Lenovo Yoga C940 frequently does thermal shutdown

This change was made by a bot.

Seems like it's Ice Lake, the support was added to thermald 1.9.

20.04 includes thermald 1.9.1, but it is not running by default.

$ systemctl status thermald
● thermald.service - Thermal Daemon Service
     Loaded: loaded (/lib/systemd/system/thermald.service; disabled; vendor preset: enabled)
     Active: inactive (dead)

What should a user like me do to prevent the laptop overheating?

Hmm, does the overheat still happen after thermald gets enabled?

On Thu, 2020-04-16 at 07:22 +0000, Anton Keks wrote:
> 20.04 includes thermald 1.9.1, but it is not running by default.
>
> $ systemctl status thermald
> ● thermald.service - Thermal Daemon Service
> Loaded: loaded (/lib/systemd/system/thermald.service; disabled;
> vendor preset: enabled)
> Active: inactive (dead)
>
> What should a user like me do to prevent the laptop overheating?
>
Go to each /sys/class/thermal/thermal_zone*
write
echo "disabled" > mode

You can make a script.

If I start thermald service, I still can reproduce thermal shutdown relatively easily.

What is cryptic is that while I monitor the sensors, CPU core temps are around 85deg, but when shutdown happens it says that critical value was 80C.

Where does this difference come from? Laptop runs above 80C for quite some time before it shuts down...

By the way, I have only one thermal_zone with mode file in it - thermal_zone0 (other don't have mode)

If I do:
sudo echo "disabled" > /sys/class/thermal/thermal_zone0/mode

I get: Permission denied

The thermal zones that usually cause the shutdown is thermal_zone3 and thermal_zone4 (rarely)

By the way, these are the types I get:
/sys/class/thermal/thermal_zone0/type:acpitz
/sys/class/thermal/thermal_zone1/type:SEN2
/sys/class/thermal/thermal_zone2/type:SEN3
/sys/class/thermal/thermal_zone3/type:SEN4
/sys/class/thermal/thermal_zone4/type:B0D4
/sys/class/thermal/thermal_zone5/type:iwlwifi_1
/sys/class/thermal/thermal_zone6/type:x86_pkg_temp

thermal_zone3, which causes most shutdowns, is actually not a CPU core - it's value doesn't match any of the values shown by the sensors command.

What could be the mysterious SEN4? How can I trace it further?

Moreover, the thermal shutdown is more easily reproduced when laptop is charging (battery below 100%) on AC power.

I have traced it to int3403_thermal module.
If I do rmmod int3403_thermal - the thermal_zone3 goes away an no thermal shutdown is happening.
I can use it as a workaround - but is it dangerous for the HW?

https://github.com/torvalds/linux/blob/master/drivers/thermal/intel/int340x_thermal/int3403_thermal.c

From what I see in the source, it is probably related to battery or charger or both.

So I guess thermald should take this sensor into acccount when throttling?

When I quickly modprobe int3403_thermal back, the full output is here (it is seen that temp is 79C, which is really close to the critical 80C):

thermal_zone3$ grep . *
available_policies:power_allocator user_space step_wise bang_bang fair_share
grep: emul_temp: Permission denied
integral_cutoff:0
k_d:0
k_i:0
k_po:0
k_pu:0
offset:0
policy:user_space
grep: power: Is a directory
slope:0
grep: subsystem: Is a directory
sustainable_power:0
temp:79000
trip_point_0_hyst:2000
trip_point_0_temp:0
trip_point_0_type:passive
trip_point_1_hyst:2000
trip_point_1_temp:0
trip_point_1_type:passive
trip_point_2_hyst:2000
trip_point_2_temp:80000
trip_point_2_type:critical
trip_point_3_hyst:2000
trip_point_3_temp:75000
trip_point_3_type:hot
trip_point_4_hyst:2000
trip_point_4_temp:65000
trip_point_4_type:passive
trip_point_5_hyst:2000
trip_point_5_temp:60000
trip_point_5_type:active
trip_point_6_hyst:2000
trip_point_6_temp:50000
trip_point_6_type:active
trip_point_7_hyst:2000
trip_point_7_temp:40000
trip_point_7_type:active
type:SEN4

