[SOLVED] Thermald fail 14.2, Round II

phalange · 01-10-2019, 08:29 PM

This post:https://www.linuxquestions.org/quest...-a-4175643733/ fixed it, but on a new install of 14.2 using the newly patched slackbuild from SBo, I'm getting an issue.

The patch seems to apply fine in the build sequence, but on restart thermald doesn't launch despite executing it from rc.local, and doesn't launch manually either.

willysr · 01-10-2019, 09:30 PM

have you installed all the patches from -stable?
we always test using the latest -stable updates

phalange · 01-11-2019, 12:25 AM

It appears so, and I just double checked to be sure. No updates available and I have the slackware.com 14.2 mirror enabled.

anon231 · 01-11-2019, 01:48 AM

which error message do you receive?

Quote:

thermald --no-daemon --loglevel=info

phalange · 01-11-2019, 01:05 PM

Quote:

Originally Posted by Arrabiata

which error message do you receive?

Code:

bash-4.3# thermald --no-daemon --loglevel=info
RAPL domain count 0
RAPL domain count 1
20 CPUID levels; family:model:stepping 0x6:4f:1 (6:79:1)
 Need Linux PowerCap sysfs 
failed to open /dev/acpi_thermal_rel 
failed to open /dev/acpi_thermal_rel 
TRT/ART read failed
I/O warning : failed to load external entity "/etc/thermald/thermal-conf.xml"
error: could not parse file /etc/thermald/thermal-conf.xml
Unsupported cpu model, use thermal-conf.xml file or run with --ignore-cpuid-check 
THD engine start failed

I'm sure the 'unsupported CPU' a red herring, being that thermald has run on this CPU before (intel i7-6850)

I'm not running ACPI - this is a stock install and I haven't added much to it yet.

bassmadrigal · 01-11-2019, 01:23 PM

It's complaining about not being able to find /etc/thermald/thermal-conf.xml

There is a /etc/thermald/thermal-cpu-cdev-order.xml and I wonder if that is what it is actually looking for. Might be worth copying or renaming it to match what the program is looking for and see if it fixes your issue.

phalange · 01-11-2019, 02:07 PM

Quote:

Originally Posted by bassmadrigal

It's complaining about not being able to find /etc/thermald/thermal-conf.xml

That didn't end up working, but I did try running it with --ignore-cpuid-check and that does appear to solve it. On restart, with that option added to rc.local, thermald is active.

It seems to implicate my device, but oddly this particular box has run thermald without a fuss in the past.

There are some git issues of cpu ids being absent from thermald, thus generating the erroneous 'unsupported' error. It's a subtle problem because without a manual check (e.g. ps or htop) it's not evident the daemon has failed to run, and the functionality is not apparent in regular use. I wonder if others have installed thermald and assumed it to be working?

Question to the group: are the frequency scaling controls in the kernel enough on their own; i.e. is thermald superfluous?

anon231 · 01-11-2019, 02:47 PM

Quote:

Originally Posted by phalange

[CODE]bash-4.3# thermald --no-daemon --loglevel=info
.....................
Unsupported cpu model, use thermal-conf.xml file or run with --ignore-cpuid-check
THD engine start failed

(..stepping 0x6:4f:1 (6:79:1))

../Thermald-1.8/thermal_daemon-1.8/src/thd_engine.cpp

Code:

static supported_ids_t id_table[] = {
                { 6, 0x2a }, // Sandybridge
                { 6, 0x3a }, // IvyBridge
                { 6, 0x3c }, // Haswell
                { 6, 0x45 }, // Haswell ULT
                { 6, 0x46 }, // Haswell ULT
                { 6, 0x3d }, // Broadwell
                { 6, 0x47 }, // Broadwell-GT3E
                { 6, 0x37 }, // Valleyview BYT
                { 6, 0x4c }, // Brasewell
                { 6, 0x4e }, // skylake
                { 6, 0x5e }, // skylake
                { 6, 0x5c }, // Broxton
                { 6, 0x7a }, // Gemini Lake
                { 6, 0x8e }, // kabylake
                { 6, 0x9e }, // kabylake
                { 6, 0x66 }, // Cannonlake
                { 0, 0 } // Last Invalid entry

Your CPU-ID > { 6, 0x4f }, // Broadwell-phalange

and ..
create "thermal-conf.xml", otherwise thermald doesn't know what to do.

Example for my Thinkpad

Code:

<?xml version="1.0"?>
<ThermalConfiguration>
  <Platform>
    <Name>Generic X86 Laptop Device</Name>
    <ProductName>Thinkpad T430</ProductName>
    <Preference>QUIET</Preference>
    <ThermalZones>
      <ThermalZone>
        <Type>TSKN</Type>
        <TripPoints>
          <TripPoint>
            <Temperature>30000</Temperature>
            <type>passive</type>
            <ControlType>SEQUENTIAL</ControlType>
          </TripPoint>
        </TripPoints>
      </ThermalZone>
    </ThermalZones>
  </Platform>
</ThermalConfiguration>

phalange · 01-11-2019, 03:51 PM

Quote:

Originally Posted by Arrabiata

(..stepping 0x6:4f:1 (6:79:1))

Thank you, this is very helpful. I am running ok without the thermal-conf.xml, I'm guessing it's not vital?

Quote:

Originally Posted by willysr

we always test using the latest -stable updates

Is it worth noting the --ignore-cpuid-check on slackbuild readme?

++

Thanks again to everyone for your suggestions; I'm marking solved.

anon231 · 01-11-2019, 04:44 PM

Quote:

Originally Posted by phalange

Thank you, this is very helpful. I am running ok without the thermal-conf.xml, I'm guessing it's not vital?

you can control the CPU better. test it with passive mode and active mode, you can see it.

willysr · 01-11-2019, 06:21 PM

Quote:

Originally Posted by phalange

Is it worth noting the --ignore-cpuid-check on slackbuild readme?

Sure, thanks for the report

anon231 · 01-11-2019, 11:44 PM

I inform the developer the CPU ID is missing. The developer will update it.