akudewan · 10-13-2005, 02:40 AM

I'll do that the next time I reboot. Thanks

akudewan · 10-13-2005, 07:50 AM

There's no thermal monitoring in the bios. How do I stop linux from freaking out? It becomes really slow in the "modulated clock mode" that it goes into.

akudewan · 10-16-2005, 02:51 AM

I disabled acpi using this command

Code:

 sudo /etc/init.d/acpid stop

But I still get the temperature warnings. Why is that?

wallie wallie · 10-22-2005, 01:30 PM

try: killall syslogd

Only if you think you can ignore the messege!

akudewan · 10-23-2005, 04:38 AM

I did a lot of google-ing, and found that this message is *not* generated by acpi, but by the kernel itself. So I did a little checking and found that my CPU fan wasn't working. I got it replaced, now the warnings have disappeared.

Thanks to linux I came to know something was wrong. Windows never gave a warning or anything. I guess windows believes that "Ignorance is bliss"

fouldsy · 10-23-2005, 05:04 AM

Your BIOS should still have been beeping like crazy if your CPU fan wasn't working. There should be an option to warn / alarm on CPU fan fail within your BIOS, as you don't really that happening too often

lowks · 08-21-2006, 11:33 PM

Following this post I went to check my CPU fan and to my horror found that the fan wasn't turning as one of the blades had been broken and the CPU was hot to the touch. Thanks !

jefkin · 10-03-2006, 03:49 PM

I have been hammered with similar messages:

Quote:

Message from syslogd@reunion at Tue Oct 3 14:41:15 2006 ...
reunion kernel: CPU1: Running in modulated clock mode

Message from syslogd@reunion at Tue Oct 3 14:41:15 2006 ...
reunion kernel: CPU0: Temperature above threshold

Message from syslogd@reunion at Tue Oct 3 14:41:15 2006 ...
reunion kernel: CPU0: Running in modulated clock mode

Message from syslogd@reunion at Tue Oct 3 14:41:20 2006 ...
reunion kernel: CPU1: Temperature above threshold

Message from syslogd@reunion at Tue Oct 3 14:41:20 2006 ...
reunion kernel: CPU1: Running in modulated clock mode

Message from syslogd@reunion at Tue Oct 3 14:41:21 2006 ...
reunion kernel: CPU0: Temperature above threshold

Message from syslogd@reunion at Tue Oct 3 14:41:21 2006 ...
reunion kernel: CPU0: Running in modulated clock mode

Message from syslogd@reunion at Tue Oct 3 14:41:26 2006 ...
reunion kernel: CPU1: Running in modulated clock mode

Message from syslogd@reunion at Tue Oct 3 14:41:25 2006 ...
reunion kernel: CPU1: Temperature above threshold

Message from syslogd@reunion at Tue Oct 3 14:41:26 2006 ...
reunion kernel: CPU0: Temperature above threshold

Message from syslogd@reunion at Tue Oct 3 14:41:26 2006 ...
reunion kernel: CPU0: Running in modulated clock mode

Message from syslogd@reunion at Tue Oct 3 14:41:30 2006 ...
reunion kernel: CPU0: Temperature above threshold

Message from syslogd@reunion at Tue Oct 3 14:41:30 2006 ...
reunion kernel: CPU0: Running in modulated clock mode

And after finally thinking I had a solution in this thread, I checked out my CPU, fan is there and running.

temperature seems reasonable.

This problem is intermittant, as I can go days with no message at all, then suddenly, I'm spammed with them all over. Rebooting rarely stops it. Turning off the computer for an hour or two, then returning... well sometimes it works.

I need a fix to this, as hitting [Cntrl]-L every 5 to 10 seconds is making me batty. Barring finding a solution, which seems unlikely, is there a way to make syslogd only report to certain ttys?

As it stands now, if I have three windows open, I have three spammed windows when this bug/error/thing manifests. If I have 23 (more common) then I have 23 spammed windows.

I'm considering following wallie wallie's advice and just killing syslogd, but my common sense says ask one more time.

Jefkin

Crito · 10-03-2006, 04:13 PM

Possible solution:
http://www.linuxquestions.org/questi...d.php?t=391581

And the actual code FWIW:
http://lwn.net/Articles/95296/

Code:

 	if (l & 0x1) {
 		printk(KERN_EMERG "CPU%d: Temperature above threshold\n", cpu);
 		printk(KERN_EMERG "CPU%d: Running in modulated clock mode\n",
-				cpu);
+			cpu);
+		send_kevent(KMSG_POWER,
+			"/org/kernel/devices/system/cpu/temperature", "high",
+			"Cpu: %d\n", cpu);
 	} else {
 		printk(KERN_INFO "CPU%d: Temperature/speed normal\n", cpu);
+		send_kevent(KMSG_POWER,
+			"/org/kernel/devices/system/cpu/temperature", "normal",
+			"Cpu: %d\n", cpu);
 	}
 }

jefkin · 10-04-2006, 12:09 AM

So after reading my own post, I decided to check my fan again. See, I'd rather be thought a bozo here than by a new CPU, so I checked the fan, and it wasn't running ... HUNH?

On more investigation, I found that one of the wires to the fan was just shy of burned through, and sometimes the fan didn't work. Replaced the fan this evening, and things look okay.

Lesson learned, if the CPU is overheating, buy a new fan, -- it's a lot cheaper than a new CPU.

Jefkin

ps. Thanks crito.

dudeman41465 · 10-04-2006, 12:14 AM

Thank god the side of my PC is plexiglass, I can just look right in there and check on things without ever shutting it down or opening it up,

Crito · 10-04-2006, 07:35 AM

Luckily modern Intel CPUs slow themselves down when they overheat. In an older mobo you would have surely fried the CPU. Had a fan fail on my AMD Athlon XP 3000+ and despite catching it early and replacing the fan my floating point unit/coprocessor was damaged. It seemed to run fine but every time I'd do something like verify an ISO checksum would get different results! Took me a while to realize what was wrong with the machine, as it only intermittently caused problems.

DyingMuppet · 10-28-2006, 05:34 AM

Hey BTW you're sensors list says that your voltage meter at 12V runs at 16+Volts

Code:

+3.3V:     +8.16 V  (min =  +3.14 V, max =  +3.46 V)
+5V:       +6.85 V  (min =  +4.76 V, max =  +5.24 V)
+12V:     +16.32 V  (min = +11.39 V, max = +12.61 V)
-12V:      +3.93 V  (min = -12.63 V, max = -11.41 V)

now I look closer and they all are a bit crappy..