On Thu, 2020-04-16 at 20:18 +0000, Anton Keks wrote:
> I have traced it to int3403_thermal module.
> If I do rmmod int3403_thermal - the thermal_zone3 goes away an no
> thermal shutdown is happening.
> I can use it as a workaround - but is it dangerous for the HW?
>
> https://github.com/torvalds/linux/blob/master/drivers/thermal/intel/int340x_thermal/int3403_thermal.c
>
> > From what I see in the source, it is probably related to battery or
> charger or both.
>
> So I guess thermald should take this sensor into acccount when
> throttling?
>
> When I quickly modprobe int3403_thermal back, the full output is here
> (it is seen that temp is 79C, which is really close to the critical
> 80C):

Correct. So this will shutdown the system. Is that sensor temperature
always in that range? If it is then sensor may be faulty.

Thanks,
Srinivas

>
> thermal_zone3$ grep . *
> available_policies:power_allocator user_space step_wise bang_bang
> fair_share
> grep: emul_temp: Permission denied
> integral_cutoff:0
> k_d:0
> k_i:0
> k_po:0
> k_pu:0
> offset:0
> policy:user_space
> grep: power: Is a directory
> slope:0
> grep: subsystem: Is a directory
> sustainable_power:0
> temp:79000
> trip_point_0_hyst:2000
> trip_point_0_temp:0
> trip_point_0_type:passive
> trip_point_1_hyst:2000
> trip_point_1_temp:0
> trip_point_1_type:passive
> trip_point_2_hyst:2000
> trip_point_2_temp:80000
> trip_point_2_type:critical
> trip_point_3_hyst:2000
> trip_point_3_temp:75000
> trip_point_3_type:hot
> trip_point_4_hyst:2000
> trip_point_4_temp:65000
> trip_point_4_type:passive
> trip_point_5_hyst:2000
> trip_point_5_temp:60000
> trip_point_5_type:active
> trip_point_6_hyst:2000
> trip_point_6_temp:50000
> trip_point_6_type:active
> trip_point_7_hyst:2000
> trip_point_7_temp:40000
> trip_point_7_type:active
> type:SEN4
>

No, normally it's below 60C.

I can get it to 80C (and shutdown) when the AC is plugged in and charging the battery and I load my CPU a lot at the same time.

Usually it means a video conference with several people and charging at the same time. A little bit of sun also helps.

So I guess the cooling mechanism here is not adequate, if thermald tries to do anything about it at all. I am attaching thermald debug log, if it helps.

Also, I usually can avoid the shutdown if I manually reduce CPU target temperature to 70C using this tool: https://github.com/erpalma/throttled This results in very heavy CPU throttling well below 1GHz.

"sudo echo "disabled" > /sys/class/thermal/thermal_zone0/mode" is not correct.

Try
   echo "disabled" | sudo tee /sys/class/thermal/thermal_zone0/mode

On Thu, 2020-04-16 at 20:54 +0000, Anton Keks wrote:
> No, normally it's below 60C.
>
> I can get it to 80C (and shutdown) when the AC is plugged in and
> charging the battery and I load my CPU a lot at the same time.
>
> Usually it means a video conference with several people and charging
> at
> the same time. A little bit of sun also helps.
>
> So I guess the cooling mechanism here is not adequate, if thermald
> tries
> to do anything about it at all. I am attaching thermald debug log, if
> it
> helps.
>
> Also, I usually can avoid the shutdown if I manually reduce CPU
> target
> temperature to 70C using this tool:
> https://github.com/erpalma/throttled
> This results in very heavy CPU throttling well below 1GHz.

Not sure if you ran dptfxtract tool on this system. May be thermald is
not doing anything because there is no target device.

Thanks,
Srinivas

>
>
> ** Attachment added: "thermald debug log"
>
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1873083/+attachment/5355770/+files/thermald.txt
>

I have now run it, but it didn't output anything about SEN4

Anyway, I have checked that Windows keeps the temperature under control on the same machine, by throttling CPU.

