ROSA Linux Bugzilla – Bug 5021
HP 8510p: CPU get hot, sensors display eye-catching numbers
Last modified: 2016-05-01 03:21:58 MSD
Description of problem:
Be aware about temp 6: Always 100 degrees. Round Numbers in general.
I run "sensors" 3 times for this pastebin, with 10 minutes interval)
Version-Release number of selected component (if applicable):
2014.1, same in the default Kernel of Open Mandriva 2015 Aplha LXQt
Steps to Reproduce:
1. Turn on the Computer (https://paste.kde.org/p7annnfh2)
2. Check Sensors
3. Check Fan Speed
Work fine with the nrj-notebook Kernel (3.14.25 as well)
Happens with the nrj-notebook Kernel now as well.
I install another Distro, to check if the Hardware is the Reason.
PCLOS KDE 64, which use the BFS scheduler as well and Mint KDE 64 work fine and stay nice to my CPUs, even after some Hours.
There is a known bug in our cpupower package that might result in such things.
Please try the following workaround: remove cpupower package and then install it again. Execute
systemctl restart cpupower.service
just in case, or reboot.
Please check then what /etc/sysconfig/cpupower contains, there should be something like this now:
CPUPOWER_START_OPTS="frequency-set -g powersave"
CPUPOWER_STOP_OPTS="frequency-set -g powersave"
Note that "ondemand" may also be set there instead of "powersave" on some systems.
If the problem persists after that, please use hw-probe as I described in http://bugs.rosalinux.ru/show_bug.cgi?id=5006#c2 to upload the hardware info and the data to us.
Believe it or not, the Issue seems (currently) to be gone.
The Temps are now cooler as in Mint :D
I do a second hw-probe: http://hw.rosalinux.ru/index.php?probe=7520936246
(The first one, in the Comment above was created, when i turn the device on, 2, 3 Hours ago. Since then was facebook,youtube and so on running, same as in the past times, i do all like before. Afaik.)
Ah, guess, it happens like that, because the manual restart of cpupower take no effect and my reboot does. Fine, thanks :)
The Issue appears again, after a Reboot and about two Hours of using the Device:
[matthias@matthias-laptop ~]$ sensors
Adapter: PCI adapter
Adapter: Virtual device
temp1: +82.0°C (crit = +105.0°C)
temp2: +45.0°C (crit = +256.0°C)
temp3: +37.0°C (crit = +105.0°C)
temp4: +38.0°C (crit = +105.0°C)
temp5: +29.5°C (crit = +105.0°C)
temp6: +100.0°C (crit = +110.0°C)
Adapter: ISA adapter
Core 0: +40.0°C (high = +100.0°C, crit = +100.0°C)
Core 1: +39.0°C (high = +100.0°C, crit = +100.0°C)
(In reply to comment #7)
> The Issue appears again, after a Reboot and about two Hours of using the
Could you please do hw-probe again? We need the logs to see what is going on.
I reboot again and currently works all fine:
(I post the next hw-probe again, when the Issue reappears.)
23:07:54 up 2:19, 3 users, load average: 1.24, 1.01, 0.74
(I keep the Notebook on a secondary sense full base for this time.)
I am now sure about it:
The Sensor #6 always start with 50 degrees, goes after around 45 - 60 minutes up to 75 degrees and then up to 100. Every Time.
hw-probe of 75 degrees: http://hw.rosalinux.ru/index.php?probe=ac2f02776c
uptime 00:44:11 up 47 min, 3 users, load average: 0.46, 0.35, 0.36
Hnn, now is it (#6) down to 60° again... Why always such round Numbers ?
So, i change the underground and the notebook gets still up to 60 degrees, even with just some tiny tabs open. Always round numbers.
Any news ? :)
The same thing happens for me in mageia 5 beta 3: in this case even from the start ! What can this be ? cpupower ? Do OM/ROSA and Mageia use the same cpupower config ? At least the sensors output is very similar:
(Which looks different in each other distro, afaik.) :(
Ok, it happens in KaOS as well.. with the slight difference, that it shut down there already at about 70 degress. Same round numbers.
As for the always round numbers - no idea.
Concerning the temperatures - it might be that the CPU frequency is kept at maximum (or even higher if boost/turbo are enabled) for some reason.