Shouldn't thermald also do it out of the box, not requiring hand-written config files?
Or maybe something is missing in my thermal_zone3, like there are no links to related cooling devices. And there is no thermal_zone3/mode to disable it...

I have posted the whole contents of thermal_zone3 in comment #10

Status changed to 'Confirmed' because the bug affects multiple users.

For me it's even more worse, my cpu would be idleing at 30-40C and suddenly it will shutdown with thermal thermal_zone3: critical temperature reached (80 C), shutting down

Could it be a sensor problem?

Wilson, does it happen when charging or the laptop is in the sun?
It seems to me this is a battery temperature sensor and it is only mildly affected by the CPU temperature.

Hi, no, its indoors while charging. Could it be a hardware issue?

It is not really used in the relationship file. So other OS may not be using this so probably never validated the critical trips. Better to disable.

Srinivas, can you please elaborate which relationship file you are referring to?

I also have the same issue - using ubuntu 20.04.

It happens usually when I have the laptop plugged in to power - maybe the turbo boost or scaling governor changes when the laptop is plugged in to my charger.

It usually happens when I am on a video call and screensharing or recording my desktop or running a game like bioshock via steam.

I don't have any of the above issues when I boot into windows (first time I have had to use windows in 5 years).

I have managed to reduce the problem a bit but not completely by following the enhanced thermal configuration section here:
https://docs.01.org/clearlinux/latest/guides/maintenance/cpu-performance.html
Enable DPTF in your bios.

First run:
sudo systemctl status thermald.service
and note it is missing "Using generated /etc/thermald/thermal-conf.xml.auto"

sudo apt install acpica-tools
git clone https://github.com/intel/dptfxtract.git
cd dptfxtract
sudo acpidump > acpi.out
acpixtract -a acpi.out
sudo ./dptfxtract *.dat

Run the following again to see if it worked:
sudo systemctl status thermald.service
You should now have "Using generated /etc/thermald/thermal-conf.xml.auto"

Things left to try:
On this page:
https://docs.01.org/clearlinux/latest/guides/maintenance/cpu-performance.html

Try buid the ThermalMonitor tool
https://github.com/intel/thermal_daemon/tree/master/tools/thermal_monitor

Seeing if changing the scaling governor makes a difference:
cat /sys/devices/system/cpu/cpu*/cpufreq/scaling_driver
shows powersave while charging for me but I may have changed it with
echo powersave | sudo tee /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor

Disabling turbo boost might help but haven't tried it yet:
echo 1 | sudo tee /sys/devices/system/cpu/intel_pstate/no_turbo

I was having thermal issues in Linux all the time before I updated the BIOS:

You can only update the bios with Windows.
https://support.lenovo.com/gb/en/downloads/ds541233

I used a Windows tool called Rufus to make a bootable windows USB drive. (Tried a number of linux tools to do this but didn't have any luck)

I also needed a fast enough USB drive to do this:
Samsung BAR Plus USB 3.1 Flash Drive 128GB - 300MB/s (MUF-128BE3/AM) - Champagne Silver

Flashed my bios from the Windows USB drive successfully.

After the bios update only had issues when the laptop is charging.

It's getting really bad. To the point where Ubuntu is barely usable. Like I would be doing nothing with my laptop just leaving it running, and the suddenly it shutdown and I concerned that the constant problems can lead to hardware damage. I hope a patch comes out soon. Is there a way to raise the thermal critical temperature? I'm no expert but I've read some articles saying 100c for a long time can cause hardware damage but 80-90c should be fine for a good amount of time.

Carolos, I'm in the same boat as you. Now I wonder if this is a bios problem with ACUN54WW. At least now I know its likely not a sensor problem.

Sorry to jump on this thread but am I to understand that the c940 works with 20.0.4? I have tried to make it work with 19+ and it did not recognize the mic right out of the box. Thank you for your reply.

Yes, digital mic and speakers now work using the new SOF subsystem in the kernel.

But the overheating shutdown is still there unless you do rmmod int3403_thermal

It seems like my sound stopped working recently after an update.

Is anyone else experiencing this?

To get sound output back I have had to add:
options snd-hda-intel dmic_detect=0
to /etc/modprobe.d/alsa-base.conf

After upgrading to 5.6.0-1011-oem it seems sound is back without the alsa change above.

This will take you away from the 5.4 kernel shipped by default on ubuntu - use at your own risk.

You will have to periodically check for and install newer versions on your own after installing this.

sudo apt install linux-image-5.6.0-1011-oem

There's a metapackage "linux-oem-20.04" for that purpose.

Hi!

I have C940-14IIL, running under Fedora 32 and I am very pleased by this laptop. Everything is OK, except shutdown/reboot (workaround i915.enable_psr=0 kernel option), fingerprint reader (do not care) and overheating shutdown.

I have resolved my overheating problem by installing the last version 2.2 of thermald from Github and automatically creating config bydptfxtract (https://github.com/intel/dptfxtract).
The bydptfxtract create 10 different thermal-conf.xml. I have tested all of them and I am attaching the most successful of them.

I did a stress test with full load for 10 minutes. Following is the uptime output:
14:47:46 up 18:57, 1 user, load average: 8.49, 9.10, 7.25

During the test temperature vary around 75-80 degree - something that I could not imagine before installing the thermald. One virtual machine and a minute of compilation or R calculations would certainly cause overheating shutdown.

Actually, as far as can explain the results, the thermald do a CPU scaling of the CPU speed in range 2.5-2.8 GHz - about 1 GHz below the maximum speed - something that I definitely agree. Scaling is done after the rise of the temperature and I do not experience any inconvenience, troubles or interruptions with the lower speed.

As per commeent #32, thermald helps here. But it is better to avoid kernel shutdown for one bad temperarure sample instead of some running average. We should have "mode" attribute so that we can avoid this by disabling the zone.
I will see what can be done to add "mode" attribute for these zones.

I installed CPU Power Manger (https://extensions.gnome.org/extension/945/cpu-power-manager/) and set turbo to off in all profiles. Prior to this running heavy games(divinity original sin 2 or MTG Arena) would push my temp to 100C then the laptop would shut off. I also tried throttle stop and that worked to keep from shutting down, but game would stutter when temp went over 90C. Now with turbo off and heavy gaming temps hovers between 55 and 70C, no stuttering at all. Like night and day. Can game without getting hot.

(Yoga c940, Fedora 32)

@ryder-tim, do you know how I could set this turbo mode off in Linux Mint 19 without having to switch from Cinnamon to Gnome?

@lolero2,

I dont know of an applet for Cinnamon but did come across the following page that should work.

https://forums.linuxmint.com/viewtopic.php?t=266897

Tim

Actually heres the way to do it: https://github.com/martin31821/cpupower/issues/128#issuecomment-655653233

Thank you for posting this solution @ryder-tim! You saved me :)

Playing with the settings, if i set the extension to 50% cpu and turbo boost ON then it will turbo up to 2.00GHZ from 1.30GHZ(when under load) and still maintain 60-70C during gaming which gives a little extra performance. Not sure what the sweet spot is yet.

The butter zone for me seems to be 60% which lets the CPU go as high as 2.40GHZ while maxing the temp at around 80C when gaming.

So now I can operate the PC and its great!

Now I need to get the microphone working. @anton-azib wrote that it was possible using the sof subsystem in the kernel. @anton-azib, could you please elaborate on what that means and how I can configure it to get it working?

@lolero2
the archlinux wiki page show hot to get microphone working.

https://wiki.archlinux.org/index.php/Lenovo_Yoga_c940

Tim

It only says requires sof-firmware. But how do I install sof-firmware to get it working?

I tried adding the lines to /etc/modprobe.d/blacklist.conf and /etc/pulse/default.pa and the microphone was still not working :(

A nice safe way of getting sound to work is to install "linux-oem-20.04". This will install a 5.6 kernel for you and sound works. You may have to undo any other sound changes you have made.

on fedora its sudo dnf install alsa-sof-firmware.noarch

on Mint probable sudo apt-get install alsa-sof-firmware.noarch
or what ever your package is called.

do a search for it
sudo apt-cache search firmware
sudo apt-cache search sof

While my problem is not exactly the same, i feel that it's so closely related as to not warrant a separate bug.
I have a thinkpad x1 extreme gen3. Whenever i suspend, upon waking up, it shuts down:
thermal thermal_zone3: critical temperature reached (128 C), shutting down

almost always zone3, sometimes zone4.
This is on my focal install, but i tried on a pristine focal live cd, as well as a groovy daily live cd (from sept 26th), with the exact same result. Groovy has thermald 2.3.