You can run cpupower frequency-info (as root) from time to time to check the frequency. Example output from the data you sent us with hw-probe:
analysiere CPU 0:
Folgende CPUs laufen mit der gleichen Hardware-Taktfrequenz: 0
Die Taktfrequenz folgender CPUs werden per Software koordiniert: 0
Maximale Dauer eines Taktfrequenzwechsels: 10.0 us.
Hardwarebedingte Grenzen der Taktfrequenz: 800 MHz - 2.20 GHz
mögliche Taktfrequenzen: 2.20 GHz, 2.20 GHz, 1.60 GHz, 1.20 GHz, 800 MHz
mögliche Regler: userspace, ondemand, performance
momentane Taktik: die Frequenz soll innerhalb 800 MHz und 2.20 GHz.
liegen. Der Regler "ondemand" kann frei entscheiden,
welche Taktfrequenz innerhalb dieser Grenze verwendet wird.
momentane Taktfrequenz ist 2.20 GHz (verifiziert durch Nachfrage bei der Hardware).
boost state support:
That is, the CPU is currently running at 2.20 GHz (maximum frequency) while cpupower may make it run at 1.60 GHz, 1.20 GHz or 800 MHz as well.
You can try to set the upper limit for the CPU frequency to 1.60 GHz as follows (as root):
cpupower frequency-set -u 1.60GHz
Then you an use cpupower frequency-info again to see that the limit is set (there should be something like "momentane Taktik: die Frequenz soll innerhalb 800 MHz und 1.60 GHz liegen" there).
Then see if the temperatures go high as before when you launch some applications etc.
There was a possibly similar problem on another HP laptop we were testing, although the model was different from yours. CPU frequency was always kept at maximum there, even when the laptop was idle. It turned out, the problem was in the BIOS: the HP-specific "Deep Sleep" option did not allow to lower CPU frequency on idle for some reason. When it was turned off, frequency control started working and that problem was solved.
Not sure if it is the case for your laptop, but it may be worth looking at.
Updating the BIOS (with all the precautions necessary for it) might be helpful as well.
I set it up to 1.6 Ghz, that works. Overheat issue stay like before and there is no such BIOS option here, as far as i see:
Thanks a lot again, great support :)
I make the BIOS update to the last version from 2011, the issue is still there.
We are now testing the update for lm_sensors, you can try it too and see if it improves anything.
Unlikely, but if you changed /etc/sysconfig/lm_sensors manually, it is recommended to make a backup of that file first.
Then, please add the repository with that update and update the system:
urpmi.addmedia lm_sensors-new http://abf-downloads.rosalinux.ru/rosa2014.1/container/2473468/x86_64/main/release/
Then you can run 'sensors-detect --auto' (or just 'sensors-detect' if you want more control over what it scans).
Reboot, just in case and see what 'sensors' will show now.
[matthias@matthias-laptop ~]$ urpmi.addmedia lm_sensors-new http://abf-downloads.rosalinux.ru/rosa2014.1/container/2473468/x86_64/main/release/
bash: urpmi.addmedia: Kommando nicht gefunden.
(Command not found)
I copy and paste your command into my yakuake.
I run the commands with sudo and they work.
Issue is still the same, temp4 shows 75.0 degrees, 60 degrees and so on.
Ok, i install PC-BSD: Exactly the same issue.
(I guess because Linux and BSD share acpi ?)
When i run Windows (XP) on this device, show me the Open Hardware Monitor and CoreTemp around 50 degrees in idle and around 60, when i play a 1080p Youtube Movie (in Fullscreen, on 1024 x 786 Display Resolution) '
Windows runs in idle and with light browser usage without fan activity and turn them just slightly on, when i play the Full HD Video.
So i guess its acpi ?
Can this help, please ?
I find this interesting Post, about a similar model:
The Battery is very weak already, 15-20 Minutes with standard Kernel.
Can this effect ACPI or the overheat issue in general.
I try without battery, issue consist.
Issue is fixed with lm_sensors 3.4.0:
When happens this update on ROSA?
I install them all 3, restart and the issue still exist:
Always round numbers: http://funkyimg.com/i/2bgUN.png
(In reply to comment #29)
> Always round numbers: http://funkyimg.com/i/2bgUN.png
Where? 61, 62, 56, 48, 58, 60, 65 - looks OK. No more round numbers than usual.
i mean, there isnt a single temp with another number as 0 next to the comma.
and these temp are checked directly from a cold boot and the fan drives crazy as always..