I *know* my laptop is not in overheat at that point. I can do this from a cold laptop, on battery or on power at 100%, doing fine with temperatures way low (~40 degC), suspend and wake up 10 seconds later _and i still have that behavior. Whatever that sensor is reading did not go from 40 to 128 in 10 seconds of suspend. I suspect a single bad reading when it wakes up trips everything up.

I've tried thermal.nocrt, or thermal.crt=150, as kernel command lines (and checked when booted that they were applied). That seems to only impact the acpitz-acpi-0 zone, nothing else, and it still powers off when waking up.

removing module int3403_thermal removes thermal_zone[1-8] in /sys/class/thermal
(interestingly, sensors still returns the same set, so either those zones aren't to do with the cpu or anything else sensors returns, or sensors goes through something else to get its values?)

and suspend now works fine.

happy to do some more digging, i'd obviously prefer to un-blacklist this module but now at least i can suspend

I also have the same issue with the Lenovo Yoga C940 regarding the regular thermal shutdowns on Manjaro 20.2. They occur mainly during charging and in particular during video conferences (Zoom, etc.), which is really annoying. I have also tried different linux kernels (5.4 - 5.10) but the shutdowns remained. This way, the lenovo yoga notebook is simply not serviceable under Linux.

If there is any solution for that I would be really thankful.

There is a patch in discussion to avoid thermal shutdown from kernel

Re: [PATCH 1/3] thermal: core: Add indication for userspace usage

from Kai-Heng Feng <email address hidden>.

If that gets merged then we can disable int340x thermal shutdowns from the kernel and let thermald manage shutdown. This gives opportunity for someone to create a config file to fix the temperaure threshold.

Thank you Srinivas. This sounds at least very promising. Unfortunately, I don't feel confident enough to work around in kernel files so I am patiently awaiting the patch to be accepted or a config-script doing the changes.

Interestingly I removed the int3403_thermal module, and disabled thermald's critical shutdown using thermal.nocrt = 1. However, it still shutdown with critical temp reached thermal zone 4 (80c). Anything left I can try, cause I lost a lot of work during the last shutdown and they only get more frequenent. I noticed after deleting the module /sys/class/thermal/ no longer has thermal_zone[1-8].

What's zone4's type?

Not too sure, however, I can say this is not a bios/hardware related bug, as on windows this never happens, also an interesting thing to note is whenever I suspend my pc the cpu throttle to 400 MHz and becomes unusable requiring a reboot

Anyway, backport [1] could help. Please wait for later kernel releases.

https://lists.ubuntu.com/archives/kernel-team/2021-January/116639.html

The workaround is to unload the thermal kernel module that overreacts to
short-lived temperature changes due to CPU load while charging.
sudo rmmod int3403_thermal

On Wed, Dec 2, 2020, 01:21 Magdalena S <email address hidden> wrote:

> I also have the same issue with the Lenovo Yoga C940 regarding the
> regular thermal shutdowns on Manjaro 20.2. They occur mainly during
> charging and in particular during video conferences (Zoom, etc.), which
> is really annoying. I have also tried different linux kernels (5.4 -
> 5.10) but the shutdowns remained. This way, the lenovo yoga notebook is
> simply not serviceable under Linux.
>
> If there is any solution for that I would be really thankful.
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/1873083
>
> Title:
> Lenovo Yoga C940 frequently does thermal shutdown
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1873083/+subscriptions
>

Then the fix in the kernel will help. Currently only linux-oem-5.6 has the fix.

This bug has been fixed in the kernel as per: https://bugs.launchpad.net/ubuntu/+source/linux-oem-5.6/+bug/1906168/comments/9

Marking this a fixed released.